"John A. Thomas" wrote:
Perry Metzger asks if the hardware random-number generator I described in a post just before the list crashed could be modified to use serial output.
You don't even need to build a serial interface, per se, on the RNG device. So long as you have output levels consistent with RS-232, you can push a clock down DSR and the random bits down CTS. The PC (or any computer for that fact) has no problem looking at these signals, so although specific software is required, it doesn't require a special async device driver. Just as with a parallel interface, you take the next sample when an inversion of the clock signal is detected. If the computer in question is too slow, the only side effect is a loss of bits as clock inversions as missed. If the computer is too fast, it just hangs around a while. If the RNG device has a negative rail (>=5v), no extra components are needed, otherwise you'll need to generate it, something like a Maxim RS-232 line driver (which only requires +5v/gnd and generates +/- RS-232 voltages) will do the trick. Certainly cheaper than a UART (which would require line drivers anyway). I remember a ham friend of mine telling me of a certain syncronous TNC that communicates with a PC device driver using CTS/RTS, I thought it was a cool hack at the time :-). cheers, Matthew. -- Matthew Gream <M.Gream@uts.edu.au> -- Consent Technologies, (02) 821-2043 Disclaimer: I'm only a student at UTS