Debugging a DOD585-A Analog Delay
I've had a DOD585 on my bench staring at me for like two months. A local friend asked me to see if I could fix it and I'm always confidently raring to debug stuff right up until the time comes to actually disassemble at which point I'm always finding reasons not to start. But I've started and now I'm mostly finished.
The problem was, when the pedal was on, the signal went to zero when you dialed the mix control all the way up. It only passed clean signal and no delay came out. Fortunately there's a schematic online, so tracing the point of failure was fairly straightforward. Originally I assumed the problem would be with the MN3005 (the bucket brigade part) or MN3101 (the clock driver for the bucket brigade part), but everything looked healthy, and I even saw a delayed signal passing all the way to the far side of the AC bypass cap on the output half of the compandor NE5711.
Most of the time hardware debugging is an iterative spiral of guessing what the problem could be, ruling that out, using what you learned from the ruling out process to shorten your list of possible problems, and then firing off another guess and check cycle. Given that delayed signal was making it almost all the way out, and clean signal did make it all the way out, my first guess and check had narrowed the list of possible problems way down. The problem was with the JFET toggle circuit.
For a bit of context if you're not a pedal dork: when we talk about whether or not a pedal is "true bypass" we mean that a non-true bypass pedal routes its signal through a semiconductor controlled buffer path on the PCB instead of through a physical clicky switch (or, more recently, through a microcontroller managed relay). Boss does this on almost all of their pedals precisely because it makes them way faster to assemble. ElectroSmash has a nice run-down of how Boss's toggle circuit works. The one in the DOD585-A is implemented differently, but it achieves the same effect: it generates a voltage that toggles high and low using a single momentary connection to ground initiated by a footswitch. That voltage goes on to control circuit routing behavior.
HOWEVER: The DOD585-A makes use of this toggling voltage a little differently. Most Boss pedals I've worked on use the toggling voltage to route the signal through one of two sets of semiconductor switches (they're JFETs most of the time). When the voltage is high, the signal goes through one set of switches and when the voltage is low, the signal goes through another set of switches. See again the ElectroSmash analysis above. That is very much not what the DOD585-A does, probably because it's a delay and some sort of trails-like fade out is more desirable when you turn off the pedal. You can't just chop the delayed signal path off when you turn off the effect or else the delay suddenly disappears in an an unnatural way.
Incidentally, this is why it's not always better to go true bypass, because out of the box a true bypass design would just chop the delay signal away the second you stomped.
The DOD585-A is using a CD4007 inverter logic IC to create a pair of complementary toggle voltages, one is high when the pedal is on while the other is low while the pedal is on. That's just out of frame in my screenshot above.
When the pedal is on, a logic high goes to the gate of the J113 JFET circled above, turning it on, and passing the virtual ground (around 9V on this pedal) to the middle of the mix control. The mix control sweeps that voltage between the clean signal and the delayed signal. At extremes, it's effectively shorting either the clean or delayed signal to "ground," i.e. in this case, the same voltage that's going to the non-inverting (+ labelled) input to the output op-amp TL072. The pedal on state also keeps the gate voltage to the BJT attached to pin 1 of the NE571 from conducting, so pin 1 happily sits at whatever voltage the compandor wants it to be and signal passes.
When the pedal is off, a logic low closes off the path through the J113 and shorts pin 1 of the NE571 to ground, removing the ability to sweep connection to virtual ground via the Mix knob and shutting off the audio path for the delayed signal through the NE571.
And so: my problem is that the J113 in my delay is fried. JFETs are known for having exceptionally low leakage current on their gates (pico amps, usually), but when I apply 12V from the toggle circuit to the 1M resistor, I get about 1 volt of drop, meaning I see 11 volts on the other side of the resistor. Ohm's law tells me that means about 1 micro amp of current, which is two whole orders of magnitude more than I should be seeing. The JFET's borked. I've got a replacement on order. Maybe I'm wrong about all this and I'll update if that turns out to be the case!
One last fun note: a lot of modern pedals use an inverter IC to create bi-polar voltage rails to power their high headroom analog circuits. They take in standard pedal 9V and invert it to create +/-9V for 18V of headroom. Not this pedal. It brings in 20V and creates a virtual ground that, in practice, ends up around 9V. It achieves the same thing as a bipolar supply, and is probably ultimately quieter because you're not putting an extra switching regulator on your board, but it sometimes makes reading schematics less intuitive, at least for me.
Anyway, the fun part is that this pedal, for some reason gets power not from the standard issue barrel jack but from a 1/4" audio-style jack. There's 20V on the tip and ground on the sleeve. That's totally fine from a current supply perspective, but boy howdy does it create a spark every time you plug it in and accidentally touch the tip to any metal on the pedal. As far as I can tell, there's plenty of protection everywhere to prevent the spark from damaging much, but weird design.