I have received some HUB75E panels with very little additional information and no markings regarding the input pins and was wondering if anybody else has seen and/or used these panels. I have been trying to run them with a Teensy4.1 and the SmartLED Shield for Teensy 4 V5 but no luck with getting anything to light up so far. I have tested the shield and teensy on another panel I have bought before and that one worked without an issue, so I am sure it’s not an issue with the shield and teensy. I have so far not found any related post for the chipset they are using. The panels are P2 panels with 64x128 pixel dimension and the chipsets are marked with ICND2153. I am pretty new to HUB75 panels, so maybe I am missing something obvious here.
There are some markings on the display, such as HUB-75E right below the connector, K716-V3.2 and KSL 2045 B23 right next to each other. And on the other side (seen on the image) it says K716-128X64-32S-V3.2 and a sticker with P2SSMD210112-H106J/2A/343-A1.
You might try getting them to work with the hzeller Raspberry Pi project, or look around in their forums for any leads, as they have more support for exotic panels than this project. Please post back if you find anything
So there is an open issue on the project repo. Some person seemed to have made it work with an ESP32 by reverse engineering the protocol from analysing the signals from a working controller (I believe I read somewhere it was the HD-D16), see this comment. Unfortunately the video explaining everything is in Russian so not sure what exact information was provided. It does look pretty detailed to an uninformed person like me, though. There is some example C code for ESP32 linked in the video description and there is this repo which is an Arduino port of that example code with some more info.
Most others that tried reproducing it on other hardware seemed to still have issues such as flickering with it and so far there was no official support added for that chipset to the hzeller project. I don’t think anybody found any official documentation on the protocol of this chipset either.
So I guess ESP32 should be doable, but not sure if that is something the SmartMatrix library could support. Hope this was helpful
Thanks for the research. It’s unlikely that I’ll find time to dig into this anytime soon but I updated the wiki to show this panel type isn’t supported and added a link to this thread
Hi all!
This is my first message in the forum, I hope I’m not breaking any rules by replying to an old thread.
About ICN2153 - I played with a similar driver, FM6353. The main difference of such drivers is the presence of built-in memory. The driver stores 16bits brightness levels for each color channel of each pixel and can refresh the picture itself. Instead of uploading row data at regular intervals as on the traditional case, you need to upload the entire image at once and then only send a clock signal on the OE pin. This saves the controller from most of the work especially on the steady pictures, since on an unchanged picture it is enough to load the data once and then the driver will refresh it itself.
In addition to memory, as I mentioned above, the second characteristic of the driver s the need to supply an external clock signal through the OE line. The driver switches rows automatically upon receiving a certain number of OE clocks from controller.
Unfortunately, there is practically no documentation for these drivers. I managed to launch a 128x64 panel with FM6353 drivers on a stm32 bluepill board by analyzing HD15 controller signals in a logic analyzer, using the code from hzeller issue discussion, mentioned above. But I was not able to fully understand the control commands and just emulated those that I saw with the chinese controller. For example, I could not find which command changes the brightness.
As a conclusion - these panels are very interesting, but to work with it, a serious reworking of the library code is required.
Regards, Dmitry
very cool, thanks for writing this up. Yes, it’s about time for those panels to have memory and do self-refresh. The current way they’ve been working is suboptimal in so many ways…
