Whether or not this is debatable, this thread is about the relationship between Vcc capacitance and risk of regulator damage.

There are numerous reasons why one might want to have a large capacitance on Vcc. There are also several reasons why one should avoid that, not the least of which is the risk to the regulator if Vin drops quickly, perhaps due to an intelligently designed rectifier/preregulator circuit.

BTW, I'll bet YOU don't know how long it takes your circuit to pull Vcc down below the MCU's minimal Vdd/Vcc (whichever they call it) once power is switched off. Even with only 10 uF of capacitance on Vdd, if your load is just a few CMOS circuits, it may take quite a long time. If a guy is sitting at the bench, with a circuit that's "acting funny" he may not be willing to wait the requisite 3 minutes between power cycles for the Vdd level to drop a couple of volts.

I know that Avnet advertised their XPLA3 demo board would run on power supplied by two series-connected oranges and XILINX advertised their board would run on three series-connected apples, one of which apparently happened to be of a different color than the other two. My XCR256 board (The Avnet EVB) continues to display a count 32KHz down to 1 Hz for about 80 counts at 1 Hz after the unregulated 9 volt power is disconnected. It has two 10 uF tantalum cap's and one 33 uF electrolytic on the positive 3V3 regulator.

The reasonable thing to do to limit such effects is to switch power through a SPDT switch with one side grounded through a low-value (< 10 ohms) resistor and the other connected to the raw supply. Unfortunately this then requires one take steps to protect the regulator, e.g. a diode conducting from regulated to unregulated positive supply. For some folks, though, this is too much trouble.

RE