Power supply build quality pictorial. part 2

[momaka]

Re: Power supply build quality pictorial. part 2

Well, I'm not going to ask too much, I'll have it for a gtx8800 with a core 2 duo.

The Core 2 Duo should be no problem, but the 8800 GTX is a power hog at 170 Watts, if I remember correctly. So with a standard 65W C2D, you're looking at around 230 Watts max from the 12V rail, which may be a bit high for it without a more moderate load on the 5V rail.

I will recap the primary capacitors and give them a second life.

The secondary-side caps are under more stress in this unit, so I suggest replacing those instead. The primary caps should be okay - even crap brands live a long time when there is no APFC circuit involved.

I'd treat that like a 250-300W unit with max 20A on +5V/3,3V and 15-18A on +12V combined.

Well, technically since this is a half-bridge design, you could pull the max. current that the output rectifiers are rated for. Of course that may not be possible only due to cross-loading. So this is where experimentation would be needed to determine that. But your ratings are probably not too far what the PSU can do if cross-loaded.

[momaka]

ThermalTake Smart M Series M850W

Here is a ThermalTake Smart M850W PSU (also goes by SP-850AH3CCB?) based on the CWT "PUQ" platform. I'm sure this has been posted here before, but here are the pictures of my unit anyways, in case there is something slightly different...

The unit itself:


Top case:
https://www.badcaps.net/forum/attach...1&d=1504989516

Label:
https://www.badcaps.net/forum/attach...1&d=1497722153

Side sticker:
https://www.badcaps.net/forum/attach...1&d=1504989516

Front (modular connector panel):
https://www.badcaps.net/forum/attach...1&d=1504989516

Top popped off:
https://www.badcaps.net/forum/attach...4&d=1497722153
In case you are wondering... YES, that primary cap is bulged and leaking. It's actually the PSU from this thread I posted a few months back:
https://www.badcaps.net/forum/showth...707#post753707

Primary side:
https://www.badcaps.net/forum/attach...5&d=1497722153

Secondary side (and caps):
https://www.badcaps.net/forum/attach...6&d=1497722153

Sync. rectifier "board":
https://www.badcaps.net/forum/attach...1&d=1504989516

Bottom / solder side:
https://www.badcaps.net/forum/attach...7&d=1497722153
https://www.badcaps.net/forum/attach...8&d=1497722153
https://www.badcaps.net/forum/attach...9&d=1497722153

DC-DC buck regulator board for 3.3V & 5V rails as well as modular connector:
https://www.badcaps.net/forum/attach...1&d=1504989516
https://www.badcaps.net/forum/attach...1&d=1504989516

Fan:
https://www.badcaps.net/forum/attach...1&d=1504989516
.
.
.
Detailed Breakdown
... well, I was too lazy to look up everything and do a proper post here, so I will link to a review of this PSU instead:
https://m.hardocp.com/article/2014/0...upply_review/1

And here is also a review of the 750 Watt version of this PSU:
https://www.techpowerup.com/reviews/.../Smart_M_750W/

The only part numbers I have for the parts inside my unit are as follows:
All of the big MOSFETs on the primary side are IPA6R190E6.
Primary bulk cap is a Panasonic HD 400V, 470 uF.
PFC diode is a QH08t7600 (or something like that).

[momaka]

Re: Power supply build quality pictorial. part 2

Same PUQ platform (version B), different PSU brand...

This time I have a Corsair CX750M for you (model 75-002019). No detailed component break-down again, as there are many reviews of this PSU already. Here they are, if you like to read:
http://www.tomshardware.com/reviews/...-psu,4799.html
http://www.jonnyguru.com/modules.php...Story&reid=416
https://m.hardocp.com/article/2015/0..._supply_review

And here are my pictures. I will start with an overall shot:
https://www.badcaps.net/forum/attach...1&d=1505081452

Label:
https://www.badcaps.net/forum/attach...1&d=1505081452

Front (modular connector panel):
https://www.badcaps.net/forum/attach...1&d=1505081452

Top removed:
https://www.badcaps.net/forum/attach...1&d=1505081452

Primary Side:
https://www.badcaps.net/forum/attach...1&d=1505081452
https://www.badcaps.net/forum/attach...1&d=1505081452

Secondary Side:
https://www.badcaps.net/forum/attach...1&d=1505081452

DC-DC board with modular connectors:
https://www.badcaps.net/forum/attach...1&d=1505081452
https://www.badcaps.net/forum/attach...1&d=1505081452

Solder Side:
https://www.badcaps.net/forum/attach...1&d=1505081452

Fan:
https://www.badcaps.net/forum/attach...1&d=1505081550

Comments:
I got this PSU a few months ago for free because it was said to emit a bad/burning smell. For a long time, I could not find anything wrong with it or where the smell was coming from, so I used the PSU for short testing. But then one day, I used it for about 30 minutes and that smell came back. Turned out to be the output filter inductor for the 12V rail. You can see in the above picture of the secondary side that the wire is quite darkened. Indeed this is what what emiting the smell. After running the PSU again untill it stank and then opening it, I checked the temperature on the output filter inductor, and it was showing over 120°C on my thermocouple meter.

So yeah, this PSU also has a problem, just like its bigger sibling posted above. However, this one has it with the output inductor, whereas the Thermaltake one had a blown primary cap and fuse.

Two major random failures for this platform doesn't exactly inspire confidence IMO. We also had another at work a few months back, but I didn't get to open that one to see what was wrong. In any case, these aren't that reliable, it seems. Or maybe I was just unlucky to see that many go bad.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

390uF for a 750W and 470uF for an 850W is a joke.
680uF should be the bare minimum. 390uF is something you'd expect in a 500-600W PSU, maybe even 650W...

Besides:
Do you know why the cap and the coil failed??

PS: Interesting that you see the PUQ only with some kind of 'modular' module and DC-DC.
Makes me wonder if CWT tried to sell this plattform to OEMs - and they didn't want it.
And the DC-DC module is some kind of plan B to be able to sell this +12V only unit to consumers...

€dit:
WIth the corsair unit, doesn't look the primary switcher heatsink discolored too??

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Come to think of it, are we going back to anodized aluminum heatsinks?
Back in the old days, all heatsinks for like TO-3 and small TO-220 heatsinks were anodized black... most PSUs nowadays I see are bare aluminum...

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Come to think of it, are we going back to anodized aluminum heatsinks?
Back in the old days, all heatsinks for like TO-3 and small TO-220 heatsinks were anodized black... most PSUs nowadays I see are bare aluminum...

Depends on the PSU (Manufacturer) these days.
Some (Super FLower, Sirfa, AFAIR FSP) often had black Heatsinks (at least for the primary side)...

So it depends...
Especially if you want (to pay for) black heatsinks or not...

[Behemot]

Re: Power supply build quality pictorial. part 2

…

Think it's not stock CWT platform, but a modified one. Anyway, heard there are problems with fan bearings who go so noisy after couple months of use. The input bulk cap is also small, as my review of the 550W version showed…and upgrading them caps had also interesting impact on efficiency.

[RukyCon]

Re: Power supply build quality pictorial. part 2

Come to think of it, are we going back to anodized aluminum heatsinks?
Back in the old days, all heatsinks for like TO-3 and small TO-220 heatsinks were anodized black... most PSUs nowadays I see are bare aluminum...

My 600w PSU has all of the heatsinks an anodized a brass looking color

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

and upgrading them caps had also interesting impact on efficiency.

Yes but with the for 16/20ms holdup time needed caps you have to sell the unit for at least 5€, more like 10-15€ more.

Just a quick search:
https://www.reichelt.de/Becher-Elkos...0%25B5F%2B400V
4,75€ for a 400V/390uF Nippon KMR

6€ for the 560uF KMR and the 680uF is 6,70€...

Efficiency isn't the problem, the cost for the bulk cap alone is. And that would have increased the price dramatically and you know...
People care more about other (Marketing) stuff than holdup time...

it's the same bullshit with the 8pin crap protection ICs that can't do nothing...
They are only used because they are cheap and most of them don't even have UVP on +12V...

[jarvis7]

Re: Power supply build quality pictorial. part 2

hello everyone ,

i have a basic query (slightly off topic ),

if a power supply has UVP,OVP,SCP,OCP and some of the bad caps start failing "slowly" like those Capxons and Teapo SC , will the ripple go out of spec ? and can that damage mobo and other components ? or is the PSU going to shut down safely before excessive ripple can damage other components because of UVP,OVP?

talking about commonly found IC's in budget units like WT 7527 , siti PS229 , sitronix ST9S429

thanks

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Yes, Ripple will go out of spec
yes, way out of spec ripple will damage components (over 200mV and more)
no, PSU will not shut down because there is no 'overripple protection'

[momaka]

Re: Power supply build quality pictorial. part 2

390uF for a 750W and 470uF for an 850W is a joke.
680uF should be the bare minimum. 390uF is something you'd expect in a 500-600W PSU, maybe even 650W...

Yeah, that's true.
The only way you'd get better hold-up time with those small caps is if they are rated for 450 V and the APFC charges them up to 420-430V - then the energy stored in the caps will be much higher.

Besides:
Do you know why the cap and the coil failed??

Nope. Both were customer units from what I know.

Though I can tell you that the Corsair came out of an average modern build with a GTX 1050 or 1060 video card and some i7 CPU, if I remember correctly - nothing that drew more than 250 Watts peak under load. So why the output coil burned, I have no idea. Not like anything got in the PSU either - it was fairly clean when I opened it (only a few tiny dust bunnies) and the fan was facing down (in a PC case with bottom-mount PSU). So nothing could have leaked in there to short anything. Looks like just a random failure to me right now. But I could be wrong, of course.

WIth the corsair unit, doesn't look the primary switcher heatsink discolored too??

I think that's just from my pictures, but I will have a look at it again in more light in the next few days.

For now, I am actually thinking of transferring the fuse and primary cap from the Corsair to the Thermaltake PSU so that I have at least one working unit. Not sure what to do about the burned inductor in the Corsair. I'm thinking re-heat and demagnetize, then rewind with new wire.

Think it's not stock CWT platform, but a modified one. Anyway, heard there are problems with fan bearings who go so noisy after couple months of use.

That would be the least of problems for me. I can fix any sleeve bearing to be quiet again and work for a very long time. Ball bearings on the other hand, there's not much to do about them when they go bad.

The input bulk cap is also small, as my review of the 550W version showed…and upgrading them caps had also interesting impact on efficiency.

Well, putting a bigger cap in these PSUs also won't affect the hold-up time at all, since this platform seems to an AC detection circuit with built-in PSU shut-off feature, as I mentioned in my thread. Thus as soon as the AC is disconnected, it doesn't matter if that primary cap has any energy left in it or not - the PSU will shut off after some pre-determined delay.

Efficiency isn't the problem, the cost for the bulk cap alone is. And that would have increased the price dramatically and you know...
People care more about other (Marketing) stuff than holdup time...

True.

it's the same bullshit with the 8pin crap protection ICs that can't do nothing...
They are only used because they are cheap and most of them don't even have UVP on +12V...

Yeah, I was very disappointed that this platform does not come with OCP on its 12V rail. This makes the PSU a potential fire hazard if a SATA connector decides to short out (as they seem to like doing). Over a long wire, the short-circuit protection may not engage due to the fairly high resistance of the wire.

if a power supply has UVP,OVP,SCP,OCP and some of the bad caps start failing "slowly" like those Capxons and Teapo SC , will the ripple go out of spec ?

Most likely yes.

and can that damage mobo and other components ?

Usually not, or at least not in a short time. From what I have seen, HDDs seem to be the most susceptible to damage, as they seem to fail a lot more in PCs with crappy PSUs than with good ones.

or is the PSU going to shut down safely before excessive ripple

The PSU won't shut off because of excessive ripple. That said, if the caps get bad enough that the output voltages start swinging too much so the PSU comes unstable, that's when the protections may engage.

[Behemot]

Re: Power supply build quality pictorial. part 2

Well, putting a bigger cap in these PSUs also won't affect the hold-up time at all, since this platform seems to an AC detection circuit with built-in PSU shut-off feature, as I mentioned in my thread. Thus as soon as the AC is disconnected, it doesn't matter if that primary cap has any energy left in it or not - the PSU will shut off after some pre-determined delay.

Which is exactly what should happen. However, if the cap inside the unit can not supply it long enough for the delay, you gat shitty HUT results. In my case bigger cap clearly helped to get almost 5 extra miliseconds, so, I take it proven

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

So why the output coil burned, I have no idea.(...)
Looks like just a random failure to me right now. But I could be wrong, of course.

So you should assume a manufacturing error.
From what I've heard burned coils can come from errors in the core material. So rewinding the coil seems out of the question...

And after the Coils there's only the output caps...

[momaka]

Re: Power supply build quality pictorial. part 2

Which is exactly what should happen. However, if the cap inside the unit can not supply it long enough for the delay, you gat shitty HUT results. In my case bigger cap clearly helped to get almost 5 extra miliseconds, so, I take it proven

I'm still not convinced from my tests. When I had a 5 Watt light bulb on the 5V rail and a 6 Watt bulb on the 12V rail, the PSU still shut off pretty much immediately upon AC disconnection. In comparison, I took an old beaten 200 Watt Bestec ATX-1956D and hooked up the bulbs the same way. The result: the Bestec kept those bulbs powered for almost an entire second. With bigger load, it shut off quicker, but still beat the Thermaltake. With the Thermaltake, it just didn't seem to matter what load I had.

Worth noting here on the technical side is that the Bestec ATX-1956D has two 200 V, 470 uF primary caps in series (oldschool voltage doubler), so the total capacitance on the primary is half of that, or 235 uF. Compare that to 470 uF of the Thermaltake (actually 500 uF, as the original cap in the Thermaltake was blown, and I had it replaced with a 450 V, 500 uF cap).

So I think the AC detection circuit intentionally shuts down the PSU before the primary cap has even had a chance to get depleted. I'm not sure why they do this - perhaps to protect the APFC? Or maybe something else? Either way, it feels like this PSU/platform was designed too cheaply. I still wouldn't trust it as much as I trust my old non-APFC tickers.

So you should assume a manufacturing error.

I don't like to assume things, though. Prefer to find out, if that's possible.

As the saying goes, when you ASSUME, you make an ASS with U and ME.

From what I've heard burned coils can come from errors in the core material. So rewinding the coil seems out of the question...

Well, if it burns up again, then that might be an indicator of something. Though there are still a lot of things that can cause the coil to heat up like that. One would be if the synchronous rectifier is not synchronized properly, in which case, you may get some or all of the transformers secondary output AC current into the coil. Another reason could be if there was a defect during manufacturing and there were shorted turns or shorted windings. Or if there is a conductive path to ground before the coil, that can cause a reverse current in the coil between the rectifiers' OFF cycles, thus making it misbehave.

[Behemot]

Re: Power supply build quality pictorial. part 2

I'm still not convinced from my tests. When I had a 5 Watt light bulb on the 5V rail and a 6 Watt bulb on the 12V rail, the PSU still shut off pretty much immediately upon AC disconnection.

You serious? We talking miliseconds here, you won't see that.

But otherwise, it makes sense. Modern circuitry works in some narrower working space, so probably as soon as the voltage drops under some level, the PSU shuts down. Old topologies could have worked with much lower actual input voltage, that does put more stress on the switching transistors though. For example FSP units often get much longer time from smaller caps so it likely works with lower input voltage.

But that still means larger input capacitance results in longer HUT, just not that long as it could as it only uses certain part of the energy stored in it.

[momaka]

Re: Power supply build quality pictorial. part 2

But otherwise, it makes sense. Modern circuitry works in some narrower working space, so probably as soon as the voltage drops under some level, the PSU shuts down. Old topologies could have worked with much lower actual input voltage, that does put more stress on the switching transistors though. For example FSP units often get much longer time from smaller caps so it likely works with lower input voltage.

But that still means larger input capacitance results in longer HUT, just not that long as it could as it only uses certain part of the energy stored in it.

That's what I am thinking too. Otherwise it just doesn't make sense what I am seeing.

If I had to guess, I would say that the AC disconnect is there to protect either the APFC or the primary switchers, particularly when the PSU is under high load and the AC gets disconnected. If the primary switchers are allowed to work with a much lower input voltage like old PSUs do, then the low voltage may force the switchers to stay on longer, thus possibly making the transformer core saturate with too much current. And then BANG.

So perhaps modern PSUs just don't have the kind of loose tolerances and headroom that old PSUs do. Probably why they blow up so often too when something goes slightly outside the norm.

[Behemot]

Re: Power supply build quality pictorial. part 2

Think it's just part of the thing, higher efficiency, active PFC, all that stuff. You *want* the controller to know precisely what's going on and shut down if something goes too out of spec.

[jarvis7]

Re: Power supply build quality pictorial. part 2

Yes, Ripple will go out of spec
yes, way out of spec ripple will damage components (over 200mV and more)
no, PSU will not shut down because there is no 'overripple protection'

Usually not, or at least not in a short time. From what I have seen, HDDs seem to be the most susceptible to damage, as they seem to fail a lot more in PCs with crappy PSUs than with good ones.

The PSU won't shut off because of excessive ripple. That said, if the caps get bad enough that the output voltages start swinging too much so the PSU
comes unstable, that's when the protections may engage.

Thanks a lot ..

so it depends on luck whether the cap is going to kill components or not with these cheap caps ? slowly vs instantly

Stefan and Momaka sir

How long do you think should a PSU with these capxon,aishi,junfu caps like Corsair VS450 or fully Teapo Sc based unit like HEC Corsair Vs400 should be able to safely last without ripple increasing high enough as to cause damage to other components in normal average PC's without graphic cards ?
rough guess ?

Thanks

[momaka]

Re: Power supply build quality pictorial. part 2

so it depends on luck whether the cap is going to kill components or not with these cheap caps ?

It is, a little bit.
Chances are, it's not, though. Typically the PSU quickly becomes unstable once the caps fail, so you will notice.

How long do you think should a PSU with these capxon,aishi,junfu caps like Corsair VS450 or fully Teapo Sc based unit like HEC Corsair Vs400 should be able to safely last without ripple increasing high enough as to cause damage to other components in normal average PC's without graphic cards ?
rough guess ?

With cheap cap brands, it's all about luck. Sometimes you can get 10 years and sometimes even as little as 10 months. This stems from the fact that their quality control is too loose.

Of course, there are cap brands and series that are known to fail in certain ways, like for example Chemicon KZG failing when sitting unused on the shelf or Teapo SC failing in hot spots. So based on that, sometimes it is possible to predict a failure in certain devices (like for example, a crappy cap next to a hot spot in a PSU - typically the first 5VSB filter cap).

So I think the best route for you would be to use the PSU until the warranty expires, then open it and recap it. Also, while the PSU is still under warranty, it still wouldn't hurt if you take it out of the computer once a year or so and blow the dust out of it + give a visual inspection of the caps through the fan grille.