You've already forked torvalds-GuitarPedal
mirror of
https://github.com/torvalds/GuitarPedal.git
synced 2026-07-20 17:42:56 +00:00
244641747b
This was something I did for the usb-stomp switch with the MIDI connectors and no UI as a "just because" thing. I had used that cap on the board for the audio jacks, so it was on my radar, and I was "let's do it on the USB board too in case it helps with USB power supply noise". I just compared the older pedal with the new smaller one, and they are both very quiet, but the pedal with the UI clearly has some noise when rotating the effect knob and moving between effects. It could be just the result of doing the "effect->seq++" when a pot changes, and the resulting work on the audio core, but I had some similar noise issues long ago when doing screen updates. Those seemed to go away when I switched over to the background i2c DMA updates for the screen updates, but I wonder if it's the screen itself that just uses more power when updating. Because it's not noise from USB power: this happens with the a plain and fairly quiet 9V guitar pedal power supply. I don't have a power meter for the 9V side, but when powering the pedals over USB, the small MIDI version without a display draws around 55mA according to my USB power meter thing, while the UI version draws between 72..80mA. Perhaps even more relevant is that the power draw seems to be related to how many pixels are lit on the screen: so just switching the display to an effect with less text on it - without changing which effects are enabled or not - will consistently make it be around that lower power draw, while a more active screen will consistently be in the upper range. So it doesn't sound entirely crazy to think that it might be some power supply noise - even if this all should be on the digital 3.3V side, not the separate analog 3.3V rail on the audio jack board. Those things *should* be very well isolated from each other, but there's obviously the digital i2c and i2s lines that cross that boundary. Of course, it might be some cross-talk issue with the i2c lines, but the screen is on a different i2c bus than the one that goes to the audio jacks and the TAC5112. So add those same caps I have on the small MIDI-only board - that probably doesn't need them - to the board with the UI. Another "just because". I'm just adding it to the schematic as a note for any future version of the board, I haven't bothered to add them to the PCB (which would likely get other changes anyway before I'd re-spin it). I do wonder if I need to look at signal integrity of the i2s lines. They are wiggling at "only" 3MHz, but with he FPC cable and connectors, maybe I need to start wondering about ringing and slew rate control? But I'm not competent enough for that. Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>