Probability of generic china PSU failing in a way that destroys your circuit.

[Psi]

I bought a Helios4 NAS PC to be my new file server and it came with the expected 'generic' china ACDC psu. (100-240AC to 12V 8A DC.)

Since i will be trusting this thing with 4 multi-TB drives i thought it would be a good idea to crack open the AC-DC adapter and see what the quality is like. Maybe replace all the electro caps with long life Panasonic caps from digikey or whatever.
I expect Helios4 box will draw about 41W under heavy loads on all 4 drives. (9W per drive + 5W for arm PC  but in reality i doubt it will be over 15W most of the time.)  41W is only 3.4A @ 12V so having a 8A psu shipped with it is definite a good thing for longevity and lower internal temps.

Anyway, this got me thinking about the failure mode of generic no-name brand AC -> DC power packs.
What is everyone experience when it comes to these generic power packs failing?
Do they usually just die gracefully without any damage to your device
Or do they go unregulated and put out 30+DC or 220V AC and blow up your device?

[Circlotron]

Make yourself an external over voltage crowbar circuit.

[wraper]

It's not necessarily no name. Very likely you simply have no clue it was made by big manufacturer (post a photo). And how gracefully they fail does not depend on caps as such (impacts time to fail though). But how protections are realized and overall topology.

[digsys]

Coincidence or what ? :-) I was just about to make a post about this -
Bought a couple security monitors last week, and as I usually do, threw the supplied power-packs away and use known quality Meanwell units.
Thought to myself - "Self, I wonder if the chinese are getting any better at making stuff", so I opened them up -
First one - did the usual - hold in a vice and tap the seam - it shattered with very little effort - a little shocked at pic1 - not a heck of a lot of filtering / protection !!

Second one - held in vice and tapped the seam - splintered into dozens of pieces - the insides were even more shocking ! and that was the expensive unit ! pic2
Oh well, more customers for e-waste

[wraper]

First one - did the usual - hold in a vice and tap the seam - it shattered with very little effort - a little shocked at pic1 - not a heck of a lot of filtering / protection !!

Good enough, even has common mode inductor.

Second one - held in vice and tapped the seam - splintered into dozens of pieces - the insides were even more shocking ! and that was the expensive unit ! pic2
Oh well, more customers for e-waste

Nothing shocking actually. There are tons of recycled PSUs in the wild. Often "modified" to output something like 9V when it was originally 5V PSU. I have one of such, broken glass zener on the output (guess desoldering was too much effort) and feedback resistor soldered on top of original, and voilà.

[coppercone2]

ebay.com -> search? HP

[NiHaoMike]

ebay.com -> search? HP

Reminds me of the time in college when my friend Allie Moore asked if I could fix her laptop PSU. I was surprised how bad the build quality was - not as bad as the ones in the picture above but nowhere near the quality I expected from HP.

[Psi]

It's not necessarily no name. Very likely you simply have no clue it was made by big manufacturer (post a photo).

Here's pics from the PSU in question.

ST Controller says
UC38428

FET is Hangzhou Silan Microelectronics
SVF10N65F

[digsys]

It's a lot better looking inside than the cr@p I opened up, but it is really hard to trust anyone these days. It's always a risk, and why I buy from suppliers I know, well critical stuff anyway. This is the Meanwell insides I just bought today - pretty much same specs -

[wraper]

Here's pics from the PSU in question.

ST Controller says
UC38428

It's UC3842B. It is quite cheap PSU, but not that bad. Capacitors are crap but not the worst. And probably it was recycled. Could you post a closer photo of area where output cable comes from PCB (top and bottom side)? It looks suspicious.

[soldar]

Do they usually just die gracefully without any damage to your device. Or do they go unregulated and put out 30+DC or 220V AC and blow up your device?

The two modes of failure I have seen most often are (1) the thing just stops working because it blew a gasket or (2) the output capacitors fail slowly and you start getting bad regulation and switching frequency at the output but still never really put out more voltage than they should. I have one that started failing like this and I just added some capacitors outside temporarily and its been quite a few years now and it is still working like that.

Overvoltage at the output, while not impossible, is extremely unlikely.

220 V out is pretty much impossible.

[chickenHeadKnob]

Do they usually just die gracefully without any damage to your device. Or do they go unregulated and put out 30+DC or 220V AC and blow up your device?

The two modes of failure I have seen most often are (1) the thing just stops working because it blew a gasket or (2) the output capacitors fail slowly and you start getting bad regulation and switching frequency at the output but still never really put out more voltage than they should. I have one that started failing like this and I just added some capacitors outside temporarily and its been quite a few years now and it is still working like that.

Overvoltage at the output, while not impossible, is extremely unlikely.

220 V out is pretty much impossible.

I would add a third failure mode that has bitten me as often as bad caps. Cold, intermittent bad solder joint on one the inductors/transformer, especially on single-sided boards. Dave made a video on this type of failure with the power supply for his overhead lighting.

Not fatal for the down stream components. Symptoms involve  processor resets, flickering operation ect.

[soldar]

I would add a third failure mode that has bitten me as often as bad caps. Cold, intermittent bad solder joint on one the inductors/transformer, especially on single-sided boards. Dave made a video on this type of failure with the power supply for his overhead lighting.

Not fatal for the down stream components. Symptoms involve  processor resets, flickering operation ect.

Interestingly I have seen this failure mode several times in monitor internal PSUs but not in brick type external PSUs. It really does seem like mechanical ultrasonic vibration causes the connection to some loose. I do not think it looked like bad soldering in the cases I have seen.

[chickenHeadKnob]

Interestingly I have seen this failure mode several times in monitor internal PSUs but not in brick type external PSUs. It really does seem like mechanical ultrasonic vibration causes the connection to some loose. I do not think it looked like bad soldering in the cases I have seen.

I have had it happen in a variety of consumer items, including 2 microwave ovens, on the small supplemental supply that drives the keyboard/display cpu. In none of them did the joint look particularly bad, they could only be detected with gentle flexing. That is pressing down on the board or component with a non-conductive rod like a plastic pen. And this failure only exhibited after many years, 15 in one case, of service. Which suggests thermal cycling is the culprit.

[soldar]

Yes, that's why I say I do not think they were bad solder joints originally but many thermal cycles and ultrasonic vibration over the years made them fail. I have repaired a few by just resoldering all of them.

[schmitt trigger]

@digsys

The first pictures that you show have paper-phenolic PWBs. Literally the cheapest board substrate money can buy.

The second group of pictures has a CEM-1 board. It has woven glass, and even though it is inexpensive, it has superior mechanical characteristics as compared to paper-phenolic.

For power supplies, which employ heavy inductors and transformers, the paper-phenolic boards can crack easily with mishandling. Also because they tend to bend easily, it should not be used for SMT components, although your photos clearly indicate that the manufacturer could care less about this fact.

[wraper]

For power supplies, which employ heavy inductors and transformers, the paper-phenolic boards can crack easily with mishandling. Also because they tend to bend easily, it should not be used for SMT components, although your photos clearly indicate that the manufacturer could care less about this fact.

Phenolic boards are used for PSUs even in quite expensive stuff.

[TimNJ]

The majority of switching controllers, these days, implement a low cost over-voltage protection (OVP) function via primary side auxiliary winding on the transformer. Whether or not the engineers cared enough to properly calibrate it and/or design the transformer for good repeatability...well, that's anyone's guess.

That said, a few years ago, my brother bought a $5 turd power supply for his laptop off eBay...and while we can't definitively point a finger at the power supply, the laptop stopped working a few months after. So, when it really matters, just get something decent, if you are worried about reliability. They aren't that expensive if you get something from Meanwell, Delta, CUI, etc..

[schmitt trigger]

For power supplies, which employ heavy inductors and transformers, the paper-phenolic boards can crack easily with mishandling. Also because they tend to bend easily, it should not be used for SMT components, although your photos clearly indicate that the manufacturer could care less about this fact.

Phenolic boards are used for PSUs even in quite expensive stuff.

PSU have become commodities. And with all commodities, the least common denominator is what becomes the norm.

Paper phenolic boards are nevertheless a poor choice when heavy components require to be supported.

[digsys]

I think the point of all this is that if I purchase equipment with a supplied P/Supply, I expect the P/Supply to behave. I expect it to be stated +/-5% with very little likelihood of failure beyond that. Definitely not 110V !! I do NOT want to hope that my (often) expensive equipment can withstand an OV failure, that is my BACKUP !
I'm NOT willing to HOPE that the P/S failure (ie wire coming off the AC Input in 2nd picture), simply resulted in a clean DC outage ! For the price, and mitigated risk, when I see manufacturing like that, they go to e-waste.

[TimNJ]

I think the point of all this is that if I purchase equipment with a supplied P/Supply, I expect the P/Supply to behave. I expect it to be stated +/-5% with very little likelihood of failure beyond that. Definitely not 110V !! I do NOT want to hope that my (often) expensive equipment can withstand an OV failure, that is my BACKUP !
I'm NOT willing to HOPE that the P/S failure (ie wire coming off the AC Input in 2nd picture), simply resulted in a clean DC outage ! For the price, and mitigated risk, when I see manufacturing like that, they go to e-waste.

I work for a power supply company. You'd be surprised about how many (seemingly) high end companies are willing to scrape the bottom of the barrel for their power supplies.

Still, I think the vast majority of bundled power supplies are sufficiently reliable and safe. It's the dodgy "replacement" power supplies that you really have to look out for.

[Psi]

Ok so there has been some developments 

I have not replaced the caps just yet.
But i did power it on to check the output voltage and confirm it's all working. Then this happened.....

[Psi]

Could you post a closer photo of area where output cable comes from PCB (top and bottom side)? It looks suspicious.

See pic below

Helios4 has all the schematics available. I didn't expect it but they also have the power supply too!
https://wiki.kobol.io/files/power-supply/YCZX_Schematics_12V8A.pdf.

[digsys]

COOOOL !! A cheap power supply AND a light show !! Good deal :-)

[wraper]

IMHO it's fairly safe form over-voltage on the output, though there is no OVP. The biggest chance to get high output voltage is when TL431 or optocoupler fails. You could add TVS to the output for a peace of mind.