DC offset ± 48V situation on Yamaha AX 750

[xavlaur]

Hello!

On the diagram below I checked as diodes all transistors Q619 ... Q643 and they are ok but I always have either -48V or + 48V starting with the collector of Q619 and up to the base of Q640 / 641. Something escapes me, Any idea?
I attached measurements, I think the problem passes "through" the input (IC601A), at different restarts DC offset appears or as -40V, or as + 40V as I noted above, with orange I put the voltages when DC offset occurs "-", with green when it goes on "+". The voltages are lower compared to the normal ones because I supplied them ballasted by 2 lamps 48V / 5W (they light up poorly) because Q635 or Q636 heat up one by one (Ubc = close to supply voltage due DC offset), depending on how it goes, on + or -.
Something seems irregular with the voltages on the IC601A inputs / outputs.
Somehow it is unclear to me the role of IC601A combined with (Q602 & Q603)

[shakalnokturn]

IC601 is the "DC servo", check both supplies to it (I only see a note for the negative rail), then check it isn't open looped somewhere (R601, R727, R602, R604).

[xavlaur]

Due on pin 7 are -5V, IC601A internally it seems to be out of order. I'll come back with more details.

[xavlaur]

Q632, Q635, Q636 replaced with equivalents.
Q640, 641, Q642, Q643 replaced on the radiator but not connected on the board. If I connect them, the consumption increases and I have nothing to measure. If not connected, the emitter from Q635, Q636 becomes asymmetrical to the output rail. In this topology I think it must be 2 resistant as a load against the table.
All transistors Q602 .... Q651 checked as diodes, are ok.
The 48V / 5W bulbs - in series with ± 50V power supply, connected after electrolytes - light up very weakly but they also supply the functional channel (on the mainboard there is no power separation between the channel).

### As additional information, with R603 = 47K disconnected (IC601A seems ok now with R603 disconnected), the system starts at + 40V (bases Q635, Q636) so that after 20 seconds it starts to decrease rapidly (in about 3 seconds) until + 30V, then to quickly reverse the polarity (in about 2 seconds) decreasing to -40V, a value that remains constant for the rest of the measurements. Once stabilized at -40V, Q621 seems saturated.


I changed C616 = 100µF / 16V to 100µF / 100V, but no any changing in the above transient mode.
C603 47µV / 16V and C605 3.3µF / 50V are harder to reach, I haven't changed them at the moment. I don't see any other electrolytes to delay reversal.
The timing of the reversal is not related to protection relays triggering.
R673 = 220 ohms slightly discolored but at the nominal level.