Power supply build quality pictorial. part 2

[Togepi]

Re: Power supply build quality pictorial. part 2

Hello,

here I have a Cooler Master power supply MasterLite 600Watt. It is new and unused. All capacitors in the device appear to be from CapXon. I don't know if this is good or bad now. The device is new and unused. I would have been interested in who the right manufacturer of the power supply is. Can someone see that from the pictures?





Regards

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

CapXon is a brand not respected here. However it looks like some of the output capacitors are polymer types.

The power supply puzzles me. It is rated as simple "80 Plus", but those heatsinks look insufficient for a 600W power supply. The main transformer also looks odd. My best guess is that the switch frequency is somewhat high, and phase modulation is used rather than pulse width modulation. But if that's the case, I would think it would be rated as at least "Bronze" or "Silver".

[pdavid]

Re: Power supply build quality pictorial. part 2

A saved an old ATX power supply from recycling last week.
The casing is damaged, has dents, a broken fan but the insides are intact.

EverPower brand with a 235W Channel Well Tech guts from 1997.


There is a seperate transformer attached to the casing that looked like passive PFC at first but it isn't. The 5V standby has 100mA current capability and it is generated using this trafo, rectifier and a 7905 regulator. I haven't seen this "cheap" method before.
The standby trafo has primary winding taped for both 110V and 230V.


The unit has ok EMI filtering, 4 Amp bridge rectifier, 330uF Matsushita input capacitors.
It's a half bridge topology with two 13007 BJTs controlled by KA7500 pwm.



A 339 comparator is used to supervise the output voltages. There is also a seperate fan controller circuit.

https://www.badcaps.net/forum/attach...5&d=1637581493


The 3.3V rail is also unusual. No mag amp circuit.
A UTC34063 is set up in step down configuration to make the 3.3V work.


Secondary electrolitic capacitors are all Fuhjyuus so that's bad.


I've cleaned and tested the psu. Turns on, voltages are a bit low-ish and fluctuating, ripple is probaly high also. Kind of unusual early ATX unit.

[Agent24]

Re: Power supply build quality pictorial. part 2

What a fascinating design! I want one just for the sheer novelty.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

There is a seperate transformer attached to the casing that looked like passive PFC at first but it isn't. The 5V standby has 100mA current capability and it is generated using this trafo, rectifier and a 7905 regulator. I haven't seen this "cheap" method before.

The current rating for the +5SB is just .1A, some the effect on efficiency is minimal. A 50/60 Hz transformer, recifiers, and 7805 3T regulator might have been cheaper than a flyback switching regulator circuit. .1A was pretty minimal for that era. I would expect 1.0-1.5A.

A TL494 clone and MJE13007 clone BJTs would easily have been adequate for the ATX spec of that era and the 235W rating; the heatsinks were OK, if not spectacular heavy metal. Looks like the PCB was laid out to allow higher power switch transistors and output transformer. The 330uF input caps might have been marginal for hold-up time. The fluctuating output voltages and high ripple could be due to the 24 year old output caps having increased in impedance. Replacing the F-yus with Nichicon PJ or PW or NCC LXV or LXZ series caps (which were in common use in the late 1990s) might result in better performance.

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Earliest ATX 1.0 spec indeed only specced only 100mA so I suspect they may have built to spec though whether or not they exceeded or not is another question. My question is why is it using an overspecced and odd 7905 negative rail regulator when a cheaper 78L05 or 78M05 would do? Incidentally I'd not call this "cheap" but rather "old, inefficient" ... those transformers are not cheap to source or ship in volume...

I found a couple of old Fuhjyyu's in some device that predates the old Antec Smartpower Fuhjyyu plague and they actually are still okay... Not sure if's just that one period with the scandal that they produced so many poppers...

[pdavid]

Re: Power supply build quality pictorial. part 2

My question is why is it using an overspecced and odd 7905 negative rail regulator when a cheaper 78L05 or 78M05 would do?

Sorry, 7905 was a typo. It is a L7805C positive voltage regulator. It is visable on the pictures. The ic has that simple piece of metal screwed on for heatsinking. The pcb silkscreening suggests a TO-92 device instead of TO220 that is crammed in there.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

I found a couple of old Fuhjyyu's in some device that predates the old Antec Smartpower Fuhjyyu plague and they actually are still okay... Not sure if's just that one period with the scandal that they produced so many poppers...

The reason those caps failed a lot in Antec psus has to do with how hot they got due to the design of the parts location inside and also the fact that the fan controller was configured very aggressively to be silent. And the fact that their efficiency was nowhere close to what we have today didn't help either (~74%)

[eccerr0r]

Re: Power supply build quality pictorial. part 2

I'd imagine back when these Fuhjyyus were installed it was before the Antec era and efficiencies were even lower...

We'll see I suppose, so far they're surviving the time.

[Pentium4]

Re: Power supply build quality pictorial. part 2

That's a very interesting unit. The longer you look at it, the more odd it is. Never seen anything like it. Thanks for sharing. What brand was the fan that failed? What are you going to do with it?

[momaka]

Re: Power supply build quality pictorial. part 2

A saved an old ATX power supply from recycling last week.
...
EverPower brand with a 235W Channel Well Tech guts from 1997.

WOW!
This must have been one of their earliest ATX-compliant designs.
I mean, it has a regular transformer for the 5VSB instead of a switching circuit - that just says A LOT about this PSU's age.

And yet even this early design seems to have OPP (over-power protection) on the primary side, as is evident by that small toroid between the main transformer and the BJT driver transformer.

The 3.3V rail is also unusual. No mag amp circuit.
A UTC34063 is set up in step down configuration to make the 3.3V work.

Interesting.
But I only see one MOSFET and no free-wheeling diode(s)... though it is a bit hard to see. So perhaps this is done in a linear fashion. Kind of strange why there would be a separate toroid for the 3.3V rail then. Seems the 3.3V rail is generated by the rectified and filtered 5V rail output (two two red wires going from in front of the main output toroid, under the heatsink, and out in front of the 3.3V rail's MOSFET.)

It is a L7805C positive voltage regulator. It is visable on the pictures. The ic has that simple piece of metal screwed on for heatsinking. The pcb silkscreening suggests a TO-92 device instead of TO220 that is crammed in there.

CWT probably figured at some point that the TO-92 regulator would not do too well. If the 5VSB transformer is outputting 9-10V (or possibly even higher when unloaded), then at 100 mA of power draw, the dissipation of the regulator would be P_d = (9 - 5) x 0.1 = 0.4 Watts. For a small TO-92 device, that's probably pushing it really close to the max allowable P_d (at least as far as internal junction temperature of the silicone is concerned.)

I've cleaned and tested the psu. Turns on, voltages are a bit low-ish and fluctuating, ripple is probaly high also. Kind of unusual early ATX unit.

Cool!
Yeah, this is definitely worth saving, even if just as a "museum" piece if nothing else. Though I suspect some new caps (that PeteS suggested) would certainly bring it back to a fully serviceable ATX PSU.

The reason those caps failed a lot in Antec psus has to do with how hot they got due to the design of the parts location inside and also the fact that the fan controller was configured very aggressively to be silent. And the fact that their efficiency was nowhere close to what we have today didn't help either (~74%)

Yup, I think that's pretty much it.
Well, either that or CWT designs just pushed Fuhjyyu caps harder somehow?
IDK, but I've seen early Fuhjyyu caps last for years and still read in spec - but never in a CWT build. Could be just what I happened to encounter, though. I am by no means suggesting Fuhjyyu caps should be trusted, of course. I'd say the are on par with CapXon... or maybe worse. Depends which series of CapXon we look at. Generally, they are all not good, but some are more "not good" than others if that makes sense.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Recently took a Fuhjyyu cap to reuse it on an old TFT LCD monitor that I didn't care too much about.

The Fuhjyyu cap have been removed from Antec power supply 10 years or more ago.

Even though the cap tested fine on micro esr meter and chinese transistor meter that shows also V leakage as soon as it was soldered and power was applied on monitor, the cap failed without venting after this.

It had no capacitance and only showed some quite high resistance. Out of curiosity, tested more Fuhjyyu caps and found at least 1 more sitting in storage that had gone too high V leakage and capacitance also appeared higher that the nominal.

They can go bad even in storage if they have been abused previously

[TechGeek]

Re: Power supply build quality pictorial. part 2

They can go bad even in storage if they have been manufactured previously

ftfy

[eccerr0r]

Re: Power supply build quality pictorial. part 2

A non-ATX PSU that still a PC PSU....
This apepars to be an ePower EP-450CD UNI, a "450W" PSU.
This is an auxiliary PSU that fits in a drive bay to power your mining cards when your main PSU won't cut it. Unfortunately it's modular and I hate modular... since I have no cables.

Well I was able to find some way to hook up power with a piece of the monitor I dropped, I believe. With it and an old AT PSU it seems to power up and i got 12.66V unloaded on one of the outputs. Anyway, sly thing, says 20A per 12V rail but it appears to be the same paralleled TO247 diodes and another diode that "isolates" the two rails.
This supply does not supply +5v or +3V3, it's purely 12v.

It appears to be a APFC cm6800 based PSU with a 7510 monitor. I didn't check to see how it dealt with missing 5v but it must someway. To turn this supply on, a 4 pin molex drive cable is plugged into the back though I don't think it back feeds it.

Power was the annoying thing as it had one of those "square and D' type 3-prong plugs. Not sure what the proper name of it is.

Did not carefully check all the caps but they do not look like they've leaked despite the fuhjyyu's. Also did not load test due to lack of connectors.

So...450W? eh... possible. Though I'm not sure what I can do with it until somehow a bunch of gpus drop on my desk so I could start mining or whatnot... It'd be nice to at least power hard drives with it, but the lack of 5V might be a problem (I do have a 50W 12V to 5V step down that i could use...)

[Pentium4]

Re: Power supply build quality pictorial. part 2

Interesting, I've had one of these before but the design was quite different. Mine had a switch on the back. As long as it was plugged into the wall, the switch would always turn it on.

How many A are the parallel diodes rated for? Looks a little sketchy for 450W but all 12V would help with efficiency and the output starting at such a high voltage is a promising sign. I bet the fans would be annoying as hell if it could do 450W lol. Then it would turn the glue conductive and it would explode anyways!

[momaka]

Re: Power supply build quality pictorial. part 2

A non-ATX PSU that still a PC PSU....
This apepars to be an ePower EP-450CD UNI, a "450W" PSU.
This is an auxiliary PSU that fits in a drive bay to power your mining cards when your main PSU won't cut it.
...
So...450W? eh... possible.

With a solo 400V 220 uF primary cap? (And Fuhjyyu at that :\ ) - Probably not. I give it 300 Watts max and maybe 350W peak for very short periods of time. The output toroid also looks a little too small for more than that. But at least it uses Micrometals -52 mix core, so it should be a little less lossy. My guess is it's a T106-52 part.

Anyways... even 300 Watts would still be impressive for such a compact form factor. But again, it's the primary cap that worries me a little, it being Fuhjyyu and rated for only 400V. I imagine with high load and extended use, it would likely bite the dust in a few years. Other than that, the rest of the PSU's build quality looks good. BTW, are those original Panasonic caps on the modular board? That would be interesting if they are.

Though I'm not sure what I can do with it until somehow a bunch of gpus drop on my desk so I could start mining or whatnot...

Bah, don't waste your time (or GPUs) with that crap. The world would be a better place if people stopped buying into this "distributed computing" crap. A good chunk of crypto currencies are used for illicit uses anyways.

It'd be nice to at least power hard drives with it, but the lack of 5V might be a problem (I do have a 50W 12V to 5V step down that i could use...)

Or crank up the output to 14.x Volts (if it let's you) and charge lead-acid batteries?

Unfortunately it's modular and I hate modular... since I have no cables.

I hate when that happens too.
Sure wish there was a cheap source of generic modular cables that I could rewire and make the PSU useful. Unfortunately, even the "cheap" garbage 6-pin / 8-pin PCI-E "splitter" cables on eBay and AliExpress regularly cost as much as a junk PSU (from which I can scavenge such cables and other useful parts), so I haven't bothered to buy any. Otherwise I also need quite a few myself. I've even considered unsoldering the modular connectors and just soldering PSU connectors directly to the modular board. TBH, that's what I'll probably end up doing at some point. It sure would make the PSU harder to take apart if needed to be taken apart again... but I suppose that's not too important if nothing fails. Otherwise, I build my PCs to be functional and not necessarily pretty... so if it looks ugly or "ghetto", I could care less.

[kaboom]

Re: Power supply build quality pictorial. part 2

With a solo 400V 220 uF primary cap? (And Fuhjyyu at that :\ ) - Probably not. I give it 300 Watts max and maybe 350W peak for very short periods of time. The output toroid also looks a little too small for more than that. But at least it uses Micrometals -52 mix core, so it should be a little less lossy. My guess is it's a T106-52 part.

Anyways... even 300 Watts would still be impressive for such a compact form factor. But again, it's the primary cap that worries me a little, it being Fuhjyyu and rated for only 400V. I imagine with high load and extended use, it would likely bite the dust in a few years. Other than that, the rest of the PSU's build quality looks good. BTW, are those original Panasonic caps on the modular board? That would be interesting if they are.

It's only made to "feed" the frenzy: "mining" for suckers.

Bah, don't waste your time (or GPUs) with that crap. The world would be a better place if people stopped buying into this "distributed computing" crap. A good chunk of crypto currencies are used for illicit uses anyways.

The whole thing was a stepping stone to tenderize the masses into "accepting" bitscam and others with similar "flavor."

Deliberate sarcasm is deliberate.

Or crank up the output to 14.x Volts (if it let's you) and charge lead-acid batteries?

If it doesn't let me, I'll force it to! Where's that feedback resistor?

I hate when that happens too.
Sure wish there was a cheap source of generic modular cables that I could rewire and make the PSU useful. Unfortunately, even the "cheap" garbage 6-pin / 8-pin PCI-E "splitter" cables on eBay and AliExpress regularly cost as much as a junk PSU (from which I can scavenge such cables and other useful parts), so I haven't bothered to buy any. Otherwise I also need quite a few myself. I've even considered unsoldering the modular connectors and just soldering PSU connectors directly to the modular board. TBH, that's what I'll probably end up doing at some point. It sure would make the PSU harder to take apart if needed to be taken apart again... but I suppose that's not too important if nothing fails. Otherwise, I build my PCs to be functional and not necessarily pretty... so if it looks ugly or "ghetto", I could care less.

Depending on location of drive bays (yes, drive bays), I'm known to make custom extensions from cables of gutted PSUs or systems.
The connectors end up where I want them, and decent cable management is possible.
Bonus: I can have "flying plugs" on short cables for temporarily powering backup or otherwise "rotatable" hard drives.

[lti]

Re: Power supply build quality pictorial. part 2

I haven't seen those 5.25" bay power supplies in at least 15 years. They were made for people who didn't want to upgrade their power supply for some reason when they got a new high-end graphics card.

[momaka]

Re: Power supply build quality pictorial. part 2

I haven't seen those 5.25" bay power supplies in at least 15 years. They were made for people who didn't want to upgrade their power supply for some reason when they got a new high-end graphics card.

Yeah, I think mid-2000's is when they first started popping up - around the time of Athlon 64 X2/FX, Pentium D, and SLI. Not a coincidence, for sure.

[momaka]

Deer DR-250ATX - revisited 2021

OK, time to re-visit an old PSU of mine… it's the Deer DR-250ATX PSU that's still installed in this Pentium 3 PC. I use this PC very infrequently when visiting family overseas. It's stashed away and disconnected from power for the majority of the time. It gets used maybe 2-3 weeks in a year or less (pretty much only when I visit) to store pictures from my camera on it or to set it up for my nephews to play old games.

Anyways, last summer when I went to visit, it was no different. However, after about a week of use, the computer became unstable – would sometimes boot and work OK, and other times not. After some troubleshooting, I found that if I removed one of the RAM sticks, I could get the machine to boot. Being a Pentium 3 PC with SDRAM, I started suspecting either capacitors on the motherboard (still original OST and Chhsi, IIRC) and the PSU. I didn't have caps for the motherboard, but I did take with me *one* Rubycon YXJ 10V, 2200 uF cap (don't ask why) among some others (but nothing compatible with the motherboard ones.)

Anyways, long story short, I decided to replace one of the caps on the 3.3V rail of the above DR-250ATX PSU – the 2nd one after the 3.3V rail's output toroidal inductor, as it looked slightly raised on the bottom.


Turns out, that cap wasn't bulging on the bottom or anything weird. It was just mounted slightly higher. However, once I got back home and tested the original CS-logo 10V, 2200 uF cap that I pulled from the 3.3V rail, this is what I found:


Ooof!
Sure ESR and Vloss seem “good”… but capacitance is 60% above nominal value! Yeah, it's definitely breaking down, just like the CS-logo caps shown here from the Casing Power MPT-301 I rebuilt not too long ago.

After replacing that one CS-logo cap, the PC ran stable afterwards, but I still couldn't use all 3 RAM sticks. Not sure if this one particular 256 MB stick got zapped by ripple from the PSU or if it's extra sensitive to noise/ripple (it's the only 256 MB module), but I can't run the PC with it for the time being. I suppose next time I go visit, I'll just prepare some caps to take with me and recap the rest of that PSU. I'm pretty sure the other CS-logo cap on the 3.3V rail is bad too. The 1st cap after the toroidal inductor always seems to be under more stress and more likely to fail.

Apart from the failed caps on the 3.3V rail, this Deer PSU is still doing OK in the PC with my partial (actually, nearly full) recap, which I did back in 2010… which now that I think about, is close to 12 years ago. Heck, the PSU itself is over 20 years old now!
Not bad for a Deer. Granted it doesn't see much use… but I don't know if that helps it much or not. After all, caps sitting unused and disconnected from power for extended periods also degrade over time.

So this one is to be continued, once I get to it. As posted back in 2012, I do have the 5V rail and 5VSB rail recapped with good Chemicon caps, along with the “critical” 5VSB cap and many of the other small caps. So I don't expect this PSU to really fail on me any time soon. The whole PC uses maybe 60 Watts of power under load - tops. So the PSU isn't under too much stress, really.