P10 Help - Well really 5mm Panels - Need Help Configuring

[derektoews]

Hey Everyone! I ended up buying some 5mm panels, and I'm trying to figure out how to configure them but I'm running into some issues that I can't quite figure out so any help would be much appreciated! At this point, I figured I would just try to figure out how to configure 2 5mm panels and once I figure that out it would be easy to understand how to hook the rest of them up and configure them... but even 2 is proving to be difficult. I also attached screenshots for all of this so you can see the setup to see what I am doing / doing wrong.

So in Xlights, my start channel is 15397 and my end channel is 27684 for a total of 12287 channels. In the Xlights layout, I created a Matrix and the number of strings (which is the vertical number of pixels) is 32 and then the Nodes/String (horizontal pixels) is 128. Then on the Setup page, I have 25 Universes setup for the 2 panels to get the 12287 channels, with the first 10.0.1.32 Universe starting at 15397.

Now on the BBB Octoscroller I have FPP installed, currently running in Bridge mode so I can configure this a lot faster and that's working fine... again, just not the configuration. In FPP, on the E1.31 page, I setup Universe 34 as 10.0.1.32 universe 1 starting channel 15397. Then it goes to Universe 58 as 10.0.1.32 Universe 25. 

On the FPP LED Output page, I have the start channel at 15397, LED output checked. As for the panel layout, I've tried a few configurations, but nothing has worked. I tried just 2 panels, both with O1 and then P1,P2... that didn't work. I tried 4x1 with all O1, then P1,P1,P2,P2, .... Nope, I tried 4x2 (to get the 12887 channels) with all O1, P1,P1,P1,P1,P2,P2,P2,P2 .....Nope 4x1 with all O1, then P1,P2,P3,P4, .... Nope, I tried 4x2 with all O1, P1,P2,P3,P4,P5,P6,P7,P8..... Nope. Nothing I've tried configures the panels correctly from Xlights.

But here's the weird part.

When I put the panels in test mode through FPP, I can get both panels at full white with only channels 15397-18468 assigned... which is only 3071 channels. So that's I guess what I'm not understanding. How is Xlights saying these 5mm panels need 12287 channels and FPP can make these panels light up full white with only 3071 channels... Obviously this why it's not configuring correctly, but I also don't know how to fix it either. Any help would be greatly appreciated!

Here's a link to the pictures... The upload doesn't seem to be working.
https://drive.google.com/drive/folders/0B2T4nFevA24vRmhKTFdtVHYxY3c?usp=sharing

[jnealand]

I think the upload is not working because you are still in moderated mode with only 4 posts at this time.  You also have not stated which version of FPP you are using - ie. is it the Dan Kulp image or the old original.  P5's need the Dan Kulp image.  I have P5s but I am running them using a Pi and the Pi Matrix adapter.  Also my P5's are 32x32 type size (square) some are rectangle shape.  What do you have?  I am asking for clarification for others to be able to help as I do not have any P5's using a BBB Octo configuration.  I do use Dan's image for my P10 setup that uses the BBB and it works great

[derektoews]

The panels are 16x32. I'm running FPP version 1.9... I'm not sure if it's the Dan Kulp image? I downloaded it from the Falcon site, installed it as version 1.5, then it prompted an update to 1.8, then 1.9. I'm assuming 1.9 isn't the Dan Kulp image?

[dkulp]

The panels are 16x32. I'm running FPP version 1.9... I'm not sure if it's the Dan Kulp image? I downloaded it from the Falcon site, installed it as version 1.5, then it prompted an update to 1.8, then 1.9. I'm assuming 1.9 isn't the Dan Kulp image?

No... it's not.

See:
http://falconchristmas.com/forum/index.php/topic,7354.0.html

[dkulp]

The panels are 16x32. I'm running FPP version 1.9... I'm not sure if it's the Dan Kulp image? I downloaded it from the Falcon site, installed it as version 1.5, then it prompted an update to 1.8, then 1.9. I'm assuming 1.9 isn't the Dan Kulp image?

Looking at your pictures, I think you are using 64x32 panels, not 32x16.   P5's are usually either 32x32 or 64x32 (which, in reality, are 2 32x32 panels stuck next to each other). 

[derektoews]

My bad, these are 64x32 panels. So if I put the Dan Kulp image on the BBB, there will be 5mm configurations on the LED output page and this will all make sense?

[dkulp]

My bad, these are 64x32 panels. So if I put the Dan Kulp image on the BBB, there will be 5mm configurations on the LED output page and this will all make sense?

Yes and no...   There will not be a "P5" option.   There will be a "64x32 1/16 Scan" panel option that you would select.    Whether the rest makes sense or not... well...  not really sure. 

That said, if using my image, I would SUGGEST using two outputs on the octo if you can from a wiring perspective.   You'll get a slightly higher refresh rate.   It's also a bit easier to configure as it would be O1/P1 for one panel and O2/P1 for the second.   Then it's just a matter of getting the arrow/direction correct.

[derektoews]

Got the Dan Kulp version installed easily and configured... however the panels still aren't looking right. Here's a link to a video to show what they are doing:

https://drive.google.com/drive/folders/0B2T4nFevA24vcURPUV9HTTE0WUk?usp=sharing

I've also attached a screenshot of the FPP config for the LED Panel page.

Are the channel counts per universe just way off or something?? Just not understanding what's going on. It looks like it's breaking up each panel into 3 separate panels.

[derektoews]

Ok, so I have been doing a little bit of testing through the FPP Channel Testing page on these 5mm panels and have found something interesting...

In the channel range, I assign the start channel to 15397 and then end channel to 15398, and then at the bottom, I select Fill, and only have the Red on, I would expect only the Top Left pixel of panel 1 to turn red. However, the top left pixel turns red, along with Column 1, Row 8 and Row 9. The second panel actually does the exact same thing, turning on Column 1 Row 8 and Row 9. So even though I have the two panels assigned down separate outputs out of the Octoscroller (O1, and O6), somehow they are kinda getting the same signal on some of the channel outputs. However, when I go from 15398 to 15401 (which should be the second red channel), the red would then light up Column 2 Pixel 1 as well as Column 2, Row 8 and 9, but only on panel 2, nothing on Panel 1. And So.. I'm a bit confused... However, I've tried this on 6 different 5mm panels and they are all doing the same thing.

Maybe all of the panels are bad?

[derektoews]

May have figured out my problem but I'd need someone to confirm this is the issue..

My 5mm panels are 64x32 1/8 Scan rate and the Dan Kulp profile supports the 64x32 1/16 Scan rate... but would a Scan rate really cause the panels to not display correctly? I thought the Scan rate was more of a refresh rate...

[dkulp]

The scan thing is definitely the issue.   The ribbon cable only has two sets of RGB lines in it and an "Address".   A 1/16 scan panel would mean, at any given time, the two RGB sets are each outputting one row each.   With 1/8 scan, each of the RGB sets has to output two rows at a time.  Thus, twice the amount of data has to go down the line, but you only have to advance the address 8 times instead of 16.   Anyway, the way the 1/4 scan P10's send two rows of data on one RGB line is to output 8 pixels from one row, then 8 pixels from the other, then 8 from the first, etc...   back and forth.   

If you can edit files on the BBB, you can try editing the /opt/fpp/www/channeloutputs.php file, around line 2155 or so is where the drop down is created.   You can add a line of:

Code: [Select]

$values["64x32 1/8 Scan"] = "64x32x8";
If your panels work the same as the P10 1/4 scans, that should make it work.  Let me know and I can officially add it. 

Actually, I just noticed the 32x32 1/8 is there.    Select that and then double your matrix size.   The 64x32 are really just two 32x32 stuck together.   

Alternatively, if you want to send me two panels, I can try and get it working.

[derektoews]

The Scan rate definitely makes sense as why it's the issue; thank you for explaining that! I went in to the php file on the BBB and edited it so that it now has the 64/32/8 Scan rate on the drop down, but it still isn't working. On the FPP testing mode, I go down the line with pixels and it works great for the first 8, then skips 8, then the next 8 are good, skips another 8, till it runs out of the 64 on that row, then moves on to the next row and does the same thing until the halfway way point in the panel and then I can't  really follow a pattern of what it's doing... I'd be glad to send you some panels if you don't mind to see if you can figure it out.

[derektoews]

Question about this... figured instead of making all of these 5mm panels go to waste that I have (even though they refunded me my money on them)... even with a correctly working Octoscroller, it's just the pixel mapping on these 5mm panels that is all over the place. Is there a way to remap the pixels within the Matrix in the Layout? (I've seen a few forums on submodels / subbuffers but I don't think that's what I"m looking for) or would I have to create a custom model?... (I'm aware, this will probably take forever)

[dkulp]

On the BBB side, use one panel.  Even though it's 64x32x1/8, set it for 4x1 of 32x16 1/8 scan.

Can you create a simple 64x32 matrix in xLights, put a SLOW chase across the first strand, and video record the result?     The slow chase should allow me to see the order of the pixels that are needed.     Then put a slow chase on the second row.   

I'm back from my trip abroad so if you want to send me one panel, I can take a look.   I'll PM my address.

[dkulp]

I received the panel today.   Got a bunch of good news and some bad news... 

Good news is I was able to reproduce the problem you had.   I was also able to diagnose what is going on with this panel and update the code so that it works correctly.   The bad news is that the update will break the 32x16 1/4 scan.    Thus, I'm going to need some time to add some extra parameters and feed those all the way into the LEDScape library.

Basically, the 32x16 1/4 scan panels that I've seen all to an 8 pixel interleave (8 pixels from row 1, then 8 from row 5, back to 8 from row 1, etc...).  These panels do ALL 64 pixels from row 1, then ALL 64 from row 9.  It's kind of strange because that would mean the configuration is really dependent on panel size.   For example, 32x32 1/8 scan would have to have a different setting.    Anyway, it looks like for the panels that do NOT have twice the number of rows as the scan rate, I'll have to add a pixels per interleave setting or something.

Couple other notes:   this panel is BRIGHT.   That said, on brightness 10, I'm seeing some ghosting.  I'll try to track that down, but setting it brightness 9 makes the ghosting go away and is still VERY bright.   Not sure I'm going to worry about it too much.   

Currently, the "6 bit" and "7 bit" options are apparently broken for the non 2x scan panels.   I'll try and get that fixed as well.  Not sure it's needed though as these panels are SOOOO bright already and have very good scan rates.