> Occam's Razor says that the preferred explanation is that the voltage
> regulator is oscillating.
>=20
> Whether this is the actual problem is another matter.
> (Occam only tells you the preferred answer, not necessarily the
> correct one :-) ).
>=20
> In this case I'd say there is a very high probability that Occam's
> simplistic reductionist dangerous approach has actually supplied the
> correct answer :-).
>=20
> What regulator are you using?
> Be CERTAIN that the regulator capacitor in and out specs are known and
> met. On some regulators Cout may not have too low or too high ESR.
>=20
> A look at the power rail with an oscilloscope would show what is
> really happening.

I tried several different regulators, including the LT1086-3.3V and
L7833. I'll make sure to try the specified capacitors shortly, but I
checked with an oscilloscope and didn't notice any oscillation at all.
The signal was extremely clean with no ripple I could see, even when
switching a small load with a power MOSFET on the same power supply
rail. I'm using a 0.1uF ceramic and a 100uF electrolytic on each supply
rail at the regulator, and a 0.1uF ceramic at each pair of IC power
supply pins.


> Another possibility is that something on board is introducing bad
> supply rail swings.
> Oscilloscope finds that too.

Checked again, and removed the switching power MOSFET and the switched
load. No change as far as I could tell to the quality of the power signal.


> Then there is possibly excess voltage on a non psu ic-ic path.
>=20
> If you have an external energy source whose voltage is higher than
> Vpsu it may be externally sourcing energy via some connection to it.
> This may pump up Vpsu (maybe via body diodes on some IC as a bonus) or
> inroduce AC energy.

Just a ~4.8-5.5V battery pack and one 3.3V linear regulator. Double
checked connections - all ICs are powered by the 3.3V regulated line
only. I also always use an appropriate Schottky diode with acceptable
reverse leakage (but perhaps a silicon diode would be more appropriate)
pointing into the regulator's Vin. Everything seems to be pretty well
isolated.

Just a note that these "hot IC" events are not common. They're rare
(maybe once or twice per experimentation session), but mysterious.


> As always a circuit would be a really really good idea and a
> description of intended functionality and connections may help. eg if
> this is an RFID transmitter that drives an inductor then what you are
> seeing would not be too unexpected. If this controls and connects to
> an RF plastic welder then what you are seeing would almost be more
> expected than not.

No RF or inductors involved anywhere. Mostly digital logic like PIC
micros and shift registers, with I2C and SPI interconnects, some
4000-series CMOS ICs, and small power MOSFET's like Olin's favourite
IRLML2502, and some of its P-Ch brothers.

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.