You say halting problem, I say banananana.
Nanny Ogg knew how to start spelling ‘banana’, but didn’t know how you stopped.
-Terry Pratchett, Witches Abroad
4 Likes
No changes to Volts or Pulse?
No changes were needed for those two, though I suppose I should change the names. I should also check the silkscreen labeling.
3 Likes
I made some adjustments for the panel to match my other modules 
Also moved some knobs and jacks a little bit, so I need to change some bits on the main pcb as well.
7 Likes
I like the LEDs closer to the jacks, and the increased separation between jacks/LEDs and scale knob. Though output jack above input jack is contrary to my usual practice…
1 Like
Good point! I will try some other options, I guess…
edit: compromise 
maybe I just leave out the noise output…
I’d leave it in, always good to have a noise source nearby. Why not move CV up and noise down?
2 Likes
Before anyone goes on ordering some boards, I just noticed some weird behavior with mine. It is a weird module, so it is hard to say what weird behavior is, but as it cycles through the bits, the bit 6 led does not light up in sequence with the rest. It gets skipped as the rest blink through, but does light up sometimes? I don’t know what is going on. I looked through my schematic, and don’t see a reason that this should be happening. I did order a new run of the main board to fix my issues with the first batch, so I may build another to see if it acts the same way.
3 Likes
Good to know! bit 6 led is the one labeled UNK3? can you check if there is a signal at pin 12 of SHIFTREG_MAIN1 that is more consistent than the LED? All of the other LEDs seem to behave as expected, so I would suspect some bad solder connection at the LED or its resistor (R15).
2 Likes
It isn’t that there is no signal, it is skipping bit 6. It goes through 7 steps when set to 8. Same on the bit output to Pulses.
Weird! Have you tried switching the shift register chips, to check if it happens at another bit then?
2 Likes
Just tried it, and same result. Bit 6 remains the issue.
Did you check with a magnifying glass there isn’t any short around stuff connected to bit 6 ?
Sometimes there is a solder bridge thinner as a hair…
4 Likes
It gets even more weird with two bits.
With one bit, it follows the sequence 1-2-3-4-5-7-8
Bit 6 just gets skipped.
With two bits, it follows the sequence 81-12-23-34-45-56-78
So six gets used when 6 is the second bit, but not when it is the first.
Then with three bits, one of the bits gets lost when it passes through 6 and the sequence goes to the two bit version.
With four bits, two bits get lost when they pass through 6 and the sequence goes to the two bit version.
With 5 bits, two bits get lost in the first pass, the the third gets lost in the second pass.
With 6 and 7 bits things don’t seem to be consistent, dropping bits sometimes.
With 8, it stays 8.
What would cause that?
Edit: I left it running with 8. Eventually it drops all bits, even the two or one. Not so predictable when that happens.
1 Like
The strange thing is, that the missing/skipping happens within one shift register! If I were between two of them, I would assume a connection problem, but like this?? And if it were just skipping I would look for a problem with the clock (maybe bit6 track runs close to the clock track, so it could instantly trigger another clock) but with the bits going missing this also makes no sense! 
Maybe it makes sense to build a second board and check if the problem is also there?
5 Likes
Can you look at the clock signal with a scope? Check the shape and timing? Are there extra ones?
I don’t know much about shift registers but I know there’s a maximum clock frequency. What happens if two clock pulses arrive too close together? Is the second one just ignored? Or can the second one screw up the handling of the first one? If as @sebastian says “maybe bit6 track runs close to the clock track, so it could instantly trigger another clock”, could that maybe cause bit loss?
(Though in fact net B3, which as I understand is the one the 6th LED connects to, crosses the clock trace on the opposite side at a perpendicular angle, so it’s probably not that. But maybe there’s a fast ghost clock from some other cause.)
4 Likes
And to simplify things and reduce error sources, I think you could remove the AND ICs, since they are only used for the pulse generator, right?
If you want to go really crazy: a usb logic analyzer costs less than 10$
4 Likes
I built a second, and got the same result.
I am now thinking this may be a power issue…
3 Likes
Nope, back to a clock issue. I was clocking it from an LFO. When I swap it to my keystep, it seems to work (just too fast of a clock). I really need to build a clock module…
2 Likes