big blue flash could indicate shorted bridge rectifier. or switching transistor.

 

I think there's no real difference using either an inrush limiting resistor or a lightbulb of enought wattage at the input, except of course that if there's a shortcircuit the lightbulb won't tend to be destroyed. Leave the variac out though.

By the mention of two transformers it seems your smps is a half bridge forward topology, maybe look for a basic diagram on that topology and compare it to your circuit so you can understand what's going on?

A lightbulb has a positive temperature coefficient - that is to say the filament resistance increases with temperature, and conversely a very low resistance when cold - that's why the turn on surge causes most bulbs to blow at the instant of switch on.

Most manufacturers of electronic equipment (the decent one's anyway) put a negative temperature coefficient thermistor in series with the mains in, it has a high resistance when cold but quickly heats up once current is flowing, at temperature it has low resistance but enough to drop a small voltage - just enough multiplied by the current for enough watts to keep it heated.

 

Come on, man, yeah, but it's not gonna be in a final design, it's just for a test in a probably shortcircuited board, plus the drop in a lightbulb of enough wattage, say 150w, 200w is negligible and well within the limits of the input voltage of a smps.

 

A trick I was taught was to open the high voltage lead to the transformer and use an external supply to back feed a DC level equal to the rated output of the supply to the output of the UUT (Unit Under Test). When the power is applied to the UUT through your light bulb current limit, the high voltage DC should come up. If the light bulb just comes on full bright, the input rectification/filter has a problem. If the DC comes up normally, the back fed DC will be supplied to the voltage control circuitry and you should be able to see the PWM signal going back to the high voltage drive device. Look at the LED side of the opto coupler. At least the trouble shooting can continue without the flash of the fuse.

 

Come on, man, yeah, but it's not gonna be in a final design, it's just for a test in a probably shortcircuited board, plus the drop in a lightbulb of enough wattage, say 150w, 200w is negligible and well within the limits of the input voltage of a smps.

Just stating the facts - what you do with them is up to you.

A series bulb might save the fuse, but may not save your expensive chopper transistor.

On the plus side, a bulb can prevent the vapourised PCB traces that often happen between the MOSFET punching through and the fuse letting go.

 

Just stating the facts - what you do with them is up to you.

A series bulb might save the fuse, but may not save your expensive chopper transistor.

On the plus side, a bulb can prevent the vapourised PCB traces that often happen between the MOSFET punching through and the fuse letting go.

A 200w bulb draws about 1.6 amps at 120v, there would be more stress to the transistor from the instantaneous current from the input capacitor.

I think the safest thing to do would be to take off the transistors and verify the proper switching of the driver ic and any possible shortcircuit in the primary side be it in the transformer itself, not likely, or if it really is a half bridge, the capacitor on the other leg of the primary.

 

A 200w bulb draws about 1.6 amps at 120v, there would be more stress to the transistor from the instantaneous current from the input capacitor.

I think the safest thing to do would be to take off the transistors and verify the proper switching of the driver ic and any possible shortcircuit in the primary side be it in the transformer itself, not likely, or if it really is a half bridge, the capacitor on the other leg of the primary.

Its only half that current on our 220V system, the point is that with a cold filament at switch on, the bulb draws something like 10 or 12x its steady running current.

The old favourite was to run it up gently with a variac, but most SMPSU control chips have UVLO and suddenly burst into life just a few tens of % less than rated voltage.

When I used to service CRT PC monitors for a living, most of my customers were too stingy to pay to replace the chip just in case, so I'd replace all the damaged components I could find and run it up without the chopper transistor, then scope the chip waveform to check its OK before refitting the chopper.

Many had either a 20V zener or 20V SB diode to catch the gate when the MOSFET punched through - every once in a while it actually saved the chip.

 

...
I think the safest thing to do would be to take off the transistors and verify the proper switching of the driver ic and any possible shortcircuit in the primary side be it in the transformer itself, not likely, or if it really is a half bridge, the capacitor on the other leg of the primary.

That sounds good, but the reality is that the driver IC probably requires feedback from the output to keep oscillating and running normally.

So disconnecting the mains switching transistors and watching the oscillator was a technique that worked ok in the good ol' days but these days it will almost certainly leave you with a non functional oscillator.

My general technique is to remove or opencircuti the switch transistor, then power it up with a small fuse and check the rectified mains voltage is available on the big cap; approx 1.4x the AC mains voltage. That DC voltage should be available on the collector or drain of the main transistor.

Then I check the high-ohms start resistor(s), to see if they are providing some start voltage (usually 24v or less).

Then if both those tests are fine I put in a new swtiching transistor and power it up, still with the smaller fuse in place.

Obviously, before running these powered tests it has been thoroughly visually checked, all the semis short-circuit checked, and small electros ESR tested or just replaced. Those initial checks and replacing the electros fix 50% of PSUs.

 

That sounds good, but the reality is that the driver IC probably requires feedback from the output to keep oscillating and running normally.

So disconnecting the mains switching transistors and watching the oscillator was a technique that worked ok in the good ol' days but these days it will almost certainly leave you with a non functional oscillator.

If you mean the bias voltage you can use a 10k 3w resistor (at 120v) from v+ and a zener to provide the vcc, it will oscillate. Also you can use this scheme to make the output of an smps vary from optocoupler-diode-drop voltage to full rated voltage. Not very wise for a permanent design, but fun

 

Am I OK uploading a photo or two here...? I didn't know that...

Cheers, I'll try a larger fuse as well. But I wouldn't like to destroy the new transistors I fitted..!

Tom

Did you try a slow blow 0.5A or slightly larger fuse? Maybe try a 1A slow blow type. I suggested 0.5A as you said "a small SMPS".

Some photos of the SMPS might help a lot at this point! You've got access to expert fixers but are just spoon feeding us little bits of information at a time.

Photos and info on what you replaced and why, would be the next step.

 

Am I OK uploading a photo or two here...? I didn't know that...

Cheers, I'll try a larger fuse as well. But I wouldn't like to destroy the new transistors I fitted..!

Tom

Buddy, stop plugging the circuit as it is right now, there's a short circuit somewhere in the board and you need to find out where, if you put a larger fuse it's probably not gonna spare your transistors.

 

That sounds good, but the reality is that the driver IC probably requires feedback from the output to keep oscillating and running normally.

All the SMPSU chips I can think of have an oscillator as a distinct functional block, they should try to carry on as normal with the chopper transistor missing - but the chip will think its not producing enough output, and will max the mark space ratio in an attempt to compensate.

That's usually sufficient while doing a repair, to check the chip is switching OK before replacing the chopper.

Most SMPSU chips have an error amplifier for voltage error sampling, and usually another for current sensing - these are pretty much just like op-amps, there is also usually a Vref pin. With the chopper transistor removed, it should be relatively easy to set up a pot supplied from Vref and the wiper to the V error input and manually vary the observed PWM ratio on the output.