P10 Panel Row Offset by 1

[silver_ice]

Trying to understand if there is a software cause for this or if it is more likely a panel fault - the rows seem to be offset by 1.

I've got a single P10 HUB75 (1/8 scan) panel into a HansonElec Octoscroller on a Beaglebone GREEN, running FPP 6.3. It lights up fine but you can see on the test pattern its just not right. Anyway, I modeled it up in xlights and sure enough, running through the nodes, if I test "Node 1" (channels 1,2,3) the pixel in row 2 column 1 lights up. And then when I get to node 225 (ch673-675) which should be row 8, that lights up on row1. Then rinse repeat for what I guess is the other grouping of rows 9-16 with the same offset issue.

I can fire it up later and video it but Im wondering if anyone has seen this before and might be able to validate if its a hardware fault on the panel vs a software issue. It happens no matter what port I throw it into on the Octoscroller.

Also, I can mostly fix it using pixel remapping in FPP (basically, remap channel 673 count 96 to start ch1 and then remap ch1 to ch97 count 672) which makes me thing its a panel manufacturing defect.

thanks

 

[Indigogyre]

Can you post the screen shot on how the panel is configured in FPP?

 

[silver_ice]

Yup and also attached is the test pattern.

 

[Notenoughlights]

Yup and also attached is the test pattern.

Is that the FPP test pattern or something you are sending/displaying via sequence?

 

[silver_ice]

Thats the test pattern. You can see how row 1 is offset a row (and row 8 shows up on row1). I'm currently digging into the FPP source, to see if I can catch a misconfiguration there.

One thought I had - Im using a Beaglebone Green and not a Black - but they are pinout identical right? except for the lack of HDMI output. Anyway, im starting to feel like this is more and more a software issue, since the physical pixels work fine, they are just mapping correctly.

 

[Notenoughlights]

The test pattern from FPP though, not the board you are using? i.e do you see this screen when turning on test mode
View: https://i.imgur.com/FVJikCy.png

 

[silver_ice]

Its the test pattern on the LED matrix page here...but yes, the full channel test output fires up every pixel just fine and you can't tell there is a problem, until you tell it to light channel 1 - 96 and it lights the 2nd row.

More info - I note that FPP is using the pin mapping /opt/fpp/capes/bbb/panels/Octoscroller.json and there is an /opt/fpp/capes/bbb/panels/Octoscroller-v2.json with obviously a different map. If I change to using the v2 mapping, the panel doesnt work at all, so I assume the FPP settings (v1) are OK.

 

[Notenoughlights]

Earlier on you posted a config screenshot, with the layout showing O2, P1, C-Def. Are you plugged into output 2 on the controller?

 

[silver_ice]

Yep, Ive tried all outputs on the Octo, each with the same issue.

 

[Notenoughlights]

May be best waiting for someone else who's had this issue with an octo board. I use the colorlight cards. I ended up just trying random settings until something worked.

 

[silver_ice]

No worries, Im going to bang some bits into it with LEDScape to rule out a hardware issue.

 

[Skymaster]

Try the output by rows setting.

 

[silver_ice]

No output when I used that option

 

[silver_ice]

Interestingly, this is some kind of hardware and software issue I suspect - testing with FPP 3.0 (the earliest I could find an image for), firing up "Node 1 (channel 1,3) results in the left most pixel in the top 2 rows lighting up.... unfortunately LEDScape hasnt been maintained - does anyone have a firmware that I can use to just bitbang into the panel to test it directly, without having to go through the channel -> PRU mapping that FPP does?

 

[silver_ice]

Nope, turns out I was just dumb. After reading through the code, I re-tried the "Output By Row" option and it worked. Previously, when there was no output, I was encountering another bug where the ffpd service wasn't restarting properly.