Hello.
Having problem with Matrix OLED.
It flickers few pixels left and right.
https://photos.app.goo.gl/gZ7ArNZaw3fpjRUQ7

Tried to add a bigger cap for display, switched PSU, tried to use linear regulator instead of switching. No luck so far.

Anything else to check or test another display?

Otherwise it’s getting together and this thing is a beast.

What PSU are you using? We found that the regulation can make a difference, at least for the LED driving. But the power will actually be derived from the 3v3 on the Waveshare board.

I would suggest to reflow the header pins and try to clean with isopropanol or a contact cleaner. Maybe some flux residues are causing issues. You could also try to carefully reseat the MCU board (making sure to not bend all of the header pins, so lift it up straight).

Hey, sorry to necro this old thread but I have similar symptoms.
The boot animation is almost stable, but the test screen goes all wobbly. The contents look correct.

So far I’ve added C1/C2 caps, removed caps, bigger C4, reflowed, cleaned, tried various 12V PSUs, injected 5V from a bench supply via J5. The only significant change was bodging a 4R7 + 1000u between PCB and display supply pins but that won’t fit when mounted This is with the core PCB on it’s own.

Seems like something might be off but other than replacing display, core, and/or all the headers (ugh) I’m fresh out of ideas what. Thanks!

latigid on What brands of 12V PSU have you tried? We recommend one that seems to work better, though we didn’t find out the exact reason for this.

So I have a random 4A one and the recommended one (Leike?). I may have a suitable Meanwell from another project.
(the Leike isn’t grounded so I was mostly using the other one to avoid leakage currents)

On its own means without the neo boards connected? Can it be that your 12V supply has a minimum current requirement? Sometimes these are only stable if they go over a certain draw.

It’s just the core PCB with neither axis nor neo boards. The minimum load thing has bitten me before, but after injecting 5V with a bench supply it seemed like the 5V/12V isn’t the source of the issue (but you know what they say about assumptions…). What I didn’t try is 12V off the bench supply.

Is there any chance that the display was overheated or otherwise mistreated?

I’d like to say “no” because I try to be careful but It’s been in-and-out a few times now though…

You could scope the 3v3 line from the display or the waveshare board, and also compare what happens with the OLED connected. Do you measure high resistance from the waveshare 3v3 to the display 3v3?

I’ll check that. I did a quick scope of the rails and and display pins for activity but not much else.

It might be possible to cut the 3v3 supply to the display and rather try an RC filter like that (not sure exactly how you wired your one?), or wire up a separate regulator.

Yeah, I cut the header to the OLED pins 1 & 2 so the RC filter was between the PCB and display.
Separate regulator might be an idea, I should have some somewhere…

Thanks for the input!

You’re welcome!

pld Yeah, I cut the header to the OLED pins 1 & 2 so the RC filter was between the PCB and display.
Separate regulator might be an idea, I should have some somewhere…

Keep in mind that the 3v3 is connected at two places on the header

Doing a board-level bodge might work here, basically interrupt the 3v3 at the large unplated via (drill it out?) and see how the 3v3 could be improved.

So FWIW: in a quick hack I added a separate 3V3 vreg (some LD1117 I had leftover) and on it’s own that had the same result. Of course that’s introducing even more variables…

I did get to wondering if it’s just tolerances stacked against me, or if there’s maybe some timing parameter in the OLED startup sequence that might help. For now I’ll stick with the 4R7+1000uF (or 2200 if it fits) which is mostly stable and re-evaluate later.
I still have all the RGBs to solder

Hmm, strange! It’s really not just flux between the header pins? I can’t think of anything much else,
If you really need so much capacitance, perhaps you could put several in series so that they fit?

After a bunch of experimenting I couldn’t get it better than mostly stable with occasional wobbliness, so I tried a new display (helped push an order into free shipping ) Behold: stable image as it should be. Even with the original cap value. I’m kind of curious what I did to cause the issue but oh well…

Not sure what the lesson here is, but it’s closure at least Thanks again for the ideas!

Very glad to hear that!