With fourteen or so sticks hooked up, I started fitting 400ohm resistors and input power capacitors to the sticks. I also replaced one of the wires from the board to the panel mount socket with shielded wire (earthed at one end—you can see the black one going from the screw terminals to the top right socket.)
Things got better, but no individual action seemed to fix anything conclusively.
But I think I might have some idea what’s going on. I think there is some crosstalk between the wires coming from the control board. The clue is that everything’s solid when it’s set to sequential output—when adjacent pins/wires aren’t carrying signal at the same time. When parallel output is on, that’s where things start to go haywire.
If I can get hold of some tomorrow, I’m going to try putting some ferrites on the signal wires (good idea or no?); if I can’t, I’ll replace them all with shielded wire. I’ll also fit the data line resistors and input power caps to the rest of the sticks as a matter of hygiene.
Can anyone see anything in my layout that sticks out as particularly poor design interference-wise?
I would not bother with the ferrites myself but with the problem you are describing, I would definitely change all the internal signal wiring to a shielded type.
I do not know if this is possible for you at this stage but I would consider completely re-doing that little perf board layout to shorten all that signal wiring as much as possible.
The shorter the wire the less potential for crosstalk and noise pickup !!
I would definitely place the terminal blocks much closer to the level converters outputs and wire them via the shortest possible wires on the bottom of the board.
I would also relocate the level converters to be closer to the Teensy and wire from the bottom with the shortest possible wire.
To be fair, the distance you currently have between your Teensy and the 2 level converters is not bad but if you redo the board, I would also do that !
Also, I do not see is a decoupling cap close by the level converters. I would definitely add 1 small 0,1uF cap as close as possible to the 5Vdc and GND pins of each of these IC’s.
Are you using MIDI connectors or 3 pin Mic Cables? In the audio industry, when they use an audio connection snake, high voltage signals/power are routed external to the snake. If power needs to cross over signal wires, a 90 degree crossing is best practice. Are you running 12v power thru the cable along side the low power 5v data signal…? Right now, I have a rig where I can run 5v power alongside 2 data lines for my LPD8806 LEDS, through a 10’ MIDI cable, with no ill effects. I’m just thinking out loud here BTW… In the post where I linked Dan’s Digi project, his wires are quite short to the panel, but most importantly, he is using the twisted pairings for his wiring to the pixels. Hope you figure this one out!
Time is one thing @JP_Roy (this is already the second iteration of the board), but space is another. There’s just not enough space on the board in between the two power supplies to fit the screw terminals for the signal wires. There’s also the issue of parallel output using a certain subset of non-adjacent pins, thus the space required between the Teensy and the level shifters. If adding the data line resistors and shielded cabling doesn’t fix it 100% I guess I’ll have to look into it though.
You say to connect things on the bottom of the board—why is that? I’ll also see if It’s possible to add a cap on the level converters’ inputs, that’s a good suggestion.
@Jon_Burroughs , they’re Neutrik PowerCON connectors, with, as you guessed, 12V running along Live and Earth, next to the 5V signal on Neutral. I chose those because they’re stage–proof, can carry the current I need and they use screw terminals everywhere. Very nice connectors.
My first thought is that it could be simply the location of the control board right between the two power supplies. There’s bound to be a lot of switching noise. I’d also try to route the outputs of the supplies out around the sides so they aren’t running directly over the inductors.
@John_Burdett , I could do that without too much trouble I guess, worth trying. Not happy how that turned out in the end, in fact. But I do have plenty more of that wire and the connectors are readily available.
(That’s part of the battle—the lag time between “Hmm, that might be it”, ordering the parts, receiving the parts, and trying out the prospective solution.)
the only reason I suggest wiring on the bottom is to get the shortest possible wire between 2 points. Your wiring now loops on the top and that simply adds unnecessary length and the potential for more crosstalk and noise pickup.
I agree with +John Burdett, he looks a bit young… but he’s right about the placement of that control board sandwiched tightly between the PSU’s and that looping set of power supply wiring !!
For that, couldn’t you easily flip them PSU’s 180° and considerably shorten the power wiring ?
Now as you said…
“I think there is some crosstalk between the wires coming from the control board.”
You may find that the shielded wires will do the trick and I sincerely hope so !
My last recommendation would be to get your hands on an oscilloscope and probe around to see ‘exactly’ what is happening on your signal wires AND power wires.
Having the PSUs around the other way would mean the grounding between the supplies and the 240V input would get in the way of everything I think. You can only see one “layer” of PSUs btw, there’s another two underneath what you see there.
Yes, I knew that you have 4 PSUs from the pictures in a older post. They appear to be screwed down on some plywood or MDF boards. From my perspective, you could re-organise them somewhat easily.
Note that although I am certain that each of the suggestions I gave so far would help reduce the crosstalk… only you can decide if they are easy or worth while applying.
but he’s right about the placement of that control board sandwiched tightly between the PSU’s and that looping set of power supply wiring !!