Wednesday, July 22, 2009

Rebuilding a Super Psycho

A couple of years ago I bought a partially completed Catgirl Synth Super Psycho LFO from someone. The circuit board itself was complete, but none of the panel connections had been made. The person who built it had bought a panel for it from Cynthia, and all of the panel components, still in the bags, came with it.

I set about completing the build, and ran into a problem: This board has a zillion connections to the panel, and when I did the soldering, a couple of pads lifted off of the board. No biggie, I thought at the time; I put in a few jumpers to bypass the damaged pads and completed the build.

Well, when I started testing, a lot of things were flaky. Some of the oscillators worked some of the time and not at other times. There were lots of random oscillation rate changes. When I investigated, I found several other pads that had lifted when the original builder was soldering in the parts. And I noticed that the original builder had had to put in a few bypass jumpers himself. When I set about unsoldering a few parts so I could put in more jumpers, more pads lifted! Pretty much anyplace where the copper wasn't covered with solder mask, it was coming off.

I finally threw in the towel. I ordered a new board from CGS, and Tuesday night, I set about stripping the parts from the old board. Here is what I started with:

I started by just cutting all of the lines to the panel. It was a big rat's nest anyway. When I put it back together, I'm going to put headers on the board and use MTA-100 connectors to connect the panel wiring, a la Dotcom.

Big pile of indicator LEDs, which I never secured to the panel. They are a mix of white, blue, and violet types. Obtained from (RIP).

The board, ready to strip. The original builder was nice enough to socket all of the ICs. Most of the panel connections, now cut, were along the bottom edge.

The solder side of the board. Note the numerous workarounds and hacks, including the cap hanging off the board:

A bunch more of the pads came up as I was desoldering. I don't know what happened to this one; I've never seen a circuit board shed copper so badly. Although I wasn't being quite as careful as I would be when soldering normally, I certainly wasn't trying to be destructive, and I don't think I used excessive heat on anything. Nonetheless, this is what happened:

So the board's obviously a goner. At this point, I've removed everything except the resistors:

Below: the new board, together with the old. The new one is a rev 2 which accounts for the slightly larger size and somewhat different layout.

Why didn't I remove the resistors? Lot of trouble. I just laid in a large stock of resistors, so I'll probably assemble the new board from my stock. I'll only go back to the old board for a resistor if I don't have one of the right value. I got everything else off the board: capacitors, transistors, ferrite beads, IC sockets, and the MTA-156 power connector. I'm not sure if I'll be able to reuse that last item; those MTA headers always get distorted and the pins come loose if you apply too much heat to them, and owing to the size of the pins it's hard to desolder one without using lots of heat. I damaged one of the IC sockets; I may replace them all anyway, just in case. Everthing else that survived will be re-used; I'll test the caps and transistors first.

Here is the panel. (EDIT: It's right side up now.) The panel is obviously from Cynthia, although the circuit board is not. I've done a couple of things to the panel: the stock Cynthia design has hi/low speed switches for oscillators 1, 2, 5, and 6, but the square/triangle switches occupy those positions for oscs 3 and 4. I've added speed switches for those two oscs, which are on the left. I changed the banana jacks for 1/4" jacks, since most of my stuff is MOTM and Dotcom.

Before I put it all back together, I'm contemplating some mods. The first thing I want to do is add control voltage capability for the osc rates. You'll note that the Cynthia panel has a jack for this. However, the circuitry isn't on the stock CGS board. On the version of the Super Psycho they sell, Cynthia adds a mod that uses vactrols to add CV capability to the oscillators. I'm going to do that too, but I may make some changes. I want to do the same basic thing, but I'm not sure if I'm going to use vactrols or a Fairchild optoisolator IC. I've got some Fairchild H11F1's, which I'm going to experiment with; I've read that their response curve is not very linear. Since I bought those, Fairchild has released a bunch of new parts, some of which are specifically labeled as being for analog applications. If I'm not happy with the H11F1's, I may order some of those and give them a try. (Note that Fairchild has since discontinued the H11F1.) Why not just use vactrols? Well, I do have some 5C3/2's on hand. But six vactrols takes up a lot of room on the circuit board, and the design of the oscillator circuits necessitates having a separate optoisolator channel for each oscillator.

Where will that board go? Well, I'm going to build an auxiliary panel. That panel will have a daughter board on standoffs, which will contain all necessary circuitry for my mods. And since I have extra panel space, I'm going to extend things a bit. Since each oscillator has to have its own optoisolator channel to have CV capability, instead of just one CV input, I'm going to have two, and each oscillator will have a switch to select CV input A, B, or neither. The extra switches and jacks will go on the auxiliary panel.

The other idea in my mind is to provide some capability to split out some of the oscillators. At first, I thought about just providing individual outs for each oscillator. I can pick off the individual signals from the main board, but they aren't well buffered. And anyway, I thought of something that will be more fun: the ability to have a second mix output, which individual oscs can be added to or removed from at will. The controls and the output will be on the auxiliary panel. I'm not sure how I want to do it yet: should I just have switches to choose one output or the other for each oscillator? Or should I have pots to mix the oscs into a second mix out? I'll have to think about that bit a little more. I might also add a DC offset capability to the second mix.


G said...

Nice, saw this post on Matrix. I actually just started my CGS21. I'll refer to this post if I run into any issues. Thanks :)

Dave Cornutt said...

Thanks! I'll be posting more on this as I rebuild it.