Spectacular AC/DC Adapter Fail

[SewingYard]

Ok, I'm quite a novice as some of you have probably already understood from some of my other posts.

BUT... I had a faulty Power adapter that I thought I could quite comfortably fix... how wrong was I!!

Basically I looked at the board and came to the conclusion that one of the transistors was faulty so I figured I would swap them for new ones, the problem was that the two that were there were ones I didn't have so I substituted one of them a C945 for a 2n3904, the pin out was different so I resolved that by crossing the pins over before soldering them in (no they didn't touch)

I plugged in the supply, waited for the bang and it didn't happen, the LED came on and I was overwhelmed with joy, I HAD FIXED IT 

The amazement was short lived however because 10 seconds later as I was about to check the output was 7.3 volts with a multimeter all hell broke loose, the whole supply went BANG, not once but 3 times, first the cap blew off, then a transistor or two then a resistor, not a single failure but 4 separate components blew off the board!

I attach 2 pictures to show the damage and hope that someone can show me where I went catastrophically wrong.

[Ian.M]

DC bus voltage in PSU running on 240V UK mains: 340V
2N3904 V_CEO: 40V

The result is quite predictable - I'm surprised it lasted as long as it did.
Either it over-voltaged the transistor you replaced or the other transistor was already damaged (e.g. leaky).   When dealing with a low parts count SMPSU you have to test *all* primary side semiconductors out of circuit, and test all resistors and small electrolytics to have any chance of success.  Next time try salvaging transistors from a CFL bulb - they are far more likely to have a suitable voltage rating.

The secondary side damage is fairly conclusive proof of catastrophic loss of regulation, possibly caused by a defective Zener on the primary side (D3?) or possibly by electrolytic C3 having gone high ESR.

[SewingYard]

DC bus voltage in PSU running on 240V UK mains: 340V
2N3904 V_CEO: 40V

The result is quite predictable - I'm surprised it lasted as long as it did.

Hi, The C945 that it replaced was only 50v as far as I can see.

[Ian.M]

Yes, and it was probably running right on the bleeding edge of its rating. This sort of cheap PSU from the far east is usually marginal on 220V - maybe 1000 hours if you are lucky.  Feed them 240V and they die even sooner.  Yours actually looks to be almost acceptable quality - they've used a 400V cap for C1 and used a resistor as a fuse (should have been a non-flammable one though)

[Gyro]

Trying very hard to keep a straight face, but that's a classic 

Given that the capacitor that blew is on the secondary side, the supply must have gone way out of regulation before it expired. Maybe that's a clue to the chain of events.

[SewingYard]

Yeah, I was having problems breathing!

I've let out enough magic smoke tonight to bottle and sell 

I'm trying to fix the supply now, I've replace the cap and the resistor now for the transistors, ive now got some higher power transistors so should be ok.

Can anyone tell me from the pictures which pins would be BCE ?

I'm not quite used to working it out yet but am having a think to see if I'm right if someone tells me what they actually are.

[Gyro]

Sorry, I'm not going to risk transistor pinouts at this time of night (not sure you should either!) 

I think at this stage you need to treat every component that still managed to cling onto the board as 'stressed'. I wouldn't be too sure about the state of the diodes for instance, including the secondary side rectifier and possibly the transformer - it must have been a heck of an overvoltage to blow the capacitor that fast. They were also using that blown resistor at the AC input as a cheap fuse, A metal film one might not fail in the same way.

I'd stop at this stage and check everything before continuing (or deciding whether to).

[SewingYard]

Sorry, I'm not going to risk transistor pinouts at this time of night (not sure you should either!) 

I think at this stage you need to treat every component that still managed to cling onto the board as 'stressed'. I wouldn't be too sure about the state of the diodes for instance, including the secondary side rectifier and possibly the transformer - it must have been a heck of an overvoltage to blow the capacitor that fast. They were also using that blown resistor at the AC input as a cheap fuse, A metal film one might not fail in the same way.

I'd stop at this stage and check everything before continuing (or deciding whether to).

Ideally id like to try and build one of these from scratch but I'm still learning, could you tell me what type of power supply this is?

I'd call it switched mode but when I see switched mode supplies they generally have lots more components and an opto coupler, this one seems to just have basic components and a couple of transistors.

Any links with examples would be great

Cheers

[Ian.M]

An alternative hypothysis to loss of regulation would be if the output rectifier had failed shorted.  That would apply HF AC to C4, cooking it till it blew, while the LED acted as its own rectifier to illuminate.   It would also screw up the feedback causing the regulation circuit to go to maximum duty cycle, and the sudden loss of the load when C4 blew would then over-voltage the chopper transistor Q1, and cause a cascading primary side failure.

Q1's collector goes to the transformer, its emitter to the big low-ohm resistor (current sensing) and its base to the collector of Q2 and a pullup resistor to +340V.  Q2's emitter goes to 0V (reservoir cap -ve) and its base is fed from the feedback network.  You *MUST* test all the primary side diodes after a blow-up like that.   Q1 needs to have a high V_CEO rating, probably >500V.

*IF* you get it going without further silicon shrapnel, test it with a resistive load of 80% of its full power rating  and add a beefy Zener diode with a rating a volt or two greater than the output voltage across the output to clamp it so that if it looses regulation again it doesn't destroy whatever it powers.

Personally, I'd simply replace it with a new better quality one with real opto feedback and an actual chopper control chip!

[SewingYard]

Thanks Ian.M really appreciate the full simple explanation. I'll be sure to follow your suggestions and see if I can rebuild it. Fingers crossed.

[roffvald]

I love when things like this happen, great opportunity to learn, and you got some excitement in your day
As "Make" says: "Burn things out, mess things up-that's how you learn".

[Ian.M]

To further understand it, trace out the circuit and draw a schematic, with component values.   You can then post that here and ask questions about how it works.

[NiHaoMike]

It looks like a cheap PSU that's not worth fixing. Just replace it with a better quality one.

[SewingYard]

It looks like a cheap PSU that's not worth fixing. Just replace it with a better quality one.

Thanks for everyone for their replies.

With regard to it being a cheap power supply that's not worth fixing...

I started with some more complex ones and really couldn't figure them out easily so I noticed this one had stopped working and figured it looks really simple, its probably crappy but it looked like a really simple implementation of a power supply that I could figure out given a little time and patience. (and a small amount of smoke)

Ill be sure to try and trace it out and figure out even though it may not be worth repairing its still worth it for the education factor.

[Ian.M]

Cheap&nasty PSUs interweave functions in a way that makes them much much harder to understand.  e.g. in this case, voltage feedback via the aux winding on the primary side of your transformer (rather than by the classic TL431 + optocoupler),  chopper current feedback mixed with the voltage feedback(rather than on a separate pin of a control IC), and AC feedback from the aux winding to make it oscillate (rather than an oscillator in a control chip clocking everything at a well defined frequency), with the bias point of the chopper driver being shifted to modulate the overall duty cycle.  At this point I'm not even sure if it does PWM or PFM or some ill-defined mixture.

[Gyro]

Agreed, something that is designed to only just work and pared to the bone isn't a good teaching model.

[SewingYard]

Cheap&nasty PSUs interweave functions in a way that makes them much much harder to understand.  e.g. in this case, voltage feedback via the aux winding on the primary side of your transformer (rather than by the classic TL431 + optocoupler),  chopper current feedback mixed with the voltage feedback(rather than on a separate pin of a control IC), and AC feedback from the aux winding to make it oscillate (rather than an oscillator in a control chip clocking everything at a well defined frequency), with the bias point of the chopper driver being shifted to modulate the overall duty cycle.  At this point I'm not even sure if it does PWM or PFM or some ill-defined mixture.

Ian.M

Thanks for getting back to this topic, I'm intrigued, you say that the primary winding is being used for the feedback, is this why there are two windings on the primary side of the transformer? I was assuming this was because it was compatible with 110 / 240 and was using one or both when required?

The model you describe would make much better sense to me as I was confused as to how this plug was doing the feedback, but its something I never considered when I was trying to learn about the circuit.

I am currently looking for an isolation transformer so that I can probe the primary side but finding that they're particularly expensive so have to wait for a second hand one to come along.

 

[SewingYard]

I love when things like this happen, great opportunity to learn, and you got some excitement in your day
As "Make" says: "Burn things out, mess things up-that's how you learn".

well, the RCD just came in handy as did the rest of the fuses running from the main board to the workshop, I really need to look at battery backup!

I tried some different transistors but it didn't seem to work, result attached!

Suffice to say... I think ill give up on this board now! 

[Gyro]

Well that's a very interesting smell that you've invented! 

I'm glad you had RCD and supply fuses (though it sounds as if they were the consumer unit ones!).

I think you've discovered why the reliance on a 1/4W resistor as a mains fuse in these things really isn't safe.

That's one valuable lesson that you can take away from this board anyway.

[jitter]

Thanks for getting back to this topic, I'm intrigued, you say that the primary winding is being used for the feedback, is this why there are two windings on the primary side of the transformer?

Yes, it is, but I would like to call it tertiary rather than primary as it will likely be a winding separate from both the primary and the secondary windings.
If you didn't blow up the transformer in your experiments, you should be able to measure with a DMM between which pins there's resistance and which are open and draw down the windings. Should be educational.

I was assuming this was because it was compatible with 110 / 240 and was using one or both when required?

On a transformer operating directly off the mains, you will indeed find ones with separate primary windings that you can put in series or in parallel depending on the mains voltage.
On an SMPS things work a bit differently. Look at the sticker on the housing of your supply, does it say 100-240 V?

Trace out the schematic of your SMPS, as has already been suggested, and you will discover that the windings on that transformer are neither wired in series nor in parallel.
The control loop in this particular power supply does not measure the actual secondary side voltage and feed it back to the primary side controller via an opto coupler. In this case a winding on the transformer provides a feedback signal. Even though it's not measuring the output directly, the voltage that's generated on that winding depends on the load that the supply is presented with.