Monday 14, Jan , 2013 , Robert Stansbridge, ISVR Southampton uni : 'The Joy
of Digital Signal Processing.' How DSP techniques are used in audio effects,
MP3 coding, virtual acoustics, sound synthesis, communications and much,
much more. 2 hours, split talk, distributed and end-block Q&A, 28 people.
(followup to his popular Sound of Music talk here on 14 Nov 2011)
(transcript only, a lot of the content was audio via amplifier and live audio spectrum video feed and animated
graphs to video projector)
A brief look at a number of subjects. I fist came across DSP, I'm an electronics guy , opamps, capacitors ,
resitors and that sort of thing.
a fellow researcher along the corridor I was talking to about stomp boxes, various guitar effects units and he
said he had a chip that can do all of that. Its a dsignal processor , like a microprocessor but more
analogue. This is how I got interested . Guitarists use a great range of these effects units, phasers,
flangers, chorus, echo , distortion etc.
Does anyone not know the difference between analogue and digital - so everyone is ok on that.
How about AD conversion - a graphic for that.
Basically for DSP I can demonstrate with this piece of freeware software. It will record
my voice onto my laptop, from a microphone.
Doing a zoom in on the recording we can see finer details until we start to see small dots instead of
a line. The computer is taking a snapshot of my voice, changing pressure, via a light piece of metal in the
microphone that is vibrating, and a changing capacitance which converts to electrical
signals. The pc doesn't record the continuous waveform but loads of snapshots. So its dot dot dot ...
and then its join the dots for a curve. There are 2 things important about that. How often it does
the snapshots , you want a lot of them and do it fast to get a good representation . If you did one
now and then you would get a rubbish reprentation - the sample rate. And then convert them to a
digital binary number. A demo grahic on AD converter. Some voltage as a vertical sliding
point of a 1 volt scale . This one .63V, the pc says is it greater or less than half the voltage range ,
its in the top half so its a binary 1. Then looks in the next half and is it in the top or bottom half, and now in the bottom
half and calls it 0 . Then repeats for the next half and a 1, and so on.
So the more bits of resolution you get , the more accurately it is able to determine what the true
value is. This example is 8 bits , the more bits , the more precision.
CD quality is 16 bits , so high precision. Beyond that a human cannot really tell the difference.
This is the transfer characteristic, a measure of how well it is doing the conversion. There isa clock that
runs and beats and for CD quality, that's at 44,000 samples per second which is very fast sampling.
So fast sampling and precise binary interpretations of the analogue values - that is your starting
point. Thats what those dots are in my voice recording, the dots constitute a graph.
Basically this is what DSP is. We are taking analogue signals, they could be audio from a microphone
, could be pressure or temperature, force, acceleration , normal everyday variations. Any physical
property and some sort of transducer, converts that to a voltage to the AD converter.
What we then do is a mathematical process on those numbers. What you have is a memory
full of numbers. Then we do some fairly simple maths on them to get some desired effect. It
could be altering the frequency response to something we like. It could be communicating,
sending something through the phone network or could be medical applications.
An example of a DSP developement circuit board. The biggest block is the DSP chip, underneath
are several hundred pins . A little chip is the A to D converter and D to A converter, doing all
the timings and doing all the convertions. Some of the other chips are extra memory .
If you are sampling at 44.1 kilosamples per second then in 1 second you need 44000
The heart of a DSP is a delay line . An animated graphic of one way of doing a delay line.
a collection of numbers coming in from the left, travelling through to the right. Each division is a memory
location. Memory location 0 , 1 , 2 ... 32 here. An input on the left slider is read into the first
memory location . All ticking through at the clock rate, 44000 times a second. Each tick the
memory in the first location is moved to the second. In fact it starts at the far end.
The contents of position 32 is moved out first. 31 moved to 32 etc back to the input posistion.
They all shimey along. The output follows the input but it takes a time to do it - a delay.
You can alter the delay time by altering how many locations there are . If I had less memory locations
, it would go faster, if more memory locations then slower. Then a matter of how fast it clocks
through . So 2 ways of controlling the delay.
We will first look at an echo.
For an echo we have a sound source , hits a far wall , and both recorded on a microphone .
If you go to the Grand Canyon you can shout and the return sound and measure the time between
initial and echo. Then if you know the speed of sound and the temperature , pressure etc , you can work out the
distance across the Grand Canyon. In this example the amplitude of the repeat is half the
original. If you use a gunshot instead of a voice you get a nice sharp crack echo back.
If you stored that , that is an appriximation to a delta function, a mathematical term. Theoretically
zero width and 1 unit high and enormous amount of energy, all of which is impossible.
People do use starter pistols to give delta functions, or burst a balloon. If you then get an echo
and 1 second delay and half the amplitude . The delta functio nis a click , called the impulse
in this particular case and what comes back is the impulse response and that is very important to DSP.
So with the Grand Canyon , you can measure that distance, via time in seconds. The amazing thing
about this is - if you take that recording of the Grand Canyon , saved that little file, and take it home via
memory stick, put it in your pc , do a particular perverse form of multiplication , called
convolution , then you speak through your microphone into your pc and convolute it , then
your voice will sound as though you are speaking in the Grand Canyon.
Your voice would be travelling along the delay line through those memory locations and you
pick off number 1 and further along , taking a tap off. The weighting factor and 2 rows of memory.
One lot is the delay line and the other is what you are going to multiply it by. If it happens to be
the impulse response of the Grand Canyon then it will sound like the Grand Canyon. Which
is jolly clever.
There isa problem with mobile phones. You have loudspeaker that you listen to and a microphone
right next to it , those 2 should feedback, especially in a hands free car setup. You speak intop the
microphone , its transmitted, might go via satellite , eventually someone else at the other end
says something. Their microphone will hear what you have said and it should send it back to your
speaker, go round and around and be impossible to use. That doesn't happen .
A simple way of getting rid of the echo is effectively if you send out a click through the system, wait for
the click to come basck and that will tell you how long the delay is around that loop. Now anything that comes back
as your voice with that delay , not a true signal, is an echo of your voice. You can turn it upside down ,
giving a delayed version of your voice, the correct amplitude for anti-sound. Sound and anti-sound
added and the sound will go away - theoretically. I tried doing a demo of that and I
couldn't make it go away.
Q: is that for all the audio frequencies, cancellling the whole rangre or just a short band?
Telephones tend to have a bandwidth of only 3.5 KHz , they sample at 8KHz a Shannon's ? cutoff frequency of
4 KHz and so .5 KHz to do some filtering. Bandwidth limited to about 3.5KHz , not full audio.
In principle it would work for full audio but not implemented generally in phones I would think.
If I record a sharp clap , quite a sharp impulse there , not bad for a delta function. When we loook
at the recording its not a delta function at all. Where has all the rippling come from.
It is you lot because I do my clean delta function then it hits the walls , hits all of you and what I've done
is recorded the positions of all of you. So you could probably identify yourself individually in this
trace. I've recorded the reverberant characteristics of this room.
Audio demo of a concert hall impulse response . They actually go on the stage with a starting pistol
, fire it and record the acoustic reverberant field of the hall. You can download these off
Another audio demo whare there is obviously a wall at the far end and gives a very strong clear echo, compared
with the recording I made in this room. You could take that or even go to the albert hall say , set off your
starting pistol, rrecord it, take it back , convolute it with your voice and you will sound as though you are
singing in the Albert hall.
But there is more. It turns out that it does that for anything. Anything that you hit , different surfaces
and objects , has a certain mass , a certain stiffness . When you give it an impulse it disturbs the
equilibrium and will give a boi.. .oi....ng . Depending on the mass of the object , the shape and stiffness,
if you record that then you can simulate that physical item. You can do it with filters in particular.
Going to lectures on digital filters and me an analogue man , I did not understand a word of it.
How can you build an 84 pole Chebychev or Butterworth filter - but a digital filter you can.
A small guitar amp with DSP effects built in. For this demo it has digital tone controls, tapping different parts of an acoustic guitar
body and using its transducer , and varying the tone controls. Its always a
collection of sinusoids and a collection of exponentials , depending on the complexity of the system.
So say you start with building a 4 pole Butterworth filter from op-amps etc , go tick, record
the output , store the voltage that comes out . Then you never have to build another Butterworth filter
because once you have it stored you can convolute it with whatever you want to go through there.
It will sound the samer as a Butterworth filter. For a high pass filter, build one, get the impulse
response and never have to build another high pass filter.
You can get Mathlab to calculate what that would look like for an 84 pole filter.
You could build an 84 pole filter, but it would be a labour of love and would be unuseable
as there would be so much noise. Mathlab would work out the impulse response, store it in your
computer and there is your 84 pole filter.
It can copy amplifier and speaker types. This small guitar amp has amplifier and speaker types
built in , I could choose the sound of a Vox AC30 , the impulse response of an AC30
at the turn of a switch on the amp. Or select a Classic Cab stack for the speaker sound. Its not
fuzz boxes, its impulse responses, they just nick them and build them in.
The tone controls, inside there is a microprocessor and a power amplifier. No op-amp
based filters in there . The controls bass or treble are not altering in an analogue fashion.
Demo of delay and reverb . One control does the different guitar sounds , another one the tone ,
another does all the delay changes. They are real potentiometers in there but the A to D converter
just reads the pots , turns them into digital values. The volume control is probably the same. If fully
up then it says multiply by 1. If half way up it says multiply by .5 , quarter up then by .25,
all done digitally.
A toy here for monitoring blood flow. Some jollop on the forearm and trying to find the correct
position and orientation. My blood flow via monitoring Doppler Shift. Uses an ultrasonic
transmitter and receiver going at 8 MHz . Basically mutiplies the 8M by itself and then you
don't get an audio signal out , but if there is a shift to 8.0001MHz, then the output comes out in audio
and the makers don't have to do much more.
As the blood is going forward you get the Doppler Shift positive and reverse flow then negative.
Q: Were you pointing the sensor head across the blood flow or into it?
I don't know how the transducer is arranged inside the dome housing.
How a C-T scan works basically. Properly called a CAT scan but that term is frowned on in medical terms.
For DSP it doesn't matter whether the source is ultrasonic or X-rays or Gamma rays on nMRI .
A transmitter and reciver, a narrow beam , that is scanned along. Depending on what its seeing you get a varyng
waveform . Then it moves through 1 degree or so , gives another trace , repeated. It sets everything to
grey and then it says how do I modify that grey until it looks like a certain wavefor, and repeated .
Ending up with a consensus between all the angles. This gets over a major problem
of X-rays that you cannot see some things that are hiding behind something else.
You can use different frequencies and different transmission methods to differentiate things
that respond differently at different frequencies. Thats how you get the slices of 3D, all
done with DSP.
There is another use where you just make a click and listen for echoes, how you go prospecting
for oil. Set off an explosion , getting reflections off any changes
of surface and reflections of reflections. When they tell you that an earthquake happened
46 miles off the coast of Japan , 8 miles down. From numerous detectors , syncronised
by accurate clocks , can tell where it came from.
Similarly for mining you can tell what is hard or soft or layers of water or oil or whatever, again
Ultrasonic welds, if its flawless and you ding it then an impulse response comes back . If there are
flaws then the cracks will reflect back and so can do crack detection using this sort of technique.
Back to medical and add doppler shift and 2 sorts of info comes back. One is the intensity of
the reflection , tells you how hard or soft the tissues are and if stationary then they just take the
grey scale. The bone will show up as white say but anything that is moving , via the Doppler , if
its going one way then colour it orange , if the other way then blue. They can look ast the movement of
fluids in your body.
Time reversal. You can't reverse time but you can in DSP. Destroying gall stones, they use DSP.
You want to smash the gall stone so it will pass out in your wee. They use a burst of ultrasound
but there are problems. First you need to know exactly where it is and also it requires a lot of power
to smash a gall stone and its not advisable to have such a powerful beam inside a human.
You have an array of transducers and you give it a little ping of ultrasound and because the stone
has mass it bounces around and the transducers can detect that bounce.
Graphic showing 3 detectors, detecting a bounce , the ringing will hit the nearest sensor first.
They each have a time history, each have ??? time history , put it onto blast with controlled
delays and they all meet at the stone at the same time. So you get a large power in and its
at the right resonant frequency of the gallstone as the gallstone created the signal and signature .
I want to show you how stomp boses are done. If you have a delay line with stuff transferring
across it, if thgis is the tap that is coming off, then if the tap moves towards to oncoming sound then
you effectively get a doppler shift , because the values are being read off , coming out faster than
they went in. So a frequency shift. If the tap is running away from the signal then the values come out
slower than they went in, stretching them. Ultimately you can't cheat time,
it has to catch up eventually, but they have ways around that. You can do a pitch shift by storing and
runniong the tap . Intended demo of stretching a string on an acoustic guitar and then simulating it
with a pitch shifter and they sound just the same.
That is just one tap to provide a pitch shift. If you do it like an echo, push through the original
signal and a tap near the beginning and one farther alonfg , thats how you can do a vibrato.
You make the tap move back and forth along the delay line you get a sweep.
If you have 2 taps one at the beginning so the sound almost comes straight back out again
andadd it to a moving one then you get interference between the 2 signals . If itsa
very short delay , like 2 violins both at concert pitch, 'A' , perfectly in tune , you can still
tell that its 2 violins because although at the same pitch , they won't be in the same phase.
So you get a beat frequency that your ears can hear.
So if you have a method by which you can wobble the pitch of one of the sounds ,
it won't sound like 2 guitars . But give it a short delay you get a phaser. If you feed it back
around again then sometimes the 2 signals reinforce each other and a peak in the response,
sometimes they cancel out and a trough in the response. You havea filter thats moving
back and forth through the spectrum so a wah-wah sound . If you feed it back around again
it gets deeper and that is called flanging . These work best if you have a really rich harmonics source.
A pure sine wave and then there would be no difference.
So give it a rich harmonic source like in a Marshall Stack , and a lot of frequencies to work
on , Demos with an acoustic guitar as source.
If you give it a longer delay then it sounds more like separate instruments - the chorus effect.
Perhaps not multiple instruments , perhaps mor elike a mandolin with 2 strings
playing at the same time. So like 2 echos and one of the echos is changing its position in time.
Q; Other than looking more cool, what is the advantage of an electric guitar these days then?
Electric giutars are built with light strings for speed , they do have a very different sound.
You can get a virtual guitar , any guitar, put a pick up on it and if you know the dimensions very
carefully taken , the makers already know what the response will be, the characteristics of the
pickup etc. A defined input and then they can take the characteristics of a Gibson Les Paul ,
Telecaster, Stratocaster , Burns? , anything. They can take the dynamics, the resonances
of the guitar . Tapping on the body , the noise response depends on the thickness and stiffness of
the wood , the shape etc. You only have to give it an impulse , the guitar is sending a series
of pulses out. When you pluck a string it sends tension doen the string and back again
, sending a series of pulses into the bridge , then the body . So the body an appropriate tap
and the sound you hear will be a function of the whole acoustic cavity . So if you have a guitar and get its
response, take away this original response , so not coloured and replace it with the character of
another guitar. So via this virtual mechanism it could become a National Steel Guitar , Telecaster or
Rickenbacker and then put it through virtual Vox AC30 amplifer , Fender twin reverb and then
feed it into a virtual Black Widow speaker cab if you wanted. The whole rig can be virtual
except for some guitar string input. You can look-up on line for "virtual guitar" . Roland are very
big in virtual guitars .
Q: Has anyone done one for an air guitar?
You can do a completely synthetic guitar. A thing called physical modelling, you can store the characteristics
of a violin say , or make up your own. The strings of the violin bow , pluck the strings in stick and slip
fashion. Bowing a string sends a series of shockwaves to the bridge . The key is getting a good model
of the impulses , then you put it through the mathematical model of the violin body ...
loss of recording ...
I can make that pulse go backwards and forwards along the delay line.
If I shorten the delay line then its the equivalent of shortening the string. If I make it faster then its equivalent ot
increasing the tension. So I can treat that as a virtual string. The time that it takes to go round that loop ,
determines the opitch of the guitar. Some of the energy is returned along the strings and some is used
to vibrate the guitar body as it hits the bridge. Te guitar cavity tends to work like a pair
of belows , it pushes puffs of air out towards you. If it doesn't loose any energy then you won't
hear it , so you have to loose some energy. If you loose too much energy then the note
will die away too quickly, so its a balancing act, between sustain and volume.
You can send that output to another stage , some you could send to an acoustic model of the
guitar and the rest to simulate the sustain. That is virtual modelling.
Its the same with a pipe in the woodwind. When you blow you can simulate that with
white noise or pink noise because there is no striking of the note. Thats your starting point,
you energize the equivalent of the pipe , shorter the delay line , the shorter the pipe and higher
the pitch. Some is lost to the outside and some is a standing wave , a discontinuity in the
impedance of the atmosphere in the pipe. When it hits that discontinuity , some bounces back and that
maintains the note.
The frequency response of strings, lower frequency ones may sustain better than high frequencies ,
so that determines which frequencies will sustain. Diagramatic representaions of
a string instrument and a reed instrument. The bell of a reed instrument couples acoustically
to the outside.
Nowadays with Home Cinema you have sound bars. In front of me is a demo of virtual acoustics.
Its called a dipole . The normal way that stereo works is that you have 2 separated speakers , left and right.
If the signal is the middle then both speakers will have the same signal. If you want to pan to the
left then you switch off the right speaker. With the demo the speakers are right next to each other.
It works remarkably well. Its very tolerant of the room, you don't have to be sat in one particular
spot, you can even move about. Same signal into both these close speaker sthen it would
sound like one speaker. The waveforms add up. You can't stop time but if you slow one of them
down with our delay line , you can skew it. The wavefront goes off to the side, and do it to the other side
and the sound will pan across to that side .
Q: You could even make a Lesley speaker, make it rotate?
You probably could.
Q: What sort of delay is there relative to the audio for that effect?
It depends on the wavelength. Different for different frequencies , I'm not sure what it is for this demo.
The system directs sound that is wanted for your left ear, to that side of your head, its
noot throwing sound around the room. Its not bouncing off the walls . 2 things about that,
one is the time delay and the other is what do you send to the 2 channels.
So for the middle the same signal to both channels. Skew it over and the right ear hears something
of the wavefront,
but not the left , so you interpret that as coming from the right, much further out than
the right speaker.
So what is the optimum spacing , what is the optimum delay and it depends on the
wavelength of the signal and so different for different frequencies. There is a lot of maths
involved, quite a big subject, although conceptually quite simple, once you've seen
my wondrous graphic.
There is a standard for MP3 but it doesn't tell you how to do it. Does not tell the
manufacturers how to acheve the results. It just gives the specification of what
has to come out. There are 3 parts to it. One is human hearing , tens of thousands
of test individuals to determine what they can hear. Such as if there is a strong
sound of one pitch , can you also hear a weaker sound of another pitch simultaneously.
How far away can you distinguish 2 pitches.
Another is , if there is a sudden bang , your ears don't actually register anything
for a few milliseconds afterwards . And in stereo , if both signals are the same , then
it would send the same thing to both channels . What they are doing is chipping away at
the data-rate. What they would like to do is reduce the sample rate , then the bandwidth of the audio
goes down. The 44KHz is because they have to be able to go to CD quality , which is
just over twice the limit of human hearing. You need at least 2 samples for every cycle,
if not 2 then you can't interpret that as a cycle, so you loose the information.
So 44.1KHz, taking 20KHz as the limit of human hearing and the other 2.1KHz
to do some filtering , to get rid of any ultrasound basically. That is huge amounts of data.
16 bits is the CD specification, an enormous amount of data every second.
So they chip away at the data. No human can hear that, so chop it out. As no
human has ever been detected being able to pick out that piece of information ,
it still might be 10 times better than any human has ever detected. People say that MP3
has ruined hifi , just rubbish quality , it doesn't have to be rubbish quality.
It depends on how savagely they apply the chop-downs. It does have the capability
of stuff that no human being can perceive. Do seriously savage cuts and you will hear the
difference. So that is one thing an understanding of the human ear/brain - what it can hear
and what it can;t.
The next thing is DSP. A standard that all manufacturers can agree to, a major achievement,
and what the player will expect to receive. The player will receive a
stream of data. Every player has to get that data stream and start maybe part way through.
Its got to be able to pick it up and run with it. If the manufacturer can create that data stream
, the standard does not tell them how to do it. A can't hear above 8KHz. Prediction -
if its a sine wave , don't bother to record it , just say sine wave and frequency in KHz,
There are 5 stages in the specification. Its not a real time thing. You did not used to get
MP3 recorders because its not real-time. What they do now is save the data , raw data and
then process it afterwards. It appears to be real time but its not. It breaks it into frames
a bit like cine film . It takes a chunk , looks at that , analyses a spectrum of it, applies
a computer model of the ear that is basically a look up table that says - don't want that,
don't want this . If you have a tiny signal it won't matter if its bad processing.
Then it does something like a zip file on it , then puts the digital all back together again.
Diagramatically - chunks of audio travelling across. It takes 26 milliseconds , through a bank
of filters , typically 32 filters, low frequencies, midsdle frequencies etc. If you had to
build 32 filters from op-amps and coils and resistors, they would have to be very narrow band ,
you can't do it analogue, you can do it digitally.
It then processes each channel and looks at what is in there - if nothing it won't go further.
If there is something in there , it will apply the model - is it possible to hear , is it important
if so then give it full band, full resolution. If its trivial then use perhaps 6 or 8 bits, not 16 .
Each channel is amplified up so they have the same amplitude. Then applies quantisation
depending on rules , then zip file Huffman? coding so it throws away a lot of zeros ,
looks for patterns, separates the patterns , puts them in an order and a lookup table
for the patterns. then it stiches them back all together . Each frame has a header , that tells
it where it is in the stream, error checking/cyclic redundency, now many bits there are this part
8 bits, this part 8 bits , 12 bits, replacing the scaling, the actual data , and some general info
Sony copyright stuff etc.
How do they get stuff back from Pluto
The Voyager spacecraft and other craft going out into the solar system and beyond .
They are transmitting pictures and data back . Its travelling hundreds millions of miles and using
solar power to send its transmissions. And there is not much Sun out there, beyond Pluto.
So at Pluto its like flicking a bicycle headlamp and trying to pick up the signal.
Really they could do with a megawatt transmitter on board.
Simulation demo. There is signal but also a lot of noise. There are 2 accurate clocks , one on
Earth and one on the probe and they are syncronised . The probe sends repeated data.
They record some data then
records a second lot of data and averages the two lots then averages it again a running total.
All the space noise is random , sometimes added to the signal ,sometimes less, and the
average will tend towards zero. If you send it a million times, then its as if the
transmission was a million times more powerful. The demo is gradually recreating the
shape of the original.
Q: The freeware digital audio processing that you were using, can you tell us the name?
Audacity. Its really good stuff, I only discovered it quite recently, just for this talk.
We usually use Audition but that is too complex for this sort of use.
Q: Does Audacity have a spectrum analyser function built in, otherwise all you need is a laptop and a microphone?
Yes spectrum analyser or spectrogram with frequencies as colours. Even phone DTMF as colurs,
identifies all the frequencies from the audio.
Q:Why is there so many pins on a DSP chip?
It communicates to the A to D via serial , a couple of pins for the data coming in , a few
more that atre looking at control keypad type functions. But most of it I would say was for
memory. You can do serial memory of course, but this thing has to go fast. For 1 second of data
at 44 Ksamples per second and CD quality and if you wanted a concert hall reverb of 1 second
you have to do 44000 multiply and add operations within 1 over 44000 of a second a few
microseconds. Thats all the time you get to do all that maths. You can do effective DSP
with just a few bits , for a filter you only need about 3 multiply and adds. For a 50 Hz
notch filter you only need 3 multiply and adds. So could you do it with a Pentium.
The Pentium is refreshing a screen and interrogating the keyboard and running Facebook or
whatever. These DSP chips are no good for word-processing or emailing , they are
very fast multiply and add. They take data, multiply and add, multiply and add repeatedly
very fast. They have hardware multipliers , It doesn't have an algorithm that
says shift left/ shift right/ add this etc . It doesn't do anything fancy just multiply and add.
Q: You haven't mentioned Fast Forier Transforms?
Instead of doing the filters for the pitch shifting, you could analyse the data . A FFT takes a block
of data and try and find all the frequencies in that block of data. Fourier series says thast
any waveform is made of sine waves. You can make a square wave , eventually , out of
a shedload of sinewaves, or a triangular waveshape . Any repeated data you can make out of
sine waves . Theoretically it has to be repeated to infinity.
You could take a snapshot of the data and say this is my waveform repeated to infinity .
Just pretend that it is , looks inside it and how many of this kind in there . Then repeat for
other signatures . A number of data points and a number of frequency buckets .
The more datapoints , then the more buckets . For a simple sinewave it would put the
amplitude into the corresponding bucket . Then the next bucket , how many in that one.
A few buckets and a rubbish job. Running through an example on the screen of changing the
number of buckets and improving resolution.
There was a recent screenplay by Victoria Wood on the later life of Joyce Hatto
who was a classical pianist . She had horrendous stage nerves and never went on to be
a performing concert pianist. Her husband a record producer , not a technical
producer , got hold of some of your audio software , take some recordings of non mainstream
pianists , analyse copy note be note and meld into a form that his wife used to play
then sell them on the internet. There was a buzz went around about this new artiste ,
even on Radio3 , until someone put one of her pieces in the analysis part of I-Tunes which tells
you not only what tune it is but what performer it is apparently. It kept coming
back has 3 different artists, one Hungarian. Not the Lang Langs of this world . He'd been doing that
for years. I got the impression that that was quite standard in full scale commercial sound
recording to almost do that routinely. in that they never have to take one take in one go.
They can edit it , pitch change it a bit , pull a bit from there recorded 2 weeks previously
and it seems almost as phoney. ?
For a chunk of data you can. For pitch shift , for singers who can't sing , then you can take your
mouse and shift from say 244 Hz to where it should be at 245 Hz. The infornmation that is in there is
the same as in the original , no data has been lost . If you do the inverse, you take a sample
of somebody singing , get a spectrum , see the peak of what they actually sang . If its not
right, with your mouse you can shift it to middle-C say and reconstruct it . Its now standard, they have
all the notes of the scale , run it over by computer , so everything is shifted and now in key.
When someone sings, and you speed up the tape then you get Pinky and Perky.
You might like to hear say Diana Ross singing something in a different key, for whatever
reason. you can speed it up and reset the key but it won't sound like Diana Ross , it
will sound like Pinky and Perky , someone else , not Diana Ross. Because your voise is a mixture
of your voice box which is a tone generator , sending out a waveform and its modified by the
shape of your throat, tongue, teeth, lips and that modifies the spectrum . What you want is
the throat structure of Diana Ross but want her to sing higher. With DSP you can do a
thing called Cepstrum Audio data , produce its spectrum with peaks and troughs which is
your voice . These squggly lines can look like the audio data you started with . Somebody
did a spectrum analysis of a spectrum analysis , probably by mistake as they thought it
was audio data . What it did was it separated out into 2 lots , now called Cepstrum , one block
caused by your throat and the other lot by the tone generator, 2 spectra.
What you can do , get your mouse, select this lot , move those , do the inverse to give the spectrum
then the inverse again fo rthe time history , now you have the vocal tract of Diana Ross
now singing at a higher pitch, thats called liftering, variation of filtering.
Q: You reckon an ordinary individual who is just general computer literate but not an audio engineer
, once they get the software , can do this magic manipulation?
I went to a lecture in the music department. You can have a recording of a long dead
composer . Using this analysis you can do a spectrum of it and assign that to a Midi
keyboard, that records the keystrokes. , doesn't do any synthesising just records
that you played middle C and you played it quite hard. N ow you can take that and play it through
a synthesiser and it would play in the same manner of a particular pianist.
So get a top quality pianist, record their keystrokes and you can play it through your synthesised
top quality piano. But the music department did more analysis of the keystroikes.
They chose a piece of music that had a regular pattern , like a piece of Bach,
say the Ode to Joy. They took the earliest old recordings of known but dead pianists , of
varying quality including amateur players. Then plotted the pitch against the tempo.
Every time they play a note, a point goes on the graph. For half-way good
piano players produced a nice straight graph. Top players the graph went all over the
place . Then they subjected it to blind testing , how do you rate this piano player.
Average piano players do things very regularly . Top players speed up and slow
down , putting more expression into it. They don't play the same thing every time.
An undistinguished, average player , plays very straight and consistently,
because that is what they were taught to do.
Q: The music industry has complaints that they do a thing called loudening?
Perhaps you mean compression. For radio etc, and a classical concert has too
much dynamic range. Loudness control is basically a bass boost and also
compression because the dynamic range especially of piano music , piano forte
has a huge dynamic range in terms of dB. If you're driving a car , quiet passages
are not suitable. So the station copmpresses it has only say a tenth of the range of the
original. In compression you have sound playing for a longer portion of the time, and
although the amplitude has not changed, it will sound much louder because its there for
Q: This relates also to the compressor found on guitar effects units also?
Yes, very much so. DSP stuff is so good , the dynamic range of the A to D converters is so
good those become less and less necessary, they can handle that dynamic range.
If you only had 8 bits of resolution then you need a good strong signal. Otherwise
it looks like a staircase. So they amplified it up
Q: Same as they use on adverts?
A small signal and your only using a few conversions and not a good sound. So they boosted
up the signal . Adverts drive you nuts because they come on strong . There are regulations
relating to this. They are not allowed top make the advert audio loud. If you are listening to
a play then most of the time it is silence. Dialogue is mainly silence. Even if its music then
there is a lot of silence. Most of the things you are listening to on TV are not full
spectrum. For an advert they compress it very heavily , because they can;t excede
the limit for the height , but they can fill the rest of the time history with sound , thereby
making it much louder .
Q: Presumably it has influence on speech to text transfer?
Yes very much so
Q: Do they have lookup tables of the wave-shapes?
They look for dipthongs like sh , tt, th, tr, transitions between two sounds, g-o of so or g-o
of go, try and identify those. When they do speech synthesis they can do a model
of the throat , impulse response of the throat , get a model of all the resonant
cavities and then do the tone generators and white noise to simulate the ss/sh and you can
synthesise speech. Another big subject for another talk .
Is the new digital radio going to be as good as the analogue radio?
Well it could be but they don't necessarily want it to be.
Do they have a lower quality in the UK than other parts of the world?
To be honest I don't know about that.
( From the floor , yes we do. The channel widths aren't available for full DAB ,
partly due to so few people actually listening to it , even now 15 years since it was
introduced . The other problem is that we were the first country to adopt DAB radio
and every other country has adopted DAB v2, so we are 1 generation behind everyone else.
Q: If we go to DAB v2 then would?
We won't adopt DAB v2 as we'd have to throw away all the kit .
We could have inproved version DAB v1 but have to use more frequencies, more bandwidth, to
get the same quality . But they are not even prepared to spend out for quality equivalence to
even standard FM radio. Look at classic FM, the bandwidth that they purchase is tiny.
Another whole subject area for a talk.
Q: Will they turn off the FM?
The intention is 2017 , whether they actually do is questionable. )
The bad news is that the quality is likely to worsen but the good news is that our hearing is
getting worse day by day . Any of you listening to Radio 3 are probably incable
of hearing the high frequencies, sad but true
He will be giving this talk sometime in the next year to the Botley Probus group.
His previous talk on
Monday, 11 Feb 2013, Dr David Rusling presentation : The use of the base pairing
properties of synthetic DNA strands to construct objects that have
dimensions - DNA nanotechnology.
These non-biological structures offer applications as scaffolds for
generating micro and biochips, for drug delivery and even computation.
3/4 hr talk, 1/2 hr Q&A, 20 people
I've worked at Soton uni about 10 years. DNA encodes information that makes us what we are .
Its the chemical that makes all living things what they are. For this talk I want you to forget
that DNA encodes for information and just think about it as a molecule. We will try and build things
with this molecule . Hopefully by the end of this talk you will see that its quite a useful
molecule for building things at a small scale, which can be used in the field of nanotechnology.
We say happy birthday to DNA. 60 years ago Watson and Crick elucidated the structure of DNA.
They did this ingeneously , not doing experimental work themselves, using other people's
work and then inferred from their work , the DNA structure. We would call them theoretical
biologists. They published their paper 60 years ago. They firstly tell you not to believe what
anybody else says about DNA . There is material they got wrong, they are not infallible.
They showed the structure of the double helix from 2 DNA strands that interact around
each other. Since then there has been a lot more thorough study into the precise
DNA in more detail , one strand and the other and they are held together by what is called
Watson-Crick base-pairing. The 2 strands are related to each other in an antiparallel orientation .
The numbers at the ends of 5 and 3 means that we give it a direction , we normally
label DNA from 5' (5 prime) to 3', don't worry why we call it 5 and 3, it just means we have a
direction there. So if you have one strand from 5 to 3 then the other has to go from 3 to 5.
The reason these 2 strands associate together is that a position there will be a DNA base ,
a chemical structure , and on the other side you have another base . What makes it so clever
is if we have a G (gualine) in that position it will automatically interact with a C that
is found in the other strand . It does this by bonding between the 2. You don't need to know the structure,
just that a G will interact with a C and the same with an A which will automaticaly pair up
with a T and wrap around to form the double helix. It doesn't matter which side the
Cs and Gs are or As and Ts. What is useful is that if you have one of these strands and we know the sequence
of these bases, you automatically know its complementary strand that will bind with it. So if you mix
the bases up , it will no longer bind to that precise strand and form the double helix structure.
That is the very essence of this talk.
The other thing to know is that it is extremely small , the width of DNA is 2nm , a head of a pin
is a million nm in diameter. 10 or 11 bases down a strand , because they intertwine around
each other then they will be on the same side of the helix that you started on , that distance is
about 3.5nm . In a cell there would be 3000 million of these bases all lined up in one
long DNA strand. For today's subject we won't be using sequences that are that long. Just short
strands that we can design and make synthetically in a lab in a machine and use these synthetic
Why would we want to use this double helix as a building block to construct things . Its small so
we can build on an extremely small scale. If you know the sequence of one strand you can automatically
produce its pair, it assembles predictably. Its very rigid, so over about 150 of these bases that link
next to each other you get something that stays linear. Go further than that and you get bending
and other movements. Its very stable , its in all your cells, it stays there . There are processes that
try to repair it but it is a very stable molecule. Its really easy to synthesize , it costs about 10p per base
at themoment. So for 20 bases , 2 pound, it does depend on how much of it you want ,
Its obviously non-toxic otherwise you'd all be dead. Not at the moment
but eventually Because its a biological molecule you
might be able to use cells and enzymes that are found in cells, that maipulate DNA , to modify
these scotches? that I will be talking about in the next few slides.
It forms a nice linear molecule but how will you build anything with something that is straight.
2 DNA duplexes next to each other. You can design DNA sequences so that starting at the top,
it will entwine one side of the helix , pair up with the helix adjascent to it . You design 4 DNA strands that just by
DNA base-pairing can link the 2 helices together . One strand stays on one pair of helices , another
stays on the other pair but one crosses over as does the fourth. We can pair up adjascent
pairs. Its simple to design such structures, just by simple base-pairing. Designing the sequences so
they will only come together to form that particular complex. So we've moved from linear to
2 adjascent strands.
If you hafe a piece of helix and then a piece of single sttranded DNA attached to the end of it ,
so nothing to pair with. Then take another molecule with a single strand at that one's end .
You can design it so the G of one wil pair up with C on the other, A to T, etc. We call these
sticky-ends. The ability to piece 2 molecules together , just be base pairing.
Taking something that is smalll and making it twice as big. Even cleverer you can use an enzyme that
is found in cells, it covalently joins backbones together to form something that is twice as big .
So 2 concepts , you can piece these helices endwise or sidewise.
Ned Seaman? first came up with the idea of using DNA in nanotec . Like any good scientist he used
an appropriate catalyst to try this. He was actually in a bar , should have been at work , and he
took inspiration from a painting by Escher called Depth, of flying fish. Cross-like fish structures
that are arranged in a periodic manner. Knowing he could join DNA together , he proposed
to design DNA strands so they form these cross-over like structures . You can draw this structure
and it can form into one of 2 configurations. Add the sticky ends at one end of an arm ,
a complementary one on the other arm and the same in 2 other posistions then you can generate
some kind of an array , a lattice , that pieces these DNA molecules together.
So the idea that you can use DNA to design an extended structure that might have some use.
We now call these tiles, tiles that associate with each other. This was a great idea but it didn't
work because you can get 2 different types of molecule formed from the same thing.
They are also flexible and when you try to piece them together you just get a mess in the
test tube. They had to come up with something slightly more complicated.
He decided to get these tiles and design the 2 duplexes are held asjascent to one another
but this time instead of having one of those cross-overs between the helices , you now have 2.
This rigidifies the DNA , stops it from moving around . These are planar? and if you look at them
in side view there is no kinking or bending. So again you design DNA strands, anneal them in
a test tube and they come together to form this structure. He reasoned that if you take one of these
structures , drawn now with the bases there , you don't need to know the sequence , and the sticky
ends . Design 2 of these tiles and make sure of complementary sticky ends . You form each type of tile
separately and then mix them together and then through W-C base pairing they will self
assembly forming an extended complex.
Because you have ends that can hybridize , to bind together , what we call DX-A and DX-B types
you get a tile=based array. You can then build up an extended structure out of DNA.
This does work, about the turn of the century. This is the first time you can visualize DNA on some
kind of surface. You can't visualise via microscope as too small to see, you have to use
an Atomic Force Microscope AFM. You can now see DNA on a slide.
So an extended lattice, having gone from what was 2nm in diameter to something that is now
100nm in diameter.
This is a2D structue, a flat lattice on a surface. You can take it one step further and try
to generate a 3D structure. To do this they use a tensegrity motif, meaning tense and integrity.
Somthing used to build objects in real life, in big life. You can design DNA to mimic tese forms.
One helix held to its adjascent by a DNA strand , made from 7 strands now instead of the 5 as before
and 3 helices. The idea is to have your sticky ends on the corners of your molecule, in this
case 6 different ends. Pairing up of triangles. So you can have something extremely long and in 3D.
The first DNA crystal that came out about 2008. Completely made of DNA and instead of
having to use some fancy instrument to look at this , you can see these with unaided eye
in a test tube. DNA that is now pieced together in a pre-defined fashion to form a crystal.
Now something that is 200,000 nm or 1/5 mm in width. So you can fit 5 of these on the head of a pin now.
You can now try and solve what the crystal is made of . You can shine particles on the crystal and when they his it
they get difracted , and a difraction pattern and from that you can determine what and where the
atoms were that difracted those particles. So prooving that the structure in the crystal is
a specific repeating unit. A triangle pairs up with its 6 partners and this goes on 10,000
times in the different directions. So you've taken a linear molecule and made a 3D object.
So you design a motif and piece together tiles that interact with other tiles.
There is a different approach by a Paul Rodmend ? of Caltech and he has called it
DNA origami. Instead of trying to piece tiles together , how about taking one long DNA strand
and fold that strand into something that we want to design. You can think of the long strand as a
scaffold , looping linearly back on itself in a long loop . The reason you can do that is that you're adding
in short segments of synthgetic DNA that are appropriate in design , that they can cross over ,
between adjascent helices and generate the structure in one go. Perhaps 100 papers since 2006
, using this technique.
What can we design by folding this long DNA strand . A square , a star or even
a Smiley face. Cover of Nature had a DNA smiley on it. These are the predicted structures that
we want to form . Obviously you have to proove that you have formed those structures and again
you use a fancy viewing technique to check you have the square , star or Smiliy face.
I would love to have been the guy who first used this technique and saw 10,000
smilie faces. The highest concentration of happiness he'd seen in his life. It was this image that got me into this
field initially. It is quote simple to generate these today, in a scale that ranges between 2nm to
100 nm and to 100,000 nm like the big crystals.
So can you piece together one large structure with another large structure. Yes , a neat way of
making slightly bigger structures but not as big as the regular 3D structure.
Instead of putting adjascent helices together , we can say piece them together but also design it
in such a way that one section associates with a completely separate section of the underlying
scaffold . You start to fold these structures into something that is 3D.
In this case a hexagonal design. So starting to look more like something you would see in the real
world. Examples of railway bridge structure, slotted cross structure, again design the structure
and then check by specialised viewing techniques that they have formed the intended structures.
Thing on the 10 to 50nm scales.
We don't have to make things that are layered. You can now design things that come together to form
something like a Bucky-ball , and design it so you can image it . You can design small containers.
You can design them so they twist , lots of DNA duplexes stacked together, Both senses
left and right . You can make structures bend , circle structures again by tweaking the rules
we've discussed before. You can make little DNA nano-flasks , something like a zeptolitre ? of content in these
flasks. There are applications for these constructions as well as looking neat.
You can construct small on the DNA scale. A DNA tetrahedron . a helix on each edge.
Design your DNA sequences and mix them together and they form and again so small that you
need a technique called cryoNT? to image it. A cube made of RNA, which has completely different properties
to DNA in cells but when you are building with it , you can do the same things.
A structure generated totally by a cell. Posistioned not in the genome but another area of a cell that
allowed it to replicate. A long DNA strand made by the machinery found withing the cell and
designed in such a fashion and as soon as its made it will fold into the intended shape , in this
example an icosohedron. And again imaging to proove it has formed.
So a lot of different shapes but what can you do with them. This is what my research is for the
future. You can use these biological molecules to include drugs and prolong the amopunt of time
that that drug is within your body. Some drugs get degraded within the body very quickly.
So if its in a cage it can be there for longer. You can design these containers to target to
specific cell types. Say you want something to target your lung then design something that
goes to your lung.
You can do computation with these DNA tiles . You're not going to use them as a calculator one day
but for some simple calculations you just will stick DNA in a tube and it will work out an
answer for you - but that would be a subject for another whole talk.
My interest is using these structures as a kind of scaffold , to position chemical and biological
components. If you take that crystal structure , posistion molecules in that crystal in
precise positions , then difract particles to work out what the structure is in there,
you can then determine the structure of biological molecules that you don't currently
know the structure of. We have lots of proteins in our cells and we want to know the structure of all
of them. Take one of these proteins and try to get it into our crystal and then try to determine the structure.
If you put 2 such molecules close together in such a structure you could study how they interact
with each other, how they may do a particular process or reaction . Not just biological molecules
but chemical molecules also. the idea is that eventually you will make little factories , with
inputs and outputs , greatly facilitated by having everything in such close proximity.
If you had those just in solution , they have to find each other to do something but as you are
tethering them to a scaffold . You can speed up the reaction . You could use molecules to stick
particles on and use them for electronics perhaps. They won't form an amazing new computer.
You form these structures in water so not compatible with computers. Eventually could be
used for nano-chips . The scale that IBM is working on at the moment is about 64nm , they can
piece things together with a 64nm at the moment but with the DNA technique could perhaps go
down to the 1nm scale, much closer than any current techniques.
One of the problems is how do you introduce these components into these structures. DNA is inert ,
how to you attach things , what my research is looking at.
Because these structures are made from DNA , double helical DNA , we have evolved a
system that uses molecules that recognise DNA and then go on to another purpose.
With our synthetic structures , we define what the DNA sequences are within these structures.
So we might be able to use molecules that react with specific DNA seequences and design the
sequences that they recognise. Then posistion them at precise locations within that structure .
If you take those molecules that recognise DNA and attatch something useful to them then
you can posistion these groups within the structures.
There are a variety of ways you can recognize DNA , lots of small molecules. Lots of the
anti cancer drugs for example are small molecules that bind to DNA in different fashions.
There are a number of ways of recognising bases within these sequences , or the backbone
of the DNA or binding between base sequences. You can use DNA itself to recognise
a double helix and generate a triple helix. You know what sequenc ethese things bind to , so you can design
those within your structures and position at precise locations. Attatch something to the
end of the DNA strand and it will incorporate it within these structures. Your cells work
because proteins recognise DNA and they control how your genes are expressed .
There are a number of methods that we can interlay or interface with these DNA nano-structures.
My research is looking at trying to do that.
One of the things I'm researching is using formulations of this triple helix , a strand comes into
the double helix and binds with base triplets now . An extra base that comes along and recognises
a base pair and we know the rules to form this kind of structure and we can take the sequences that
bind to and introduce them to our nano structures.
Going back to tiles that form arrays , you can design sequences that add in the extra strand ,
recognising double strand regions, binds at precise locations . In this example designed to be 32nm
apart , so compared to IBM scale of 64nm we are looking at the 32nm scale. You could make these
closer if you wanted to . Then you try to add something to this extra strand of DNA and
position precisely in a repeating fashion .
I've targeted these third strands that have a protein attached to them and you get a repeated
protein with a periodicity along one of these structures. Each one of these is an individual
protein attached the DNA strand that I've targeted to this array. Run a line across
them and the spacing is an average of 32nm.
Bind in the extra dNA strand within the motif and attach something to the extra stand ,
now forming your crystal and each one of these groups incorporated within the crystal. I spent some time in New York
last year doing thse experiments.
Crystals with no added group , so clear . Then attached a coloured dye to the third stand
that is being targeted into , turning the crystals a blue colour. The intention is to replace the blue with something
more relevant like a protein and then determine the structure of those that you've incorporated.
Experiment next week in fact is taking a long DNA strand , in scientific paers call the disc with holes
but its a smilie face . Unlike the tile based arrays , where you design the sequences that you want in those
tiles , in this one you are restricted. Because that long single strand that you are folding back oand
forth on itself has a predefined sequence, you can't add another sequence into it.
By the formation of those triplet structures , where on the intended smilie face they might be able to bind
and introduce a group on that structure and target to those empty regions. Maybe I cold give it a nose
or an eye-brow or beuty spots. It sounds klike its not real science but it is , I assure you.
Previous work has confirmed I can produce the underlying smilie face and one day there will be
some added blobs.
Q: There are 4 chemicals in natural DNA , are there any other pairs you could use for synthetic DNA?
There is a whole department that works on analogues of these bases. The bases that we have ,have specific groups that allow
them to come together . You can put those groups in different orientations, different positioons and
get them to come together in different ways. There is a lot of analogues of those bases, not
natural, completely synthetic in the lab. Put in DNA and examined afterwards.
I was thinking of other planets, different style of DNA, therre are others that could be uused?
Yes there is potential for that, but I don't think it is as simpole as that as other things wqould
have to change . The DNA backbone and everything about it would be different on a different
planet, if there was life elsewhere.
What about the ? medical uses, ? similar to yours?
Yes and no , yes in some ways because they would not be able to encode
anything if they got into your cells. Not that these structures would do anything in your
cells at the moment. But if you had something different then your body would say , I don't like this ,
lets get rid of it. ... (end of recording, operator error, hit the pause buttton instead of record level button)
Monday, 11 Mar 2013, Roger Munford - wide ranging presentation , from the
technical aspects of solar photovoltaic generation, the grid, storage, politics and economics
all of which have changed dramatically in the last year.
Interlaced talk and Q&A of 1.5 hours, 30 people
Why is solar important. It would decelerate climate change if we can harness it. a lot of people
think that you put more energy into making solar panels (SP) than you get out , not true, you get a lot more
energy out. This is the most important thing. In the UK , the energy pay back is about
2 to 4 years. If you recycle your solar panels which you wiill do in 30 or 40 yesars time then the
energy pay back will be less, about 1 year. There is a company in Norway that makes
the Silicon (Si) bars using hydro electricity and they claim an energy pay back of only
1 year. A panel has a lifetime of about 25 to 40 years , they should last , once in place as low
mainainence. They should be producing energy for about 25 years.
History. The effect was discovered in the mid 1800s by Becquerel in France. It only really
became viable in 1954 at Bell labs when they invented the Bell Si battery . In 1954 it did
cause quite a stir . There was a conflict between those who thought that solar would be the
energy of the future and those who thought atomic energy would be the future.
A quote from th eNew York times in 1954
At the beginning of a new era , leading to the eventual realisation of one of mankind's mopst
cherished dreams - the harnessing of the almost unlimited energy of the Sun.
But they were up against atomic power and Robert Oppenheimer assoxciated with the Atom bomb ,
he said to the solar enthusiasts.
If you really want to get this adopted then do what we did in the atomic field .
You've got to classify the Sun as top secret . Establish a commission to manage for the
benefit of all mankind . Give appropriate high level indications of a super weapon , based
on th esun and then make a policy pronouncement , wishing to use the sun , not for
devastation and war but for the betterment of mankind.
That was 1954 but not much happened, it was used in space but not used seriously on the ground
for generationg power. One of the co-inventors said , what shall we do with our new baby.
I think his baby was a very slow developer but I think it is now developing.
Basically , silicon is used in most solar cells , when a light particle falls on the atom an electron
is emitted. Normally this would recombine but if you add 2 types of dopant and make a p-n junction
you can prevent the recombination and get an electric field . Connect it into t a circuit and a current
will flow .
The grid on the front of a cell is a sort of bus-bar to collect the electrons and then make a circuit.
There are now over 400,000 solar panel installations. tHere are 2 types of Si cell , depending on how the
Si is produced. The monolithic crystal type , achieved by drawing a Si crystal out of molten Si.
The advantage is that the Si is very pure throughout , the disadvantage is that it takes a lot
of energy. The second is polycrystaline , produced by depositing Si in a Siemens Reactor .
Gas is deposited on bars to make the Si rods . You don't get a nice uiform crystal . Look at some SP
and they have an attractive effect. " panels here , one is mono the other is poly.
The process is becoming so good now that they look much the same now.
The poly is not as efficient because of the interfaces betweeen the different crystals.
In a cell the generation is right at the surface of the wafer, so most of the wafer thickness is
wasrted in generation terms and only used for mechanical strength. A couple of years ago
I did a course in Germany, sitting in the front row. The instructor got one of these cells , gave it to
me to look at , thinking it is very expensive . When he gave it to me he twisted it deliberately
and shattered into about 20 pieces. He said he had thousands of them . Picking up the pieces
he then showd that the generated voltage was the same on any piece of it. Small or
large entire piece then about0.6 volts.
To make better use of the active layer they have a technology called thin film . Deposit
Si on one side of a sheet of glass and try and build a cell on that. It does work but is no where
as efficient as the solid Si. The way the market is going, thin film is becoming less and less
important because the crystaline forms have become cheaper and cheaper.
The nice thing about thin films is that you can priont it on glass windows , getting good
visual effect if you want dsome shadow . Instead of neutral / grey glass then have a solar array
printed instead. I';ve seen some in Germany and they are really effective visually.
The fourth type is amorphous , with no shape and is very flexible. You can roll such
a product out over the roof . If you have a fragile roof , you can glue such a material
to the existing roof and makes little additional weight. Again it is no where near as
efficient. So a last resort, but you can always stick it somewhere. An example of it here.
For generation, one cell gives you about .6V , packaged into a panel , 60 cells in this
one giving about 36 volts. Then up to about 24 such panels in a single array, giving about 720 volts.
Domestic installations would be smalller , but the larger you can make the array, the higher
the voltage and the more efficient the generation.
All these cells are joined together in series and they will only pass current at the rating
of the weakest cell. Say you have 1200 cells and there is one bad one then the whole array is
going to work at the efficiency of that bad cell. When they build these panels , they test each cell
individually and they match them together. The best cells go into a high power
panel and sold at a premium price. The weaker ones are put in different graded panels.
So when you are building an array and if you have plenty of room thenb you go for
cheap panels , but for a roof then you want small and efficient.
You always have to mount the panels in the same plane. If they were in different planes
then one section would be receiving more sunlight , creating more current but that current
is limited by the cells facing another direction . So when you see an array they are always in the
Another big problem is shading. If you see a big array and there is a bit of shadow in one corner
, you think well that is almost nothing in terms of the overal area of the SP. But that shaded cell
is going to reduce the generation of the whole array. There are a couple of techniques for avoiding that,
that I won't go into here.
The second part of solar generation is the inverter. This is where 1954 would have been to
early to start the solar revolution as the electronics was not there then. A modern inverter
is quite an amazing device. An inverter has 2 basic functions. A solar array has a sweet spot
where it produces its maximum power and this changes all the time. Continuously
changing with different light and temperature, the sweet spot changes.
The inverter has a maximum power-point tracking function , continually monitoring the power
produced by the whole array and every few seconds it changes its internal parameters just to
see if there will be more power going off in one direction rather than the other direction.
Like trying to tune into a radio station where they are changing the frequency all the time.
When the signal fades you have to try retuning. The inverter will do that automatically.
The inverter can switch the array off electronically , so it is isolated and the inverter won't be putting any power into the
mains or it won't be absorbing any power.
The second immportant function is the DC to AC conversion. Its more efficient , domestically ,
if we add the solar power that we are generating into the mains.
Its a 3 step process. The first thing is to switch the DC at mains frequenbcy so you get a square
wave . The second process is to chop that square wave a bit finer into pulses so the power is much the same .
Then through a filter to smooth it into a nice sinusoidal wave , exactly the same as the mains.
At the heart is a microcontroller or microprocessor that is doing all this. Its running MPPT ? and
monitoring the mains frequency. It spends 3 minutes synchronising itself precisely with what the mains
is doing. Once its locked into the mains it will continually track the mains.
Once its synchronised it feels safe to switch on and it will close its relays and the
inverter power and the mains power will be connected.
In most situations you have 3 connections. Consumer unit that supplies power
to the rest of the house , the mains coming in with the meter , and your inverter.
For the consumer unit the power has to go out , the inverter has to transfer the power from the
array to the mains . So it will either go out to the consumer unit and used domestically
or if too much than go into the mains. A 2-way street, you can put power back in or
you can take power out. All the power going into the 3-way connection has to be equal
to the power going out.
For pushing the power out into the mains, it raises its internal voltage just enough to
transfer the power inside it , into the mains.
So the inverter is monmitoring the power coming off the array, monitoring the mains frequency
and voltage , knowing how much powe rit has to put into the mains , and gives just the
right elevated voltage to ppush that amount of power into the mains.
The inverter will run near enough cold , it is very efficient and very little power is wasted.
Once the inverter is running how do we use it.
At night there is no power being generated , its not nice like nuclear that you have switched on
and a nice steady power flow.
Curve of generation on a really nice day , no cloiuds . The other curve is the normal
household usage. So evening comes along and lights go on , TV, cooking etc so power
useage and that will stop when people go to bed.
During the day the other curve is the self-consumption , power you are generating from
your array and using in your own house. Perhaps just a radio and a computer , but on a sunny
day your array is generating a lot of power and the excess is exported into the mains.
On a less nice day , same consumption but less generation , but just enough to
cover internal consumption. This snapshot was February rather than summer and if we'd
put on an electric kettle then the power demand would have been off the graph.
A kettle takes a lot more but only does it for 2 minutes. In that situation the power off your array
won't go into boiling your kettle. In the summer , with a lot moire power , then perhaps
you could boil your kettle all day . So there is a huge mismatch of your power requirements
and what the sun is delivering. And its changing all the time - a huge problem
for solar PV . It helps running a household because it reduces the amount of power
you're using but cannot be relied on to do anything.
The obvious thing to do with such a mismatch is storage , lets put batteries intop the
system say for the winter . After 30 minutes of winter sun I could have enough power
to boil a kettle but my array is not generating that amount . Batteries are expensive
and to size the battery required which would make sense fiunancially would be
difficult because your requirements in the winter are completely different to the summer .
In the winter you could generate no more than cover for a radio and a computer
but will never cover thing slike a washing machine and would never recharge the battery
for the required demand. In the summer you could generate 20 KW in the day and useage
be only 8KW , you could charge the batteries but you would not need the 12 KW, perhaps
only 2KW of storage to see you through the night. But you would not need a 12KW battery
because you could not do anything with the power. So batteries are very tricky.
What we need is a storage mechanism that will tide you over just for when the
washing machine is running . There is a technology using super-capacitors that is
exciting because they are cheap but no products available in the market at the moment.
The only product in the market , are battery systems . The reason thoise are
available is because the Germans have changed their feed-in tariffs to support battery storage
So Germans with solar PV for 10 years have been paid for the power they export but have just
changed it to being paid for the power they use themselves. So in Germany it is worth your
while to store any power and use that because you get extra subsidies. In this country
there is no reason for storing any power.
Grid Parity - where the generation costs are going to be the same as costs from utilities . Just about
there in the next year or so. Personally I believe we were there a few years ago as anyone
with a PV array has bought 25 years of electricity in advance at a rate that is quite low
and after 25 years we will be paying so much for electricity that the costs that we incurred
would be almost insignificant.
There is another interesting way of supporting solar and that is called net-metering .
That is what people think they will be getting when they get PV installed. That the meter will
turn backwards. So every kW that they generate , they don't use and goes intop the mains,
that will turn the meter backwards. So exporting power for the same price as they are buying power, effectively.
If that was extended to everybody then there would be no need for any storage as there would
be no financial insentive to have any local storage.
Feed-in-tariffs (FITs) they are credited to Germany as Germany started this industry really.
The idea was to support the initial installations so the price of the technology would come down.
It was actually a Swiss canton that got in there a few months earlier. But the Germans came up with
the idea firstly. Brittain adopted the FIT about 3 years ago , we wer eabout the sixtieth
state or country to adopt it. Waiting until loads of others had got some experience of it.
We stepped in and made all the mistakes that every one else makes. They started the FIT with
shedloads of money and it was difficult to believe what a good deal it was. You were getting
so much money and the installers kept a lot of the money back . The installation costs were higher
than they really had to be because there was so much money sloshing around.
They realised quite quickly that something was going wrong but were legally
obliged to keep the FITs at that level. Lat year there was kerfuffle about the govt wanting
to slash the FITs ahead of schedule and the likes of FOE taking them to court to keep
the FITs at the same level. It is now reduced to a sensible level which makes it sustainable
because you can have an installation that does what you need it too do, get a decent return
and your payback period would be about 10 years.
You get paid for every kW that you generate regardless of whether you use it yourelf
or you export it. You get a small payment for any exported power . The utility
companies say its far too difficult to measure, requiring another meter, so lets say 50%.
So we will give you , from your generation reading, say that 50% is exported and you
get 4.5p for each unit exported . The utility will take 50% of that . They are quite
happy about that . The average house could not use half the power that is generated.
In our particular case it is about 18% and we are home most of the time 2 pcs running .
We thought we would be a good candidate for free electricity but most of our generated
electricity was exported and the utilities are onmly paying us 50%.
So around the country the utilities are making money on this export tariff.
The FIT depends on the size of your array. 4Kw , about 16 panels , on a large
domestic roof . 10Kw , about 40 panels on perhaps a small factory . Larger than
that would go on large barns or be field mounted. At the moment the payments are
guaranteed for 20 years. Still a good thing to do but not as worthwhile as last year.
We've nerar enough reached grid-parity and any new build hous e should have PV
built in. It will save a lot of money.
Q: How do they determine the size of the array. What is there to stop someone
having an enormous great array and truncating at 4KW for the maximum FIT .
A: The utilities would be checking because they can calculate quite easily from the size of your array
which has to be registerd . They compare with what others in the area generates .
Another factor is what direction your array is facing. If facing south with an
inclination of 30 degrees thats perfect . Facing anywhere else and you generate
proportionally less. So on your registration certificate is all those details and they
know how much you are expecting. If you're cheating then they will find out.
I know someone who has a flat roof and he will build an extension on top.
At the moment the panels are laid out on the roof and shaded by balustrades and then
his panels will be in the perfect position on the extension. So he would be
generatng a lot more . The utility contacted him , saying your array is not performing
well , a large array but it is permanently shaded . But it was necessary to
install that array to get the FIT from the previoius regime that was much higher.
A lot of people throwing arrays up , just to get them registered before the cutoff period.
Q: I'm surprised about your 10 year payback period. A relativ eof mine had PV
fitted 7 years ago and then it was a payback period of in-excess of 15 years.
And that assumes that none of your kit breaks down over that period.
The current regime came in about 3 yrears ago, previously they gave you grants to
support PV, so very different then. So in many ways it is unfair to the real
pioneers . So now it is much better. Your friend paid a fortune and he was doing it for the
right reasons, all a bit of a shame. 3 years later the govt comes along giving out sheds of
money and the pioneers were left in the cold.
Q: Is that because they have had to drop the price of these systems because the FIT was lowered
or freed from manufacturers / installers that mafde it expensice to start with.
It was expensive tyo start with in the early days because it was new technology. You have to see the UK
in a global market because we were 10 years after the Germans and 6 years after the French.
There isa huge PV market around the world . It was basically German companies providiong the kit
but this is the thing the Chinese can do well because it is medium tech but there is a lot of metal
in there and can produce by hand. So China invested heavily into PV production . They make about
90% of the wolds PV and that has put founding German companys out of business.
Every manufacturer says there is a 25 year guarantee on out panels but the Germans had panels that were
over 25 years old. So you could see how they performed over 25 years. And a lot of
those companies don't exist any more and a lot of lost experince and knowledge gonme with them.
The Chinese undercut the German and USA an dNorwegian industry so much that a lot went out of business.
Last year USA imposed import duties on Chinese panels and the EU is looking at it .
Last week they said that it is now necessary to register the import of all panels now. If there is any duty
to be paid then it would be effective from this 06 March. The effect would be a rebalancing of
prices. The mad 2 years where alll the PV companies were fighting for market share and they
were undercutting themselves . These days most panels are sold at below cost price ,
madness in the market. 2 years ago the panels were 3 times as expensive . Even Chinese
companies are now going under . The EU stepping in to rebalance is too late for a lot of these
companies. So you can expect the cost of PV installations to go up now in the next few
Q: I was at the Eco-build exhibition last week . There was a company there involved with storage.
And they feed it into an immersion heater and domestic cylindr so storing as heat .
That is the obvious thing to do for someone with a solar installation/. They see that they are exporting loads of electricity and say to themselves what can I do with this as its my power.
So day 2 they ask can I connect it to the immersion heater. Thereis a series of companies that have
done that. So monitors what you are generating , what you are using and surplus will
go into heating your water. But an Irish company has got a patent on this idea . The patent is just the
idea of taking surplus power and heating water with it . There must be prior use in Germany
for this ten years before.
Q: You've shown us a number of styles of panel there . I'm living in a building in a conservation
area , the chances of me getting planning permission to put such things on my roof are about 0.
Is therer anyone looking at producing panels that look like roof tile or slate.
If you are in a conservation area then you may find that English Heritage is your friend.
Because they take the view that climate change , which they accept , which is going to make
their job of looking after historic buildings much harder. So EH are behind anything that can
help. Get the EH document on micro-generation.
Marley tiles produced one from about 10 years ago but could not sell it.
There are tiles on the market , a company called Solar Century make them , much more expensive .
They are a tilish sort of thing, fit better into a roof but does not look great .
I have come across a company that makes incredibly expensive tiles which do the job.
If you can put the panels on a roof away from from what they call a key-view.
Resume: Temperatuire has quite an effect on PV. The hotter it gets then the less efficient
they become. The best conditions are at the top of a mountain in winter for getting the maximum
power out of PV. But few hours to do that due to dreadful weather. The solar cells are
black and mounted in mini greenhouses so they get incredibly hot just sitting there whether generating
much or not. Its a problem when you're designing these systems because the lower the temperature,
the higher the voltage. You can have problems if your design is close to the limits , if there is a cold
period and then a sunny day you can have a higher voltage than your inverter can cope with.
If you are interested in PV then there is something called SuperHomes , people dotted around
the country getting the technology and energy efficient homes . So PV, solar hot water
etc. Book a visit to one of these homes for a tour.
I read a report by Offgen last week , prepared in November talking about the energy mix . They did
a larger report about 3 years ago and the govt asked them to report concerning our electricity generation for the next 20 years. Over those 3 years it has changed drsamatically . 3 years ago they thought 35% of our
energy from gas , now it is 65% . The coal plants wil be shut down soon and at the moment they
are shovelling in as much coal as they can to get as much use as they can from the old coal
fired power stations. What was 35% from coal is now 45% and that will go down dramatically in the
middle of this year when they have to shut down the plants. In that recent report there was no mention
of PV as so minimal a percentage. Talking about gas , wind and nuclear. On the other hand Shell
did a visionary report for the next century , they thought solar would provide 50% of the world's needs
by 2070 . But with the mismatch between generation and demand , how can we provide
power when there is no solar available. These problems are still there and they have to be
solved if they are going to get the best out of solar.
End of Talk
Q: I discovered that there was an industrial scalle array of PV near Ryde, is that nearest
one of any size near here?
there are several big ones in the New Forest . The South Coast is the best place
in the country for sunshine. There is one enclosed by forest . they had to lay a cable
for about 7 miles so a lot of work involved with installing it. There is another at Calshot
on the Cadland Estate maybe off public access. They don't like people wandering around them. The one in the Forst has a massive fence around it.
Q: Why did the Offgen report not mention PV.
Its not really significant at the moment. The Germans had a day last summer when they produced
half of German usage was produced by solar, that was a Saturday. The Friday before they produced
1/3 of the power demand. Its part of the mix but over the whole year , in sum, it is not much.
Yearly it turns out to be about 7% . The Germans have about 30 times the amount of solar that we have in this country.
Q: Weren't the Germans concerned that they might produce more than they could cope with
or is that just a press story.
It is causing problems. If you have a load of solar and because you are paid by the KW , everyone
in Southern Germany has their arrays pointing due south and at 30 degrees so the solar generation
in midday goes straight up. There has been an idea to give people different tariffs
, so your array was in north Germany and it was pointing a bit to the east balanced
by some more pointing a bit to the west, your overall generation curve would be much
flatter. Its all driven by money , money before engineering . That current situation of one direction power
surge has to be controlled because it may not be there tomorrow. Its causing grid problems .
Not that problem here as nowhere near that amount of solar. In southern Germany they have
water storage using surplus electricity to pump water up for hydro-generation later.
Q: You mentioned the French but when we went to Nice last year we saw hardly any PV.
They're not as advanced as Germany . They have a lot of nuclear generation but their FITs are
arranged so its very good for domestic PV . Then they dropped the FIT to such an extent that
it stopped any instalations so they had to increase the FIT to encourage again.
Italy is the second largest market in Europe.
Q: I asume by problem you mean that say you have 90% generated by solar you have some
frequency problem and instability.?
I can't really tell you but one thing is the following. I visited one of these huge solar farms with
5 megawatt and it is row after row of panels. The thing that amazed me is they had row after
row of inverters as well . The same inverter that goes into a small building and just hundreds
of them. I could not believe that all these inverters could syncronise and work to
each other. Each inverter is monitoring its local environment and just put their individual power into
the grid. I could not believe you could connect independently , all these , and they'd all work.
I was expecting some extra big piece of technolgy to have overall control and handle
any problems , but nothing extra.
Q: Do they need to clean these panels or is rain sufficient to keep them efficient.
If they're over about 20degrees then its sufficient . They have a special surface to
stop moss growing. Some solar farms next to landfill sites are covered in seagull guano
and there is nothing they can do about that.
Q: With al this changing of FIT levels and now EU surcharge coming in it does not make for
a stable market,. Is it very up and down.
This technolgy is not rocket-science and it boils down to installlation work no more complex
than putting a freezer in your garage. The same job time after time . You get computer
programed gdesigns for arrays , everything is near enough done for you. You don't need specialist
brains behind each install. There were a lot of companies that started up just to sell solar
and they came and went but the install jobs can be done by companies that
are not specific to PV so another arrow to their bow. They have to have NCIS accreditation that costs a lot
and they have to sign up to a financial scheme .
( ... loss of record due to disc change. Q: about whether combining PV and solar hot
water in the same array due to temperature intolerance of PV? , A : yes .... )
Q: Recent snow event. The tiles hold the snow but with our PV it was like an avalanche.
I was wondering if neighbours are aware of what may come their way.
There is always unseen consequences. The person I know with a large array was out that morning
wiping all the snow off , to squeeze a few more KW off . You get obseesed when you've
Q: Are all the panels rtectangular or square, triangular ones or ones that will fit awkeard roofs. ?
Company called Romad? , who make bespoke ones , I've seen triangular ones but they are
Q; If you're an average householder and an average house is it a nobrainer that you put PV
on because of the money saving? say you have the capital to play with.
If you are planning on staying there for 10 years its worth doing. Last year it would have been
crazy not to.
Q:If your house faces east to west ?
Then you have to have 2 inverters or an inverter that has 2 power-point trackers. With an E-W house
with a nice ridge you can put half your array on one side and half on the other and that
will give you a more even generation . Because the panels are relatively cheap now, then you can have all
sorts of different configurations and it is worth it. You can have one inverter to just one panel
, a micro-inverter. In that circumstance you would likely be using it all as low in comparison
to house demand.
Q: has the reduction in FIT caused the demise of companies that installed for free and taking the
I hope so . The worst one was saying that you would get 50% of your electricity free. And you
only have to pay us £5 a month for mainainence. Not just taking the FIT but passing on
demand for monthly maintainence. The govt did take the step that if you have more than 20
instals that belong to you then you get a lower FIT whiuch is justifiable.
Q: I was thinking of the wind today and PV. Relatively large unsupported areas of
glass on roofs. Is there any chance of them popping due to vacuum . negative pressure
over the ridge of a panel and sucking out the covering glass. Compared to small uniot area tiles, where
eventually the wind if sufficientrly strong would lift them.
I hope not. You need to leave a gap around the edge of the panel, used to be 30cm , now 40cm , around the whole perimeter. Not the case of standard panels not fitting to non-standard sized roofs. You
need that gap so the wind cannot get under the panels. The Germans sorted out the mechanics and they
stick these things on mountains so they must happily take strong winds. The big thing in the data sheets
is the snow-loading . The mounting fixture companies show videos of failures other firm's mounts
but generally they are over-engineered. As long as they are screwed properly to the roof.
Something else you have to do is that you have to get a structural survey to make sure your
roof is solid enough to take the extra weight.
Q: A more general question relating to renewable electricity that is available from electricity
supply companies. Mine is from Ecotricity. Unless there is boundless amounts of renewable energy
waiting to be used , I don;'t think one of these companies ever says we cannot sign you up
because there is not enough renewable generating capacity. There seems to be too much available
to sign up to green tariffs. What is going on.
All the electricity is mixed in , in the mains.
From the audience - The power companies have a certain obligation to take a proportion of renewables anyeway. For their more
generous clients they just charge higher tariff for it but get the samne electricity as everyone else.
One thing I'd like to leave you with is tech that is coming in noiow is that the inverter technology is mixing energy from
different sources. With the communications systems we have today you can make very small sensors and hundreds of sensors all over the place and can monitor and control energy supply and use much better
than the old days. I think that will be significant in the near future. The energy efficience of houses
will go up. We now have all these tools and energy efficiency is the first thing to have a go at.
Q: Do you see the manufactured price of PV panels coming down?
Not for a while because everyone is losing money. There is huge oversupply
in the market. All the panels you buy now are at less than manufacturing cost.
Q: Is there any new technologies in the pipeline that will bring the price per watt down?
There are some technolgies but the price hiccough has had effects. A few years
ago everytone thought that thin-film would be the future but it is not
price/energy efficient compared to recent crystaline production. If the pricing was normal
, without the problems , things would happen much slower and these other technologies
would not have been left on the shelf. Everything is driven by the cheapes panels of the momnent.
Q: Any quantum technologies coming into the scene?
Not silicon based but titanium oxide and platinum , sounds expensive but they are meant to
be allow cost future as the world is full of titanium oxide (basis of white paint)
and such a tiny amount of platinum . You can also dye them different colours so probably
able to colour match to existing roofs. But again it is something that has been put
on hold because no one needs to make a cheaper version. Just awash with cheap panels .
Q: Am I right in thinking that it only takes one leaf to land on one cell and its corrupted
the whole system. Is there anything that tells you that you have one cell being blocked othe r
than looking up and noticing , oh we have a leaf stuck on there ? Does the system tell you
that your system efficiency has dropped significantly?
You can use a micro-inverter , where every panel has an inverter is one way round it.
Another way is inside the panel they divide thew panel into 3 sections and in a sense each
section can work independently of the other 2 sections . An automatic process
that bypasses a section if one cell in that section is low in output.
Q: The cost of micro inverters?
About 100 pounds , some companies say 70 pounds and then £30 for the connection
cables. Also you need a small monitoring box that costs 2 or 3 hundred pounds which
gives you individual reporting on each panel.
Q: Can you opt for say 12,24 or 36 volt system or are you constrained to having a
240 volt system?
For off grid use you can get inverters , say for caravans . Thats where PV emerged in
the caravaning/sailing area . They made panels that provided 12V.
Now we have the grid connection then as large a panel as you can.
Q: Could you parallel up a load of standard 36volt panels rather than series them up or
is there a current hogging problem or something? Can you limit yourself
to 36 volt system for charging batteries say, using standard panels that would otherwise
be part of a 240 volt system?
Yes . Because of the change in the last 3 years there is surplus stock. So you can pick up
surplus panels for something like £50.
Q: I believe the efficiency of PV decays over the years, what is that sort of figure?
I assume its improving over time. ?
Theyare all guaranteed to be 80% after 25 years. I should have mentioned that the
individual solar cells will decay over time . The guarantee also stipulates to be over 90%
of their initial value after 10 or 12 years. In Germany some of these
panels were measured after 25 years of use were down to 92%. Another cmpany said they
had lost 2% over 10 years.
Q: How could they decide how quickly they were going to decay?
They have tests using accelerated aging . Something I lioked about the German panels
is that they had history and you could tell how they were performing after a period
of time. Nowadays everyone has near enough the same guarantee. Perhaps
meaningless because the solar company may not be here in 20 years time .
The guarantee means you have to take the panel off the rof, send it to a lab for
Q: Is the aging linear in that you could measure it over 1 year and that continues equally evey year?
There ar emany factors. One factor is the clarity of the glass . Apparently in space there is hardly any degradation , with the panels on satellites.
15 Apr 2013, (third Monday) Prof Lindy Holden-Dye: optogenetics
21 people, 1 hour
I'll start the talk with the big question - how does the brain work. How do we remember things, how
do we incorporate our past experiences to behave in the way we do. And what goes wrong, diseases like
dementia or Parkinson's disease. These are the big questions in neuroscience. For decades now
neuroscientists have used a range of different techniques to try and understand how the brain works.
One route in is to see what happens if a bit of the brain is damaged. The story of Phineus Gage?
a rail;way worker back in the 1800s , working on the railway and there was an explosion which
blasted an iron bar straight through his brain. Amazingly he survived but it completely changed his
personality. He became very disinhibited as though drunk all the time . Before his accident a likeable and reliable bloke
but after the accident became unreliable , abusive , getting into fights . It was devastating for him , but
it told us something about the region of the brain , the pre-frontal cortex is involved with controlling our
behaviour. It stops us acting on impulses, one of its roles. Lots of other similar stories in the medical
literature. A famous story of a patient called HM and he had a region of the brain called the Hippocampus
removed because he had intractable epilepsy. He kept getting seizures so he had surgery on where
the seizures originated . It worked in curing the seizures but unfortunately HM could not form any
new meomories. So an important bit of the brain for making memory. he had to live the rest of his life
in an institution as a result of that surgery. So lots of stories like this where people have suffered damage to
a particular region of their brain , we get an understanding of what each area of the brain does, all
More recently, we can image the brain , Functional Magnetic Resonance Imaging , placing a person
in a big magnet and the brain is imaged while they ask questions . Again its a bit observational,
some cynics call it a new phrenology , like feeling bumps on your head. It gives some insight
but its not very discrete. Animal studies have been important in understanding the brain. Some
of my favourite work has been done by John O'Keafe? and Edward Moser? . They have looked
at a particular region of the brain called the enterinal cortex and the hippocanthal network , a part
of your brain that is involved in spatial navigation. How do you find your way around the
environment. They put electrodes in those parts of the animal brain and monitor while they are moving
around the environment. Some cells fire specifically when the animal is in a paticular place in its
cage. Those are called place cells and others fire when making a particular head direction - head-direction cells, and others
fire in a very geometric pattern , whenever it reaches a particular place in the cage , that maps out
like a grid and called grid-cells. So a map of the environment , a geometric mapping, like a cognitive
sat-nav. What we can't do yet is specifically activate a particular network or particular cell and asl
does it produce a specific behaviour. Could we perhaps activate specific memories by activating
specific circuits within the brain. Can we understand how the brain encodes information.
The other thing biology and neuroscience has done is to take tools from nature for different
experimental approaches. Using flourescent proteins that occur naturally in nature and using
them in another system to give some insight into the biological organisation. The first was a green
flourescent protein found in jellyfish, that glow in the dark. So biologists took the gene that encodes
the protein that makes the jellyfish glow and transfer that gene to another animal . So animals
like mice that now glow in the dark. You can use it as a tag to find out where specific molecules are
within the nervous system for example and then lok to see where it is.
Now to introduce my experimental animal , that I have in my pocket. This is C-Elegans Cenordap - Elegens?
a little nematiode worm just amm long. A popular animal in biology and neuroscience because
it is genetically very tractable , and a simple nervous system of just 302 brain cells. Although a simple
nervous system it can still do some quite clever things. It likes food . When it is growing up and you put it
on a surface of a particular temperatuure so it associates food with that particular temperature
then when an adult you put it on a temperasture gradient it will go to the temperature it associates
with food. It can make that association between temperature and food. It has a nervous system that
can learn and behave in quite a clever way so a useful tool for neuroscience.
This worm is used with the technology that I will describe to you. So magnified video of the worms.
They are translucent so you can see their internal organs . Wriggling around on a background of bacteria
that they eat. Because they are transparent you can visualise the structure of their nervous system
in the intact wriggling worm. So we can add the green flourescence to the worm's nervous system
and see where it is. That flourescent protein was given a "postcode" a message to say just put
that protein in the pharynx say of the worm and you can see it flourescing in the worm.
You can ask particular questions about where particular molecules are expressed.
Scientists scoured nature for other flourescent proteins that might be useful and they
found a red flourescent one from coral. Added to the worm and an anterior section flourescing as red
flashing. So difernt flourescing proteins can be used for these approaches . The ultimate has been to
produce a colour palette , the Brainbow technology
They transected the brain witha palette of
flourescent proteins and they'd been taken up randomly by different cells to give a whole spectrum of
colours dependent on whether taking up red or green or pink or blue . It creates beautiful images but also
a very useful purpose for tracking pathways in the brain. You can follow the pathway to another part of the
brain , mapping out the circuits in the brain. These flourescence techniques have been used more than a
decade now as a tool in biology to mark tissues to better understand how tissues are organised
and what molecules they express.
What would be a really fantastic tool in neuroscience is if you could activate particular neurons within a
surface and what behaviour does that control. This new tool uses algae , green algae that
respond to light. They express a proteing that responds to light much like the protein in your
eye responds to light called rhodopsin. These green algae also expressa rhodopsim that eables them to respond to
light and affects their movements. The diffence between your eye rhodopsin and the algae response is
that its an iron-channel . A molecule that sits in the membrane and when it responds to light it opens
the channel and that channel allows an electrical current to flow across the membrane and activates the cell.
That how these algae move essentially. The light shines on it , activates the rodopsim molecule , an
electrical current flows the channel and the algae can move. The people working on that project thought
wouldn't it be neat if you take that protein that is actually responding to light and you could use that
as a light sensitive switch in neural circuits. So introduced to a membrane and when light is shone onto
it, it changes its shape , a conformational change and a current flowing through it. This current can activate neural
they then discovered another switch that is a microbial protein that passes a current but now instead of being an
excitatory current that will switch cells on it is an inhibitory current that will switch cells off.
So 2 proteins identified in nature, one that could be used an an on-switch for nerve cells and another
proteingn for an off-switch. So we have a genetically encoded, light activated , on and off switch
for neurones. The next step was to show this working in a functioning nervous system.
So a nerve cell that has been transformed so it expresses this on switch and the off switch, these
channel ridopsims . This is what is meant by optogenetics because they are genetically encoded
sensors and can be controlled optically. Simply shining a light on it will change the activity of the cell.
A recording of where a an electrode has been inserted into such a nerve cell . When a blue light is
shone on and excites with a lot of activity in the cell and then yellow light activating the off-switch
and the cell is becomes silenced. Switch the yellow off and active again and switch both lights off
and it recovers. So you can remotely control the activity of a neural circuit just by shining light.
Since then a lot of work into optimising this kind of approach has been done in invertebrate
models. We've ben using C-Elegens (C-E) to optimise and use for our work. Also a lot of studies in using
Drosophola as well.
So how are these genes introduced. Transgenics where you put a gene from one organism into
another. We can make C-E transgenic simply. Take an adult worm that is with eggs on a microscope
stage with an agar pad then a needls with the DNA that you want to inject into the worm
aimed at the gonadal region of the worm , the DNA gets taken up by the developing germ-cells .
When it lays its eggs then they will be carrying the genes,fairly straightforward.
Collect the the progeny from that worm , hermaphrodite self-fertilising so no crosses required.
Then on your plate lots of trangenic worms , that the way we've made the worms that are expressing
the channel rhodopsins (C-R) . given the Cr a specific postcode so they will go to a particular cell .
Q: Does it always work or is it statistical , say 1 in 10 or something.?
If you're good at it then you can get about 30 percent taking it up. You have a thing called a dominant
marker and that will show whether your gene of interest has gone in or not. Often its a flourescent
protein. The CR with a flourescent protein as well and look for the green worms say.
Q: Then the next generation will be 100percent?
It won't be 100% unless you do something called integrating , because the DNA is outside the
normal genome , called an extra-chromosomal array. Give them a shot of UV light , it makes
breaks in the DNA and the exogenous transgene gets incorporated into that break and you can
make stable lines and 100%. This is what we've done with the CR CE. all the progeny will have the CR.
There are different approaches, you don't always have to do it that way.
Work from Alec Skopshalk's (?) lab in Frankfurt , the early work on CR . He expressed CR
in the worm , in cells that control the motor activity of the worm. With no light shone it is moving away and then
as soon as the light goes on they become like little sticks.
Instead of the command-motor system like us , CE have nerve cells that tell them when to go forward
and when to go backwards for food foraging . We put the CE in the neurones that determine
whether or not the worm goes backwards. To see if we could make the CE go backwards on command.
Not just for fun , but trying to understand how the circuits work and changes in response
to its environment. Video showing the remote control worm. They go backwards a bit , spontaneiously .
Put the blue light on ond 2 worms go back at the same time . The backward activity neurones are activated
because the CR has opened in the cells and the worm goes backwards.
We can record these light induced reversals. Flash, how long after the flash before the worm goes backwards .
The CR activity requires a co-factor called retinal? , same as in your eyes, required for rhodopsim function in your eyes
also. For the worrms with the Retinal , when you shine the light they go backwards almost immediately.
THe ones without Retinal go backwards when they feel like it. We can discretely control the
activity of the worms by expressing the CR in the sub=circuits within the animal.
One of the projects we are interested in , is how the brain responds to alcohol and the adaption
to continual presence of alcohol . What happens within the circuits of the brain when a person
becomes tolerant to alcohol , thebrain adapts, becomes less responsive , then with prolonged use,
show withdrawal. The circuits change, showing some plasticity . We have a model for alcohol
dependency in CE. We treat them with alcohol and they get drunk. We can take the alcohol away,
after they've been exposed to it for some particular period of time , they show tolerance and then withdrawal.
Becoming quite hyper-active. We wish to understand what occurs in the circuits that underpin
We can use them for discrete experiments that are electro-physiological . Cut the head off the worm ,
and expose the 100micron of the pharanyx. Stick an electrode in there and record the potentials
in the muscle. We put CR in the circuit that is driving the activity of the muscle and when we shine the
light we get activity within the muscle and contraction and relaxation.
With these CRs we can very discretely activate particular classes of nerve-cells and try and understand
how its activity is being controlled.
Back to the big question as to how the brain works. The CRs will be really important tools in
understanding this. Being used now in lots of topics within neuroscience. Using mouse models
there have been studies into anxiety , depression , Parkinson's Disease, fear memory ,
The one I like is fear memory . We've wanted a better understanding how memory is
encoded within the brain. Wgat we haven't been able to do is activate specific neurons and ask
if that gives memory recall. We know that if we ablate particular neurons an animal
cannot make a memory. But we've not been able to activate that particular neuron and see if that is where
the meomory is stored. Some nice work in that area using a quite simple approach. A conditioning
experiment on a mouse, like a Pavlovian experiment . Fear is a very robust and long-lived memory from a
traumatic event . Sound a tonne and give them a foot shock , enough to be uncomfortable and make them jump. They then associate
that tomne with the shock. They will then respond to a foot shock by stopping in their tracks
and also respond to the tone by freezing in anticipation of the foot shock . So the qustion is wher eis
that memory stored in the animal brain. They used CR . During such a conditioning experimnent
neurones that are activated like that will express a particular kind of protein . They tagged the
protein with CR so when they carried out the experiment, the neurons i nthe brain , that are active
in the memory formation will be loaded with the CR . Getting the CR into just htose cells in the
fear conditioning . Now they could generate the freezing respons by the sound but could also elicit
the freezing of the animal if they shone light to the CR nerons via a fibre optic cable and the animals
froze. That activation of those neurons was recalling the memory that had been encoded
during the conditioning.
Q: If you find in one individual a particular neuron , stores that memory, would the same neurone store
the same memory for other animals of the same species or different location for different individuals?
How consistent is it . It might be different for different kinds of behaviours for example. This is always
in the hippocampus region . Subsets of neurons rather than just one. They would not be exactly the same .
In CE because it has such a simple nervous system with its 302 cells , the only animal
where the wiring diagram of the brain has been completely mapped. You can go back again and
again to the same nerve cell in the same animal. But you can't do that for a mouse as billions of
neurons and much more complicated. But you can look at subsets of neurons.
Q: Don't memories move around between short term and long term?
It is distributed. Short term in one part of the Hippocampus . Like patient HN who could not
make short term memories but his long term memories was still intact. All his procedural
memory like playing the piano , riding bikes was retained . Memories are usually consilidated in the
There is also the issue of forgetting things . Does it have to do with reassigning memory to new
neurons ? If we don't use these memories then it is lost?
I think its more than that. I have a colleage in the US who works with snails . They have this idea now
that forgetting is an active process. Its not just that you've lost the memory but that you've
actively destroyed it. And forgetting is an important part of learning., Sometimes you can't learn
new things unless you destroy the old things.
So it seems to be an active writing and erasing process?
Are there any paralells with computers? In the early days of computers they didn't have much memory
and used to have to keep erasing stuff before they could replace with new programs. ?
I find it fascinating that CE have 302 neurons , precisely? do they all have 302?
Yes, all the ones that have been counted , yes 302.
Is there any suggestion that higher animals have , a bit higher than CE , have such regimented brain
CE is probably the only animal that we have that degree of detail . Things like jellyfish have simple
brains but have nerve nets that are extraordinarely complex. It was extraordinary
perseverence in the 70s , taking CE and slicing it into little bits and elctron microscopy looking
at each scell and reconstructed cell by cell, the wiring diagram of the whole worm.
The paper is called "The Mind of a Worm" in Proceedings of the RS, on the internet .
If you're a neuroscientist interested in animal behaviour you have the whole circuit. Which are involved
with egg-laying, which for locomotion , which for ? sensations . Then ask questions on how an animal adapts
to its environment then Ce is a wonderful organism for that.
Are all CE neurons in one place or are they distributed?
Distributed , but more in the head than the tail apart for the males which have more in the tails.
No spine but a motor-nerve cord with neurons in it.
Again like a computer where you often hve a control chip on a peripheral raher than on the main board?
When did humans and nematode worms have the same common ancestor? I find it surprising
that both have the same reaction to alcoihol, physical and psychological . ?
They live in rotting fruits, they're preferred environment.
I know elephants are supposed to react in the same way as they often come across rotting fruit . Is it just coincidence
of using the same chemistry, the same cell structures then the same effects.?
These sorts of things have been conserved right through evolution. The neuro-transmiters , the proteins are
all there just maybe in a slightly different arrangement. The CE only has 302 nerve cells but the
way they work is much the same as how a human brain works.
Ids there any mathematical significance to the number 302? its not a prime number?
Thats a good pub question.
What sort of intensity of light is required for the switch on and off, presumably more than room light but
not necessarily laser light?
When we first did it we went down to Maplins and got little blue LEDs and placed on the side of a microscop e
and that was sufficient. We can use UV light but have to be careful of heating as an artefact ,
maybe getting an activation from that which was non-specific. Usually the problem is expressing
enough of the CR in the cells , getting the transgenics right to get enough of the protein
into the cells.
You r graphics showed very much on and off cycling , but I assume there is intermediary state
or is it binary process?
The channel is a binary state, either open or shut , no subconductance states that I'm aware of
with the CR. And pretty rapid within milliseconds, thats why its quite neat.
Deep brain stimulation has been developed for Parkinsons Disease where electrodes are implanted
and act like a pacemaker to the brain to alleviate movement disorders . You can imagine that down the line these
sorts of approaches offer an alternative and mor ediscrete way of controlling such circuits in our brains.
You have a problem there of finding a way to intrduce genetically modified material.
With deep brain stimulation you are stimulating a closely associated group of neurones not
targetting specific ones?
Yes . They've used this in animal anologues and Parkinsons Disease . DBS is not always effective and if they could
tweak it to activate specific circuits rather than bluntly activating all of the circuits in that part
of the brain , the sub-cepholanic ? nuclei , you might get more fine control . With animal analogues
they can get much finer control via the genetics approach and better outcomes in movement.
But for humans you need to find a way of safely getting that gene into the human, quite
a long way to go for that. Even if that route comes to nought, at least it will give us a better
understanding of how those particular circuits in the brain work.
Monday 13 May, 2013 Prof Chris Rhodes : What happens when we run out of oil.
3/4 hour talk, 3/4 hour Q&A, 35 people
A petrol pump from a few streets away from where I live, garage didn't fail from peak oil, it
just went out of business. 90% of the world's transportation depends on liquid fuel . So does just about
every other aspect of our lives. If there is going to be a slow down of production of crude oil
then it will change human civilisation to some considerable degree.
The world use of energy over time , 1965 to 2005. The top curve at the end of that 40 year period,
humanity was getting through about 2.5 times the amount of oil at the start. Twice as much coal,
3 x as much gas and the rise of the nuclear industry. By the end of that 40 years, 3 x as much
energy than at the start of it. And that is just for a doubling of the population, tha tis more
energy per capita.
From the BP statistical review, the latest info.
The lion's share of energy use by humans comes from cruse oil, coal is a close and rising second ,
then natural gas then avbout 5% each of nuclear and hydro and less than 2% in the form of renewables.
So about 90 % of our energy comes from fossil fuels.
The "good old days" the Humphrey Jones gusher in Texas a century ago.
Then all that was required to do was make a hole i nthe ground. An oil well doesn't only contain oil ,
gas under pressure , a hydro-static pressure of water pressing against the oil. Once you drill through the
rock at the top that is holding it all in place , then the pressure forces the oil out.
In those days you only needed to expend one barrel's-worth of energy to get 100 barrels in return.
Now the figure is probably somewhere between 10 and 20 , payoff falling all the time. We have to work
harder to ge tthe oil out, for various reasons. 30 billion barrels of oil produced a year, 84 million barrels
a day. Major producers are Saudi-Arabia and Russia , between them about 1/4 of the world oil.
Different designations of oil , according to its viscosity, heavy or light, sweet or sour depending on its
sulphur content. The sweet oils are favoured as cheaper to process to fuels. The light and sweet oil is the
easiest to refine to petrol. Lots of it still in Saudi, Iraq and Iran.
Oil is not a single material - an exhibit at the Norway Museum of Science and Technology .
The differecnt Norway fields are not that far apart from each other , geographically ,
but adjascent fields pull up entirtely differnt kinds of material, some very light some you can handle as they
are that viscous, like tar macadam. Entirely different molecules and different properties.
Petroleum is far from being a simple substance.
Much of the oil remaining is the high S variety. Needs very costly processing and not all is
recoverable. People often make claims that they have so many billion barrels of oil , discovered
somewhere. And it will turn out that a majority of that may not be recoverable at all.
People say there is something like 3 trillion barrels of shale-oil unde rthe USA but actually
its not shale and its not oil, it contains an organic material called Keragin ? a solid that you have to use a lot of
energy to heat it up to produce a product that in any way resembles crude oil. Very poor return
of energy on energy invested. Oil is the raw chemical feedstock for plastics , chemicals ,
pharmaceuticals, computers, telephones etc. Try looking around a room and find something that does
not depend on oil. The bricks of this old pub have probably never seen any oil, but modern
bricks would have. We're totally dependent on crude oil and the rider is that without oil and
natural gas to make fertilisers, we couldn't grow any food.
Modern industrialised farming , a field of soya being harvested by an array of combine harvesters
in Brazil. The fuels for them is refined oil, then transported around the world again using fuel.
The great cloud of dust kicked up by them , topsoil. We're loosing topsoil at the rate of about 1%
per year. If we could get around the shortage of resources, we are running out of land.
The UK imports something like 40% of its food and then moved around the country.
UK food production uses about 1 million tons of oil for tractors etc. Transporting food in the UK
uses about another 3 million tons of oil and about a fifth of that is for car journeys to
shops to buy it. Add it all up and about 7 million tons of oil per year to feed Britain. About
one sixth of all the energy used in the country yearly, ie farming, packaging , refrigeration, transport.
5 million tons of oil for plastic is used in Britain every year , much of that for food industry packaging.
Total oil for transport in the uk is about 60 million tons per year and about a fifth of that is for planes.
A man in 1956 , people should have listened to a bit more but they didn't. A geo-physicist , Hubbard,
working for the Shell developement Corporation in the States. No computers or analytical
models then. He had an expert system . He looked at oil production and the peak discovery year ,
the year when more oil was discovered in the USA before or after , was 1930.
Forty years later the peak in production would happen and everyone laughed at him . Because in 1956
Texas was awash in oil. He was spot on , production peaked in the USA in 1970 and now the US
imports about 2/3 of its oil. So if Hubbard is right for world production and peaking in 1965 , add 40
and then 2005 , we're not running out of oil. But light sweet crude oil production did peak in 2005.
We're not scraping the bottom of the barrel but going that way, more of the heavy crude , high energy
to deal with stuff.
In 1980 we started using oil faster than we found it. So now for every 4 barrels of oil we use , we only
discover one new barrel. So like spending your money 4 times faster than you earn it.
Of 98 oil producing nations , 2/3 are now past their production peak. Saudi is still ok , Canada seems to be doing ok
as does Venuzuels. But Russia , America , the North sea (peaking in 1999) is a global
Energy Information Administration , part of the USA Dept of Energy, so not some crank outfit.
Produced in 2010 , the analysis reckoned that about 2013 there should be a peak in oil production.
Once oil production is peaked then there will be a terminal decline and by about 2013 only
half our current production of oil. They euphemistically identified the gap as "unidentified projects"
A big hole, where is it going to be filled from.
The French equivalent, the Paris based , International Energy Agency , world energy outlook
published in 2012
Projected oil supply out to 2025. The bottom set of bars , actual oil being produced at the moment.
So by 2035 producing half what we are producing now.
Fields yet to be found - they know where they are but unless the price of a barrel goes up to 130 or 150
dollars , its not worth it.
Natural gas/ liquids are not the same as petroleum , diferent molecules. You can't refine them into petrol,
you can use them for other purposes.
Other unconventional - coal to liquid, tar-sands, like tight-oil is what people generally call
shale-oil, recovered from fracking. They don't believe its going to take over the world. It
will be fairly steady. So if you add the light tight crude oil , unconventionals then probsably about the same
scenario as the USA counterpart of this agency - within 20 years about half the amounft of
crude oil than at the moment. And no obvious means to replace it.
So we need to find alternatives to oil. We're running out of the cheap and easily won stuff. Need to
find alternative fuels, carbon feedstocks for industry. The other suide ot it , burning oil
contributes about 3 billion tons of carbon to the atmosphere, about 1/3 of all emissions that can
be blamed on humans, and the issue of climate change. So as the oil runs out, we will be releasing less CO2 is
one argument. We are a global civilisation that is entirely dependent on moving goods and people around.
Without some alternative then a fairly difficult situation. People talk of hydrogen as the perfect green fuel.
In reality it is . Mix hydrogen with oxygen from air in a fuel cell , produces electricity to run a green
car , pure water drips out of the exhaust. But you can't mine hydrogen out of the ground. It has to be
made. Most of the hydrogen is made from steam and natural gas but for every 4 lots of hydrogen you
get one lot of CO2 . Most of the hydrogen, presently , is used to make fertilisers . Used i nthe oil industry for hydro-
cracking, hydro-desulphurisation . The process whereby you get the sulphur out of the sour heavy oils.
You have to do that or you can't really process htem . Capturing the CO2 , CCS, would cost about
1/3 of the energy of the hydrogen that you get out at the end. This also applies to coal-fired
power stastions. If you want to adopt CCS, for every 2 coal-fired power stations you need to build a third
one to cope with the C emissions from those 2 and itself. I think that is why noi government has
taken it seriously, so far and maybe they never will.
An alternative is water-electrolysis, pass electricity through water, split into H and O , but where
does the electricity come from. To generate enough H in one year , to match the 60 million tons
of crude oil that we use to make our liquid fuels at the moment would need 61 GigaWatts of power,
assuming you can use H at twice the efficiency of petrol. So 61 new standard sized power stations.
So if these were coal or gas fired , for the UK, that would treble our C emisions. Its seriously
suggested that we could use nuclear power as CO2-free, when they are running, a lot of CO2
for making the concrete etc of them. But you still need 61 of them. How many can you bring
on in a year, say 5. It takes 10 years to go from drawing-board to producing electricity.
So 20 years time, by then we will have lost about half of our conventional oil production,
so a loosing battle, tiomewise. There is only one factory in France that can make the new
generation reactors, making 2 per year. For any of these, requires a huge amount of manufactuiring
capacity, True for solar or any other innovation. Its not just the UK thast would want new nuclear generation, everyonne
will want them. Perhaps use renewable energy from wind-power, we still need 61GW.
So take the biggest available wind turbine, 5Mwatt. Rated capacity is not what the generator gives out.
More an average output, due to wind speed etc, usually about 30% capacity factor so about 1.5MW
on average. So need about 40,000 , so put them in scale. They would be clustered rather than evenly
around the English coastline. But if they were , the coast about 2500 Km around , so placing the
recommended 10x rotor diameters about , so 1.25km , a single band would be occupied by 2000
turbines, so for 40,000 would need to be 20 turbines deep, about 15 miles wide around the
entire coast. How quickly could you supply all this. Assume making and installing one a day , it would
take over 100 years. Another aspect is rare-easrth elements. There are shortages . Very important
in wind-turbines and hybrid cars because they make very powerful magnets . A rare-earth magnet can
have 10 times the strength of an iron magnt. So for the same magnetic field a tenth of the weight.
But for each Mwatt of wind power you need one tone of rare earth element, 1/4 ton of neodymium ,
so about 135,000 tons of neodymium . That is over that we need currently 5 times the rare earths currently in proiduction
, to meet existing targets because we've signed up in Britain , to the EU to provide a lot of renewable
energy in the form of wind power. I do really wonder if we can meet our C emission
targets. The Toyota Prius contains about 1Kg of neodymium , 10 to 15 Kg of lanthenum
for the battery. So hybrid cars need a lot of rare earth materials. The joke rin the pack os that
practically all the rare earth elements in the world, at the moment, come from China.
China itself is undergoing unprecedented industrial expansion and need all the energy it can get its
hands on, including renewables. They are holding back on rare earth exports for their own
purposes. So for the UK and other countries to meet the targets
then we need to find other sources.
A field of sugar beet in Norfolk. Grow your beet, extract the sugar, ferment it , separate the
ethanol from the water , via distillation or membrane technology . the UK has only about half the acreage
required for this, even if we stpped growing food crops , only 50 % of our fuuel via this process.
Its even worse for bio-diesel . Being charitable , in assuming we convert all our engines to diesel ,
so less fuel required. We would then need about 400,000 sq km of arable land but we still have only that
65,000 . So grow rapeseed only, stop food production , we'd only meet 1/6 of our current fuel
requirement from crude oil. At best land-based fuel crops are never going to make more than a few percent.
We already import about 40% of our food , we should be growing more of our own food rather than turning it over
Water use. If you look at water use in Litres/Megawatthour , for extraction and refining of oil
you're in the tens or low hundreds. For corn and soya beans for ethanol you're up in the millions.
So not only pressure on land for one kind of crop or another , huge pressure on the amount of fresh
NMaking bio-diesel from algae. Its green because it contains chlorophyl , undergoes photosynthesis
, absorbs CO2 from the atmosphere. Then process the algae into biofuel. This has recommendable
points to it . Once you get the alga and filtered , it is still incredibly wet. The energy required to dry
it before further processing, will be pretty large. Advantages are it can be grown in
tanks that yield of about 100 tons per hectare, compared to corn or rape seed of about 1 tone per hectare.
So if you could have 4,000 sq km of these tanks , that would produce your 40 million tons of biofuel.
We could run the UK on about 1.5 % of the land area. Huge amounts of engineering and at least 10
years off. You don't need good crop growing soil, you can stick the tanks anywhere, so no
compromise with food production. You don't need to use fresh water either. It runs well on
saline water and waste water , and so no competition with general water requirements.
You can feed your alga what is usually rubbish, agricultural run off water and sewage water
contains a lot of N and P and avoids the otherwise artificial input of N and P , actually
cleaning the water which would otherwise cause algal blooms. Cleaning this water
artificially is very expensive and difficult but if you can integrate this into one collective,
would be impressive. Tanks with alga growing in them, waste water treatment plant,
fossil fuel plant emitting CO2. You feed your C emissions and your polluted
water to the algae , you get rid of 2 problems and solves a third one, which is what to do to replace
the crude oil.
Fracking. All very controversial. Take a fluid that is largely water containing all sorts of
other materials like sand which acts as a ?, it also contains various chemicals acting in various
ways according to the situation. You pump the fluid down at 15,000 psi .You have some rock that
contains some gas or oil bu tthe rock is impermeable. The pressure is high enough to crack the
rock open so what it contains can flow out. A very energy intensive process. A single
well can produe a million to 5 million gallons of waste water. This water contains not only the
original fracking fluid but also the likes of Radon . These processes are done at depth
and in some locations you will have radioactive materials and all sorts of other things.
You can never know what will be in the mix at the end of the process . To meet our projected
targets of 2035 we'd need to dril about 1 million new shale gas wells.
There has been something like a million fracking operations conducted across the world already
and about 9 out of 10 oil and gas wells are stimulated by fracking processes. The production drops
, so they can be jigged-up by fracking . Not a new technology , been known of for 60 years.
There is a big concern about contamination of water . Video clip called gas-land? a guy in the
USA goes to his kitchen , turns on the water tap and he can ignite the gas coming out of it
s its so heavily contaminated with methane. A University of Texas research showed there can be methane
in the ground for other reasons than fracking. Old timers will say they were igniting the gas here long
before they were fracking. A complex issue. I don't think the claim that the USA has enough gas to
last a hundredyears , stands up. Look at the figures and its more like 20 years worth.
Can solar, wind , geo-thermals save us. Yes, but not entirely. What we get out of these sources is
electricity. Looking at primary energy use it is about 1:1:1, for electricity, primary fuels poarticularly petroleum ,
oil for heating etc. Even if we could make all our electricity from renewables we still have to
somehow to provide the equivalent of liquid fuel and the heating. Liquid fuel supply is declining at
about 3% per year. Lests say by 2030 there will be this big holeof only half the production we
have now. There is no easy way around the transportation problem . 34 million cars on Britains
roads at the moment, powered by liquid fuels. I don't think there will be 34m electric cars any
time soon. Personalised transport will have to go for a Burton. The sensible way to move people
around using electricity is light-rail and tramways.
There are limitations to these various elements. I've touched on rare earths but other materials
like indium, cadmium , telurium etc which are critical to solar power and electronics
generally. They are called hitch-hiker elements , meaning they are not extracted in their own rights.
They are side extracts from the mining of other material such as copper, iron , aluminium , zinc etc.
To ramp up a lot of the renewable energy resources anyway , we would need to find alternative
and deliberate sources of these other elements.
It seems unlikely that we will be able to replace oil , in the enormous quantities that we use today.
We can't solve the supply problem from the supply-side so we have to start to look at the demand side.
Looking at the available oil and all our uses for it. It looks like that in 20 years we will have had to cut
our liquid fueled transportation by maybe 70% . In effect likely a return to a more localised
society. A process called Transition. Because we will no longer be able to move food around the
UK. let along bring in from abroad at the level we do now, about 40 %. We will need to produce at a much
more local level. Yes electricity for tram transport but we also must not ignore the foot and the bicycle.
A third-world basket maker , himself and his family taking under his own steam on an overloaded
bike to a local village. Producing more local food, materials, and the things that we actually need,
to cut down the amount we need to import from elsewhere.
Need to work closer to home. I live in Reading a well known commuter town. You think , the people
who live in Reading commute to london but an almost equal number commute into Reading
from London and surrounding areas to do high-tec jobs. So perhaps reskill the Reading people
to do the Reading jobs an dthe London folk to do the London jobs , would save alot of energy.
No mor echeap foreign holidays.
Last time I came here was to hear a talk on peak phosphate by Dr James Dyke . Modern
agriculture depends on artificial fertilisers , particularly rock phosphate. There is evidence that it
will peak in 2030 . I reckon it won't be that bad but there will be an issue some time this century.
farming is dependent on synthetic N fertilisers from natural gas and fresh water. There are
techniques like perma-culture which enable you to gtrow on a small scale but very effectively,
with minimum inputs. No good for mega farms with 20,000 pigs or whatever.
The future world will require more recycling and capture of P , N , composting human waste etc.
The one-time use of everything is incredibly wasteful and will not be viable into the near future.
I think a realistic future will not be where we've thrown away all our technology but producing
on a much more local level.
What if we don't do anything at all.
Up to 2005 line his history in this 2005 slide, beyond is prediction.
The red curve is the production of oil and then the first of the oil shocks , a slump, in 1973 when the
arab oil producing nations decided to make a show of strength against the west. They reduced the production of
oil by 5% and the price shot up 400%. A wake-up call and the west brought on some
interesting and useful projects. When the cheap oil came back ,there was no longer the incentive.
The aquatic species programme where they grew algae , to lead to fuel, that got shelved as
considered too espensive. The second shock was the Iran-Iraq war end of the 1970s,
then Gulf War , terror . Then we have price spikes, profitering , financial markets collapse
, unemployment, recession chilling familiar prediction those few years on.
So a not too slow descent into chaos and mayhem if we continued as business-as-usual. The elephant
in the room is population. 50 years ago there was less than half as many people on the planet.
Our use of resources is not a direct relation to the number of people on the planet but people of developing
coutries are aspiring to a consumer lifestyle . The western consumer lifestyle is unsustainable
even in the west. This is the cliff that we are trundling off.
Assuming a couple of trillion barrels of oil are to be had, altogether, you can make some guess as
to what the decline rate of oil use will be.
Population from the WHO of 9billion people by 2050 , who knows 12 billion
by the end of the century this plot is not indicating that. The population in this prediction will peak
at about 7,3 billion. I've seen various analyses and they're not all as Draconian as this , but they
all seem to predict a population peak sometime this century. Then it will fall, perhaps 1/3 the number now,
at the end of the century which some reckon to be the carrying capacity of the planet.
We are living ona planet with finite resources. A bit like bacteria growing in a petri-dish.
Only so much food available to them so you get your S-shaped curve. Whether bacteria or humans,
with finite resources, something has to give. Its likely to be limitations in resources that will
determin the peak. A happy ending - transition towns. Aim to achieve resilience etc. Driven by a
belief that the elected leadership will not sort out everything for us. What can we do at grass roots
level in terms of food, materials production etc. A great deal you can do. Insulate your house properly
saves about 20 % of our energy. If you start along the line of thinking that you are
going to have 5% less energy per year then you do things differently . Sustainable jobs.
Of those 70,000 from Reading to London every day what sort of jobs will they be doing in 10 or
20 years time, will those jobs even exist with say global companies , probably not. Perhaps
reapplication of practical skills and a total rethink of how we live. Quite a challenge.
I quote Charles Kingsley a clergyman and uni professor a commentator on the human condition .
Effectively he said that we act as though what we actually need , is lots of stuff and what we
really need is a reason to get out of bed in the morning , in modern parlance.
From all the previous it looks like game-over , but I think there is real optimism and real
opportunities to be had in this time in hand.
Q: No mention of fusion power
No one has so far got more energy out than they put in , so far noone has passed the break even point.
There are other designs like inertial confinement fusion , using high power lasers to compress
the target , they reckon they will do better with that . I can't see that we will have fusion powered
sources on the time scale of these present problems. Its not that clean , still produces radioactive
waste. You're exposing the containment to high energu neutrons , so not as clean as people
would think. Even if you had fusion what you wouldn't solve because of the liquid
fuels crisis which is what peak oil is, when they run out what is their replacement. Even if you had
fusion you cannot totally electrify your transport system. No chance of 34 million electric cars
quite soon which require all those exotic materials which are close to peaking also
Q:What about using electricity to produce hydrogen and use in engines as you do petrol?
Probably a better way of doing it. You still need the huge electricity infrastructure .
You have a distrubution problem then , getting the H to locations as well. Petrol stations
wouldn't suport H, would that be a reasonable route?
Are you thinking about generating at a local level rather than a massive distribution system.
I would not expect H production in people's garages or even at village level .
There may be something in what you say. But you still need an enormous number of wind
turbines or solar panels if from green sources.
This is where fusion would come in , ideally. Its always been on the back-burner no one sems to have
put real effort into it. If you compare it to someone like Intel , they can design and build a
massive manufacturing complex in 6 months. Around CERNE they have spent 8 years on the new IATA building
a big hole and some footing but nothing like a complete building yet.
I don't think anyone really beleives that fusion will come to our aid anytime soon.
Q: There will almost certainly be energy rationing in the future , what impact do you see
that haviong on society?
Perhaps a wartime like situation and petrol rationing. I think that will be the only way.
Q: I think you're painting far too a rosey picture. Gibbon ? and his 3 killer facts, 2 degreesC,
565 Gigatons of CO2 we can release without exceeding that and 2800 Gigatons of CO2
that will be released if we burn all the proven resources. We cannot even run down that declining
curve without exceeding 2 deg C and then who knows what. An even greater
threat to civilisation .
I think peak-oil will hit us first and then future generations are going to fry as a result of that.
2030 is pretty soon. By 2050 perhaps a toss up between the two.
We will still see pretty drastic effects with 0.8 deg C
We won't know until the experiment has been conducted in reality.
Q: The situation is so depressing perhaps we will have people sliting their throats. With
economists we can run politicians and between them perhaps form an alliance. How much of
it is an economic and political problem as opposed to a scientific problem. Is there anything else
that scientists could do .
Are you referring to peak-oil deniers or climate-change deniers?
All sortts of deniers. In the American Republican party there are people that will deny
virtually all the science. Then the more moderate ones say everything can carry on as it is.
A range of deniers across that spectrum. Is the science all there for people to see
I think the science is all there but it needs to be presented more widely. The governments
of the world are clearly well aware , from advisors etc, they do know what is going on .
There isa n obscure website of the UK government that does say something about
peak-oil but you don't hear Mr Cameron shouting about this. Governments don't want to make a fuss
about it. How politically popular is petrol rationing going to be. Its very tricky territory .
Pretend things are ok and they will get re-elected.
Q: I was surprised about your statement that population would peak and start dropping quite
dramatically. I would have thought it would grow and level out.
Even the WHO latest analysis seems to be indicating a peak, perhaps not as soon as 2025 but
certainly 2050-2060. The mechanism could be rather unpleasant , starvation, disease, wars over
resources . Population control is a very sensitive issue, very difficult to impose on people.
When Russia went through its fall of communism changes the population fell because there
were terrible hardships and shortages. I think that will be the case across the world.
Q: Why is population such a delicate issue. I've grown up over a time where its become culturally
unacceptable to be racist or even a smoker so why is it not culturally unacceptable to have
lots of chidren? I don't understand why we have not had incentives to have less children .
Its part of some religions , to have lots of children . It might be perceived by say a parliamentary
figure as discriminating against a certain religious faction . Difficult, it is the elephant in the room.
A lot of estimates reckon the carrying capacity of the Earth is 2 billion, what it would be if
we did not have all these fossil fuel inputs and so forth. We would not have these problems.
There is a group calle dPopulation Matters , Jonathen Porrit gave a talk in Winchester
a few weeks back . It is about prejudice, eugenics and things like that .
Q: What do you make of Transitions and the like?
Its a positive way to address out resource issues . I don't know if we can pull it off entirely but the more
we can do at the local level then the less demand we have on external inputs. That has to help in
all sorts of respects. The biggest issue is that we won't have so much transportation available
to us in the future and that will change all sorts of things. That drives the local dimension
Q: Just to go back the population thing. Chuina had the one-child policy and that is now under
great pressure because of so-called affluence and a lot of that pressure comes from the West .
I beleive the population of China is growing at about 0.6% ,half that of the USA say.
This is part of the consumer life-style, you want mor estuff , you want more kids too,
the freedom to hav ethe things that the US or Europe has. Understandable but our way is not
sustainable either, not a good role model.
Q: The Bilderberg group is meeting in June , do you imagine this would be on the discussion agenda,
I'd be surprised if it wasn't , though I don't know.
Q: Maybe democracy is a problem. If you want to get elected , you cannot say my government
will cut the use of private cars , stop holiday flights, and restrict foodstuffs to English grown only.
You wouldn't be elected.
You wouldn't . I was interviewed by George Galloway on his radio show a few years ago and he
said precisely that. That is in a capitalist economy, you cannpt make these legislations. In a command
economy you could. So its not in our democratic structure to do this. It would be perceived
as fascism to impose such controls.
Q: So they have to do it surrepticiously by forcing up the prices ? its the only thing that changes our actions.
The driver for change , such as when it costs 3 quid for a litre of petrol , that will change behaviour
considerably. It already has, in that people are now driving fewer miles than they used to .
Q: The cost of fuel is pretty inelastic, we are addicted to it , and will just keep paying more . But there
does come a point where people who had a large car will go for a smaller ans only become more so like that.
The price of fuel has gone up something like 4 times in the last 10years , that is reflected in
costs at the petrol pump. Some people take the view that as oil has gone up from 25 USD to
100 USD a barrel , why has it been flat, why haven't they been pumping mor eout of the ground. Because
they cant , they're at the limit now. If they could have done they would have done, they're
not charities . I think we are at the ceiling. To maintain that rate of production needs all the
new kinds of technologies which all cost a firtune . The Keynsian economists say , if the price
goes up then you can find more of a commmodity . I don't think it fits in with the idea of peask oil.
There are geological and technical limits to it. Its not what is down there, the size of the tank,
its the size of the tap. Its how quickly it can be got out.
Q: A thought experiment, assuming very roughly that at peak oil we've used half of it . If tomorrow some
miracle happened and we got free enregy, fusion perhaps, do you think we'd survive on the liquid fuels
tha twe have left. Or are we at a crisis point even if we suddenly got free energy and free water.
Because our transport infrastructure is entirely built on liquid fuels then a problem. It would take
a long time to come up with a parallel transportation system that was all electric for example.
Within the society and civilisation theat we have then quite difficult. If by transitions type
structures we can build up transportation based on resilience , sustainability all very good.
But I'm not convinced we can do that at the level at which we can stay . Our future may become
sustainable but it will be a lot lower energy job. Using less stuff in the future. A huge shift in
perception as much as anything else. THere is a part of Transitions that is to do with hearts
and souls , basicly how we will get our heads around the whole concept. The new order without
slitting our wrists as the man over there said earlier on.
Q: Something like 30 or 40 percent of the jobs are only there to support consumerism .
I said of those 70,000 Reading commuters every day, those jobs won't exist in the future. What will they
all do. On allotments, planting fruit trees , I don't know, Certainly for practical skills , carpentry,
permaculture , all sorts of things but it will all be a different way of living. I look a tthe university
system and the whole eductation system . We seem to be educating a generation for a status quo
future which iss differnt to what they are likely to find. It seems a woefully neglected duty,
So when will the shit hit the fan?
I think by 2020 things will be wobbling quite a bit. By 2030 , erm. Do you watch Aljezerah .
The other night they were interviewing an american academic . The interviewer asked, what was
Iraq about? , it was about oil. What was Afghanistan about? they don't have any oil. Its where they
are placed next door to Iran , Khazakstan, etc surrounded by oil. She was explaining the point
of Afghanistan was for building military bases . So I think to myself is this the map of world war 3.
Something of that magnitude . Say there was military action on Iran , and the Strait of Hormuz was
shut , about 1/4 of UK gas goes through there from Qatar. One of those tankers contains enough gas to keep
Britain going for 6 hours and they have to arrive continually. Any interuption in supply .
Germany has about 10 weeks supply of gas, We have a couple of weeks. A month or 2 back and the cold
weathger we were using 370 million cu metres of gas pe rday , 100 million more per day than normally for that
time of year. It was touch and go, They reckon that on the day that Margaret Thatcher died , whether the lights
would go out. We are that close, we are so vulnerable. North Sea production has fallen considerably
1.3 of what it was 30 years ago . So importing oil and importing gas for the same reason , for
gasfield decline, importing food. It was worse with the food production issue during world war 1.
Which is why the convoys were so vulnerable. During world war 2 the nation produced a much
bigger proportion of its food. Heavy use of fertilisers, heavy tillage,heavy machinery which in the USA
caused the Grapes of Wrath situation made us much more productive in food. But how viable and sustainable is that likely to
be without hte cheap oil. Its not , a huge game changer.
Has anyone produced a working large C capture and storage facility, say 1Gwatt?
Not to the best of my knowledge.
Are there serious technical difficulties or just a matter of scaling things up?
I think the main problem is the expense. If you were makingt electricity in coal-fired power plants
, for 2 you'd havre to build a third to provide the energy to power the C capture of the first 2 and itself.
I think that is the reason that no government has seriously adopted it so far. As a technical exercise
it could be done , but the best so far is taking th e CO2 and pumping it down into old oil wells. A lot of
the CO2 in the oil industry is reinjected asa secondary production strategy. There are differnt aspects to
oil production. There is tyhe easy way , when you first strike a well, you dril through and its under pressure,
after you've produced about say half of it , the pressure drops . So you have to keep the pressure up , which can be done by pumping gas , can be CO2, or methane that came up with the oil before. The tertiary recovery
is where you increase the fluidity of the oil , generally by heating and most common is by steam injection.
All these extra procedures cost a lot more energy, that is why the energy return , across the world ,
is falling. A Prof Charles Hawe? in the USA worked out that a society like ours complex
and industrialised needs an energy return on energy invested of about 10. For agrarian societies maybe about 2.
In between would be like the Victorian times and a lot of coal burnt. The return is now coming down
to about 10 for oil, for what we require to function as we are.
The alga technology sounds too good to be true in a sense. It requires fairly low-tech
processes and I see jus tlast week its been scaled up to about an acre sort of size on a commercial
basis, and it will work, there is nothing wrong with the process. Can the output be used
as a feed stock for the plastics industry or purely for energy production?
You can use algae for various purposes , you can certainly make plastics out of bio-mass and alga
is biomass. You can convert the algae of other biomass to simgas? and then you have an established
means for making plastics and hydro-carbons generally speaking. It does not have to be
for making fuel. Again, its not going to give you the energy return that you get out of oil
and natural gas. There was something on the cover of Chemistry World the chemical
industry publication and the title was Getting away from Oil. About making plastics etc
from lignites , interesting but takes an enormous amount of energy to do it.
I think this will be our failing, we will just fall off this energy cliff. There are a lot of
very clever technological innovations around , just scaling them up and thats where many
of these schemes fal down.
If you had 3 practical tips trhat people could do , from the Transistions point of view, easily, What would they be ?
Insulate your house, good draught insulation
Drive less, try to earn a living without having to drive into London and back
Get into perma-culture, using low input methods and good designs , grow as much as you can
My 3 critical ones
One thing that has come up is food waste, in that we waste a lot in this country, the energy and resiources
that go into making them .
Is it the Bogoff thats the problem or is it the supermarkets telling farmers , we need tomatoes exactly
that size, no blemishes ? Personally I find nothing wrong with bogoffs, I wouldn't buy perishables
that were bogoffs ,but 3 or 4 cans of beans for the price of 2 is fine by me? At the farm stage and the
overarching control of the supermarkets that the waste was at principally.? In the case of tomatoes ,the
farmers cannot even can the "waste" tomatoes as they ar ethe wrong sort for canning. Italy produces the
sort of tomatoies that can be canned. Would you happen to know the ratio of wastes from the 2 sources of
bogoffs and farm waste?
A figure I heard recently was about 1/3 of food crops get left in the fields. Perhaps you can attribute
the other 2/3 to the bogoff.
A figure I've not been able to lay my hands on is the percentage of commuters who commute to work
and they sit at an office at a computer, fax or telephone for 80 or 90 percent of their prioductive time,
that could equally be done at home?
I was asket that some years ago. I did manage to find a figure but its complicated. Its something like
60% of people who work in offices doing processes that could be done at home. As distinct from the
police or fire service say , who have to be out doing something. Its a sizeable fraction.
I know the effects of animal farming on CO2 , what about the effects of animal farming and peak oil ? Such
as transporting soya from Brazil to here , to feed animals .
If you look at just the trasportation issues, that is where you can make your big savings. We'll hav e
to as we won't have the affordable oil to move it as we do at the moment.
And moving animals from country to country to be killed . aslso? The export of horses over to France for killing
That too .
Move over to eating insects?
You say that the UK is not going to meet its commitments as far as renewables, so how do you feel
about the government giving subsidies for PV cells etc , is it a waste of time?
Its not a waste of time, everything helps . When they say produce 20% of out electricity
by 2012, not that far off , I don't know that they'll manage that. We seem commited to winfd power
for whatever reasoin , even 1 turbine erected a day for the next 7 years we will still be
behind the targets.
If we'd have done the Severn Barrage then we would have been well on the way then, but not close
to that target?
A figure of 14Gwatts comes to mind , pretty high. It would certainly cut a hole in it. Decided
against it, ironically, for environmental reasons.
Monday 10 Jun , 2013 Dr Ivo Tews
Title: Preparing for the International Year of Crystallography
Covering the science from the beginnings of crystallography to the more recent successes that
lead to the nobel prizes in Medicine and Chemistry, and will focus on the underlying biology of these successes.
3/4 hour talk, 1/2 hour Q&A, 34 people
The School for Theoretical Physics, Zommerfelt? , founded the school, a very important
movement in Germany. A very influential school for physics. He had a student called Peter Voliber?
who had a problem with rays and waves and X-rays, wave theory was new.
Electromagnetic spectrum Infra red, radio , micro-wave , visible and much shorter wavelength Xrays.
Trying to understand how waves interfere and with a contribution from Maxwell Loewe? . If the waves
are getting smaller and smaller , what we could possibly get is the same thing with interference
within crystal lattices. What they knew at the time that if you put light waves through slits then
you get patterns of dots etc.
Some "toys" for the audience to play with, from a market place in Rome , pen-torch size green LED laser pointer with a little grid inside ,
making an array of dots , and you can turn the grid to change the pattern projected on the wall.
Maxwell Loewe thought that a crystal could difract waves with patterns like these laser projectors.
From the patterns we could then say what was within the crystals, but no one believed him.
Loewe went on to prove this. People believed at the time that there would be too much
motion in the crystal and there would not be ordered paterns. So Loewe with his
2 students Clipping? and Friedrik ? went on to show that this indeed worked. X-ray
tubes were available since Roengen . A crystal of copper sulphate , got the pattern,
and the Academy of Science audience went , Wow!. A dramatic picture, fifth of a series ,
a milestone of crystallography (CY), in the early 20th century. So xrays can be difracted by crystals.
How do we interpret the images. For the copper sulphate crystal ,
there are copper atoms , oxygen , very complex at the time to know the structure. Another
step is required to do this. This is why we now consider the homeland of CY is Britain, because of
father and son Bragg, developed the theory. From the pattern to then calculate back
to say what is inside, absolutely crucial.
A bit like the importance of E=mc^2 is to relativity then the fundamental one for CY
A crystal lattice with light from one angle then you get difraction at a diferent angle , angle theta ,
n(lambda) = 2d sin (theta)
The waves coming out are shifted with respect to each other and the formula from sin(theta)
, the spacing and the triangle , a simple geometric construction. Sum them up and you get interference
of those waves and you observe a spot on your photographic plate.
Bragg awarded the Nobel in 1950. That is the physics, now we go into the aplied sciences.
We ask the question , what has CY ever done for us!, sanitation, medicine,
public order, irrigation, fresh water, roads, education.
What does CY help us to understand in biology. The central dogma in biology is that the central
information that we have is DNA , the genes, they have to be put across to proteins and enzymes.
There is a step in the middle that is RNA, the dogma tells you a direction, DNA to RNA to
messenger RNA to proteins. There is transcription and translation. Transcribe the DNA to RNA
and translate into proteins. We can replicate the RNA, replicate the DNA , we have viruses
that are based on RNA and DNA genomes. We have machinery in our cells to do this.
There is reverse transcriptase which is the enzyme that HIV has , so RNA can reverse
transcribe to DNA. CY has helped tremendously in understanding these processes.
Some Nobel prizes from the endpoints of these processes, the DNA and proteins.
With proteins there was another Nobel for Linus Paulin, a CYer in the 1930s.
Starting with salts and minerals and then to biological materials.
Francis Crick and James Watson and Maurice Wilkins doing the structure of DNA.
These go back to photographic images produced by Dorothy Hodgkins, she never
got a Nobel prize for this. Paulin also at the time tried to solve the structure of DNA but
his proposed model was completely wrong. He did not have access to Hodgkins Xray images.
Max Perutz? and John Kendal determined the structure of blood myoglobin and later
haemoglobin. This was a time when biology intensley used CY to its best and the first images
for bio-molecules using crystals to look at things.
We do this because we cannot see the atoms of individual molecules . tHe Nobel
prize for Dorothy Hodgkins was for anti-biotics, she discovered the structure of penicilin.
In the 1950s a number of things were happening. People who would grow crystals
for instance from viruses . Bacterio-phages , plant viruses . Not only CY to view them but then also
A lot of structural knowledge of biological molecules , like viruses. The next Nobel prize
winner in this area was Aaron Fluke? . That virus has a genome that is inside like a big
staircase and otside are the proteins. It does have a very defined length, a virus that
affects plants, tobacco mosaic virus. A numbe rof other viruses we know today the structure
of through evolving methods of CY. Since the 80s we now have the methods to do this.
Beautiful structures of Polio , plant and avian viruses and a cancer producing virus.
A step up now in compexity, recent break throughs . A virus is always a repeat of the
same protein and we are now addressing transcription and translation. A very
complex process, a number of differenrt proteins put together, very difficult to do.
But you can put all this in a single crystal. A picture of the ribosome , the molecule in your
cells that make proteins. Molecules coming in at one end , Ductor ? molecules and proteins
out at the other side. Remarkable achievements through CY , tells us the inner workings
of these bio-molecules. 4 more Nobel prizes , recently awarded.
The first in 1988 and the last in 2012. We now look into membranes, the boundaries of the
cells. And look at how wwe cross the boundary. The photo-synthetic reaction centre ,
the information in this molecule shows how to take light and convert to chemical energy.
It tells you by yhe location of the chlorophyl molecules , a cascade of these molecules and you
have to know the exact spatial arrangement of them to understand how the chemistry works.
That is why CY is so important.
John Walker of Cambridge, the molcule that produces energy in your cells , the molecule
in your mitochondria . We have fantastic machinery that sits in a membrane , with a
spinning disc, so always some motion in your cells. Frightening, because it spins quite fast.
It spins because you have protein gradients. A difference in charges across the membrane
that drives it and produces ATP, the universal currency of energy.
So we know how this works. Then in 2003 Peter Ager ? and Bob McKinnon? explained
how water crosses cells in a directed way. How we can regulate this in membrane proteins
and how ions cross . A student of his , Deklan Doyle? is now at Southampton as a lecturer.
The final piece in the jigsaw is the GPCR which is the most important thing,
Lekowitz and Kobilka ? looked at these and are pharmatological targets for nearly half of all our medicines.
They tell you how a signal is received on the outside , changes the cell inside.
So plenty of progress in structural biology . We have other techniques as well, including spectroscopy
and other ways of looking. Its so visual, to finally see a molecule and what it does via this
technoque of CY.
For a while now we've had the human genome, a milestone as well. Structural biologists
dream of not only knowing all the genes but all the proteins in cells. That's called structural
genomics. Now looking at every dsingle protein. The cell, the genome of that cell, then select
a number of molecules , some already in a database , put them together and make the circle,
adding to the database. Finally you will be able to explain how the cell works and how all
the molecules work together because you know their structure.
A flavour of what I'm doing to get data like this. First we purify bio-molecules like
proteins or DNA or whatever and we make crystals. Crystals in little drops. The next thing we need
, about 2 weeks for them to develop, we have to take them into tiny little loops.
About 0.1 micron in length, requiring a very steady hand to do this, a manual operation.
I get the crystals out of solution and mount them on an Xray generator.
In front of the tube where the Xrays emerge. An image plate or a detector where we take the data.
The crystal is frozen in a cryo-stream . You can turn the mount around and do the Xray
difraction. Usually this takes overnight , something like 20 hours or so, to se from all
But we have more intense radiation sources these days, and much smaller samples , something
like 0.1 micron droplets, tiny crystals of the size 2 microns that cannot be mounted manually.
This is where we push the boundaries. We also need a big machine that produces Xrays .
The Diamond Synchrotron in Oxfordshire, light source of intense Xrays. We use this quite
often. The power consumed is that of a 50,000 inhabitant city. Particles accelerated
in an electron gun , run them around a ring requiring bending them. Every time you bend them you
get Xrays from them, called Remstrada? . There are 2 ways of producing this
A magnet and a beam of particles, perhaps electrons , travel as bunches , bend them
from their path and off comes Xrays. synchrotron radiation. You can build dipoles of
different magnets and even more intense radiation. Whether at Soton uni or at the
synchrotron we have a crystal , shine in Xrays, they get difracted , collect the image.
Then the structure is determined by some computing , from this image you can compute
that. The maths just works. Very fast detector , chilled down , a robotic unit with 50 crystals
sitting in there and mounted in the cold air stram. Once happy with the setup , you leave
the room , seal it the lead doors and control everything from the outside.
There is the European synchrotron in France , same design, in Grenoble.
The new kid on the block is the free-electron laser , the one in Hamberg, costing
1 billion , Soton uni is part of the consortium that runs one of the stations for the UK.
If you put ever more intensity of Xrays you will simply destroy matter. You cannot
deposit that much light into a bio-molecule . 2 femtoseconds , 2x 10^-15 seconds , 5, 10,20
and after 50 femtoseconds the whole sample explodes. But if you can collect data quicker than
it explodes then you can still use it. But you have to be very quick, seconds,millis, micros, nanosec,
picosec, femtoseconds. We have technolgy that does that. The amount of data out of such a
detector is gigantic, imagine the frame-rate. Some real problems in computing for this, but we
can . The proposal in 2000 , being built now and we will operate our station in 2017 .
In 2012 a paper came out showing the feasibility of this. For this we will be using a stram of nano-crystals
, ever smaller samples and we blast them to bits. So a lot going on in this field.
Why the picture of a kindergarten. Kindergarten building blocks and crystals are similar.
Go back in history to the 19C and Froegal ? born 1872 a polymath
"I could perceive unity and diversity and interconnections in all living things and matter
and the principles of physics and biology"
This is the age of enlightenment , greatly influenced by Prof Lotze a Swiss educator of the time,
Rosseau the philosopher, and Weiss who was a CYer. Rousseau in Emile, how you grow up and look
at nature and what you learn from nature . A hypothetical book, but the idea comes from it that
everything in nature is good and everything that man touches goes bad. Everything degrades in the hand
of man. He tried to learn from nature. He was a CYer , studying with Weiss, at Berlin University.
He was given a collection of crystals , at the time of the Prussian kings, they had 75%
of all known minerals. Given the task of go and sort it. What system to use to make sense and sort the
different shapes and morphologies . Weiss developed what we still use today with our biological
crystals , the goniometer ? . Place the crystal on it, and view with a microscope , the shape.
The shape on the outside is very much telling you what the shape is on the inside.
So classify what they look like, then what is in there , how they could grow and exist in nature .
They can make planes inside the crystal and describe what it is built from. Weiss was
translating a book from France , CY theory . Froegal then developed these and a book the
Crystallography Kindergarten. He made gifts like little balls for entertaining children . These
gifts 3,4,5,6 are the building blocks to make crystals with . The idea that a child could discover
things was a new concept at that time. The child learnt by experiment. A grat gift that
he has given us , the 20 gifts. They don't really exist these days .
A number of influential people have been through this kindergarten. Kandinski , you
can deconstruct one of his painting sby Froegal gifts. Influential in modern art and architecture.
Geodesic domes , Buckminster Fuller went to a Froegal kindergarten.
Frank Loyd Wright went to a Froegal Kindergarten, organic abstract geometries in his
plans. The Prairie House style in the States, each brick in the wall he drew in the plans.
The mechanics of the freezing. The beam comes in , the beam goes out , manipulating the
crystal around, the environment its in , is it an aqueous environment.?
You've seen pictures of crystals in solution. We take a solution of a protein or DNA , concentrate
it up until it can no longer stay in solution,. So 2 possibilities , it crashes out and destroys
itself which happens in about 99% of the cases or it forms a crystal inside the solution. So
a crystal is the lower energy state of that molecule . If the concentration is too high , it
has to come out of solution but if it preferentially goes to a crystaline state rathe rthan just
drying out and destruction. This crystal from a protein is not like you are are familiar
with in NaCl , points and atoms and all neat and tidy. It has huge channels and pores in
there . Half of the crysatal is water. If you touch it you will destroy it.
Movie of droplet with a crtstal in solution , ad a little bit of liquid to make it happier,
then a little loop to fish for the crystal that does not really want to go in.
Requires manipulation and this particular one has a crack in the middle of it. With luck and skill
the crystal hops into the loop. Then you have less than a second to freeze it. With the crystal
in the loop it will otherwise dry out very quickly. Vapour pressure from the outside will destroy it.
So drop it into liquid H , very quickly, frozen and solid. If you freeze water you will
get water crystals and ice crystals will interfere with our experiment. We put a little
bit of sugar solution or similar as a cryo-protectant.
Freezing human bodies, you can conserve yourself if you have shedloads of money.
Put yourself in a liquid N tank and hopefully someone in the future will find a way to
With the sugar it doesn't form ice crystals because the sugar is in the way and it
helps the target crystal to freeze in situ. Its amorphous like a glass , glass is a state
of molecules that is not really a crystal , you can see through them. Actually forming a glas.
So crystals in proteins have to be made, they don't just happen in nature?
Many of the things we do in CY goes back to discoveries of crystals. Whales have myoglobin
that is crytalline. Open up a whale, take the crystals from the blood , known fo ra long
time. Catalase? is another and in your cells you have compartments the so-called
peroxysomes? to detox your body. In the peroxysomes there are crystals of proteins ,
know for 100 years. We could never analyse them but since the arrival of these technologies
we can. Some of these are natural. Even membrane proteins , some bacteria
at the outside of the membrane , purple patches and they can do photosynthesis and these are
crystals, not 3D but 2D in a membrane a regular organised molecule. Actually we use them
in electron-microscopy , using electrons rather than Xrays but the same principle.
These membrane crystals, can you crush them like copper sulphate crystals and they will crack.?
Yes and you can take a piece, and if still ok , you can still use it but you'd rather take the
whole crystal. Cracking and mechanical shearing changes the structure , a protein cryatal
may not survive. The buffer solution to assist the freezing has to be fine tuned to the
conditions, and if not right the crystal explodes, shatters into pieces. Thry're very sensitive.
When you produce Xrays from bending particles , what angle do you have to bend to?
Not much , forming an end-station as we call it, for the detector as you have to be away from
the curve . The lines ar e30 to 50m long and then the hatch/hutch? the lead casement for the
experiment. In Diamond has 30 segments .
If you change the angle do you get a more intense Xray or less intense?
The way we do this is by using mor ethan 1 magnet , a number of dipoles about 50. The intensity now is
so much greater that instead of the 20 hours for a data collection now about 20 seconds.
The partices are going on a wiggley path and one such manipulator is called a wiggler
producing the Xrays. Every time they wiggle back and forth intensifies them. The Diamond
is now so intense that we can collect data in less than 1 second. We usually attenuate the beam
and use 20 seconds. The upgrade for the new French one in 2018 will produce 3 orders of magnitude
more Xrays. We had a meeting in February 2013 where the designer stood up and asked
what we can do with these 1000 fold increased intensity Xrays and we do not really know.
We will destroy matter very quickly , so will we be attenuating that beram to the level we
already have, we don't know yet.
The matter wil lbe destroyed, is that because the atoms are ionised?
In electron microscopy you can do much the same thing. You can do difraction . What do you
think is more damaging to do the experiment with electrons or Xray beam. Photons are the
most intense particles , they hit an electron and the electron goes off . There are 2 events really
primary and secondary damage. Primary in terms of chemistry creates a radical
which can now go on and make reactions inside the crystal. Radical reactions are fast
and dangerousRadical chemistry happoens at 10^-10 sec, nothing survives a radical reaction,
they attack near enough everything around. These radicals travel and destroy the molecule . We freeze them
down to cryogenic temps , 100 deg K, because we don't want these molecules travelling that fast.
We want the radicals to stay where they are. It helps us a bit to prevent the secondary damage.
But the primary damage to the atoms we can never prevent, the photons will alwys damage matter.
Why do you hav ethe blast of very cold air going over the crystals rather than thermally connected
via the mount.
To construct a way of cooling like this is very difficult. The cold stream instrument costs
60,000 . It vaporizes liquid N, blows it across and blows another stream outside of it
that is just slightly faster of completely dry air . What would happen if you just cooled something
you would just get condensation and ice which is not what we want in our experiments.
You always do these experiments in a linear way?
Jus tone end station is 10 million in cost. The designers of these are physicists and machinery
To get your crystals , what we want is the structure we expect to se in a cell. We assume it will
have lots of water around it . Your crystals will have some water there and ions . But to get
a crystal to form we have to drive the water away. Will that structure going to be really
the same as you'd expect in a cell.?
A question that is often asked. How realistic is what we actually observe . We are making something
that is in your body , taking out of your body and making a crystal. I've givenm 3 examples
where we observe crystals in real life. But the molecules we are looking at don't look
like that. We have to take the steps through the crystal to analyse the atoms
tha tmake the structure. If you look into a cell you have so much liquid . True for a
jellyfish with something like 99% water but not true for humans. We use the terminology
molecular crowding , thick matter inside the cells. Very little water that is just floating about.
Inside the cells there are high consentrations of proteins and diferent molecules ,
the concentration is near enough the same as we have in one of our crystals. We are doing the
process of only taking the same molecules.
Can you give us an insight into how the pharmaceutical companies are using this knowledge
to make medicines. ?
We go back to GPCR , of the Nobel prize last year. 30 or 40 % of all drugs target GPCR.
Always this molecule and its elluded structural analysis for so long. For decades people are
interested in this. You have molecules flying in, binding and then change the molecule .
The drug coes from the outside , in your blood say, and it has to go through the membrane.
So it signals a change insid ethe cell, to understand that you need to see how it binds ,
and different states to the response. Then insid ethe cell there is a change. With the structural
knowledge of the atoms , we look at the drugs , change a methyl
group here or a sulphur or something and look to se ewhether it can still bind or
provoke a different change. A Nobel prize fo rthis as nature has engineered this to
be s flexible tha tthis is quite opposit eto what we want to do in CY, something that is
always the same. Its elluded us a very long time, and now it is possible.
???, if you can get results in solution ??? ?
CY is not a technique that stands on its own. If I take a result from a protein structure it
doen't help me a great deal. As a scientist I would not be able to publish the data on their own.
I can deposit this into a public depository on a database but this does not tell me the story.
What I have to do is describe the reactions within the cell. You put things in solution and other techniques
like spectroscopy greatly helping in understanding these results. And using ??? how molecules
interact . Without that knowledge the structural work is kind of shakey.
If you turn the molecule does the pattern change?
Why do we get these spots on the film is the underlying question. The molecules are repeating
inside the crystal. We look a tthe crystal in 3D ,get the difraction pattern and the
diffraction pattern is in 3D , from this rotation. The difference that is appreciable and that we
have to work with . From say an original Bragg picture. If you look carefully
the degree of blackness of the dots varies and that is where the information sits.
If you turn it , one gets more black and another gets less black.
Can impurities in the crstal mislead you?
Impurities are a great problem. We try to work as pure as possible. For a chemist
you know you can recrystalise substances and that is a purification process.
We take out only the things that fit into our crystal. Inside a crystal we think we have a
pure substance, even if the original sample is not pure. Sometimes the molecules themselves
have different conformations , they change and that is a problem. Some molecules can
not be interpreted from the data . Conformational Heterogeneity , if you have 2
types of the same molecule in a crystal then it doesn't work.
Some of those early iconic images . I thought Rosalind Franklyn was the
uncredited person behing the DNA Xray images but you mentioned a Dorothy
Hodgkins in that role. What is the connection there?
No it was Rosalind Franklyn, i meant, in that role
With the new very high power free-electron laser generators and images in a couple
of femptoseconds , can you do it while in solution , without crystalising? I
appreciate you would have no choice about the orientation, but could catch it before
it had moved. ?
The dream of the original proposal is that you could see a single molecule , that was
exactly what they proposed , to observe single molecules in solution.
Technically not possible, why they use crstals. These crystals are pretty small .
Calculations now show a limit of 5x5x5 molecules is sufficient for a crystal
So how do you define a crystal?
A crystal is a lattice which has a repeating unit .
Is there a minimum number that you can still say it is a crystal?
I suppose 4 repeats in X,Y and Z then you could call it a crystal, 2 is too few.
A reserve email account is diverse9(commercial at)fastmail.fm.
Please make emails plain text only , no more than 5KByte or 500 words.
Anyone sending larger texts or attachments such as digital signatures, pictures etc will have
them automatically deleted on the server. I will be totally unaware of this, all your email will be deleted - sorry, again
blame the spammers.
keyword for searchengines , scicafshadow, scicafsoton, Southampton Science Café, Café Scientifique, scicaf, scicaf1, scicaf2
, free talks, open talks, free lectures, open lectures ,