Hello, It's been a few weeks since I posted my first question to this list. It was a dumb one about interrupt handling, but I was amazed by the number of helpful responses I got (and no flames :-). If I missed acknowledging any individual answer (I think I got most of them), then apologies and thanks. Thanks also to Eurodis HB Electronics Ltd, who sent me an Embedded Control Handbook free of charge, even though I explained I am only a hobbyist. If I ever go commercial, they'll be at the top of my list of preferred suppliers. Here's this month's dumb question:- How precise does the bit timing have to be for serial ASCII output (e.g. to a COM port on a PC). The examples in the handbook seem to use simple equations like / , but I have worked out that 1 bit at 9600 bps is only 104uS. With a 4MHz xtal, each instruction takes roughly 1% of the bit time. There has to be a different path through the program depending on whether the next bit to be sent is a '0' or a '1'. With KISS coding, this can be up to 4 instructions (i.e. 4% of bit length at 9600 bps) - more for start & stop bits. It's possible to put a few extra nop's and goto's in the program to balance things up, but they look ugly and take up memory. Which should I do; go for the KISS approach and accept variations in bit timings, or write ugly code and get the timings right? Thanks in advance for advice, Mike ================================================================= Mike Jones. || e-mail (home) mike@newjay.win-uk.net Newport, Gwent, UK || (work) mike.jones@ons.gov.uk I thought PCs were a pain in the DOS, till I discovered Linux... ================================================================= =