shimweasel

Power’s out to the DAGOC building, and as pigdog (the awful, awful Dell laptop) has a battery life of approximately 27 picoseconds, there’s not an awful lot I can do until it comes back on. Hence, blogging.

My current problem is that the Fat Controller is operating 8 times more slowly than it ought to. It has an 8MHz internal oscillator, allegedly, but if I pause for 1000 milliseconds, it takes 8 seconds to process it. This wouldn’t necessarily be a showstopper: the obvious hack would be to scale time intervals to an eighth of what you actually want. The snag is twofold:

1. Pulse Width Modulation is dependent on high-speed switching, both to look smooth and to get up to full brightness. (This is a common technique for getting a simple on-off light to simulate in-between brightness levels - if you want 10% brightness, you just have it on 10% of the time and off 90%. If you can’t switch fast, though, it’ll stay on too long at the low levels and not long enough at the high levels.)
2. Serial comms are very sensitive to timing: if it thinks it’s running at 9600 baud, but the world outside is passing by at 8 times normal speed, the little chip gets very confused.

I’m hacking around with asm to see if it’s a problem with the PicBasic compiler: will hopefully have an answer after lunch. This has all become somewhat urgent, as we thought this part of the problem was totally solved, and my boss is … somewhat worried, shall we say. Still, nothing like programming in a blind screaming panic to get the adrenaline pumping.