Power supply build quality pictorial. part 2

[370forlife]

Originally Posted by goodpsusearch

Hec Windmill Pro 385

GBU406 4A bridge rectifier

K3767 5vsb mosfet

UC3843B controller & WT751002 supervisor IC

STP10NK60Z 10A continuous MOSFET

S7912PI for -12V

STPS30H100CT 30A schottky rectifier for 12V

STPS4045CW 40A for 5V

MBR3045PT 30A rectifier and 09N03LA 50A power transistor for 3.3V

I had one of these, except with a dual 80mm fans. It was actually a pretty decent 300W unit. They have OPP that steps in at around 350W.

[Pentium4]

And it can actually do 350W? I'm guessing the one with two 80mm had bigger heatsinks, cause the one pictured here has pretty small heatsinks for 350W

[chowder]

Seasonic 250 SU Active PFC (1U FlexATX)

This thing is tiny, ~80x40x150mm, much shorter than most other 1U supplies I've seen.

I wasn't able to get great pictures due to the crowded internal layout. The grey cover you see on the picture with the lid off is stuck with a mild adhesive to a couple of the heatsink surfaces - perhaps to aid heat transfer to the outer case, therefore using it as an additional heatsink?

NCC 150uF 400V primary, the rest are Teapo and Su'scon. I really dread the day I need to work on it, you'd have to remove a heatsink or two to get to the caps underneath.

Having said that, voltage stability is excellent, and it's a bronze rated PSU.

[Behemot]

Yes, casing is often used for cooling with these small/server units.

[Pentium4]

Yeah, I hear ya. Looks like a pain to work on. Even with high efficiency and good design, I wouldn't trust those Teapo and Su'scon much longer in there. I'd check the Superred fan to make sure it has adequate oil in it.

[goodpsusearch]

This is interesting...

I got another HEC psu for recap. The exterior of the psu including the label is 100% identical with this psu, but the interior is different:

The heatsinks are beefier, the PCB layout is similar, but not the same, and the sleeve bearing Globe fan is slower.

As you can see, 3 caps are bulging. The 2 green bulging caps are Jamicon, while the black bulging capacitor is Capxon. All the other caps are Teapo SC (green colored) and Teapo SEK (small brown caps) and a small black Teapo SH.

All the Teapo capacitors are visually fine, including the 5vsb caps. I am going to test every capacitor with ESR/ C meter and update with the results. This is a 2005 made psu, sold in 2006 and servicing PCs for 10 years...

The primary caps are Capxon 680uF 200V and their real Capacitance is 707uF and 720uF! Nice!

This version of the Hec windmill pro uses TNY267PN for 5vsb instead of the classic 2 transistor circuit

The power mosfet is TO-247 W12NK90Z instead of P10NK60Z. Excelent.

3.3V uses MBR4045PT rectifier, 5V uses 2x MBR4045PT (total 80A!) and 12V uses a 30A rectifier as well.

Definitely worth a recap, I think it can do 385W.

What a shame that they replaced it silently with a stripped down version..

[Behemot]

Jamicon, CrapXon, Teapo. Whou would have thought them to fail.

Originally Posted by goodpsusearch

The primary caps are Capxon 680uF 200V and their real Capacitance is 707uF and 720uF! Nice!

Check also ESR. It should be very close to each other both on ESR and capacitance as they are in series, otherwise you get very different voltage on each (ohms law).

Originally Posted by goodpsusearch

2x MBR4045PT (total 80A!)

Does not really work like that.

[momaka]

Originally Posted by 370forlife

I had one of these, except with a dual 80mm fans. It was actually a pretty decent 300W unit. They have OPP that steps in at around 350W.

So kind of like the (old) Thermaltake TR2-430W / Purepower XP550-NP, I guess.
(Thermatake revived the TR2-430W PSUs again, but this time with a 120 mm fan, and I don't know if they are still made by HEC or not.)

Yes, the OPP/short-circuit protection works nice on those PSUs. I'm using mine in a test PC that I use for checking video cards. Had a few with blown MOSFETs, and the PSU shut right off and latched. The only time the short-circuit/OPP didn't work is when I had a video card that had shorted 12V rail through the PCI-E 16x connector's power pins. I guess the resistance of the pins in the PCI-E 16x connector made the PSU not detect the short-circuit.

Anyways, I've had the at PSU since 2007 or so. Never used it that much, but its caps still appear good, despite often sitting on the shelf without use for at least several months at a time.

Originally Posted by Pentium4

And it can actually do 350W? I'm guessing the one with two 80mm had bigger heatsinks, cause the one pictured here has pretty small heatsinks for 350W

Yes
This is a review from a while back:
http://www.hardwaresecrets.com/therm...supply-review/

Originally Posted by chowder

Seasonic 250 SU Active PFC (1U FlexATX)

Nice PSU there!

Originally Posted by Pentium4

I'd check the Superred fan to make sure it has adequate oil in it.

+1
They do dry out sometimes, but are pretty good fans otherwise.

[Behemot]

Which TR2? They name every second PSU with TR2-something. There are the Munich models, first made by HEC (80+), than Sirtec (80+ Bronze) and now there are some CWT scheduled (Bronze as well). They are switching OEMs like socks!

[goodpsusearch]

Originally Posted by momaka

Yes, the OPP/short-circuit protection works nice on those PSUs. I'm using mine in a test PC that I use for checking video cards. Had a few with blown MOSFETs, and the PSU shut right off and latched. The only time the short-circuit/OPP didn't work is when I had a video card that had shorted 12V rail through the PCI-E 16x connector's power pins. I guess the resistance of the pins in the PCI-E 16x connector made the PSU not detect the short-circuit.

Anyways, I've had the at PSU since 2007 or so. Never used it that much, but its caps still appear good, despite often sitting on the shelf without use for at least several months at a time.


Yes
This is a review from a while back:
http://www.hardwaresecrets.com/therm...supply-review/

Thanks for the link, this is really helpful! I have been wondering about this platform's performance since I got my hands on Arctic A-430BX. I spent several hours recapping it and modding it.

After reading the review in the link above I think it was worth it. Excellent voltage regulation, in specs ripple till 350W and working OPP! Only the efficiency is surprisingly low.

BTW I tested every capacitor in HEC Windmill Pro 385 posted above. All the Teapo capacitors including the 5vsb filtering caps and the little caps tested good! This proves that Teapo is among the best non Japanese capacitor makers. I would put only Taicon above them. Ltec and OST, I think they are of similar quality.

Capxon always fail! No matter where they are used or the capacitor series, they are terrible...

And remember, this psu is in service for 10 years.

[Pentium4]

That's a nice HEC PSU Looks like a lighter version of an "HEC Raptor" I have. HEC usually make good units with bad caps. I would consider Teapo on a lower tier than OST and Ltec, honestly. At least SC series

[Pentium4]

Power Win makes their own units, and they all seem to be well put together and honestly rated. This one was made in October, 2002 and was ahead of its time as far as low end PSUs go, being forward topology.

Input filtering consists of 2 X, 3 Y, 2 coil, 2 MOV. 6A bridge

Switching FET is Toshiba K2607 rated at 9A. Plenty for 330W. Might get warm though

Controller located on the primary is ST UC3843B

5V and 3.3V use 30A schottky, 12V uses 16A fast recovery. Controller on the secondary is Weltrend WT7510

All caps are Su'scon and Jamicon. All of them are GP...Su'scon SD and Jamicon TK. However, they all tested good! Does this mean I can use this PSU to rid of my Rubycon PX stock? They're exactly on par spec wise and I'd like to use them for something. These crappy brand GP caps lasted 14 years and are okay. Even the ones with discoloration on the sleeve from heat tested good.

What do you guys think?

[goodpsusearch]

Seems a good idea! Make sure they don't get hot, I would replace or move minimum load resistors if needed.

Is there a fan controlling circuit? Does the fan keep the psu cool during higher loads?


I would replace the 16A fast recovery @12V with a rectifier of 20A or more.

[Wester547]

Originally Posted by Pentium4

All caps are Su'scon and Jamicon. All of them are GP...Su'scon SD and Jamicon TK. However, they all tested good! Does this mean I can use this PSU to rid of my Rubycon PX stock? They're exactly on par spec wise and I'd like to use them for something. These crappy brand GP caps lasted 14 years and are okay. Even the ones with discoloration on the sleeve from heat tested good.

Sus'con SD isn't actually general purpose (about on par with Chemi-con LXV/LXY/Nichicon PM/Rubycon YXF according to the datasheet), SK is their general purpose series. The forward topology was actually around for decades before the 2000s, but it's true that el cheapo power supplies continued to use half bridge topology for years to come. And yeah, it looks like some misplaced minimum load resistors are what caused the discoloration (unless the fan seized). If the sleeve was discolored, I'm surprised it didn't shrink as well.

[momaka]

Originally Posted by Behemot

Which TR2?

TR2-430W, as per my other post.

Originally Posted by goodpsusearch

Thanks for the link, this is really helpful!

You're welcome!

Originally Posted by goodpsusearch

After reading the review in the link above I think it was worth it. Excellent voltage regulation, in specs ripple till 350W and working OPP! Only the efficiency is surprisingly low.

Yes, that unit does put out a bit of warm air, especially with old, 5V-heavy PCs. As long as the PSU is reliable, though, I don't really care that much for the efficiency . For a system drawing around 100 Watts DC from the PSU, the difference in what the PSU pulls from the wall will be less than 20 Watts between a 80%- and 70%- efficient PSU. If I really wanted to save those 20 Watts, I'd switch to an LED monitor (no I won't! ). Besides, I don't keep my PCs turned ON if I am not using them, so I save on electricity there.

Originally Posted by goodpsusearch

BTW I tested every capacitor in HEC Windmill Pro 385 posted above. All the Teapo capacitors including the 5vsb filtering caps and the little caps tested good! This proves that Teapo is among the best non Japanese capacitor makers.

Depends which Teapo series we are talking about here. SC and SZ are the worst, IMO. They regularly bulge or go bad when sitting unused, especially 8 mm diameter ones. Their other series aren't that bad. Actually, I'd say they are okay for PSU use.

Originally Posted by goodpsusearch

Capxon always fail!

Not always, but very very frequently compared even to other crap-cap brands. And it's regardless of the series, too.

That said, I still have a handful of good KMs from a PSU made around 2000. Back when I recapped this Macron PSU, I didn't have the right caps for the -12V and -5V rails, so I used a few of those CapXon KMs from the same PSU I mentioned above. They all still test good.
Of course, I also have a lot of failed CapXon KM and (especially) KF from various other stuff. So it seems that with CapXon, if they last for a long while, they may turn out to be okay. But I still wouldn't trust them for anything I care for.

Originally Posted by Pentium4

All of them are GP...
However, they all tested good! Does this mean I can use this PSU to rid of my Rubycon PX stock? They're exactly on par spec wise and I'd like to use them for something.

In general, many older PSUs tend to run at a lower switching frequency (especially the half-bridge designs). Hence why it is usually not a problem to use GP caps in them.

That said, I always recommend entry to mid-range low ESR caps for the 5VSB, due to flyback topology. Something in the class of Chemicon LZX/LXY/LXV to KZE/KZH and Nichicon PW/PM/PS/PJ to HD. Of course, the good thing about flyback topology is that it usually isn't picky. So both general purpose and ultra-low ESR (motherboard CPU VRM) caps will work okay with it.

Speaking of the 5VSB, I see that Power Win PSU has a 2-transistor design. And with those small caps next to the 5VSB switching transistor, I'd say one of them is the "critical cap". So make sure to get those replaced.
... Well, not sure why I say that to you, because I know you always do a proper recap and replace everything meticulously .

Originally Posted by Pentium4

Even the ones with discoloration on the sleeve from heat tested good.

I wonder if that's due to heat or just the plastic sleeve degrading. I don't see any big low-Ohm resistors.

Also, those output caps look pretty small, like 1000 uF. I'd up that to two 1500-2200 uF for the 3.3V and 5V rails (so at least 3000 uF total capacitance for each rail). And at least a single 1500 uF for the 12V rail (but preferably 2200-3300 uF total capacitance).

[Pentium4]

Originally Posted by goodpsusearch

Seems a good idea! Make sure they don't get hot, I would replace or move minimum load resistors if needed.

Is there a fan controlling circuit? Does the fan keep the psu cool during higher loads?


I would replace the 16A fast recovery @12V with a rectifier of 20A or more.

I wish it had a fan controller. It doesn't, which is probably why the caps did so well. I think I will just permanently run the fan at 5V. I plan on installing a 20A schottky Since it's single forward I will have a larger selection on rectifiers to use.

Originally Posted by Wester547

Sus'con SD isn't actually general purpose (about on par with Chemi-con LXV/LXY/Nichicon PM/Rubycon YXF according to the datasheet), SK is their general purpose series. The forward topology was actually around for decades before the 2000s, but it's true that el cheapo power supplies continued to use half bridge topology for years to come. And yeah, it looks like some misplaced minimum load resistors are what caused the discoloration (unless the fan seized). If the sleeve was discolored, I'm surprised it didn't shrink as well.

Woops. You're right. I think 1200µF 6.3V Chemi-con KY 8x20mm for 5VSB filtering. They used a Su'scon 1000µF 16V and 100µF 25V Jamicon TK for 5VSB filtering... the Jamicon cap amazingly tested 103µF, 0.13Ω ESR. In my experience, I think Jamicon and Su'scon make better caps than CapXon and Teapo. I think those other brands would have failed here.

In general, many older PSUs tend to run at a lower switching frequency (especially the half-bridge designs). Hence why it is usually not a problem to use GP caps in them.

That said, I always recommend entry to mid-range low ESR caps for the 5VSB, due to flyback topology. Something in the class of Chemicon LZX/LXY/LXV to KZE/KZH and Nichicon PW/PM/PS/PJ to HD. Of course, the good thing about flyback topology is that it usually isn't picky. So both general purpose and ultra-low ESR (motherboard CPU VRM) caps will work okay with it.

Speaking of the 5VSB, I see that Power Win PSU has a 2-transistor design. And with those small caps next to the 5VSB switching transistor, I'd say one of them is the "critical cap". So make sure to get those replaced.
... Well, not sure why I say that to you, because I know you always do a proper recap and replace everything meticulously .


I wonder if that's due to heat or just the plastic sleeve degrading. I don't see any big low-Ohm resistors.

Also, those output caps look pretty small, like 1000 uF. I'd up that to two 1500-2200 uF for the 3.3V and 5V rails (so at least 3000 uF total capacitance for each rail). And at least a single 1500 uF for the 12V rail (but preferably 2200-3300 uF total capacitance).

It's a little conflicting though...because I've been told not to stray away from original cap specs, but that GP caps aren't suitable, higher capacitance is better, but not too high??? So I'd love to use 3300µF 10V Rubycon PX (1100 ripple current) in place of 1000µF 16V Su'scon SD (1050 ripple current) but wouldn't such a large leap in capacitance be bad for the feedback loop? I also think the identically specced 2200µF 16V Rubycon PX would be good for the 12V since that rail has the huge PI coil. All I've checked so far is 33Ω resistor on 5VSB, which is too low!

Hmmm, I wonder how it would react just swapping the 12V and 3.3V rectifiers

[goodpsusearch]

I have seen more than a couple psus and ups with a selection of Teapo and Jamicon caps and the Jamicon caps were bulging or just high esr, while Teapo caps were ok.

I have replaced 1x 1000uF 16V capacitor with 1x 3300uF 16V Panasonic FK or FR for 12V output most of the times without oscillation problems. But I had bad luck 1 or 2 times with picky psus

[goodpsusearch]

Just finished partial recap and I am testing it now on a P4 3GHZ LGA775 computer

For info about capacitors used check attached pics

[momaka]

Originally Posted by Pentium4

It's a little conflicting though...because I've been told not to stray away from original cap specs, but that GP caps aren't suitable, higher capacitance is better, but not too high???

It's not uncommon for a PSU manufacturer to have one PSU design/platform, and then make several "models" off of that, usually just putting cheaper components in the lower-end models. That includes smaller caps too.

And other times, manufacturers are just cheap - they make a good design and then cheap out at production and use smaller components than called for.

Either way, when I see only 1000 uF caps on the major rail outputs of an older PC PSU, I know I got cheated. Anything that uses a half-bridge or forward topology with UC384x (and similar) IC usually runs at a fairly low switching frequency - typically 22-50 KHz. The capacitance and ESR of the output filter caps is directly related to that: the lower the frequency, the bigger the output capacitance has to be. (That's why linear wall warts still have fairly large capacitors, despite being only rated for 0-1 Amps.) Conversely, higher switching frequency usually *requires* caps with lower ESR and *allows* for smaller capacity caps. (Same reason why motherboards nowadays have so many phases on CPU VRM and so little output capacity.)

And then there is controller feedback compensation, which can "tweak" some of the above requirements at the expense of others.

But all in all, higher capacitance almost never hurts. IIRC, only Sirtec had problems with oscillations when you deviated too far from the original cap values. So those Rubycon PX caps of yours could probably be used in a lot more power supplies.

Originally Posted by Pentium4

Hmmm, I wonder how it would react just swapping the 12V and 3.3V rectifiers

The 3.3V rail may not be too happy with an FR rectifier instead of its usual schottky.

[Behemot]

Sirtecs cheaper mainstream plaforms have problems with ripple, voltage regulation etc. even today…brand new from factory.