Analog: Patch Panel, Cabling, Case (publ. 2025-05-08)

I'm now in the final stages of assembling the analog computer. I came up with a wire list and have wired up cables to the back of the patch panel. I also mounted the potentiometers, the IC/RUN switch, the ground lug, and the power binding posts onto the main computer case, an aluminum cookie box. The main computer case and the patch panel are separate and will be connected by two signal cables. I need to get a few rubber grommets to slide the signal cables through, but I'm planning to pick those up after work. Fortunately, my local AiH sells them in small quantities. So, It looks like I could have it up and running within a day.

The patch panel looks a little funny for a few reasons, one being that I mixed in two different styles of banana jacks — the least expensive option for me due to what I had on hand. Also, I did not realize it at first, but the plastic on the cheapo Chinese banana jacks I bought from Amazon have almost no heat resistance. So, when I was soldering the wires onto the back, I didn't realize that the plastic around the plug was melting, causing the jacks to bend sideways. I messed up about 20 or 30 of them this way before I realized what was going on, and some of them are bent around 20 degrees of center. Fortunately, though, none of them ended up being shorted to the chassis, and all of them will still take plugs if I bring the plug in at an angle. This is rather embarrassing, but money is tight, so I don't know if I'll ever replace those or not.

The computer currently makes available the following components:

4x integrators

1x +10V reference

1x -10V reference

2x multipliers

2x amplifier/inverters

1x adder (three terms)

5x coefficient potentiometers

I limited myself to a 48 jack patch panel, which is really a very small number of jacks for an analog computer. But I felt like I needed to set some limits or I'd never finish the project. The metal sheet I found had space for 48 jacks, so I settled on 48. Here are various shortcomings of my design:

(1) Five potentiometers is a very small number of potentiometers for an analog computer. Ideally, you want one potentiometer for each integrator's initial condition line, and then even small models generally have 3 or 4 coefficients at various other places. But one can simplify things by tying initial conditions to 0V or +/- 10V, limiting the starting conditions of your model.

(2) No comparator is available. Quite a few interesting models use one or two comparators, from what I've seen on the Internet.

(3) I'm not certain if two inverters is going to be enough in practice.

(4) My computer board actually has connections available for 10x scaled input on most of the components. But to save on jacks, only 1x connections are available on the panel.

(5) There is no "repeat mode" built in, so one has to flip the RUN switch manually after adjusting coefficients. Also there is no "hold" mode.

One could argue that I should have reduced the number of integrators to make room for other components or connections. But I really wanted to be able to do some modeling involving force, acceleration, velocity, and position, and I didn't want to run short on integrators. I suppose just three would have been sufficient.

I perhaps could have got by with one less multiplier, but the multiplier chips were $20 each, so I really wanted to make use of both chips.

I'm still saving up to purchase a THAT analog computer, but it still could be another two or three months. I had to reduce my savings plan on that from $100/month to $50/month, due to running short on grocery money, especially around the end of the month when most of the bills are due.

Copyright

CC BY-SA 4.0 Deed