The winter of our discontent

December through early January was not my greatest time building.

I have two Kosmo modules under construction. One is an FM Drum, built around a circuit board from Barton Musical Circuits. It needed two SPDT toggle switches, and I had a ton of them in my stash… right? Turns out, no, I didn’t; I had two. So no problem? Except they had no nuts and washers on them, but I have toggle switch nuts and washers lying around loose or in their own little bags all over. Except they don’t work. Apparently these switches have a different diameter or thread than the usual.

And at the time I found that out, I’d just sent off an order to Tayda. And I decided it’d be cheaper to order new switches from AliExpress, so I did. And waited.

About three weeks later — on Christmas — I got an email from AE saying “Your package was delivered!” And of course it hadn’t been. So I asked for a refund and they refused, saying “The shipping company says it was delivered, if there was a delivery problem you have to take it up with them.”

The shipping company is something called On Track, and I kind of assume they take delivery from China and break it out to individual orders to be mailed to the purchasers. But I subscribe to USPS Informed Delivery, and if I’m getting anything via USPS normally they notify me when it’s on the way and again when it’s delivered, and I hadn’t heard anything from USPS. The only tracking number I have wasn’t recognized by USPS (or UPS, or DHL, for that matter), only by On Track, which cheerfully told me the package had been delivered. When I notified them it wasn’t they said “If it wasn’t delivered you have to take it up with the seller.”

I went back to AliExpress, requested a refund again, and it was denied again. This time they said “You have to go to USPS and get official proof of non delivery.”.

Official proof of non delivery.

Hello, USPS, can you give me official proof that you did not deliver this package I have basically no information about? Right.

At that point I decided for a $5 order it wasn’t worth any further hassle. Call it tuition; what I learned is I do not want to put up with AliExpress’s sketchy practices and failure to do right by its customers any more. I am done with them. I sent off another Tayda order, including the toggle switches I needed, and it arrived within a week. I haven’t gotten around to using them yet, though.

The other module is a low pass gate, from an AI Synthesis PCB. Since I’m building Kosmo, the jacks and some of the pots are panel mounted and wired to the on-board footprints; two switches and two pots remain board mounted though. I built it, it works… mostly.

But there’s a switch labelled “Deep” and an associated Depth pot, which apparently are supposed to compensate for how a vactrol VCA (when it’s in that mode) doesn’t fully close when the CV is zero. I found that even with the Depth pot fully counterclockwise, there was a big reduction in the output level when I turned the Deep switch on. Big enough that I didn’t think it was particularly usable. I checked my components, checked my soldering, checked the pot wiring — found one pot wired incorrectly but it wasn’t relevant to the Deep switch. Other than that, found no problems.

I wasn’t sure what was supposed to be going on in the circuit so I did an LTSpice simulation of the part that handles CV and drives the vactrol LEDs. Lo and behold, in the simulation, even with the Depth pot fully counterclockwise, using the Deep switch significantly reduced the LED current. Seemed like the simulation was behaving the same as the build.

I contacted Abe at AI Synthesis and he didn’t know what the problem was. He sent me a link to a video demonstrating the output level not changing at all when the Deep switch is flipped. And I can’t replicate that behavior.

I then decided to breadboard that part of the circuit; hopefully it’d work correctly and I’d be able to use it to find the flaw in my build. And on the breadboard… even with the Depth pot fully counterclockwise, using the Deep switch significantly reduced the LED current. And the more I thought about the circuit, the more convinced I became that has to be right, the current has to be significantly lower. Build, simulation, breadboard, and understanding were all consistent with each other — and inconsistent with Abe’s video.

(Unless the LED current changes but the output doesn’t, which would make sense if the vactrol LDRs are at saturation already; then doubling the LED current wouldn’t make any change in the output. But then the output also wouldn’t change in response to small changes in the knob settings or the CV. And large enough changes to change the output should result in the Deep switch changing the level, because then the LDRs aren’t saturating any more. And presumably Abe doesn’t see either of those behaviors.)

I posted a query to the LMNC Discourse group, but it hasn’t resulted in any new ideas. Or even just confirmation that no one else’s module behaves like my build does… or that someone’s does, I suppose.

Maybe I should call it good enough, and just use it without using the Deep switch. Or maybe I should abandon it. Or maybe I should put it aside and maybe figure it out later. I’m going with option three for now.