SMPS Behavior

[PA0PBZ]

So my son bought 2 wall USB PSUs and one didn't work out of the box. Since he had already thrown away the receipt and the thing was only 5 euro he asked me if I was interested to have a look. From the outside the problem showed as a blinking led with a period of about 1 second, and putting the thing on the scope showed the following, first without a load and then with a 22 ohm (~ 200 mA) load:





As most of this stuff is ultrasonic welded I took a hammer and managed to open the thing. It didn't look as bad as expected, it's fused and has the proper isolation slots:





Now what puzzles me is where does the time constant of 1 second come from? It's obvious not the decharge curve because it doesn't change when I put a load on it. So what's wrong, does it go into protection? It looks like the overcurrent resistors are ok, and there is no overshoot on the output. Anyone that can refresh my theory on these things? I repaired dozens but this one has me wondering for now. Below is the reference design from the datasheet which seems to be what's on the pcb but unfortunately the rest of the datasheet is in Chinese...

[tautech]

Monitoring VDD will tell you if there's a supply problem or it's in tick (protect) mode. That result will determine where to look next.
Not seen a SMPS IC start supply resistor dropper straight from mains, normally it comes from the HVDC.

Isn't the soldering bloody awful. 

[PA0PBZ]

Monitoring Vcc will tell you if there's a supply problem or it's in tick (protect) mode.

You mean VDD? Looks like it is started from the net and then it feeds itself, so I guess I will see the same happening on there. I will have a look and post the results.

Isn't the soldering bloody awful. 

Looks like it has been reworked manually, it's ugly but I couldn't fault it though.

[tautech]

Monitoring Vcc will tell you if there's a supply problem or it's in tick (protect) mode.

You mean VDD? Looks like it is started from the net and then it feeds itself, so I guess I will see the same happening on there. I will have a look and post the results.

Isn't the soldering bloody awful. 

Looks like it has been reworked manually, it's ugly but I couldn't fault it though.

See edited post sorry.
The resistor dropper is the start supply and the aux primary winding is the run winding.
Check C6 and D6 if the start supply seems OK. 10 uF for C6 seems a low value, usually 20 +uF is more common.

[PA0PBZ]

C6 and D6 look fine, the voltage across C6 varies between 8 and 14 volts in the same rhythm as the rest.

[Andy Watson]

C6 and D6 look fine,

? I would expect D6 and the associated circuit to take-over after the first cycle and maintain 10-15V on C6.

the voltage across C6 varies between 8 and 14 volts in the same rhythm as the rest.

Is that a clue?
The transformer and switching parts are obviously working because you are getting a blip on the low-voltage output. Once the "forward" conversion has fired, the voltage on C6 should be maintained.

[PA0PBZ]

Well, since the thing switches off after a few cycles and stays off for almost a second I guess it's not strange that the voltage across C6 drops. I didn't do the math but there are 2 1M resistors from the mains so it could be totally possible that the starting voltage is a lot lower, somewhere around 5V maybe? My Chinese is non-existing unfortunately.

Edit: It looks like the voltage across C6 slowly rises and then suddenly drops, almost as if D6 is reversed  The slow rise could be the 2M charging C6 but why the sudden drop?
D6, R3(?) and the winding must be ok otherwise the mains would not be rectified.

Edit 2: After playing with Google translate I found the following:
Startup current is only in the uA range, working current around 2 mA. Further, if VDD drops below 8.2V it will go into undervoltage protection mode and switch off.
So... Why can't it power itself? D6 and C6 measure ok but maybe it's time for a blind replacement...

[Andy Watson]

The slow rise could be the 2M charging C6 but why the sudden drop?

C6 charges (slowly), the control chip realises that it has enough supply voltage to run, it wakes-up and starts driving pulse(s) and other functions - thereby discharging the energy stored in C6 - relatively quickly ... and repeat.

D6, R3(?) and the winding must be ok otherwise the mains would not be rectified.

No, the initial rectification is performed by the bridge rectifier via R18 and R2.

[Seekonk]

Perfectly consistent with D6 or the transformer winding being open that supplies voltage to C6.  It is interesting to note that other versions of this chip have a separate bypass pin with only a capacitor to common.  I've used these chops at much lower voltages (12V) than they are speced at for low voltage cutoff by attaching a pullup resistor to that bypass pin.

[PA0PBZ]

Well, it works! Replaced D6 and C6 without making a difference, measured everything again and it looked perfect, but then I noticed something weird between the winding and D6. When touching the meter very lightly it looked like there was no conductivity; pressing the pins a little harder and it was ok. Scraped the solder mask off the trace and gave it a good beating with the iron and look, it works! 

Nothing to see when looking at the track, and just putting the meter there fixed it... Victory at last, now I can throw it in the bin without getting sleepless nights.
Thanks all!

[tautech]

Interesting. Do you think it was a dry joint ?
Of course it doesn't matter now but it's always important IMO to know why something won't/don't work.

With it cycling the way it was there were two likely possibilities, IC supply or protect mode.
In many ways initial SMPS fault diagnosis is the same as any other equipment fault..........check the PSU first.