OLED Noise on Power Rails

[lauraz]

Hi everyone , I'm having trouble resolving a noise issue with a project I'm working on. Essentially, my project involves acquiring a photoplethysmograph (PPG) signal and displaying this on an OLED screen. The PPG signal involves a photodiode (in the NELLCOR DS-100) connected to a transimpedance amplifier, followed by filters and a gain stage on my PCB. The OLED screen is the 2.42" OLED screen from Adafruit which uses the SSD1305 driver chip.

Until yesterday, both the signal acquisition/circuitry and the OLED screen were working well together on my PCB. Yesterday afternoon, I plugged in everything as per usual and strangely there was a lot of 93Hz noise on my +3.3V power rail (when there was not previously). I am really confused as this has never been an issue before. When I remove the screen from its PCB socket, the noise seems to disappear, so I'm guessing something to do with the screen could be the culprit. I have since tried multiple PCBs, multiple screens and adding capacitors (various values) between the +3.3V and GND rails near the screen. Nothing seems to make a difference. I already have an LC filter on my power rails (as suggested by the microcontroller datasheet).

I've attached schematics for my power supply/firmware/PPG circuitry as well as some images of what the noise looks like (zoomed in and out).

It would be really appreciated if anyone had any suggestions for how to approach this problem/ideas of things to look at. Thank you very much in advance!

[station240]

So the Load Switch IC take an input from PA24 controls the 3.3V rail to the OLED and the opamps.
Have you tried bypassing it to see if the 93Hz is the result of a software/AP2281 issue.

I'd also try installing a 2nd parallel inductor from the 3.3V regulator, to supply only the switched load or even just the OLED.

[perieanuo]

Hi,
If I see correct you have 10 uF on +3v3v and 150uF on -3v3?Anyway, never put below 100uF no matter what numbers said.
Second, pcb matters also, but for this frequency I doubt it some pcb mistake like 0.1uF far away from mikrocontroller pins.if you change a little bit the routine in software, the freq changes?put some delay in code to see once I digged a such issue in 5 minutes modifying the code.
Regards,pierre


Envoyé de mon iPad en utilisant Tapatalk

[SiliconWizard]

Don't know what happened if the noise appeared "suddenly" as you state. But some possibilities. Maybe it just appearead as you started to display more things on screen? An OLED display will draw more current if more pixels are lit up. That's just how OLED displays work. You may also try and leave the display powered but displaying only a black background, and see how it affects the noise on the 3.3V rail.

Those displays need a higher voltage than 3.3V and it's generated on board with a charge pump. It may add significant noise for higher currents. If you can experiment a bit and relate the noise level to the display's content, you're going to be left with a couple options.

As a side note, I'm a bit nervous seeing that you're generating the negative supply of a low-noise amplifier with a charge pump (MAX660). That may need some additional filtering.

[AloyseTech]

I had a similar issue once. I was displaying pattern with a lot of black/white zone. Actually changing the frequency (spacing) of the white and black zone changed the waveform of the noise. I finally used a separated power supply highly filtered from the rest of the system and I tried to use smoother visual for the UI. Smoothing the visuals also helped avoid some ghosting effect on the UI (it was a passive color OLED)

[Siwastaja]

Am I reading your scope trace correctly? It seems the 3V3 line is browning out between over 3V3 and basically 0V? How can anything work at all?

Given such ridiculously low frequency, and massive amplitude, it's clearly something not typically called "noise", and any sane amount of filtering won't work. Sounds like something is taking too much current (the display shorting out?) and the supply regulator is hitting the current limit and shutting down?? Or you have a loose wire / contact issue with tens/hundreds of ohms of resistance somewhere, so that the normal, slow load changes that can't be supplied by the bypass caps cause massive voltage drops?