Debugging Sparkfun 32x64 with adafruit pi bonnet

EDIT: Item is a p4 indoor LED screen module with 1/16 scan. Spec sheet can be found here

Hi there, I’ve just been tinkering with this matrix, trying to get it operational. No matter what I try, however I seem to be exclusively seeing red and green bands:

This looks distinctly different from other issues people have been having; is there a chance my unit is damaged? I have tried switching around the col and row settings, various multiplexing and chaining settings, and so far am het to find anything that really shows me something different than red and blue patterning.

Any and all help appreciated!

Update; I doubt the unit is damaged; as I’ve managed to get some blue to show whilst using a 16x16 config with a chain of 8, however I still don’t see any kind of cohesive output and the display shows in two horizontal bands.

Since you bought this from sparkfun and are using an adafruit adapter, you may try to get support from either or both.
Without checking your panel with another solution (like teensy with smartmatrix), it’s hard to tell if the panel is at fault, or the hardware wiring/software is at fault.

Hey! Thanks for the response,

I’ve done a bit more fiddling about and found that I’d missed the --led-slowdown-gpio flag. I’m using a Raspi 4.

I can now get better output; enough to suggest that maybe the hardware is ok:

I’m now starting to see the fuzzy banding I’ve seen in some other issues. This output comes from the command:
sudo ./demo -D 1 --led-gpio-mapping=adafruit-hat --led-slowdown-gpio=2 --led-rows=32 --led-cols=64 runtext.ppm

If I increase gpio slowdown I seem to get even worse distortion. I’ve tried all the optimisations I can see on the readme (Changed to dietpi, set a dedicated core using isolcpus=3, etc.), but I still seem stuck with the flicker.

My lowest refresh rate has been 179hz, boosting to 260hz using --led-pwm-lsb-nanoseconds=90 --led-pwm-bits=7. I am logging in via SSH; is this liable to be creating interference too?

Ok, glad you got the basics working, I forgot that rPi4 requires a new settings (@hzeller if you are reading this, it’d be awesome for the library to detect the CPU and autoset the new slowdown for rPi4 which is the default for most users now).
ssh does not cause the issues you pictured, I use ssh all the time without this problem.

The library has so many settings, Try ~/rpi-rgb-led-matrix/examples-api-use/demo
look at all the settings, and read the description for each to see whether it can help your use case.
I guess it’s also possible that a bad ribbon cable could do this, but I’d say less likely

1 Like

Thanks for this! I’ll revisit this tonight and see what I can poke around with. The ribbon cable came with the board, but that by no means confirms there isn’t a problem with it. As a last resort I’ll try buying a new one. Hopefully I’ll have progress to post tonight :slight_smile:

So I did some experimentation last night, and with some messing around with the values for the demo command I settled on the following:

sudo ./demo -D 1 --led-gpio-mapping=adafruit-hat --led-slowdown-gpio=2 --led-rows=32 --led-cols=64 --led-scan-mode=0 --led-no-drop-privs --led-pwm-lsb-nanoseconds=75 --led-limit-refresh=60 --led-no-hardware-pulse --led-pwm-bits=10 runtext.ppm

This gave better results, but there’s still a lot of horizontal distortion. There does appear to be a little correlation between the number of powered LEDs and the amount of distortion. I best guess is that it’s something to do with refresh rate, as messing with that has tidied the results. However I’m having a lot of difficulty knowing where to go next from here.

(The missing bands of LEDs in the footage are simply a camera refresh rate issue - it’s not visible IRL)

Ok, outside of swapping panel and ribbon cable, I’m not sure.
Also make sure you are powering your panels with enough voltage and that you aren’t amp limited.

1 Like

After taking a second look at the adaptor I’m using, I think amp limitation is pretty likely! I’m still quite new to hardware projects, if that isn’t obvious :sweat_smile:

Problem solved! The matrix required rather more than the 1amp guitar pedal power supply I was working with (lol), and to be set to gpio slowdow 5 :tada: