What is your testing PSU and what's its story?

[Th3_uN1Qu3]

I'm just curious. The 200W Seasonic wasn't cutting it anymore, so at the moment mine is a 420W Powerlink. Or, at least it used to be a Powerlink till i got my hands on it.

Here's what the label says. For 400W (there is no "420W" model printed on the label):

• 3.3v: 24A. Bullshit! The rectifier is a SBL2040CT (20A).
• 5v: 33A. True... almost. The rectifier is a SBL3040CT (30A). It can usually handle a couple amps more than its datasheet says.
• 12v: 16A. True... in theory. The rectifier was a SBL20100 (20A), however, it was blown. It probably was a knockoff part.

I got it for free. It had the following issues:

• Bad caps on the 3.3 and 5v rails (2200uF/10v Forever and Canicon)
• Shorted 12v rectifier
• Shorted primary transistors (D209L)
• Blown (open) base drive resistors
• Blown fuse (of course)
• Incomplete transient filter (two Y caps and that was all)

However, it had several things that made me decide it was worth fixing:

• A bridge rectifier! No 4-diode treatment, thank you.
• Decent primary capacitors (a pair of 470uF Teapos)
• Complete output filtering, all coils were in place (if you didn't know, they like to omit those as well!)
• Adequate gauge wiring on the main filter coil
• 20+4-pin ATX connector
• SG6105D controller

The bulged capacitors were replaced by 1000uF/10v Nichicon PL(M). I didn't find anything official about this series, but they look a lot like the PM(M), and the specs on those look real good. They came from a Tyan Tiger 100. I'm pretty sure that the lower capacity is made up by the much lower ESR. Also, they are the same size as the "2200uF" Canicons, so it's likely that Canicon was simply lying about their capacity. The caps on the 12v rail were left alone for now.

The other parts came from a 300W Powerlink that i also got for free. This one was in working condition and had a half-decent transient filter, however it was a disaster in any other aspect. 220uF primary caps, all pi filter coils were missing, it had a grand total of THREE main capacitors, two of which were bulged, the wire on the main filter coil was the thinnest i'd ever seen (i think it's 26AWG), the main transformer was ridiculously small (first time i'd seen one of those too). It was ridiculously crammed as well, no chance of actually fitting the missing parts in. It also had a loud fan. Probably that's what kept it from blowing up...

Even after recapping it couldn't keep a 20GB hard drive running, and it happily continued powering a motherboard with a shorted chipset even though ALL voltages had gone out of spec. Luckily i noticed that the fan had slowed down and shut it down, otherwise it would probably have gone boom. Loud fan and all, it was quite hot inside after this incident. So i decided to transplant the primary transistors and filter components into the 420W. Here's what got donated.

• The two 2SC5763 switching transistors
• The 2.2 ohm base resistors
• One SBL2040CT rectifier for the 12v rail
• The input filter coil
• One X capacitor
• The fuse ()

I plugged the 420W unit in, and... nothing. I quickly found the reason: One side of the input filter coil wasn't actually connected to anything on the PCB!!! I re-used one of the straps they used to bypass the coil to wire that end of the coil to the diode bridge. I then plugged the PSU in again. 5vsb? Precisely 5 volts. Green wire to black wire.... and.... and.... it didn't blow! Yay. It also has a fairly quiet fan.

I connected the PSU to the board with the shorted chipset. It did what it was supposed to do - didn't power on at all! So this one actually has protections... I wonder why it blew up in the first place.

I left the PSU connected to a Pentium III and it's doing fine for now. I'll fire up the 'scope and check its ripple later, but i think it's well within specs. So let's revisit the ratings now, judging by the rectifiers. The input bridge rectifier is 4A and mains voltage is 230v here, so no problem with that. The 3.3v rectifier is 20A so that makes 66W. 5v - 30A, 150W. However, since 3.3v is derived off 5v, that's a total of 150W for 3.3 and 5v together. 12v - 20A again - 240W. 150+240=390W for 3.3, 5v and 12v. If you also add -5v (0.6A), -12v (0.8A) and 5vsb (2A), it looks like this one might just honor its rating! We shall see.

Things it still needs:

• A MOV in the input filter
• PCI-E and SATA power connectors

Pics coming tomorrow, i'm too lazy to make some after typing all this. So. What's YOUR testing PSU and where did it come from?

[Hondaman]

Re: What is your testing PSU and what's its story?

I think I've got a Cooler Master 430 (max. power 400W) (ATX) in the basement. Cheap after rebate, probably about 5 years old. Runs my Athlon 2900 (KT400 or KT600 chip set), 768 RAM. When I drop in power-sucking SCSI (Adaptec 2940 and 2 18 GB drives and/or a SCSI CD-burner) it powers up just fine....

I should probably swap it out with the (9 year old?) PC P&C 350W ATX (Silencer or Turbo-Cool? Don't remember).

[PCBONEZ]

Re: What is your testing PSU and what's its story?

For server boards I use a Delta DPS-700FB or an FSP FX700-GLN.
Delta DPS-700FB - OEM from a Gateway Server - good PSU but crappy efficiency so I don't want it in a final build.
FSP FX700-GLN Epsilon - crap caps I was keeping an eye on before I trusted the model in a build, good thing, they bloated, just ordered replacements.

I have 4 re-capped Sparkle FSP460-60PFN for testing regular boards.
- Had mixed Teapo, OST, and one Fuhjyyu each. The OST & Fuhjyyu were bloated.
.

[Th3_uN1Qu3]

Re: What is your testing PSU and what's its story?

Delta DPS-700FB - OEM from a Gateway Server - good PSU but crappy efficiency so I don't want it in a final build.

I knew that Delta generally makes big loud fans... Now i also know why.

[c_hegge]

Re: What is your testing PSU and what's its story?

I have 8 spare PSUs sitting behind me right now just waiting to be used, so for testing I just grab one randomly. they are:

1x Antec Earthwatts 380W (Re-capped)
3x Hipro 300W (all re-capped)
1x Dell/Newton 250W (re-capped)
1x FSP 250W (Re-capped)
1x Aywun 500W (really only 320W. Re-capped and added a fan speed controller)
1x Antec SP-400 (Re-capped, one fan replaced)

[Behemot]

Re: What is your testing PSU and what's its story?

Eurocase 500 W, recapped with Samxon's. Not sure what power it can really handle, I expect around 400 W...

[370forlife]

Re: What is your testing PSU and what's its story?

The bulged capacitors were replaced by 1000uF/10v Nichicon PL(M). I didn't find anything official about this series, but they look a lot like the PM(M), and the specs on those look real good. They came from a Tyan Tiger 100. I'm pretty sure that the lower capacity is made up by the much lower ESR. Also, they are the same size as the "2200uF" Canicons, so it's likely that Canicon was simply lying about their capacity. The caps on the 12v rail were left alone for now.

Bulk capacitance is more important to power supplies than Low ESR. Also, lowering the ESR can actually increase ripple in power supplies as the PI filters are "in tune" with the ESR rating of the caps. I like to use Samxon RS (available here or through Big Pope) as they come in 10x30 size in both 16v, 3300uf and 10v, 3300uf. Good for bulk capacitance and not-to-low of ESR.

My testing power supply(s) are a Raidmax 500W that is actually a Sunpro safety series, the other is a PowerUp 550W which is actually a RSY 400W.

[Th3_uN1Qu3]

Re: What is your testing PSU and what's its story?

Bulk capacitance is more important to power supplies than Low ESR. Also, lowering the ESR can actually increase ripple in power supplies as the PI filters are "in tune" with the ESR rating of the caps.

I have to check on that one... The PSU is performing great atm. Though i did find another spot with room for improvement - the ATX 12v cable should be heavier gauge. As it is now, a Pentium D can dip it to 11.75v which is lower than i'd like. The rest of the connectors stay over 12v, so i'm sure it's the cable.

[goodpsusearch]

Re: What is your testing PSU and what's its story?

I have to check on that one... The PSU is performing great atm. Though i did find another spot with room for improvement - the ATX 12v cable should be heavier gauge. As it is now, a Pentium D can dip it to 11.75v which is lower than i'd like. The rest of the connectors stay over 12v, so i'm sure it's the cable.

Always the 12V connector has lower voltage than the other 12V cables whatever the gauge is. It's because many watts are drawn from there.

[etnietering]

Re: What is your testing PSU and what's its story?

Mine's a 330W Logisys advertised as 480W, which I've upgraded to maybe actually be able to put out 330W. The whole saga is detailed here.

[momaka]

Re: What is your testing PSU and what's its story?

150+240=390W for 3.3, 5v and 12v. If you also add -5v (0.6A), -12v (0.8A) and 5vsb (2A), it looks like this one might just honor its rating! We shall see.

If the primary transistors are in TO-220 packages, chances are something will blow up before you even get to 350W. 250W-300W is more within the range of these PSUs.

By the way, what size is the main transformer? 35 or 33? If the latter, you're looking at no more than 250W continuous (300W peak).

You did a lot of good work on it, though, so it should be more reliable now at least. Good thing you put a schottky on the 12v rail. Much lower heat loss than a fast recovery rectifier. Definitely check ripple if you have a scope, though.
...
For my test PSU, I use a 300W Macron Power MPT-301. Got it for free as well. Had a seized fan and blown Fuhjyyu caps. Fixed the fan, got new caps, and it's running great now. 10A rectifier for the 5vsb, 30A schottky for the 3.3v and 5v, 16A fast rectifier for the 12v.

[Th3_uN1Qu3]

Re: What is your testing PSU and what's its story?

Always the 12V connector has lower voltage than the other 12V cables whatever the gauge is. It's because many watts are drawn from there.

A thicker cable should still help. They're all drawn from the same place you know.

If the primary transistors are in TO-220 packages, chances are something will blow up before you even get to 350W. 250W-300W is more within the range of these PSUs.

Yes, they are TO-220. I'm not too worried about them however, i've blown dozens when building my own SMPS with +/-70v output so i know exactly what they can do. That is, till i switched to MOSFETs in that supply - IRF740s - blown pairs count - zero.

I'll put some MJE13009s in there sometime soon so i can stop worrying about them entirely. Btw, the D209L transistors it had were in TO-3PN package, and they were still blown. If you read Hardware Secrets' reviews, they had primary transistors blow in ultra-low-end 350W Huntkey supplies, yet the very same transistors did not blow in the 700W Rocketfish supply which is also manufactured by Huntkey. It's about the implementation as well.

By the way, what size is the main transformer? 35 or 33? If the latter, you're looking at no more than 250W continuous (300W peak).

It's an EI-33. In my +/-70v supply i pull 400W from one myself, so i know it is possible.

You did a lot of good work on it, though, so it should be more reliable now at least. Good thing you put a schottky on the 12v rail. Much lower heat loss than a fast recovery rectifier. Definitely check ripple if you have a scope, though.

Thanks. I will check ripple for sure, but i'm too lazy to drag it to the scope atm. Hey, it can run a Pentium D, so it should be alright. I'll post pics in an hour or two.

[goodpsusearch]

Re: What is your testing PSU and what's its story?

Momaka is right about the transformer ratings.

A thicker cable should still help. They're all drawn from the same place you know.

18 gauge is perfect for a psu like that, 20 gauge is so and so... Anything thinner than that is unacceptable (power on and power good cables are not included).

[momaka]

Re: What is your testing PSU and what's its story?

If you read Hardware Secrets' reviews, they had primary transistors blow in ultra-low-end 350W Huntkey supplies, yet the very same transistors did not blow in the 700W Rocketfish supply which is also manufactured by Huntkey. It's about the implementation as well.

I agree. With that said, I wouldn't count on a cheapo PSU to be designed well .

It's an EI-33. In my +/-70v supply i pull 400W from one myself, so i know it is possible.

Interesting. Did you use a slightly higher switching frequency, perhaps?
I do remember reading that the limit for those size 33 laminated iron core transformers is 250W to 300W.

Hey, it can run a Pentium D, so it should be alright.

Yeah, nothing eats power like a Pentium D. Gotta buy me one of those for when the winter gets here .

[ratdude747]

Re: What is your testing PSU and what's its story?

sounds like the ultra V that i recapped, red-did the molex/sata lines on (y cabled with empty solder holes), and beefed up the 12v+ wiring... it still needs 2 caps which i need to get around to ordering... still for samxons/rubycons with nippon primaries (stole from a dying bestec 12e), the 2 bad brand ones have not been much of a drawback... it was new when i bought it

now it powers my pentium dual core machine (not in use right now, needs more ram and my desk for it has yet to be moved).

iirc, the ultra V was a crappy model... works fine now that i redid it...

i have another one still that i may do the same with and use for testing... just need to order caps.

[PCBONEZ]

Re: What is your testing PSU and what's its story?

A thicker cable won't do anything much.
If you have a drop across the cable it's more likely ill fitting cheap pins in the connectors.
.

[Th3_uN1Qu3]

Re: What is your testing PSU and what's its story?

Interesting. Did you use a slightly higher switching frequency, perhaps?
I do remember reading that the limit for those size 33 laminated iron core transformers is 250W to 300W.

Yes. The power supply that i built runs at 100kHz. I never believed the EI-33 core can do more than 200-250W (although i've read about people achieving that), till i did it myself. Btw, it's not laminated iron, it's ferrite. Perhaps you were thinking powdered iron, which is what the output chokes are made of.

A thicker cable won't do anything much.
If you have a drop across the cable it's more likely ill fitting cheap pins in the connectors.

New cable comes with new connectors. So why shouldn't it improve things? Anyway, here the pic. There's one large cap hiding behind the secondary heatsink that didn't make it in the pic but oh well. The brown caps in the middle are the Nichicons. They actually heat up quite a bit, i think it may need more of them.

I've scoped it and overall ripple/noise levels are at about 70mV for the 12v rail, with the PSU powering a Pentium D. Acceptable. However, there are some spikes that get over 600mV tall... Looking at them with a small timebase shows that they are actually very high frequency ringing, they clock in at 8.7MHz. For now i am sure they do not come from the transformer, as i've tried placing snubbers there and no result. They may be a result of parasitic inductance somewhere in the secondary... i'll check it out when i get back, as i'll be leaving for a week.

Adding extra caps to the board does not help. But strangely, adding them on the rail i am measuring (especially a 100nF ceramic), cuts the spikes in half. The cables seem to kinda act like an antenna... so i'm not exactly sure if the spikes come from the PSU at all. There is a ~200mV waveform at about the same frequency, that i see on the scope even with the power supply off, so it's interference from something. It has a distinctly different appearance than the spike, but i think it could be a part of it.

[goodpsusearch]

Re: What is your testing PSU and what's its story?

Very nice that you use a scope to measure ripple!

I would add another X capacitor and a second coil at the input filter. I would also add mov protection.

What is the capacity of your input caps?

[Th3_uN1Qu3]

Re: What is your testing PSU and what's its story?

I mentioned in the first post - it's a pair of 470uF Teapos. There is no room for a second coil, as you can see i decided to put an X capacitor there, but i can solder a second X cap to the back of the AC plug like commonly done.

I will add MOVs as soon as i have a good excuse to go to the parts store, aka as soon as i have at least $20 worth of stuff that i need from there.

[goodpsusearch]

Re: What is your testing PSU and what's its story?

You can place a X cap on the AC plug and a second smaller X cap in the empty space of the pcb near the y caps. That way you can add a second coil where your X cap is now placed.

470uF primary caps I think are good for 300watts max.