the gutless, bloated, and fried power supply hall of shame

[Dannyx]

Re: the gutless, bloated, and fried power supply hall of shame

Ok, you definitely went further than I did. I not only scrapped the PCB - I crushed it up so it'd fit in the recycling basket easier Same with the transformers and toroids. I kept some at home when I was still a kid as "souvenirs", since they looked neat and shiny

I also did the same exact thing as you did, judging by the picture: didn't bother with the supervisor IC - most likely useless. If a PSU DOES exhibit supervisor IC issues indeed, it's likely not worth bothering with and those that are expensive enough to justify such effort will very likely not use such an IC altogether

EDIT: I see something went pretty wrong on that PCB at one point. The area where the +5Vsb transistor is usually located (if I know my ATXs ) is charred, as is the area around those 3 resistors on the left. The cause of the heat wasn't likely the transistor itself, since it's mounted high up away from the board, but rather something around it like a smaller transistor or pull-down resistor...just a hunch...

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

Ok, you definitely went further than I did. I not only scrapped the PCB - I crushed it up so it'd fit in the recycling basket easier Same with the transformers and toroids. I kept some at home when I was still a kid as "souvenirs", since they looked neat and shiny

Yup, I did that a lot as a kid as well - kept cool-looking electronic components without really knowing what they were meant for or what I could use them for. Interestingly, many years later when I learned, some of them turned out to be quite valuable for various projects I did.

Then again, I have the mind for it. Most people I know who store stuff away typically forget they have them after some time. I rarely do. Probably know where a good 90% of all of my junk/scrap parts can be found in my house (despite me not being that organized.)

I also did the same exact thing as you did, judging by the picture: didn't bother with the supervisor IC - most likely useless. If a PSU DOES exhibit supervisor IC issues indeed, it's likely not worth bothering with and those that are expensive enough to justify such effort will very likely not use such an IC altogether

Well, the reason I left it on there is because it was toast. 5VSB circuit committed suicide and killed it.

On that note, the IC is not a supervisor. It is a DBL494 half-bridge PWM controller. Those can actually be fairly useful if building your own power supply or buck converter (can do a simple single-ended 2-phase VRM, like some old ECS motherboards did.)

IEDIT: I see something went pretty wrong on that PCB at one point. The area where the +5Vsb transistor is usually located (if I know my ATXs ) is charred, as is the area around those 3 resistors on the left. The cause of the heat wasn't likely the transistor itself, since it's mounted high up away from the board, but rather something around it like a smaller transistor or pull-down resistor...just a hunch...

Yes, indeed it did.

One hot summer, I came home after vacation, and the family PC we had no longer turned On. A neighbor who used to work in IT came over and looked over the PC. He tried jumpering the PS-ON to ground, and still nothing happened. So he told us it was the PSU and that it needed to be replaced. A new PSU indeed got the PC back up and running.

So this PSU ended up headed for the trash can. I was still in middle school back then and just starting to get interested in electronics (started mostly from my interest around RC cars and motorizing my LEGO Technics projects ). First thing I grabbed from it was the fan and some wires. A few years later when I started soldering, I completely desoldered every component just for practice.

Anyways, it wasn't until many years later after I joined BCN that I inspected the PCB on this PSU and finally understood what the issue was.

Turns out, this PSU used a single transistor self-oscillating design for the 5VSB circuit (a cheaper and/or simpler version of the 2-transistor self-oscillating design). Basically, with that design, the primary side of the circuit is always pulsing for a set power level (these circuits are typically meant for very low power) regardless of how much power the secondary side is using. Since there is no opto-coupler, there is NO FEEDBACK. As such, this design is extremely inefficient, and a lot of power ends up getting dissipated by the primary-side snubber when the 5VSB is not loaded.

So over time, the repeated heat output from the primary snubber probably over-heated other components and caused the 5VSB to get into thermal (and voltage) run-away.

Since the 5VSB also provides power to the PWM chip when the PSU is in Soft-off, this is why there are burned resistors on that PSU: when the 5VSB ran away, it caused a massive over-voltage on the output, and that killed the PWM chip and some of the resistors connected to it.

Now you might wonder how my motherboard survived.
The answer is, since there is no feedback on these single-transistor designs, this means the output voltage will swing wildly based on loading (9 to 15V is the norm). To get stable 5V on the 5VSB, PSUs that use this circuit always employ a 7805 linear regulator (IC701 on that PCB.) The 7805 takes the "unregulated" 9-15V output and converts it to 5V. Since 7805 regulators are rated for up 40V input, the 5VSB would really have to go insane to kill it. Meanwhile, the PWM controller is typically connected to a second winding on the 5VSB that is 5-10V higher, so that's why the PWM controller smoked before the 7805. Otherwise, the DBL494/TL494/KA7500 are all rated for up 40V max input too.

And that concludes the failure analysis here.

[goodpsusearch]

Re: the gutless, bloated, and fried power supply hall of shame

Beware! The most powerful Powertech power supply ever posted here!

Powertech ATX-550W

label states dual 12V rail and 17A on each one with no max combined output limit!

The fan brand is "Cooling fan". Cool

There is also a full input filter, except the MOVs because they are expensive!



This unit broke the chinese tradition and even has an IC controlled 5vsb :
em311Z

I searched but couldn't find a datasheet.

half bridge psu with 2x 13007 again

the rectifiers for 3.3V and 5V are MBR2545CT

and the rectifier for 12V is mur2020ct, rated for 20A.

So, it is indeed really powerful compared to the other Powertech units.

And there is a fan controller with thermistor ghetto glued on secondary heatsink

I wanted to experiment with this psu.

Since it is one of the less gutless Powertech/Hantol/aqprox! power supply, I wanted to try to make it more safe and try to improve it to be able to output some more power with ripple in specs or almost in specs.

And all of the above without spending even 1 euro on new parts.

First thing that I did to make it more safe for a pc was to remove the PCI-E 6 pin connector
Now, I can be sure that no one will try to use it with a power demanding gpu!

Then, I tested the 2x470uF Canicon primary caps on my esr meter and found that their real capacitance is 330uF!

So, I replaced them with real 470uF LCZ caps because I find 330uF to be too low even for a 200-250W psu.

I also replaced the caps on all the output filters except negative 12V line.

5vsb: 1 x 680uF Canicon "Low ESR" and 1x 470uF HQ "Low ESR" -> 2x 1000uF 10V "saturn" logo caps, taken out from Deer/Allied psu.

3.3V: 2x 1000uF 10V HQ "Low ESR" -> 1x 2200uF 10V TNR & 1x 2200uF 10V CS

12V: 1x Canicon 2200uF 16V "Low ESR" -> 1x Capxon KM 3300uF 16V

The PCB had spaces for 2 caps for 5V, but only one was populated. The other space on PCB had a ridiculously low ohm minimum load resistor, 22ohm! But there was also another minimum load resistor for 5V, with value of 150 ohm if I remember correct.

So, the actual resistance for 5v was 22ohm//150ohm! I removed both of them and put in place a 51ohm resistor, trying to bend it to be as far as possible from nearby capacitors. I did the same with all the other minimum load resistors.

5v: 1x 1000uF 10V HQ "Low ESR" -> 1x 2200uF 16V Canicon "Low ESR" removed from 12V output & 1x 2200uF 10V CS

-12V: 1x 470uF 16V HQ "Low ESR" -> unchanged

Unfortunately, while working on the recap, the output cables started to get cut from the PCB on their own. The wire was crazy thin and the soldering very bad. I had to unsolder and resolder all the cables and while at it, replace the ATX 24 pin connector with one taken from another power supply.

I also oiled the fan.

The voltage regulation of this unit without any load connected to it was terrible. 3.3V was at ~3.01V, 5V, 5.39V and 12V climbed at 12.57V approximately.

After the recap and the replacement of the minimum load resistors, things got worse:
3.3V was now ~2.96V, 5V, 5.41V and 12V climbed at 12.69V approximately.

Then I tested psu on an old LGA 775 Pentium 4 3GHZ HT machine. Voltages were measured while pc idle and found normal now: 3.31V, 5.16V and ~12.27V.

In this latest one is the PFC connector spot maybe even upgraded so it sits between live a neutral and not just off to the side not connected to anything?

On this unit there is space on PCB for -12V coil but there is no coil and it is not jumpered either:
https://www.badcaps.net/forum/attach...1&d=1592695278
So, I was wondering how come the negative 12V have any output at all. Then, checked backside of the PCB and found this:
https://www.badcaps.net/forum/attach...1&d=1592695278
They did it again

[Per Hansson]

Re: the gutless, bloated, and fried power supply hall of shame

Hahaha yea that is just awesome, shows what Momaka said on the previous page:
Like I said... old Chinese traditions must be observed.

There is also another beautiful design decision, a silk screen ground designator that is not part of the ground fill?
-Well of course, we don't want to leave the -12v coil alone!

They did part from the PFC tradition though, it looks like there is not even any spot for it anymore, not even a fake spot

[goodpsusearch]

Re: the gutless, bloated, and fried power supply hall of shame

I also oiled the fan.

The fan controller is good for a gutless power supply like this. The fan moves a fair amount of air, without getting too loud.

I am going to donor an old dual core PC:
cpu will be something like E2160 with integrated graphics on motherboard, DDR2 memory and old HDD and an optical drive. Oh, and 2 case fans.
Do you think it could handle it? Will the ripple be in specs after the modifications?

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

E2160, iGP, DDR2, 1x HDD, ODD and 2 case fans... even a L&C w/ complete filtering (I have lots of these around) will handle it, at least from my experience. That torroid looks up to it.

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

I wanted to experiment with this psu.
...
First thing that I did to make it more safe for a pc was to remove the PCI-E 6 pin connector
Now, I can be sure that no one will try to use it with a power demanding gpu!

Excellent!

As a bonus bit, you also now have a spare 6-pin PCI-E power cable that you can add to a more worthy power supply.

Then, I tested the 2x470uF Canicon primary caps on my esr meter and found that their real capacitance is 330uF!

Standing up to their name, pretending to be 470 uF caps when they aren't - trying to get us conned!

3.3V: 2x 1000uF 10V HQ "Low ESR" -> 1x 2200uF 10V TNR & 1x 2200uF 10V CS

Lol, those Fuhjyyu TNR and CS-logo caps are CWT/Antec and Deer/Macron crap cap "classics". But with good ventilation and low load, sometimes they do hold up OK. If they are still in spec after all these years, they sure as hell beat Canicon and HQ "LOWESR" caps.

12V: 1x Canicon 2200uF 16V "Low ESR" -> 1x Capxon KM 3300uF 16V

LOL, more Canicon garbage. Good to see you get rid of them. You know a cap brand is bad when even CapXon is considered an upgrade.

The voltage regulation of this unit without any load connected to it was terrible. 3.3V was at ~3.01V, 5V, 5.39V and 12V climbed at 12.57V approximately.

After the recap and the replacement of the minimum load resistors, things got worse:
3.3V was now ~2.96V, 5V, 5.41V and 12V climbed at 12.69V approximately.

Then I tested psu on an old LGA 775 Pentium 4 3GHZ HT machine. Voltages were measured while pc idle and found normal now: 3.31V, 5.16V and ~12.27V.

If I had to guess, probably the compensation circuit on this PSU was designed... err... I mean -copied- incorrectly.

That's the only thing even cheapo L&C/Deer PSUs have going for them - even when they are build like crap, they still tend to have good voltage regulation. Just never mind the tsunami ripple waves when they install only one or two 1000 uF caps on each rail with no PI coil. On that note, poor ripple suppression is probably worse for a PC than poor voltage regulation. So I wouldn't be too worried about your recapped PSU. If it works well when loaded, it should be OK.

On this unit there is space on PCB for -12V coil but there is no coil and it is not jumpered either:
https://www.badcaps.net/forum/attach...1&d=1592695278
So, I was wondering how come the negative 12V have any output at all. Then, checked backside of the PCB and found this:
https://www.badcaps.net/forum/attach...1&d=1592695278
They did it again

^^

What amazes me with these cheapo PSUs is how sometimes they have design decisions that purposely make them crap without being as much cost-effective. I mean, if they aren't going to install a coil and make a PCB revision to bypass the coil spot, why not even delete/remove the coil altogether from the silkscreen? Why waste PCB space or silkscreen ink? After all, it might just save them 1/1000th of a penny per 100 units in silkscreen ink.

Hahaha yea that is just awesome, shows what Momaka said on the previous page:
Like I said... old Chinese traditions must be observed.

It gets even better when they forget they followed a "tradition". Case in point: installing a PI coil in a spot that was already bypassed on the PCB design. I saw that in a cheap L&C PSU I got for free a decade ago. I think I even posted it here somewhere.

There is also another beautiful design decision, a silk screen ground designator that is not part of the ground fill?
-Well of course, we don't want to leave the -12v coil alone!

Good catch there!
Fortunately (perhaps by miracle), they did happen to leave the other screw hole on the secondary side still connected to secondary ground. So rest assured, the PSU is grounded.

On that note, I wonder if they did that because they are not installing the screws on some PSUs. I've seen some really cheap PSUs omit installing a screw or two (on purpose) to hold down the PCB. Same with fans (install only 2 out of 4 screws... because ya know it - screws cost a lot of money$! )

They did part from the PFC tradition though, it looks like there is not even any spot for it anymore, not even a fake spot

Maybe they never intended for this PSU to be sold over in EU, but did anyways, because... it's just business, nothing personal.

I am going to donor an old dual core PC:
cpu will be something like E2160 with integrated graphics on motherboard, DDR2 memory and old HDD and an optical drive. Oh, and 2 case fans.
Do you think it could handle it? Will the ripple be in specs after the modifications?

That's very kind of you.
Yeah, your recapped and modded PSU should be able to handle that system just fine, both in terms of load and ripple suppression. Hopefully the person you give it to doesn't get any malicious ideas of adding a high-power GPU and then using adapter cables to power it (on that note, maybe even cut out all of the SATA and Molex drive connectors, leaving at most only one spare of each?)

In any case, an E2160 won't pull more than ~100W on the 12V rail with Intel Linpack stress-testing. So with all everything else integrated, I imagine the PC will draw 110-125 Watts absolute max. The caps you installed should be fine to let the 12V rail get loaded down to 150-190 Watts with still very little ripple.

[RJARRRPCGP]

Re: the gutless, bloated, and fried power supply hall of shame

CS-logo caps

Those were an old Coolmax PSU classic, saw them bulging and leaking at a computer recycling center. (WinCycle, Windsor, Vermont, when I was taking apart PCs to put the PC stuff in a bin. I saw the Coolmax PSUs, two of them in a bin, IIRC)

It was the recycling version of a trash bin. But, in the same monster-sized room, there were used CRTs ready to be sold! Looks like those days are gone!

[goodpsusearch]

Re: the gutless, bloated, and fried power supply hall of shame

Excellent Power IT ATX 350

But I couldn't find anything excellent about this psu.

It's a Deer, the 2003 platform with the 2003 chip of the year.

The fan seized, the 12V output uses the 2 diode on a bracket treatment, some caps bulging, but the thing still works!

SB1040CT for 3.3V

sbl1640ct for 5V

2x 13007 main switchers

and 2x 330uF 200V ultra small primary capacitors.

And of course a lot of space on PCB for extra caps and coils that have not been installed

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

Excellent Power IT ATX 350

But I couldn't find anything excellent about this psu.

Well, that's because you're looking at it the wrong way.

See, this PSU is an excellent example of cost-cutting to-the-max.

That label, tho!
- It has more lies and fake promises than a political campaign.
PFC? Fan control?? All those power ratings???

It's a Deer, the 2003 platform with the 2003 chip of the year.

The fan seized, the 12V output uses the 2 diode on a bracket treatment, some caps bulging, but the thing still works!

Yup, it's their early/mid-2000's BTX design.

When fully built with all parts installed and of proper size, they actually make decent 250 Watt PSUs. Of course, those are extremely rare, lol. But with all of the missing components like this, you'll be lucky to pull more than 50-100 Watts without the ripple going out of spec.
... not that you could with those rectifiers. I mean, 10 Amps on the 3.3V rail and DoaB (Diodes-on-a-Bracket) for the 12V rail?! This thing will struggle probably even with a low-power Celeron... nevermind a P4. And with that anemic output filtering, even a Pentium II/3 can be too much. Just ask me how I know - I have a PSU based on the same platform as this, but built only a hair better (TO-220 rectifiers on all rails, at least... though not much more than that going for it.) It kept crashing with a Pentium 3 PC and a more "powerful" S3 Savage 4 card - especially when I added a second stick of SDRAM to the system. Total draw was probably 40-45W max, and that became too much for the PSU, so ripple was way out of spec. I suspect the regulation on the 3.3V rail is what threw things off, because SDRAM often is fed directly from the 3.3V rail on many old boards. So unstable 3.3V rail = unstable system. With a more modern mobo, this PSU could probably be pushed for higher draw without causing anything to crash... but the ripple would be even worse, still.

And of course a lot of space on PCB for extra caps and coils that have not been installed

I guess you can say this PSU has eXcellent airflow.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

Does it become a night light if you draw more than 200W of power?

[goodpsusearch]

Re: the gutless, bloated, and fried power supply hall of shame

Top Power 500W

Not to be confused with topower OEM

Another boring cheap psu.

But I really like the names they give their fans. This is the Bestcooler fan brand.

Did you notice anything unusual? The primary caps are Chemicon 200V 680uF .

I put those out and tested them some time ago. Those are fakes. Real C was a lot smaller than the rated.

Someone stole the input filtering because the psu doesn't have any.

Nice that they put a proper bridge rectifier...

And pi coils on all output rails (hope they are not fake ).

the main switchers are 2X d304x , probably good for 300W.

5V and 3.3V have 20A schottky rectifiers but 12V has ultrafast 16A rectifier. This is really bad for a modern power supply.

Also, notice the big resistor next to the PWM IC getting too hot and discoloring PCB.

[Per Hansson]

Re: the gutless, bloated, and fried power supply hall of shame

That nameplate is a legend, it shows the fake rating of 5v in one column and then the real 5v rating in the next column?
Also the S, N and FI ratings are very obvious fakes.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

That nameplate is a legend, it shows the fake rating of 5v in one column and then the real 5v rating in the next column?
Also the S, N and FI ratings are very obvious fakes.

I think the second "+5V" column is for the -5V output. IMO, the chances the -12V and -5V outputs are capable of 1A are pretty low if they are regulated by a 78xx 3-T regulator.

The Semko and Nemko marks are pretty badly done, .

I'd guess the realistic output power at 250W-300W. Do NCC KMH series large-can lytics have black sleeves? Or the usual NCC brown?

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

Also, notice the big resistor next to the PWM IC getting too hot and discoloring PCB.

Yup.
Bad caps on the 5VSB output will do that to a 5VSB circuit with a 2-transistor design. Usually, the 5VSB will start to "loose" voltage/regulation, but the secondary-side auxiliary that feeds the PWM controller will go up and up and up until the PWM bites the dust. The secondary-side aux. rail is typically anywhere from 12-20V on a properly-working 5VSB (that has a sec. side aux. rail.) On a 5VSB with bad output caps... well... conisder this: DBL494 / KA7500 PWM controllers are rated for up to 42V DC input and usually have a "dropper" resistor right before them and the sec. side aux. rail. So for that "dropper" resistor to burn or for the PWM IC to burn, imagine how much higher the voltage goes.

That's why I never ignore bad caps in a 5VSB circuit on a PSU I care about. Even if they test good, leaving potentially crappy brands on the output of a 2-transistor 5VSB circuit is pretty much playing Russian roulette with the PSU (and sometimes the attached hardware too.)

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

That's why I never ignore bad caps in a 5VSB circuit on a PSU I care about. Even if they test good, leaving potentially crappy brands on the output of a 2-transistor 5VSB circuit is pretty much playing Russian roulette with the PSU (and sometimes the attached hardware too.)

QFT. Almost all Deers I have sport either Sanyo, Rubycon, Nichicon or UCC KZE in the 5vSB section. At least if anything fails first, it ain't gonna be the standby rail.

[goodpsusearch]

Re: the gutless, bloated, and fried power supply hall of shame

T&P-LIW Meiji-420 ATX

No name junk psu.

The input filter is missing and the fuse is blown.

The shape of the legs of the 4 diodes is interesting.

All the output rails rectifiers are rated for 20A continuous but 12V is BYV32E-200 ultrafast, not schottky.

I can see some pin coils on the output rails.

main switching NPN transistors are 2x 13007

[Per Hansson]

Re: the gutless, bloated, and fried power supply hall of shame

There is a hole in a resistor on the primary side here, under the heatsink.
Really beautiful fuse, they cause so much mayhem when they go like that lol!

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

PhotonicInduction to fuse: "I popped it!"
Curious what the failure was.

Garbage PSU is garbage, though.
Like the rest of these, I suppose:
https://www.badcaps.net/forum/showth...&highlight=KYS

But, hey, at least you can scavenge some 20 Amp rectifiers out of the one here now.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

The thing's 13 or 14 years old (PCB datecode 0638), so it lasted a while. But the heatsinks and main transformer put it in the 200W-250W range, IMO. It's be an EMC nightmare, and the output noise probably sucked.

I'm going to guess one or both of the 13007s are silicon slag inside. That glass fuse (!!!) met a violent end.