Dot matrix display

[gregebert]

From another thread: can you make a clock out of a Williams DMD display?

The answer is yes, you can, but it will be considerably more complex than a nixie-tube clock. Refer to a typical datasheet ( http://www.vishay.com/docs/37006/apd128g.pdf ) to get an idea how the display operates. This particular display is multiplexed as 32 rows x 128 columns. In other words, each row of 128 pixels is displayed momentarily, then the next row, etc. The typical refresh rate is 70Hz, though the datasheet says you can run it up to 200Hz. The timing is pretty simple, and you will want a CPU to control it. Even at 200Hz refresh, the maximum pixel-rate is below 1Mhz.

I have one of these in my junkbox waiting to be built into something, though I dont know when I'll get around to it. Most likely it would be a clock that has digits that "roll" like an odometer or slot-machine when they change. Since I'm addicted to FPGA's, I'll probably use an FPGA to process bitmaps from an attached raspberry-pi.

I dont know the operating-life of the display, and I have some concerns about how much power they consume. 

[Golferron 54]

Thanks for the explanation ... looks like anything is possible, it is the execution is tough.  Oh well, at least I have a backup display in case one of my DMDs go bad.

Appreciate the time.

ron

[Tristan]

I started development of a clock using some similar displays (except 256x64 pixels).

https://sites.google.com/site/tristansideas/electronics/pinball-display-clock

Sadly a lack of time has prevented me from finishing the project so far. They can pretty much be driven from the SPI peripheral of most micros. I used an MSP430 for testing. This allowed each row of display data to be shifted in using one of the DMA channels with minimal interrupt/CPU usage just to latch each row keep track of the number of rows. I was able to achieve 4 levels of grey pretty easily. The chip on the dev board I used didn't have RAM to hold the entire frame buffer so I was using an external SPI SRAM (also DMA driven). This was not intended to be the long term design and I would use a part with enough RAM on board.

I've been thinking of driving them from a BeagleBone Black. Simply because the PRU's available would be capable of doing the real time processing required. Trammel Hudson used this method to drive a Mac-SE display and only minimal changes would be required to run a DMD.

https://trmm.net/Mac-SE_video

That's not to discourage the use of an FPGA. Even without one I was able to exceed the 200Hz maximum using the MSP430 (they still seem to work at > 200Hz). These displays do use a fair amount of power. They can get warm. They also have a tendency to produce a high pitched whine when operating. They did run in pinball machines in arcades pretty much non-stop but they do wear out eventually. There are LED based replacements available now but that lacks a certain something that you can only get from neon.

[Quixotic Nixotic]

I heard of a guy who is sort doing the reverse - he is taking the four levels of neon dot intensity found on DMD displays and mapping these to different colours on an RGB LED matrix, to get automatic pseudo-colouring. I bet he knows how the drivers work inside out.

As the original displays get older and fail, more people are turning to the colour LED displays, where someone has laboriously mapped every animation frame in colours. Not cheap, just shy of $400 for one of these.

John S