Homebrew Analogue Computer
 

I came across details of the Elektor "Chaos Machine" recently, which is a construction project for an analogue computer to show the behaviour of the Lorenz Equations. It seemed to me that this could also be done using Unilab System Alpha. Unilab System Alpha is an old educational electronics system comprising various interconnected modules. The modules are available very cheaply in quantity now, probably because they're being thrown out by schools and universities. The modules include most of the analogue computer building blocks so can form an inexpensive introduction to analogue computing. Personally I have put together a Unilab based analogue computer to display the Lorenz Butterfly. If there's any interest I can post some more details.

Re: Homebrew Analogue Computer
 

Please post more details - I learnt analog computing at university and it would be nice to see something again.

Re: Homebrew Analogue Computer
 

For the Lorenz project the following functions/operators are required:
Add, Subtract, Multiply by constant, Set Coefficient, Integrate and Multiply. 'Add' can be done using the Unilab 'Summing Amplifier' (module 204), 'Subtract' can be done with the Unilab 'Difference Amplifier' (module 202), 'Multiplication by constant' can be done with the Unilab 'Inverting Amplifier' (module 200), 'Set Coefficient' can be done with the Unilab '100k Potentiometer' (module 9), 'Integrate' can be done by tweaking the Unilab 'Summing Amplifier' (module 204) (tweak is to change the feedback resistor to a capacitor). 'Multiply' is missing so my approach is to use Unilab module 50 (Project Board) and plug in an AD633 multiplier chip. Here is a short video of my completed computer in action:

https://www.youtube.com/watch?v=BCBl_PxKgpQ

Re: Homebrew Analogue Computer
 

I have a commercial analogue computer (EAI1000) and would be interested in trying this sometime. It has (not surprisingly) amplifiers, integrators, analogue multipliers, pots (both free and with one end grounded) so it should easily be possible.

Can you point me to more information on said design, the interconnections of the computer elements, etc.

Re: Homebrew Analogue Computer
 

Lucky you! I'd love to have an EAI1000.

Here are some instructions for implementing Edward Lorenz's equations on an Analog Paradigm computer. As normal there's some scaling involved. Look for the second issue of 'Analog Computer Applications' :

http://analogparadigm.com/documentation.html

I should think a very similar if not identical approach would suit an EAI1000.

Alternatively, Professor Clint Sprott has included in a paper [1] a linear transformation of the Lorenz equations which can be implemented on a 10V analogue computer without scaling. I used these transformed equations which can be found at:

http://analog-ontology.blogspot.com/2014/07/

Look for the section titled 'Self Propelled Flowers'.


[1] Sprott, J. C., 'Simplifications of the Lorenz Attractor', Nonlinear Dynamics, Psychology, and Life Sciences, 2009, Vol. 13, No. 3, pp. 217 - 278.

Re: Homebrew Analogue Computer
 

Hi Gents, one of the "bibles" for analogue computers was a huge app note from Burr Brown, late 60's I think, that showed many variations of op amp circuits to achieve the many and varied functions required.
Hopefully it is available on the web somewhere

Ed

Re: Homebrew Analogue Computer
 

Have a look for information from Philbrick as well.

David

Re: Homebrew Analogue Computer
 

Pardon me for barging in but I couldn't help posting this photo of an 'analogue computer' I built a while ago. It's really a supported prototyping system with meters, knobs, buttons, etc. It is set up for a lunar lander simulation. The three meters display height, speed and remaining fuel. I succeed in landing it about half the time so I think I've got the difficulty level about right. I can post the circuit for this if anybody's interested.

Re: Homebrew Analogue Computer
 

I'd be interested in seeing that circuit.

There are some photos of my EAI1000 in my flickr account (tony_duell). I have 3 analogue 'bays' in mine, each bay has 6 summing amplifiers, 4 integrators, 2 multipliers, 2 'free' pots (all 3 connections brought out to sockets on the panel) and 10 grounded pots (one end of the track connected to analogue ground, the other end and the wiper on sockets). There are also some digital functions, a couple of comparators, a couple of Dtypes, etc.

Fortunately I got a lot of patch leads with it. You can still get suitable leads (e.g. at RS) but they are several pounds each...

I've made one modification to it. I added a DB25 socket on the back (I made a new blanking plate, there's no permanent changes) wired to the edge connector on the control module. This brings out the 'external trunk' lines, etc. I've made a box of 4mm sockets that connects there so I can easily plug in a 'scope, XY recorder, meters, etc

Re: Homebrew Analogue Computer
 

That really is terrific, you seem to have a knack of doing really outstanding work.

Many years ago I was having an argument, banter really, about the notion of how best to differentiate an electronic signal. A friend suggested that by far the better way to do it was by digitally sampling the waveform and doing it with a software protocol. His argument was that fundamentally all analog differentiators were flawed, because the better they were at their task, the more noise susceptible they became, which in theory is true.
Unless you are smart enough to make the differentiator out of an integrator. See the attached image, wireless world, 1978.

Re: Homebrew Analogue Computer
 

Except that's not a differentiator, is it? It may well detect the edges of a square wave, but I don't think it would turn a triangular wave into a square wave, or anything like that.

Re: Homebrew Analogue Computer
 

The mathematical function is not exactly the same, but it serves the same purpose, for many applications, producing a signal that is dependent on the input signal's rate change, but without the noise issues. In most cases that is what you want a differentiator to do, to give an output which increases with the rate of change of the input signal, or detect a transient of some kind on the incoming waveform and ignore slowly changing signals.

Re: Homebrew Analogue Computer
 

The problem with 'true' differentiators is that their gain rises forever with increasing frequency. You have to plateau the gain at some point. For signals whose spectra are well below this point the effect is the same as a differentiator. Stick with integrators if you can - they are much better behaved :)

Attached is the circuit for my moon lander.

Re: Homebrew Analogue Computer
 

I should have guessed you could land on the moon with 7 OP amps, very cool !

Re: Homebrew Analogue Computer
 

Indeed. Differentiators are almost never used in analogue computing as they tend to be badly behaved. It's normal to re-write the differential equation in integral form and use that (look at the 'damped simple harmonic motion' solution on an analogue computer, it uses 2 integrators for all the equation taught in most physics courses is a second order differential equation).

I must try that lunar lander on my EAI1000. I think it has all the necessary elements although I would have to link up a couple of meters and some LEDs.

Talking of 'lunar lander', am I the only one here to have played the (digital, or is it Digital) version on a GT40 (A 1970s graphics terminal with a PDP11/05 CPU inside)? I never managed to get to the McDonalds on the moon, although a friend did (running on my GT40).

Re: Homebrew Analogue Computer
 

set gt40 on