[LONG] Another switching PSU question...

Hi !
I gave up on the PCServer 325 PSU, since at 99% was damaged a custom board.
It was a standard ATX with different connector, so it was enough to adapt the
failing PSU connectors to a "standard" (I chose 285W instead of 230W,
just to be sure) and it worked.

Now, there's a another small PSU which I can't substitute, belonging to a
DECServer 700.
I followed your suggestions and hints.
The only ICs are
UC3844N PWM
SFH601-3 Opto
IRFPE40
AVS10 kit (AVS10CB, AVS1AC) Automatic voltage switch

I checked voltages (after the usual short-circuit hunt), and around PWM
and Opto there's nothing or some mV.
Voltage stops around the AVS10 kit.
The Triac and the surrounding high valued resistances are really hot.
I checked the datasheet of the AVS10CB controller against measured values,
finding something odd.
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00000665.pdf

I haven't figured well, yet, AVS10 kit behaviour.
By the way, according to the datasheet typical app, Vm (main) is connected
through a partitor to the AC main.
Following the circuit is not easy, due to plane changes
So, Vm should be an AC value, around 4 V.
I found 2.8V.
DC values are around 150 V.
I checked R1 (exactly 1MOhm) and I changed, just to be sure again.
Can't identify still R2, the scheme differs.
Removing R2, I should have no Vm: I found 4V, so the PSU circuit differs from
the datasheet typical one.
Since on the two big caps after the rectifier bridge I have 150V, I should assume
the AVS kit is working, anyway, even if the associated Triac is really
*very* hot.
On the IRFPE40 pins I found the whole rectified voltage, over 300V.
Then I lose track of the signals.
Next step would be to follow 3844 supply line, in order to see where it
comes from.
Any hint is welcome !

 

Removing R1, 1MOhm, of course

 

A little history is in order. The original PC - XT - AT power
supplies had a 110/220 switch which should be in the appropriate
position according to the local AC supply. In the simplest terms the
switch changed the rectifier configuration from a bridge (for 220 VAC
in) to a voltage doubler for 110 VAC in. Note the in either case the
electrolytic caps are in series.

The original design using a mechanical switch became a problem - if
set to 220V the power supply would not work with a 110 VAC input. If
set to 110 volts the power supply would be damaged when connected to
220VAC mains. The AVS10 kit automates that switching. Of course, in
most modern power supplies the SMPS components have been selected to
operate over a 90 - 250 VAC input range.

I tend to troubleshoot by looking at critical points. You SHOULD have
1.4 x your AC input voltage across the series electrolytics if the AC
input is in the range of 200 - 240 volts, 2.8 x the AC input voltage
if the input is in tha range of 100 - 120 volts. For a nominal AC
input of 120 VAC I would expect about 165 VDC across each cap, 330 VDC
across the pair.

PlainBill

 

I tend to troubleshoot by looking at critical points. You SHOULD have

On one capacitor there is 156V, 165V on the other, and it should be correct ...

 

Well, I made some more measurements.
The IRFPE40 shows the following voltages:
VDS=312V (max. 800V)
VGS=0 (?????)

What's the IRFPE40 function ?
Looking at the PCB it seems that the 3844 supply voltage comes from it...

 

Are you looking at this with a scope?
I've seen many instances, especially in monitors
where a cap has gone high resistance. You measure the average
volts with a voltmeter, cause that's what's being regulated.
But look at it with a scope and you see high peak voltages
that make stuff HOT...just before it explodes.

Probing around in the line side of a switcher with a scope
is a good way to make a dead psu and a dead scope and a dead
technician. I use fully isolated probes.

 

IMO you should not be experimenting with R1 or R2. This potential
divider senses the line voltage. If you make the AVS10 believe that
the line voltage is 110VAC when it is fact 220VAC, then you will have
660VDC across the two capacitors rather than 330VDC.

- Franc Zabkar

 

I would expect that the MOSFET is the chopper. Its gate would be
driven by the UC3844 PWM controller.

- Franc Zabkar

 

If your application is 220VAC only, then you could disconnect the
AVS10 section completely. Just remove the triac, or cut the link to
the junction of the bulk capacitors.

- Franc Zabkar

 

The PSU was repaired, with the aid of a friend technician.
The PWM was faulty, together with a couple of resistors.
Then it started and went down again.
All caps were desoldered, finding that many of them had leaked (the
PCB under them was wet).
Checked with the ESR meters, they looked good.
Changed all caps resolved the remaining troubles.

Thanks to you all !