PSU Repair: A Case Study

Do We Have A Definitive Fix?

Recap

I started out with a whiny power supply that would not reliably turn on, discovered that I could make it become predictable by loading down the 5VSB output and decided to start working backwards from there. In the end, I got it to function properly without crutches using $2 worth of new caps, even if they didn't fit on the board quite right. Of course, that was on top of the $3 worth of capacitors I replaced back in 2009.

Looking back at the results, I believe I can put together the sequence of events that lead to this failure: putting my Core 2-based machine into standby on a regular basis, something I rarely did on my Pentium III, wore down the under-rated 5VSB output caps. As they got worse, ripple on the auxiliary capacitor also increased until it failed while the main outputs were off. Beyond that point, the PWM's capacitor was alone and the power supply continued chugging along until its ESR became too high and the supply would no longer turn on.

Disconnecting the unit from the motherboard removed the load and board capacitors, which provided some clamping for the 5VSB output like my external capacitor did. This allowed surges on the auxiliary output to reach levels high enough to flash over inside the dead auxiliary capacitor, and once the capacitor had warmed up enough from it, contact got made and the power supply would work normally again until the next time it cooled down. Adding load on the 5VSB rail caused the 5VSB's pulse rate to increase. Once the load was sufficient to make the pulse period shorter than the voltage decay time on the PWM's bypass capacitor, the power supply would turn on properly again even when cold.

The increased 5VSB voltage before the auxiliary capacitor replacement was likely caused by the flyback circuitry using its own auxiliary output supply to shut itself off based on the 5VSB feedback. With no power to operate the primary side of the feedback circuitry after the auxiliary voltage dies off from the dead cap not holding a charge, a new flyback cycle starts even if not required and the voltage rises.

In Retrospect

Thinking back to the earliest days when this power supply started giving me trouble after my first repair attempt, I vividly remember hearing an irritating whining noise coming out of my speakers with a pitch that shifted in time with the blinking of my PC's standby LED about a year later. I actually ended up buying an EA650 for my Core 2 and put the SL300 in one of my spare PCs until I reused the EA in my i5-based system. I would not be surprised if that happened to be the very first hint of trouble on the 5VSB and auxiliary outputs. As shown in my preliminary tests, even a 50mA change had a significant impact on the flaky 5VSB's behavior. With the fixes, load on 5VSB makes hardly any difference whatsoever anymore.

Conclusion

Throw in a proper $0.50 PCB-mount fuse to replace my thin wire fuse hack, a pair of 8mm capacitors with at least one of them rated for 2A ripple that fit properly for the 5VSB output and perhaps something with the correct lead spacing for the auxiliary cap--that's what I would call a definitive fix. Would I use this supply in a PC again? Sure, but not in anything beyond a spare system until I got around to putting proper parts in and doing some more testing.

Should you attempt to repair decade-old $40 power supplies? Most likely not. Even if you have the knowledge, skills, tools and parts to do it safely and properly, buying a new $40 power supply may still end up cheaper than the repair once you account for your time. Also, the new unit will usually be considerably more efficient and contain only minty-fresh components, complete with manufacturer warranty coverage. It only made sense to me because I could satisfy my own curiosity and do something useful with my oscilloscope. Having spare known-good power supplies with high performance caps never hurts either.

Next in my repair queue: the proprietary form factor supply from my Antec Aria case. It is the worst case I have personally worked with to date, made worse by the Pentium 4 (3GHz Northwood with HT) I put in it and the effectively nonexistent ventilation through it, especially in the processor area. The CPU fan is screaming even at idle when all three covers are on. If you want to read about that, let me know in the comments!

MORE: How We Test Power Supplies
MORE: Who's Who In Power Supplies, 2014: Brands Vs. Manufacturers

MORE: All Power Supply Content

MORE: Power Supplies in the Forums

MORE: How To Build A PC: From Component Selection To Installation

Daniel Sauvageau is a Contributing Writer for Tom's Hardware. Follow him on Twitter, Facebook and Google+.

Follow Tom's Hardware on Twitter, Facebook and Google+.

This thread is closed for comments
62 comments
    Your comment
  • Nuckles_56
    An interesting read, it was interesting following the process you used to troubleshoot the problem
  • Crashman
    I used to do this for a living :)

    Don't tell my boss, I've managed to convince him that I'm only an expert at running benchmarks and writing about the results :)
  • epsiloneri
    Disclaimers won't help. The people who will likely hurt themselves trying this are the same who lack the reading comprehension and self awareness to understand those disclaimers are directed at them. I admire you courage in publishing this.
  • beetlejuicegr
    the truth is the paper clip and multimeter is all i can go in to psus. after all i haven't studied electricity or circuits or whatever.
    However i do hate to throw stuff earlier than it should, like you.
  • C12Friedman
    I like this article and I fully agree with the conclusion. I've repaired a few PSU's but, for the most part I scavenge them anymore since I can't put them in a new system (nor would I want to) IMO they aren't really worth anything other than for on a test bench.
  • Mr A
    Daniel, I know next to nothing about electronics, and yet I could not stop reading this article. Fascinating! Thanks very much!
  • Urzu1000
    This was a great article! It was informative, as well as interesting. Personally, I've only had one PSU fail on me so far. My brother-in-law's self-built computer had a really low-end Thermaltake PSU. 800W Bronze, and oh man, did that thing go out in a blaze of glory. Very loud popping, and smoke, and funny smells. When I ripped it out of the computer, there were burns inside the case. Miraculously, the other components remained unharmed, so I slapped in a new PSU (750W Gold Seasonic) and fired it up.

    Still working good, but I get black soot on my hands every time I open up that case. It's a black case, so it's hard to clean it off properly.
  • nukemaster
    Good read.

    Thanks for taking the time to document this repair.
  • Interesting article. I would have simply replaced the entire unit. You saw how to fix the failure, but how many units were damaged that you didn't see? A ticking time bomb that will eventually send some spike to your much more valuable hardware than a 10 year old PSU. Wasteful, yes. I get it. I don't like to waste either. And if it's on marginal hardware, fine. But on primary systems I'm not willing to take the risk. I'd rather throw away a 200$ part that has a 0.05$ repair solution, than risk frying 800$+ hardware.
  • kalmquist
    "Antec's manufacturer (Channelwell in this case) got the live and neutral wires backwards, which means that in the “off” position, the neutral line gets opened and everything on the primary side becomes live instead of neutral."

    That's really bad--I doubt it is even legal to sell a power supply wired like that. I've never bought a CWT (Channelwell) power supply, and based on this I wouldn't buy one, except perhaps for a high end model where you might gamble that the company would exercise a bit more care.
  • Daniel Sauvageau
    211508 said:
    Disclaimers won't help.

    They may not help the people who choose to ignore them but they help reduce the likelihood of legal issues if something blows up in their face or worse after they do so.

    2008047 said:
    Daniel, I know next to nothing about electronics, and yet I could not stop reading this article. Fascinating! Thanks very much!

    You're welcome. I try to put enough technical details in there to keep technically minded people interested but not so much as to make it inaccessible to more casual readers.

    723938 said:
    Interesting article. I would have simply replaced the entire unit. But on primary systems I'm not willing to take the risk. I'd rather throw away a 200$ part that has a 0.05$ repair solution, than risk frying 800$+ hardware.

    And a new PSU was the fix I initially implemented in my PCs too, though mainly because I could not be bothered to investigate at the time since I lacked the tools necessary to do so properly and with reasonable confidence.

    262841 said:
    That's really bad--I doubt it is even legal to sell a power supply wired like that. I've never bought a CWT (Channelwell) power supply, and based on this I wouldn't buy one, except perhaps for a high end model where you might gamble that the company would exercise a bit more care.

    The wires on the IEC plug filter board are likely inserted by hand. Without a few more units to compare it against, it is entirely possible that my inverted wiring was a one-off or otherwise uncommon manual assembly mistake. It likely happens to all other manufacturers too, the question being how often it gets through QC unchecked.
  • Daniel Sauvageau
    1526053 said:
    This was a great article! It was informative, as well as interesting. Personally, I've only had one PSU fail on me so far. My brother-in-law's self-built computer had a really low-end Thermaltake PSU. 800W Bronze, and oh man, did that thing go out in a blaze of glory. Very loud popping, and smoke, and funny smells. When I ripped it out of the computer, there were burns inside the case. Miraculously, the other components remained unharmed, so I slapped in a new PSU (750W Gold Seasonic) and fired it up.

    No need for miracles here: most spectacular catastrophic PSU failures occur on the primary side which is isolated from the outputs by the transformer. If the BJT/FET switching becomes too slowly due to weak gate drive, the junction or channel may exceed its SOA and blow up - I was surprised this had not happen to my SL300 here. If the switch stays on for too long or the transformer has a flux imbalance that causes it to move up its hysteresis curve, the transformer core eventually saturates and the low impedance destroys the switch. Either way, the amount of energy transferred to the secondary side from the fault should be negligible compared to the total capacity of output caps.

    The scary PSU failures are those involving the feedback circuits: if these fail open, you may end up with the primary side operating at its maximum duty cycle and outputting the maximum voltage the main outputs are capable of based on the transformer's turn ratio. This is basically what happened with the 5VSB output here: the auxiliary output was failing and caused the primary side of the 5VSB feedback loop powered by that auxiliary output to misbehave, allowing the 5VSB output to surge to about 8V not counting noise.

    I would be far more concerned with silent failures (like the 5VSB here) than spectacular ones.
  • chazking260
    Nice to revisit the technician aspect of troubleshooting. It's a dying art and cost prohibitive except in a lab environment. Great Problem reporting!
  • razor512
    I was able to repair a $20 power supply by replacing the dead caps, with some industrial 135C rated caps that I salvaged from some industrial equipment. (the caps likely cost far more than the power supply, but the repair worked.

    For PSUs I generally do not repair them beyond capacitors, unless it is something that is custom and cannot easily be replaced.

    After a repair, I always test power supplies with dummy loads. (mainly car headlights). (pack of 2, 55 watt headlights are generally around $7-8
  • nukemaster
    Hey I was just looking over this again in Spartan/Edge with reader mode and notice those "rubycon"caps are also fakes.

    They say rulycon

    Should have had the logo like this :)


    Instead it is.
  • shrapnel_indie
    120VAC is enough to kill some people, or at least interfere with their hearts enough to cause medical/physical issues. Some may just feel a pulsing "bite." Not worth being careless (or even intentionally) to find out which one you are. 240VAC will kill. I should be more precise though: The associated current does the damages, and each human body's electrical resistance can vary among others and/or with conditions.

    I do agree that neutral (white) and hot/live (black) wires get reversed quite often enough that it could be a problem. It's why decent outlet testers will cross-check the connections between hot, neutral, and ground.


    Quote:
    All capacitors have a tolerance rating that can range from -20% to as high as +80% for aluminium electrolytic’s affecting its actual or real value.
    - Electronic Tutorials, Capacitor Characteristics, http://www.electronics-tutorials.ws/capacitor/cap_3.html

    Sometimes caps do have other tolerance ratings, but they are usually marked when they do. The 600uF cap was more than likely within tolerance (min 544uF) while the 34uF cap was definitely not in tolerance and was bad. The 47uF cap measuring 112pF (min 37.6uF) was bad. The 22uF cap measuring 18uF was within tolerance. (min 17.6uF) (Mins given at -20% tolerance)

    Changing a voltage filter cap to a higher value has no harm other than it can have a higher "charging" current as it charges to the full voltage available to it. Changing it to a lower value will increase noise and ripple in the output voltage. Changing other caps, such as in timing circuits or frequency dividers with other values can change timings and potentially cause problems. It's usually not an issue (other than cost) to use a cap with a higher voltage rating than what is needed. It is important to observe polarity on polarized caps unless you want it to go BANG! Design, age of the design, and the materials called for by the design will affect size, leakage currents, and ESR of the caps, especially electrolytic caps.
  • Daniel Sauvageau
    35532 said:
    Hey I was just looking over this again in Spartan/Edge with reader mode and notice those "rubycon"caps are also fakes.

    In my old junk box which I trashed a few years ago, I also had "Lubycon", "Samxung" and a few other similar spoof-sounding names. The first time I came across a "Fuhjyyu" capacitor, I thought it was a spoof attempt on Fujitsu's name.

    Since many asian languages lack phonetics to clearly disambiguate those spelling variations, I bet many products ended up with these devices in them simply due to confusion or creative interpretation in the supply chain.
  • Crashman
    330834 said:
    I do agree that neutral (white) and hot/live (black) wires get reversed quite often enough that it could be a problem. It's why decent outlet testers will cross-check the connections between hot, neutral, and ground.
    I blame the difference between color codes of AC and DC circuits. Most DC circuits are "Negative Ground" and black is the "Negative" wire, so it gets connected to the chassis. Power supplies are both AC and DC, so the white wire is ground on one side and black is the ground on the other side, if you connected black to black you'd energize the case.

    I tried explaining this to an ELECTRICAL ENGINEER who couldn't figure out how to wire up a motorcycle. Seriously, he kept saying stuff about the stupid motorcycle company hooking up all the hot wires to ground and he was trying to fix it (he was actually converting it to positive ground and couldn't figure that out). The thing is, this guy designed control systems for manufacturing and it was all AC and wiring diagrams in his mind.
  • Daniel Sauvageau
    330834 said:
    Sometimes caps do have other tolerance ratings, but they are usually marked when they do.

    Technically, all components have tolerances on all parameters worth listing. Capacitors' nominal capacitance just happens to be notorious for having some of the widest tolerances in all of electrical engineering.

    In the case of the cap that still read 600µF on the multimeter, it turned out to have about 4 ohms of ESR in my step test when I revisited it. The reason it let so much noise through despite having a "good" capacitance is the ESR: the capacitor is no good at suppressing ripples at 10+ kHz if there is a 4 ohm ESR in front of it. If there is 1A of ripple current from the flyback, you get 4V worth of ripple instead of the expected 0.1V.
  • pjmelect
    I liked this article, It is important that the replacement capacitors are 105C type, as the normal 85C type capacitors would not last very long. I often repair power supplies and I use a ESR meter to test in circuit all of the electrolytic capacitors in the power supply which only takes a few minutes.
  • nukemaster
    Is low ESR caps wasted on normal 60hz power filtering(diode rectifier after a low voltage AC transformer)?
  • Someone Somewhere
    8708 said:
    330834 said:
    I do agree that neutral (white) and hot/live (black) wires get reversed quite often enough that it could be a problem. It's why decent outlet testers will cross-check the connections between hot, neutral, and ground.
    I blame the difference between color codes of AC and DC circuits. Most DC circuits are "Negative Ground" and black is the "Negative" wire, so it gets connected to the chassis. Power supplies are both AC and DC, so the white wire is ground on one side and black is the ground on the other side, if you connected black to black you'd energize the case. I tried explaining this to an ELECTRICAL ENGINEER who couldn't figure out how to wire up a motorcycle. Seriously, he kept saying stuff about the stupid motorcycle company hooking up all the hot wires to ground and he was trying to fix it (he was actually converting it to positive ground and couldn't figure that out). The thing is, this guy designed control systems for manufacturing and it was all AC and wiring diagrams in his mind.


    Oh... wow.

    I didn't realise the US used black for phase and white for neutral - I'd seen the wire colouring in the odd power cable etc. that comes with US gear, but...

    We use black for neutral here down under. Red (or yellow (was white until a decade or so ago, but UV turns everything white) or blue if three phase) is phase. Many appliances are built to European standards with brown phase and blue neutral.
  • Shankovich
    These articles, this is why I always love Tom's
  • Shankovich
    These articles, this is why I always love Tom's