Thanks to those of you who emailed me about this thread.

These statements are both incorrect. The P-ROC and associated driver boards are certainly capable of being updated just as fast as you're able to update them, whether it's at typical OS USB servicing speeds or optimized speeds (ie. 1ms cycle USB servicing times that you can achieve with an optimized linux kernel and USB driver).

The beauty of the P-ROC (and PD-LED board) is that you don't have to do any host side optimizations to achieve beautifully smooth fades on one LED or even on hundreds of LEDs simultaneously. The PD-LED is an intelligent LED controller that you can send commands to fade each LED from one color to another, even at different speeds per LED. This intelligent control happens at nanosecond resolutions, which is MUCH faster than you can get with host side commands even on an optimized system. That's why color fading with a P-ROC and one or more PD-LEDs appears perfectly smooth, whereas on systems with slower update resolutions, you can see the incremental steps as LEDs 'fade' from one color to another.

If you don't want to use the built-in functionality the P-ROC and PD-LED provide, you don't have to. You can push incremental color changes down to the hardware as fast as you want, and they'll work just fine. So if you really think you need to keep the low level coordination in software by doing the incremental color changes there, you certainly can. I'd propose there's a better way to architect your system though.

I'm happy to go into more detail, but I don't want to derail the thread.

All that said, the current version of the P-ROC does not plug directly into a WoZ without modifying the machine's hardware and/or wiring. You could replace the entire driver board with a P-ROC + PDBs if you really wanted to, but it's by no means a plug and play solution.

- Gerry
http://www.multimorphic.com

Yep, we spoke on the phone about a year ago. I remember it being a nice conversation.

The intelligence is in an FPGA, not a microcontroller. Yes, decisions are made every few nanoseconds, and all LEDs (even across PD-LEDs) are controlled independently and simultaneously. That means each and every LED in the system, even in a system with thousands of LEDs, can smoothly fade from one color to another at individual speeds at the same time with no software involvement other than setting up the fade params.

There's no concept of a 'function' in an FPGA. All of the logic is running in parallel, at the same time. Also no need to 'get time of day'. The timing in each circuit is fixed relative to the oscillator.

The P-ROC receives commands from the host over USB, and it can receive and process them as fast as the host can send them. So if you can get your host to send commands faster than 1000 times a second, receiving them and optionally responding to them is no problem for the FPGA. Multiply that by another 1000 for 1,000,000 commands a second (one command every microsecond), and you're still not taxing the FPGA.

- Gerry
http://www.multimorphic.com