16:35 <russss> think I might need to get the Glasgow out for this one... https://led.limehouselabs.org/docs/tiles/dmg1083/

16:36 <russss> "it's just a standard LED matrix interface, how hard can it be?" etc

22:19 <Attie[m]> 🤞

22:21 <Attie[m]> russss i've not looked in detail, but aside from the pinout difference, it may well be HUB75-compatible-ish

22:21 <Attie[m]> https://github.com/GlasgowEmbedded/glasgow/blob/main/software/glasgow/applet/video/hub75_output/__init__.py

22:21 <russss> yeaah that's what I thought

22:21 <russss> but please enjoy this https://led.limehouselabs.org/docs/tiles/dmg1083/#driving

22:22 <russss> the LED shift register driver chips have framebuffers. And configuration registers.

22:23 <Attie[m]> oh neat - "a whole frame’s worth of pixels is clocked into this panel at once, and held in a SRAM double-buffer on the MBI5153 chips"

22:24 <russss> yeah it's pretty clever

22:24 <esden[m]> yeah, that is pretty normal for more recent chips. The page you linked says that it is more tricky than it seems 😉

22:24 <russss> (I wrote this page btw, I have been slowly discovering the complexity)

22:24 <esden[m]> I had the questionable pleasure running into these chips.

22:25 <esden[m]> the issue is that they tend to have a bunch of undocumented features and registers

22:25 <esden[m]> just like LCD panel driver chips

22:25 <russss> we have a commercial display driver system which...kind of mostly works with them https://led.limehouselabs.org/docs/tiles/dmg1083/20230915-test.jpg

22:26 <russss> waiting on a different receiver card for it

22:26 <esden[m]> It would be fantastic if you get it to work and document it though.

22:26 <russss> the main issue I have realised is that the Glasgow doesn't have enough pins, and nor does my Saleae

22:26 <Attie[m]> is "one MBI5153 frame" the same as "one scan line", or does it deal with operating that side for things too, for a genuine 78x78 frame?

22:26 <esden[m]> I never was able to get the commercial driver stuff to work...

22:26 <esden[m]> especially for the chinese panels and driver boards

22:26 <esden[m]> the settings are extremely obscure

22:27 <esden[m]> sorry I mean "opaque"

22:28 <russss> I think for this chip it seems to be...adequately documented

22:29 <russss> like, this is actually relatively good, though it's not for that exact model https://led.limehouselabs.org/datasheets/MBI5051-52-53-AN.pdf

22:29 <russss> I think/hope it's identical though

22:30 <russss> Attie[m]: each driver chip stores (in this case) 20 scan lines of 16x 14-bit channel values

22:31 <russss> then it cycles through the scan lines

22:31 <russss> and there are 20 driver chips per colour on the board. The new data is clocked in to all the chips shift register style, and then you send a "VSYNC" command to flip all the buffers at once.

22:34 <esden[m]> Ahh yeah that is not one of the chinese chips that I ran into before.

22:35 <esden[m]> those were severely under documented

22:35 <russss> yeah it's Taiwanese, heh

22:36 <russss> the tiles come from https://www.digiled.com/ who make high-end LED screens. We have something like 800 of them. They are really nice at giving their old stuff away.

22:36 <esden[m]> Yeah ok, it is not some low cost stuff. That is nice. 🙂

22:37 <esden[m]> So virtual production studio aka. the volume when? 😄

22:37 <russss> we visited their showroom and I saw their latest screen which has 1.25mm pixel pitch. It's one single monolithic 150x150mm COB array.

22:38 <russss> and those are just magnetically mounted in the chassis, very satisfying. I failed to take a photo.

22:39 <esden[m]> Ahh yes, I saw that this is what they are doing now. The nice thing is one can use a big magnet to remove a tile out of the wall to fix/replace it.

22:39 <russss> they have these...modified Dyson vacuums for that https://usercontent.irccloud-cdn.com/file/Ef4ehhKV/PXL_20230912_122942500(1).jpg

22:39 <esden[m]> ahh that is nice!

22:42 <Attie[m]> hah, that's awesome! (vacuums)