Juno-106 Restoration story/Power Supply woes

[Mooger5]

Poly 1 selected in tezt mode so all voices sound at once, too?

[Mooger5]

Nvm, you got another calibrated unit which is flawless. Damned.

[Mooger5]

I can only recommend to learn oscilloscope basics to monitor the dac and processor behaviour and compare with the healthy unit.

I might get back to this irritating thing if I remember of anything barely useful.

[The LT]

I can only recommend to learn oscilloscope basics to monitor the dac and processor behaviour and compare with the healthy unit.

I might get back to this irritating thing if I remember of anything barely useful.

Unfortunately, my oscilloscope is a puny Jyetech DSO shell.
I've swapped the CPU for a spare and IC23,24,26. Same thing.

I give up.

[The LT]

Poly 1 selected in tezt mode so all voices sound at once, too?

During Poly 1 mode, the voices immediately fall to lower duty cycle. You can hear them jump as soon as you press the key. During revolver 1-2 mode, it gets shifted after 6 key presses.

[Mooger5]

Wait... I'm not sure I understand how it's going on afterall. It begins to sound like it's a bad or wrongly set potentiometer during calibration...
Would you mind explaining step by step what happens?
Thanks.

[The LT]

Wait... I'm not sure I understand how it's going on afterall. It begins to sound like it's a bad or wrongly set potentiometer during calibration...
Would you mind explaining step by step what happens?
Thanks.

Step-by-step:

Unit turn-on, into test mode.
Load bank 5.
Switch on Poly 1 Unison. Voices IMMEDIATELY fall to 20% and less PWM on C4. Out of bounds for VR-adjustment.
Switch to Poly 1-2 Revolver. Press C6. Press and hold C4. Normal PWM of 50%, well within VR adjustment bounds.
Continue pressing C4. Next voices also hold PWM on 50% without any problems UNTIL we hit it for the 6th time. Then, PWM jumps back to around 20% on C4, which is distinctly heard.

I will shoot a video later.

[Mooger5]

When voices immediately fall to 20%, try moving the PW fader down to Zero...

[The LT]

When voices immediately fall to 20%, try moving the PW fader down to Zero...

It's not something obvious as this. I know it sounds like it, but the PW fader is down at zero, and yes, I've pulled it down. When I do raise it, it only further lowers the duty cycle and doesn't return it to 50%. Faders are new, it's not the panel board. We've measured the CV to the 5534 waveshapers and it was 6.01, remember?

[Mooger5]

Yeah I remember.

Unfixable. Smash it!

[Mooger5]

Joking. Sort of. Wouldn´t be surprised if it was some stupid little thing so obvious that we´re overlooking. If you ever find a culprit please report.

[Mooger5]

Just noticed in the video the 75% duty cycle has a negative bias when the output should be symmetrical. So another part that´s common to all voices is TR7 and associated resistors. It´s labeled Q7 in the PCB. I´m only guessing that it´s locally regulating a negative voltage to the demux. Maybe the transistor is leaky?