Very true. And it can be as simple as a small connector which gives you access to the RS232 port. But if you don't contemplate it from the start you'll be in a bad place.

Something mildly amusing happened last year on a project I was developing firmware for. The hardware was essentially all done before they even contacted me to develop firmware for it (go figure). Needless to say I needed to ask for a number of modifications to the hardware, one of which was precisely a way to get diagnostic/error logging out of the device. They rigged a few of their prototypes with the modifications we needed. When it came time to download the diagnostic information to a computer to check out what was going on, BOOM! The instant they connected the RS232 of the device to the PC there was a huge sparking event and the serial card on the PC died immediately. They wisely concluded that it was a fluke and proceeded to attach the device to yet another computer, BOOM. That entire computer died instantly. They then realized that maybe there was a problem (no kidding!), spent half an hour thinking about it, and eventually came up with a solution. They went to connect it to yet another PC and this time they let all the smoke out of that computer, too.

Anyway, I can't even remember exactly what the problem was. We had all been up for nearly 40 hours by that point so perhaps their mistakes were understandable under the circumstances. You know you're tired when you toast 3 or 4 computers and actually find it funny; nothing quite like the smell of burnt ICs in an office full of firmware developers and EEs at 4am. I'm just glad we didn't connect the device to my laptop--probably the only time I've been glad that my laptop doesn't have a serial port.

Regards,
Craig Steiner