AC adapter faulty - where to start checking?

[Peabody]

It's a 5V 2A AC adapter.  At no load, it's 5.1V, but the voltage drops quickly under load.  At 500mA, it's under 3V.  It appears I have the circuit shown below.  I can check all the diodes, but assuming they all check out as good, what would be the best place to go next?  I understand it could technically be almost anything, but it seems to me the output electrolytic would be the most likely culprit.  At least every other time in my limited experience fixing power supplies, it has always turned out to be an electrolytic.  And yes, the date code is 2011, and it has been running constantly since then.  It powers my Ooma Telo box, which idles at about 1A, so literally on 24/365, but plugged into a UPS, so probably not much exposure to mains spikes.

[bdunham7]

Your guess would be the first thing I'd check as well.  If  you don't have the equipment to test it, replacing it should be easy enough and for test purposes, almost anything would do.  What size/type is in there now?

[james_s]

Bad electrolytic capacitor is almost always the problem, at least the initial problem, sometimes other parts blow up in the process but replacing the capacitor(s) is where I'd start.

[retiredfeline]

It's a common SMPS not worth fixing. Accept that it has had a useful life and replace it.

[bob91343]

The IC or one of the resistors or diodes or capacitors may be faulty.  As said above, it's not worth fixing except as a learning project.

https://www.ebay.com/itm/403648956557?hash=item5dfb5a488d:g:ZpUAAOSw69Vbci1g

[bdunham7]

It's a common SMPS not worth fixing. Accept that it has had a useful life and replace it.

If you can easily get a replacement that is not a complete POS, perhaps.  But an OEM power adapter that has lasted 10+ years can often be put back in service for even longer than that with just a few decent quality parts. 

[Peabody]

You mean throw it away?  Surely you jest.  Anyway, it may be strange, but working on stuff like this is my idea of a good time.  And besides, it has  an oddball barrel connector - 3.8mm x 1.35mm x 6mm long.   Of course I could get a genuine replacement from Ooma for about $30 delivered, but where's the fun in that.  So I'll spend some time trying to fix it.  My last repair adventure was a Disney coffee mug warmer - kind of a mini hot plate.  It was a capactive dropper supply with a Zener regulator.  It had exactly one electrolytic in it, and that's what was bad.  Not so much a loss of capacitance, but the ESR  had gone off the charts.  It seems all of these warmers have the same problem after a couple years.   They just chose a crappy cap from a mystery supplier, and they're just no damned good.  But this adapter has lasted quite a while, and I don't think it's unacceptable that something should fail after this long.  Anyway, I'll report back what I find, if anything.

[Kim Christensen]

Yea, 99% of the time it's the output filter capacitor. If you have a scope, you could check the ripple under load. If you're curious, remove it and measure with a ESR meter if you have one. Or just change it on spec.

[floobydust]

OP, are you sure it has an external power mosfet? In the 10W league it's usually just a single IC, which running 24/7 they age, they get weak for some reason- as if the current-sense malfunctions. So you might need a new IC if it's been switching high current all its life (unlike an external mosfet).

I bought a few cheap 5V 2A SMPS off aliexpress and one out of the box worked like yours. It had 5.0V at no load, dropping under load 2.9V at 0.44A so I sent it back and got a refund. I traced it down to the SMPS IC being defective, bad current-sense. The other one's electrolytic capacitors get quite hot at even 1/2 rated load. They're just cheap junk.

I would ESR test and replace the electrolytic capacitors, the 47uF part is critical as well. 8,760hrs/year is a lot.

[james_s]

It's a common SMPS not worth fixing. Accept that it has had a useful life and replace it.

I repair the good quality OEM ones routinely when they fail, usually it's 9pm on a weekend and my internet or whatever stops working, so unless I have a spare readily on hand it's well worth repairing the one I've got. Also when it's just a capacitor why not? Hardest part is cracking it open without damaging anything.

[Peabody]

It uses an external mosfet.  I can't quite see what it is.

I had some time this morning, so I removed all five electrolytic caps and tested them.  There are two big caps near the mains input that are 10uF 400V.  They both test good.  Then there is a 10uF 50V cap that's somehow involved in the feedback circuit, and it also tests good.  And finally, there are two caps on the output in parallel, both 680uF 16V.  I believe those are bad, one twice as bad as the other, but both bad.

I don't have much test gear, so I have to make do.  I use my DSO150 battery powered kit scope, which has a built in signal generator - a 3.3V square wave at 1KHz, which has an output impedance of about 1K ohms.  I set it to AC, then simply connect the capacitor across the signal generator output, and attach the scope leads as well.  A large perfect capacitor produces a straight line across the scope display - it's essentially a dead short.  But a capacitor with a lot of ESR will display the typical capacitor charge/discharge curves.

And I use an analog multimeter set to 1K resistance to measure capacitance.  Well, "measure" may be too strong a word.  Basically, I just see how far the needle deflects as compared to a good cap with a known value.

Below are three scope display captures with the same settings on all three.  The first picture shows a known good capacitor - a 470uF 35V that I got from Digikey a while back.  Sorry the picture is so crappy, but basically all good caps look pretty much  like that - either a flat line, or a very low amplitude square wave.

Then we have the first orignal cap, and the second original cap.  Both look pretty resistive to me, the second being about twice as bad as the first.

I also desoldered and tested the output diode, and it looks good.  It's an SB540, which is a big honking 5A Schottky.  All the other circuit diodes are also good.

I don't have any 680uF 16V replacements.  I have 1000uF's, but they are only 6.3V, and are too tall.  So I'll have to add new caps to my next Digikey order.  Meanwhile I have rigged up an alternate supply that will work for now, but isn't a permanent solution.

If replacing the output caps doesn't fix it, then I think I will have reached my limit as to how much to invest to revive this thing.  I may put one of my 1000uF caps in temporarily just to see if that fixes it - before I bother with ordering the replacement caps.

[Alex_Baker]

Nice testing method, I am a hobbyist with not much of a budget so I appreciate somebody who can be resourceful and get the job done with minimal equipment. I looks like you are on the right track with the output caps.

[No.Mad]

What you have at hand is Quasi resonant flyback power circuit - hope that will help with finding relevant tech docs. Very common for AC-DC 5V chargers/PSUs.

As it was mentioned, my bet is the output caps - they were running 10yrs and I assume, by the value, these are electrolytics. If it's some unit made in PRC these are not top brand like Nichicon. It might be dried electrolytic or changed ESR due to age. Grab two spares from Mouser/Digikey/Farnell, and replace them. Report back as well if that was the case 

[Peabody]

In fact, the caps are the "Su'scon" brand.  But they appear to be genuine Su'scon caps. not some cheap imitations.  :-)   And they are just a bit domed on top.   So I think the odds favor this being the problem.  But of course there still could be something else that's gone bad.  So I won't know for sure until I replace them.  I'm still thinking about temporarily putting one of my 1000uF 6.3V caps in just to see if it will do 1A without drooping much.  If it will, I think that would tell me that everything else is ok.  I just don't know how dangerout that is.  6.3V is awfully close to 5V.

[bdunham7]

Are the two capacitors directly in parallel or is there an inductor between them?

[Peabody]

They are directly in parallel.  There are no inductors on the secondary side.

[No.Mad]

Yeah, this a bit of thin ice situation. I woudn't put them as it's cutting too close, only 1.3V difference for any ripple. And it's possible that caps were busted by something else - in that case you would loose a good one as well.

Capacitor voltage rating at least 1.5x greater than voltage rail for caps - that's how I roll 

Maybe do you have a spare electronics assembly or unused PSU board which could act as a donor for test? You know, copy-paste with solder and iron?

[floobydust]

6.3V cap for 5V is fine, giving 25% margin there. But in this application the ripple current spec is critical.
Higher voltage-rated capacitors have higher ripple current specs due to their larger physical size. Size is a problem in a small AC adapter, and with cheap design philosophies the filter caps are too small to begin with, for long life. They'll push parts too hard.
If I want long life out of the adapter and can't fit in decent (value, ESR, ripple current spec) then I'll break the piggy bank and put in polymer caps.
It's important to check the original part's values are not too small. 2x680uF for 2A?

My lab had the stink of electrolytic capacitors on and off for a few months, searched everywhere drove me nuts but it was from a DSL modem not even 2yrs old.

[Peabody]

The schematic I posted in the original post is from the datasheet for the OB2263 PWM controller.  From the markings on the chip, I think that is in fact the part that I have.  The schematic doesn't agree in every respect, but it's pretty close.

https://file.yizimg.com/332467/2010080308450435.pdf

I suspect the use of two caps in parallel is just so they will fit physically.  The caps are 8mm in diameter, 16mm long.  The only one available on Digikey that's an exact match is this one:

https://www.digikey.com/en/products/detail/nichicon/UPA1C681MPD1TD/3129898

I guess I could also look at temporarily changing the bias on the 431 reference to reduce the output voltage from 5V to maybe 4V, and then the 6.3V cap would make more sense.  I would need to do that by adding resistance in parallel with an existing resistor.  I assume the lower resistor in the divider would be the one to modify.  Or maybe I have that backwards.

[Kim Christensen]

I guess I could also look at temporarily changing the bias on the 431 reference to reduce the output voltage from 5V to maybe 4V, and then the 6.3V cap would make more sense.

Personally, I'd just throw the 6.3V ones in there to test. Maybe even tack 'em on the solder side temporarily. Then, once you've confirmed that fixed it, order the new ones.

[Peabody]

Real life kept intervening, but today I was finally able to test the power supply with temporary capacitors installed.  I replaced the two original 680uF caps with one 1000uF and one 470uF.  And it worked like a charm.  It's 5.17V with no load, and at 1A it's down to exactly 5.00V.  After 20 minutes at 1A, it was still doing fine, and not heating up.  So I think that settles the issue. The bad output caps were the problem, and everything else seems to be ok.  I hate to pay for Digikey shipping just for two replacement caps, so I may wait a while to see if I need something else.  But the bottom line is what everybody predicted - it was the output caps ESR that dragged everything down.  Thanks for everyone's contributions.