Reply #21 appears to indicate that reset goes low only briefly.
However apart from the presence of the capacitor which is charged when RTS goes low, I can't see any reason why the two different reset methods should behave differently.
Your correct again as usual. This thread has been going so long I forgot I even posted on it before.
I've got to admit this is one of the more unsual and puzzeling symptom problems that I can recall here.
Almost remind me of the rare but repeatable DTR auto-reset 'lock-up problem' that that led to arduino adding the reverse biased diode from reset pin to Vcc on Rev3 boards, which 'fixed' that problem.