What is wrong with this vintage SMPS from 1985 ?

[rodcastler]

Can you give me a hand with this Sony Power Supply from 1985?

Symptoms:
- 5V present at the 5v line, but only 9.7v at the 12v line.
- If I apply any load to any of the rails (even as low as 200mA) voltage goes down to less than 1v.

What I have tried:
- Checked Caps.
- Checked rectifying diodes.
- Fixed a couple of broken traces.
- Checked the output power transistors.
- Checked (MB3759) with separate testing circuit: Outputs of the MB3759 chip show perfect square waves under the test circuit however,
- Outputs of the MB3579 seem odd when back in PS circuit.

Not sure of what to try next. Any suggestions?

I'd appreciate any help!

thank you

[rodcastler]

These are the output waveforms  I don't believe are right (output at the MB3759 pins C1 and C2, not to be confused with the power supply output rails). Maybe they are normal but they are very different from the good looking square waves when tested off the power supply under the test circuit.

[shakalnokturn]

What voltages do you have for primary rectified voltage and supply to PWM (pin 12)?

[rodcastler]

I wrote the readings on the photo hoping it gets easier for you.

[tunk]

  What voltages do you have for primary rectified voltage ..

I guess shakalnokturn thinks about the voltage
across the biggest capacitor (C109?) - it should
be around 300Vdc.

[rodcastler]

Yes, that is correct. The rectified mains voltage across that cap is 300v, which seems alright to me being in a 220v country.
Also, the PS itself is rated 100-240vAC.

[shakalnokturn]

I'm having trouble understanding where the PWM gets its startup voltage from, the only way I can see is through T103.

What's going on on the small vertical PCB?

Steady state supply is through D204, D205 and Q202 as regulation so I assume D212 is a 9V1 Zener.

The waveforms you show seem reasonable for an unloaded supply, the main regulation will be on the +5V as that will have the heaviest load, in normal conditions with a loaded +5V I'd expect the +12V rail to come up to specification.
I'm not even sure that 9.7V unloaded is abnormal.

Now why would it not stand any loading...

A leaky secondary diode or capacitor loading it even without
normal load? (You should notice heat in this case.)

Excessive resistance on primary side, the first culprit that comes to mind is R101 gone high. (Worth checking...)

A problem on the primary current sense that seems to be built around T102.

[rodcastler]

It seems to me that the PWM gets its startup voltage from the large transformer per this latest image.

The small vertical PCB (I colored pink) has few components: some resistors, caps, diodes and two transistors. I assume this board's mission is to drive the power transistors that are located underneath the gray rubber cover at the bottom. To that end, I believe T103's primary is at the upper side but I am not sure about that.

I followed your suggestion and I realized that when the 5V rail is loaded and goes down to 1.x volts, the 12v rail actually goes up to 16v unloaded, although I know it doesn't stand any load.

A leaky secondary diode or capacitor loading it even without normal load? (You should notice heat in this case.) 

Thermal camera shows normal heat activity, impedance between 5v and gnd is above 500kohm before the output caps begin to charge.

the first culprit that comes to mind is R101 gone high. (Worth checking...)

I checked, and it's well within spec (3.3 ohm).

A problem on the primary current sense that seems to be built around T102.

That one shows no activity at the 1-2 winding. I have not touched 3-4 with the scope, being on the high side. That being said both 1-2 and 3-4 show expected continuity on T102.

[shakalnokturn]

Do the PWM C1/C2 outputs change in duty cycle at all under load?

It seems that this supply has output over voltage protection (OVP) on both rails through Zeners D217 and D219.
Is it possible that loading 5V causes 12V to hit the OVP which also limits the power output? Have you tried loading both outputs simultaneously?
(With 5V output loaded down to "1.x Volts" will 5V output increase if you add some loading to 12V output?)

[rodcastler]

you have just nailed it my friend.

As it turns out, it's now fixed and I am not sure how.

I was following your latest advise of loading both rails at the same time. The rails were all over the place on the scope UNTIL the 5v rail was loaded with 800mA at the same time the 12v rail was loaded with 500mA. That brought them both to rock solid specs and I could successfully bring them both up to 1A with no issues.

I tried it again with the original device (which DID NOT power up before) and they fired up normally, for the first time with this PS.

These are my preliminary conclusions:

1.- During the diagnosis phase I de-soldered, measured and re-soldered lots of components, not to mention I socketed the PWM controller.
Somehow the PS got fixed at some point. I only replaced two caps out of caution, but the ones removed are still within specs.

2.- I failed to detect it was actually fixed, since I kept applying 200-300mA loads to the rails individually, which kept the rails out of spec.  Is this expected? I have fixed SMPS before, without ever requiring such as narrow load scenario.


I want to thank shakalnokturn for his dedication and help. I did learn way more about this because of his comments than I did with my prior analysis.

[shakalnokturn]

You're probably right, the resoldering and repairing cut tracks was likely the fix.

Because of the way this SMPS is designed:
Regulation loop with a proportion taken from each rail and tailored to the end use current requirements.
Over voltage protection on both rails and inadequate test loading fooled you into thinking it was still defective.