You've missed the point I've repeatedly made regarding these two possibly overlapping issues.

First of all, however, what has a supervisor to do with a 3-terminal regulator? How can it help protect it? Why would you inject a subject that's not germane to the discussion?

Just because something doesn't work, there's no reason to believe the problem, whatever it is, to be RESET-related. In this case, somebody had read some 8052.COM comments and drew the conclusion that whenever an 805x doesn't work properly, it's because there's a reset problem and a supervisor will fix it. Neither of these is necessarily true. I've generally gotten the impression that when an 805x circuit doesn't behave properly, it's because it was improperly coded, though that's not the case 100% of the time either.

Secondly, this problem, which is still undetermined, was not a reset-related problem, as far as I know, because no changes were made in the RESET circuitry, yet the occurrence of the problem was relieved. There simply was too much bulk capacitance on Vcc, however one characterizes that, for this particular circuit to operate properly, whatever that means.

I've long believed in lots of bypass members, but not necessarily in lots of capacitance on the distributed supply. I do believe there ought to be a 100 nF cap on each IC's power and gnd connections, if the driver clock is below ~10 MHz and 10 nF if it's higher. I believe that the component side of the board should be a pads-only layer with a contiguous poured plane connected to Vcc, and the "wiring" side should be a contiguous poured plane connected to GND. One can't always do that, but it really helps.

Additionally, I've not encountered any problem that was identifiably attributable to RESET in circuits the Vcc risetime and fall time of which were less than 10 milliseconds. In fact, I've actually not encountered any problem in any way directly attributable to RESET, though I don't doubt that they exist. I believe, but haven't proven, that one possible "fix" for flash-corruption as has been widely reported, can be implemented by clamping Vcc to GND within one oscillator cycle of the rising edge of RESET. One of these days, I'll find a way to prove/disprove that, but until I can find a way to cause the flash corruption thing to occur, I can't do much with it. A totem pole built of a complementary pair of mosfets might well do it. That would then allow one to use significantly more capacitance on Vcc and still have reasonable rise and fall times on Vcc, provided, of course that the circuit were properly configured. Doing that, BTW, would probably require that one protect the regulator with a diode, as I previously mentioned, but that's another matter.

Please explain to everyone why you, Erik, believe that a supervisor is at all germane to this thread? Do you think that power supplies using 3-terminal regulators should always have a supervisor? What benefit would a supervisor provide to the power supply? How would it help protect the regulator?

RE