strange behaviour of the damaged smps with uc3845

[excalibur_the_sword]

i am trying to fix a 400w rainproof smps that uses a uc3845 controller ic.
it just will not turn on when plugged into the mains.
but:
when i connect 12v to the output, the transistor sides of the two optocouplers show a resistance of under 500 ohms. in the picture, they are next to the main transformer. (So the feedback section seems to be OK).

when an external voltage is applied to the controller IC, it turns on at ~8.6V. Vref=5V and the output is stabilised at 105kHz and reaches the gate of the mosfets. (So the controller IC and the path to the gates are fine).

But when I connect it to the mains power, it does not turn on. The voltage on the Vi pin of the UC3845 is 8.9V (no fluctuation), but it will not start. Vref is ~0.5V and no square wave on the output pin.

The strange thing is: when i disconnect the mains and the bulk capacitors are charged, and i apply an external voltage of 8.6V to the controller ic, it starts pulsing and smps turns on and stabilises some voltages at the output. even the output led turns on. the voltage decreases until the bulk capacitors are fully discharged.

can anyone help me find the problem?
Thank you.

[Harry_22]

Hi!

Could you make two good PCB photos from top and bottom.

[dustooff]

Hi,
if Vref is being held low, the controller will stay inactive.
Vref is only able to source 30mA max.

from the data:
When VCC is greater than 1 V and less than the UVLO threshold, a 5-kΩ resistor pulls VREF to ground. VREF
can be used as a logic output indicating power-system status because when VCC is lower than the UVLO
threshold, VREF is held low.

[inse]

Are you sure it’s not a supply issue of the controller?
It usually has two supply sources: for startup via resistive divider from mains and by an auxiliary winding on the transformer during operation.
Typical faults are: startup resistor high ohmic, Vin filter cap dead.

[excalibur_the_sword]

Thank tou all.
main problem solved by replacing Vcc and output capacitors.
but still the output fluctuates, and needs more investigation.

[inse]

Secondary cap check

[excalibur_the_sword]

sorry for late reply.
i replaced all suspicious caps. but still it will not generate Vcc for controller.
I charged main caps and applied 9 volts to controller, it stated and smps produced stable 12 volts.
but when connecting to the mains power, it fluctuates.
I auspect the Vcc circut from axillary winding to the Vcc pin.
axillary winding is ok, as it output has a square wave of 104khz but stopes and starts periodically as controller restarts again and again. i checked the diode after Axillary winding. ok. Vcc cap ok.

[minifloat]

Hey,
Please also check these two
* the startup resistor
* the current limit resistor in series with the aux winding
See also in attachment screenshot taken from TI Datasheet.

I was assuming, you are testing the PSU in idle condition, which would be fairly low duty cycle. In this condition, the aux winding cannot help much, and a failing startup resistor can cause the issues seen.

Same happens for startup being designed high-ohmic anyway (only for the first few cycles) and the aux winding current limit resistor out of spec high.


br, mf

[excalibur_the_sword]

thanks. unfortunately both of them are Ok 

[minifloat]

What about the current feedback shunt resistors? From your pic, the PCB substrate looks a bit cooked around them...

[excalibur_the_sword]

both are ok. 22ohms.
i attached picture of axillary winding output in three zoom levels.
and Vcc capacitor voltage waveform
when main caps are charged and i apply external 9 volts to controller, psu produces stable 12 volts.
when i slowly increase the input voltage, controller starts switching at 8.63V. But when it is connected to the mains, it will not continue to work, even the voltage reaches 8.77V.
i do not know why the burning sign is there. there is nothing there except those two resistors, and they are both ok. axillary winding is also, at the other side of transformer and it is unlikely to be damaged.
 so i think the problem is in startup circuit. but i don't know where it is.

[minifloat]

22Ω seems a bit too high at the DC/secondary current sense (where the PCB is charcoal black).
For primary side current sense, the datasheet exemplary circuit uses 0.75Ω.

I did not find out what the transformer T2 is doing, looks a bit odd though...

br, mf

[Harry_22]

Both 22Ω in series with capacitors. These are snubbers.

[inse]

The normal supply voltage should be higher than barely above the UVLO, I would guess 12..15V.
Check whether it’s about the rectifier diode, the current limiting resistor or a zener diode, if present.
You have already ruled out the supply filter cap.
There have been cases with damaged transformers as well.
What is the second optocoupler for, are you able to reverse engineer that?
Something like over current feedback, maybe.

[excalibur_the_sword]

The normal supply voltage should be higher than barely above the UVLO, I would guess 12..15V.
Check whether it’s about the rectifier diode, the current limiting resistor or a zener diode, if present.
You have already ruled out the supply filter cap.
There have been cases with damaged transformers as well.
What is the second optocoupler for, are you able to reverse engineer that?
Something like over current feedback, maybe.

thanks.
typical start voltage for 3845 is 8.4v.
almost checked every component and i think the last suspect is transformer.
i couldn't figure out what the second optocoupler is for.
if thats any issue except than startup circuit, it should act even when i apply external Vcc.

[inse]

Sorry, the last sentence left me confused:
if thats any issue except than startup circuit, it should act even when i apply external Vcc.

I thought it was working with Vcc supplied

[excalibur_the_sword]

sorry. that's magic of my awful english.
as i mentioned in previous posts, when i apply external Vcc to controller and bulk caps are cherged, psu generates stable 12V until the cas are discharged. so if there is any problem in current sense or other parts, that cause restart, controller should restart not only when connected to the mains, but also when started using external voltage.
so, i think the issue, whatever it is, should be in startup circut. i checked every component from winding to the Vcc pin. everything was ok. just the winding itself remained unchecked.

[inse]

How much current is the controller drawing from external supply?
Does that match to the data sheet?

[excalibur_the_sword]

i don't remeber. but i tried new one. result was the same.

[shakalnokturn]

If you've already checked or replaced the diode that supplies UC3845 from the auxiliary winding, try loading the PSU's 12V output with maybe 10 Ohms, some PSU's just won't run correctly unloaded.
Maybe you're at the limit of UVLO and the ripple is bothering the UC3845, loading the regulated output will bring the UC3845's Vcc higher if the switching kicks in at all.

[shakalnokturn]

...and this became?
Electric death ?
A plasma ball ?

[MathWizard]

What about a scope single shot of a startup attempt from mains ?

[shakalnokturn]

Problem appears solved or OP ran into unrecoverable trouble, I'm not expecting we'll read any more here.