Analog Computer: It's Alive, WHA HA HA!!! (publ. 2025-05-14)

Assembly is complete on my DIY analog computer, and it appears to be working, although there are some components I haven't tested yet. I'm calling this the "Poor Man's Analog Computer". Here is a photo:

assembled analog computer box and panel, with oscilloscope, running a simulation

The breadboard computer is inside the red box. Mounted on the front of the box are the five coefficient potentiometers and the switch for IC or RUN mode. Two signal cables, containing 24 signal lines each plus an unused spare line, come out of the right side of the box and go to the back of the patch panel. There are three binding power posts on the left side, for the +/- 15V supply, as well as a bolt for a separate chassis ground. The patch panel has 48 jacks providing the following ports:

4x integrators: SJ, IC, and OUT

1x +10V reference

1x -10V reference

2x GND reference

2x multipliers with differential input: A+, A-, B+, B-, X

2x amplifier/inverter: IN-, 1x, RES, OUT

1x adder: A, B, C, OUT

5x coefficients: IN, OUT

As far as the actual electronics, the modules above were all based more or less on the Grappendorf analog computer schematics, though without using the provided PCB design.

Grappendorf analog computer

The Grappendorf design provides a separate operational mode switch for each integrator, but all my integrators are synchronized via a single switch that powers several electromechanical relays.

The open amplifiers become simple inverters if you tie together IN- and RES, which is generally how I intend to use them. But since that is not hardwired in, it should be possible to use the open amplifiers to implement some other functions such as division.

I'm questioning now my choice to include all the differential input ports from the multiplier chips. It might have been better to ground out the negative ports (A-, B-) and use those jacks to get more coefficients. Of course, if I had more coefficient ports, I would have had to find some way to mount more potentiometers.

Wiring up everything inside the box was a lot of work. Aside from wiring up the signal lines from the panel, I also had to wire up the potentiometers, and the connections from the switch to the two relays for the integrators. Also there were the three binding posts for the +/- 15V power supply, as well as wiring for a separate chassis ground. Also I had to buy some rubber grommets for the holes where the signal cables come into the box.

The chassis ground is provided via a bolt on the red box, which has electrical continuity with the box, and it also has a line going to a bolt on the metal patch panel frame. I'm not sure if there is really much danger of a person being directly harmed by a short to chassis at 15V DC, based on some Internet research I did — at least in the plausible scenarios of regular use. But I imagine that there is a variety of dangerous or inconvenient things that could happen if the chassis became a 15V power source for anything else it happened to be touching.

In the photo I linked to above, the computer is set up for the damped oscillator simulation as provided in the Analog Thing tutorial. You can see the results of a run on the oscilloscope. Here is a screenshot of that:

oscilloscope screenshot of damped oscillator simulation run

I can see thin spikes of noise, around 1V p-p I'd guess, all around the output signal, so perhaps I should do some more troubleshooting for noise suppression.

I'm not too proud of my analog computer, as far as the fine details go: if you look close at the drill holes and the terminal blocks and the soldering, and such like, it is not a thing of beauty. And previously I mentioned the fiasco with the bent banana jacks — usable, but odd. The breadboard and inside connections are fragile and I'm sure it wouldn't survive an apocalypse. But I'm happy to have finally got this project built. The product specs are humble, for sure, but the humble goals allowed me to bring the project to completion in a reasonable amount of time — a few months, with a few minutes of work here and there each week.

The costs for the project were more than I intended, but still affordable. In the end, I spent somewhere around 200 USD, not counting some cheap components I already had laying around, like terminal blocks and old cables. Some of that money was for comparator parts which didn't end up being used in the computer. However, I should say I did have the advantage of a local source of affordable used parts — Eric Nichol's shop. If I had needed to buy everything brand new from the Internet, the project might have cost another $100 or $150, I suspect.

Copyright

CC BY-SA 4.0 Deed