A couple errors!

 

Twice this month, customers have said "hey, there's something wrong with one of your modules," and both times, they've been right! Argh!

The first one was the Etch, which is the subject of the video embedded above, but who has time for five minute videos, so here's the too long; didn't watch summary: one of the jacks was upside down because I can't possibly stick with one jack footprint (especially the ones on the Etch -- the new fixed PCB has **another** different jack footprint, one that's not so dangerous) so at some point I swapped all the footprints, with different schematic symbols, which sometimes get inserted upside down or rotated in some other way. I didn't notice for the filter CV input, thus the tip and ring were swapped on the board. I should have caught it while designing -- the ring of the jack is always grounded after all, and never goes off to do something more important.

Thank you, random customer, for catching that error! The fix was easy enough -- my jack PCB footprint is as compact as I can make it, so turning the jack 180 degrees to use the other pads worked just fine.

The second module problem surfaced with the DumpsterDelay. Turns out they work fine, as built, for about half an hour, but then the signal going through the delay chip fades away to very quiet or nothing. Sounds like a capacitor loading up, doesn't it? Turns out that's exactly what it was -- there's two multiple-feedback notch filters built into the module, which don't actually get used for anything -- physically damping the Piezo elements works better than customizing a filter with weird-value resistors. BUT!!! If you look at the schematic in that link, you'll notice capacitors C1 and C2 go nowhere (kinda) if there aren't any resistors installed!! Oh noes!!!

In imaginary physics/electronics land, that's okay, but real world op amps leak a small amount of current out of their inputs. Without the resistors keeping the inputs referenced to some reasonable voltage potential (like, say, ground? or the output of the amp for proper feedback??) the current leakage out of the inverting op amp input would charge the capacitor, so the output of the amp would creeeeeep up (or down? didn't measure) until it got close to one of the power rail. This MFB filter is actually a bandpass filter, so needed to be mixed in to the original signal but inverted, turning it into a notch filter (ugh sorry for the technical blabbbeeerrrrr) so the "notch" getting mixed in was actually just a steady voltage, which pushed the signal off into the voltage region the amp could no longer reproduce.

Maybe that made sense to one reader, thanks for sticking with me.

Thank you, other random customer, for finding that error! The fix is even simpler than the upside-down jack, one need simply put a resistor of literally any common value from zero to probably 4.7Megohms in the R3 positions of the pair of filters, making the output sit at zero volts, which gets mixed in to the signal perfectly fine, like if you're making a recipe and one of the ingredients is "zero cups of cardamom" awesome, not gonna mess up a recipe with that :D

I played with the delay a bunch to make sure it works well (it does) and came away with some conclusions. First: the dumpster part of the DumpsterDelay is the Piezo elements, stuck together, which transmit the delayed signal through themselves. I'm not sure they do much besides just add some unpleasant high-frequency noise. For low values of "repeats" this is okay, it's a bit like extra-high tape hiss, but it reinforces itself with runaway repeats until the signal is just "screeeeeeee" and nobody likes a steady screeeeeeee except maybe harsh noise artists.

I bypassed the Piezos on one of the delays and it's kind of much much nicer to use..... . .. .... .. .so I guess that option exists for people who want to mod their DumpsterDelay modules.

The other insight I had is that the input attenuator knob isn't needed at all. I wonder if I felt that it was necessary because the delays signal kept getting quieter so I thought the signal going to the delay chip needed adjusting TURNS OUT IT DOESN'T

Okay that's all, thanks for reading something not written by AI. Writing like this takes effort and time of a conscious being, not just trillions of processor cycles from an unthinking probability engine. I hate getting tricked into spending time reading stuff nobody spent any time to write.