I'm not an EE, just a hobbyist. The circuit shown in the linked picture shows 12V power being switched on by the alarm trigger of a DS3231 RTC. The INT/SQW pin of the DS3231 is open drain which goes low when the alarm time is reached. But the datasheet is clear that a maximum of 5.5V can be applied to this pin, so for 12V the pin cannot drive the P-channel mosfet gate directly. Instead, an intermediate N-channel mosfet is added, and the INT/SQW pin is connected to its source. When it goes low, Vgs goes to 3V, which turns on the N-channel, which in turn turns on the P-channel. But before the alarm is triggered, the source will be floating if the G/S resistor R2 is removed.
It's been suggested on the EEVblog forum that R2 isn't needed, and indeed, I have breadboarded the circuit, and it works fine with or without R2. But I want to be sure I understand why letting the source float is ok, if that's the case.
As I understand it, if noise that might turn on the mosfet is picked up at the floating source, the mosfet will turn on enough to immediately bring the source back high and turn it back off. I don't know whether the source has any capacitance, but if it does, the resulting high charge there has no way to bleed off, so the mosfet should stay off, at least for a while. But in my test, there was no indication that the P-channel mosfet ever turned on, even briefly, so any noise didn't ever make it that far.
And any high-voltage noise above about 12.7V would be dissipated through the body diode, so no damage is likely to occur.
Does that all make sense, or is there more to the story? It would be nice to eliminate R2 so that no coin cell current would be needed to keep the power on - it's pretty much a free lunch. I think the circuit is also nifty in that it allows you to turn on an N-channel mosfet by bringing a line low, which is not what you normally think of.
Anyway, I'd like to know what actual designers think of this, particularly the N-channel floating source.