6522 weirdity

On Dec 30, 13:57, John Lawson wrote: Yes. Worst example I came across was a CMOS circuit which was designed by an "expert". It worked on a breadboard, but when transferred to the final PCB it misbehaved and eventually died of "dead chip". It turned out that a couple of unused inputs were left unconnected and floating. On the breadboard, that wasn't a problem, because it was dirty and there was enough leakage (a hundred megohms or so) to keep them in a reasonable and constant state. But the PCB was waterproofed, and they really did float, eventually leading to excess dissipation in the chip, and destruction of one of the gates. The reason is that a CMOS gate is basically two MOS transistor switches, one connecting the output to the power rail, and the other connecting it to ground. If the input floats halfway between logic levels, both turn on (at least, in older devices), and the current destroys one of the MOS transistors. Another example was a TTL circuit that ran fine at low to medium frequencies, but not at several megahertz. Floating input again; that sort of thing has a drastic effect on the speed of the gate. -- Pete Peter Turnbull Network Manager University of York