Slightly different from the original RGB LED panels topic.
I finally got some of the panels a bit more than a week ago and I thought it would be fun to build my own version of the colour idea on a micro computer that I own. I got sidetracked. What if you don't need a cpu to calculate the frames and whatnot? Would it possibly be the cheapest way to replace broken DMD's?
First of all , I have limited electronic knowledge, and only the parts that I have in my house for the project, if you want to get involved post your findings/improvements here. I will include as much info as possible on building the circuit. The circuit is not perfect yet because I have run out of components and there is no electronic shop near me. The picture from the pinball machine needs syncing manually, I have tested STTNG,Shadow en TFTC. TFTC doesn't stay 100% in sync like the WB machines but I believe this can be overcome.
LEDDMD.jpg
LEDbb.jpg
Here is a video of STTNG and me syncing it.
OK let me explain what is happening. All the signals are almost the same for both the LED panels and a traditional plasma DMD.
The main difference is that the original DMD scans 32 rows of data per frame and the rows doesn't have to be addressed because they are scanned in the one after the other from 0 to 31 over and over. On the LED panels the scan is 2 portions of 16 and each one of the 16 lines must be addressed 0..15 (x2). The next semi big difference is that the pixel data on an original plasma DMD gets sent to the DMD as one signal, so one wire, because only one row of data is needed at a time, for the LED panels 2 rows of data get sent to the panel at once, one row in the top half and one row in the bottom half of the screen,so 2 wires (six actually, but we will just use 2).
One more problem to overcome is the frame start signal from the pinball machine to the DMD, this is actually a problem related to my solution for the other 2 problems. But more later.
The first 2 problems is easily solved by only supplying 1 line at a time to the led panels effectively changing them into 1/32 scan devices! This is accomplished by counting the lines and sending the first 16 rows to the top half of the screen and the second 16 to the bottom half. How? By using a counter that starts counting rows from 0-15 when the frame starts. Every time 16 is reached the signal gets sent to a different half of the LED panel. This is accomplished using a buffer for ic for one half and a transistor for the other half (all I had).
A second counter is used to address the rows.
The 3rd problem, the syncing is because of the way the framestart signal is used by the plasma DMD's effectively causing us to miss the address of the first row of every frame. The solution is to subtract 1 or add 15 to the address, but for this one needs a 4 bit full adder and I don't have one... at the moment. For now the signal has to be synced manually with a micro switch, click,click,click until the picture is in sync.