Reincarnation of q3kled project (archived below). Same 128×128 (4x4x[64×64]) RGB LED panel, but based on Colorlight 5A-75B.

While said board has been reverse-engineered and can be used as a target for custom FPGA bitstream, no such bitstream has been prepared. However - using some disgusting python code (involving raw ethernet frames broadcasting) original bitstream can be controlled.

q3k's project to create a Zynq-based RGB LED Matrix driver.

Video of a 128x128 panel it action at 32c3.

This project targets the MYiR ZTurn board with a custom buffer interface board (to be released). The programmable logic exposes a 65-kbyte block RAM to Linux running on the PS via a full-speed AXI slave. The logic also loads data from the BRAM, gamma corrects it and outputs it to the RGB panel interface.

It works! But really needs to be re-written. Keep poking q3k on IRC until he does so.

Issues:

• Vertical glitches Fixed by switching buffers to 5V
• Low framerate Fixed by switching buffers to 5V and increasing pixel clock
• Horizontal ghosting
• No double buffering in hardware
• No vsync / interrupts
• No runtime configuration
• Kernel framebuffer access from userspace (FBDev/DirectFB) locks up whole system
• Gamma correction is poor, needs and actual ICC color profile correction
• Hardcoded parameters in kernel/mpv driver

Note: photo of display was taken before dank memes started becoming appropriated by hate groups.

• AXI4 slave IP Core
• Linux Kernel framebuffer driver
• mpv fork with direct LED output (bypassing kernel)