Switching PSU... Dead :o(

[LektroiD]

I appear to have a mostly dead power supply inside my Proteus 2500 synthesizer. I have been in touch with both the synthesizer manufacturer (E-MU) and the PSU manufacturer (Powedec) to see if I can get any schematics or info on this. Both results were fruitless, so I'm on my own with this one...

Here's what I'm getting from each of the pins.
1 2.956V DC
2 9.29V DC
3&4 (0V)
5&6 (0V)
7&8 Ground
9 (0V)
10 (0V)

Pin 1 appears is labelled SW on the main synthesizer PCB (labelled PWM on the PSU), I'm assuming this goes to the soft power switch... I'm wondering if this circuit is the problem, and somehow need to trick the PSU into activating the switch to start the PSU... Or if the problem is elsewhere in the PSU...

Hopefully someone here is knowledgeable with switching supplies and can point me in the right direction...

[Rerouter]

It sounds like you need to ground pin 1 to wake it up,

[LektroiD]

It sounds like you need to ground pin 1 to wake it up,

Tried that, unfortunately nothing...

[MosherIV]

Hi

Try looking up the part numbers of all the ICs in the SMPS, they should all be some kind of SMPS controlled chip.
The data sheets should have some kind of simple example circuit, you might be able to figure out each circuit from those.

If nothing else, the data sheets should give you pins out for each device and you should be able to probe each device to figure which ones are not working.

BE VERY CAREFUL !!

You will be probing 340V DC (in some cases) !!

[Chalcogenide]

Is that the infamous brown glue that turns corrosive and conductive with time? Try getting rid of it as much as you can.

[JFJ]

Try looking up the part numbers of all the ICs in the SMPS...

From picture 3, the primary controller is a KA3842A.

[LektroiD]

Try looking up the part numbers of all the ICs in the SMPS...

From picture 3, the primary controller is a KA3842A.

Strange, I'm getting -165VDC on every pin except pin 1, which is 0V... I did a continuity test and pin 1 is not tied to ground.

[MosherIV]

Strange, I'm getting -165VDC on every pin except pin 1, which is 0V... I did a continuity test and pin 1 is not tied to ground.

The 0V/GND might be isolated (through a high freq transformer) between the high voltage side and the low voltage output side.

[LektroiD]

...but should there be -165VDC on all the other pins? I looked at the datasheet and it seems to be rated for +15VDC... I also looked for a replacement, but all that comes up is a UC3842A, would this be an OK drop in replacement (if that is the faulty part)?

[darrellg]

See this video regarding that brown crap on the PCB:

[MosherIV]

...but should there be -165VDC on all the other pins? I looked at the datasheet and it seems to be rated for +15VDC.

Sorry, I did not explain very well.

Where are you measuring ? Where have you put the 0V probe of the DMM?
If you put it on the 0V of the PSU output - then it is probably not phyically connected to 0V on the input side.
Find the 0V/GND on the controller chip and use that as the 0V/GND

Hopefully you will get more sensible results.

If the device is only rated for 15V DC - I would have expected 'the magic smoke' to escape.

[strawberry]

It is standard multi output flyback smps converter. You can turn it on manualy by shorting that (unknown part designator) optoisolator on to main PCB directly.
-Remove that yellowbrownish gunky thing from PCB, sometimes it is resistive.
-Any 3842 should work
-check R17
-check output for short circuit

[LektroiD]

Many thanks for all of your help so far...

Reworked:
All visible brown gunk removed.
Recapped all the caps that appeared cold on the Flir (I'm awaiting delivery of more 470µF/25V caps before I do the rest)
Checked R17, OK

Test results:
+5V shorted to GND
power supply still dead.

[LektroiD]

**UPDATE**

I found some 470µF caps, so it is now completely recapped.

I retouched all the solder joints (the connector looked dry).

I tried shorting each of the optos at both resistor and LED sides...

Same results as before 

[strawberry]

Nippon chemicon caps are good brand.
2.Check KA3842 supply treshold voltage 16V or something between pin5 and pin7, (when PS turned on pin1 to ground). If it is false, then Q2(TO92 can) is failed or bias supply for that  IC is dead ,
1.What hapen on that opto on main board SFH615 pins 1 and 2, if PSon pin1 is lifted and set back, if true then control is ok

I dont know if this particular design use some 100k resistor from 320Vdc or standby supply to feed KA3842 through Q2 propably?

[xavier60]

Many thanks for all of your help so far...

Reworked:
All visible brown gunk removed.
Recapped all the caps that appeared cold on the Flir (I'm awaiting delivery of more 470µF/25V caps before I do the rest)
Checked R17, OK

Test results:
+5V shorted to GND
power supply still dead.

What is causing the +5V short to ground? Output diodes can go short circuited.

[LektroiD]

Hi, I checked the diodes, all seem to be ok (tested out of circuit).

I bought another switching PSU from eBay... On the meter the 12V fluctuates between +5v & +7V, about once per second. I scoped it out and this is a visible sawtooth in the minus, and a blip in the plus, possibly a spike and cap discharge. I don't know why my scope is showing way higher than the 5-7V on the meter...

I'm just wondering if this is normal for a switching PSU? I don't want to risk it inside my synth if it's going to cause further damage.

[xavier60]

From here it looks like the 12v rails are pulsing over 100V. Have you set the probe attenuation setting correctly? Try a 10 ohm load on the 5V rail.

[carl_lab]

Is that the infamous brown glue that turns corrosive and conductive with time? Try getting rid of it as much as you can.

That's exactly what I thought when I saw the pictures.
https://www.eevblog.com/forum/repair/the-brown-corrosive-gunk-of-death/msg984134/#msg984134

[LektroiD]

From here it looks like the 12v rails are pulsing over 100V. Have you set the probe attenuation setting correctly? Try a 10 ohm load on the 5V rail.

I noticed it was ticking on each spike (although there doesn't appear to be a relay anywhere in the PSU). I don't have a resistive load box in my setup, and a spare 1/4W resistor would swiftly result in a puff of magic smoke, so I attached a 13.5V incandescent lamp between the 5V rails. The ticking stopped and it lit the bulb nicely with a steady +5V output. However, the +12V was giving out +17.2V. I understand this is unloaded, but that's quite a difference. Does this sound normal?

Next step is to wire the synth up and see if it powers up. I have a 5V to 3.3V converter here, but I don't have anything to attach to the 10V rail; or the SW - although I'm assuming it won't need the SW pin attached as that is possibly just feeding the switch information back to the PSU. Maybe the +10V is only needed for the switch?

I guess if it doesn't power up without the +10V attached, I could always inject +10V from my bench supply. If that works I'll throw in a 7810 to feed it...

I'll be back with the results in due course...

[strawberry]

It is normal if one output is owerloaded, but 10% deaviation in normal operation in specs. Light bulb is nonlinear resistor.
Remove capacitors from TL431, if that help