On Sun, May 22, 2022 at 12:45 PM, John Quinn wrote:
......
?
Now my bizzare observation: upon probing the gate drive signal for Q103 from U108, I found that the problem appears to go away! Unfortunately, replacing U108 (a LM393N) made no difference. Shunting that line to ground with a 10pf cap had no effect, but adding a 1 meg pulldown resistor does appear to make the problem disappear.?
?
-----
Are resistors R122 and R143 approximately correct and can you trace connectivity from TP104 to gate of Q103? The resistors don't need to be accurate but voltage at gate of Q103 should follow TP104 closely. Since adding 1M makes a difference in this state, comparator is not pulling its output to -18V and gate voltage of Q103 is set by TP104.
TP104 voltage should follow the star connection wired to bottom of R125 (first star summing node?) within few mV. Although an error there should impact other places but perhaps there is some offset and adding 1M to ground creates a voltage divider to make the voltage about right for Q103 in the test conditions you tried.?
Ozan
?
?
