Consider a case, when the supervisor has a "better" bypass than the mcu, i.e. it "sees" a "quieter" power, with less fluctuations, and also slower changes upon powerup/down/glitch. It would then not reset the mcu when the mcu's power is outside the allowed range. Conversely, a supervisor with significantly poorer decoupling/conditioning would unnecessarily reset the mcu, which is not a good thing either.

Ideally, the supervisor should not have any extra bypass; it should "see" the very same voltage (and its variations) as the mcu "sees". Thus the supervisor should be placed as close to the mcu's power terminals as possible and both should share the same decoupling caps or whatever power conditioning is used. (This is also the reason why an otherwise *properly designed* internal supervisor is superior to external supervisor in this regard).

---
I'd like to recall here also the unfinished writeup concocted on resets and kin a couple of years ago.

JW