VOGONS


A tale of two PSUs

Topic actions

Reply 401 of 472, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
ODwilly wrote:

Here is a dumpster find I found. Still works! Had a rather nice made molex to P4 connector attached to it and after replacing a couple bulging Fuhgyuu's it seems to work fine.
Anyone tell from my cellphone pics if it is decent enough for a spare for a low-end Slot 1 rig?

Here's a summary of what I can see:
Input filtering looks complete. The daughterboard on the plug should result in an easy pass for EMI.
Component selection appears to be good aside from capacitor brand. Primary caps, transformers, and heatsinks all appear to be a healthy size. Based on what I see, the power rating for this supply is conservative.
Output pi filters are complete which should result in respectable numbers for ripple.

All good signs so far.

Here's what I can't see:
Transistor and diode selection, so I can't tell if the supply is definitely good for 250W, nor can I tell what is used to generate the standby rail.

All hail the Great Capacitor Brand Finder

Reply 402 of 472, by ODwilly

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:
Here's a summary of what I can see: Input filtering looks complete. The daughterboard on the plug should result in an easy pass […]
Show full quote

Here's a summary of what I can see:
Input filtering looks complete. The daughterboard on the plug should result in an easy pass for EMI.
Component selection appears to be good aside from capacitor brand. Primary caps, transformers, and heatsinks all appear to be a healthy size. Based on what I see, the power rating for this supply is conservative.
Output pi filters are complete which should result in respectable numbers for ripple.

All good signs so far.

Here's what I can't see:
Transistor and diode selection, so I can't tell if the supply is definitely good for 250W, nor can I tell what is used to generate the standby rail.

Thank you, Im going to assume from the rest of the build quality that they cheaped out on caps and didnt cheap out on the transistors and diodes. The amazing thing is how close to spec the rails are, the 12v reads 11.9 and the rest of the rails are either perfect or the same distance off. Now that the 2 Fuhgyuus are gone the Crapxons are next up!

Main pc: Asus ROG 17. R9 5900HX, RTX 3070m, 16gb ddr4 3200, 1tb NVME.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 403 of 472, by Nvm1

User metadata
Rank Member
Rank
Member

I opened up the PSU of the AT Tower I received for two weeks.
1 cap bulging, 2 have blown/vented alot. I am in dire need of some AT PSU's for my rebuilds.
My question, is this AT PSU worth salvaging by replacing the caps or not?

IMG_20180516_210651.jpg
Filename
IMG_20180516_210651.jpg
File size
1.82 MiB
Views
1240 views
File comment
AT PSU Internals1
File license
Fair use/fair dealing exception
IMG_20180516_210657.jpg
Filename
IMG_20180516_210657.jpg
File size
1.84 MiB
Views
1240 views
File comment
AT PSU Internals2
File license
Fair use/fair dealing exception
IMG_20180516_210708.jpg
Filename
IMG_20180516_210708.jpg
File size
1.67 MiB
Views
1240 views
File comment
AT PSU Internals3
File license
Fair use/fair dealing exception
IMG_20180516_210727.jpg
Filename
IMG_20180516_210727.jpg
File size
1.74 MiB
Views
1240 views
File comment
AT PSU Internals4
File license
Fair use/fair dealing exception
IMG_20180516_210744.jpg
Filename
IMG_20180516_210744.jpg
File size
1.65 MiB
Views
1240 views
File comment
AT PSU Internals5
File license
Fair use/fair dealing exception

Sorry for the Potato pictures, I hope they still are clear enough to see the important pieces.

Reply 405 of 472, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

I'm certainly not an expert on power supplies, but bulging or leaking aluminum electrolytic caps are fairly uncommon on devices made before 1999 or so... I've only seen them on a few very very low end Micronics SS7 AT boards. Anything with visibly defective caps before they were common may have been built with especially cheap components all around. I wouldn't bother with it.

Now for some blitting from the back buffer.

Reply 406 of 472, by DAVE86

User metadata
Rank Newbie
Rank
Newbie
Nvm1 wrote:

My question, is this AT PSU worth salvaging by replacing the caps or not?

It's '90 generic at best. Has some EMI input filtering, though not all components are the safety approved type. Has a bridge rectifier, small input capacitors and the primary transistors are plastic to-220 types. Main transformer looks like EI33 size. Good enough for 200-230W depending on the switching frequency that KA7500 is configured to.
Secondary protections are most likely handled by that other chip which is a dual comparator.
The secondary heatsink is a bit flimsy. That two 3 amp diodes bunched together are the rectifiers for the 12V rail. It might not be an issue if the 12V rail is specified 4-6 amps max. The 5V rectifier looks decent, probably 20-30 amp rated.
If you have the time and spare parts/know-how you can try to revive and even beef the unit up a bit. Othervise don't bother with it.

Reply 407 of 472, by Nvm1

User metadata
Rank Member
Rank
Member

Okay so for a 486 VLB or a Socket 5/7 system with nothing special in it it can be used.
Changing caps should be easy, recapped alot of newer units where space is really cramped.
Other things I can improve outside replacing the old Fan and the old caps?

Reply 408 of 472, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie
Nvm1 wrote:

Other things I can improve outside replacing the old Fan and the old caps?

You could improve the input filter by scavenging safety approved X and Y caps from other units and if you manage to fit larger heatsinks it may help with heat as those look quite thin. Other than that, it should have no problem powering a 386/486/Pentium machine.

Reply 409 of 472, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

The filter network isn't really even sufficient, so I'd probably retrofit it with an additional self contained filter can.
Schaffner_EMV_AG_-_Figure_1.jpg

Aside from that, it's minimally sufficient at best. I'd like to echo everything DAVE86 said (he's much better at spotting transformer sizes than me), plus the secondary filter isn't very robust, and protection is probably quite rudimentary, maybe OVP only. I would really only bother if you haven't got any other AT supplies and you're absolutely dead set against using ATX to AT adapters.

All hail the Great Capacitor Brand Finder

Reply 410 of 472, by okenido

User metadata
Rank Member
Rank
Member

Hi folks

Found this PSU, its working and quite clean, i was wondering if it's worth keeping and recapping ? Would be for some PIII setup.

IMG_20180522_173524.jpg
Filename
IMG_20180522_173524.jpg
File size
1.76 MiB
Views
1155 views
File license
Fair use/fair dealing exception
IMG_20180522_173557.jpg
Filename
IMG_20180522_173557.jpg
File size
1.64 MiB
Views
1155 views
File license
Fair use/fair dealing exception
IMG_20180522_173543.jpg
Filename
IMG_20180522_173543.jpg
File size
1.78 MiB
Views
1155 views
File license
Fair use/fair dealing exception
IMG_20180522_173623.jpg
Filename
IMG_20180522_173623.jpg
File size
1.58 MiB
Views
1155 views
File license
Fair use/fair dealing exception

Reply 412 of 472, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

If it's working, passes basic PSU tests and doesn't having swollen or leaking caps, I wouldn't bother recapping it. You're more likely to introduce problems to a working unit unless you recap boards all the time and are very experienced. It'd also depend on the brand of caps. If they're a decent name brand, they may last years without causing any problems.

Last edited by Ozzuneoj on 2018-05-22, 19:03. Edited 1 time in total.

Now for some blitting from the back buffer.

Reply 413 of 472, by okenido

User metadata
Rank Member
Rank
Member

There seems to be two brands of caps, one labelled "DON" and one with some 'Y' logo in a circle.

I can solder easliy on the simple PSU pcb's, I was thinking about recapping since i've had some issues recently with old hardware - Randomly failed IDE drive detection, motherboards and GPUs that worked before and stopped working for no reason... I was using random PSU from my spare, they all work but some have bulging caps, I'm afraid it will kill my hardware even if caps look OK (their capacity decrease over time)

cap.jpg
Filename
cap.jpg
File size
72.56 KiB
Views
1147 views
File license
Fair use/fair dealing exception
cap2.jpg
Filename
cap2.jpg
File size
93.11 KiB
Views
1145 views
File license
Fair use/fair dealing exception

Reply 414 of 472, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Ozzuneoj wrote:

If it's working, passes basic PSU tests and doesn't having swollen or leaking caps, I wouldn't bother recapping it. You're more likely to introduce problems to a working unit unless you recap boards all the time and are very experienced. It'd also depend on the brand of caps. If they're a decent name brand, they may last years without causing any problems.

You should be able to find a ~250W ATX PSU from an OEM unit that's as or more powerful than this one (in reality) with proper component selection, construction, and circuit protection. This guy is probably a two transistor standby PSU that's ticking away waiting to go *boom* all over your retro parts.

All hail the Great Capacitor Brand Finder

Reply 415 of 472, by okenido

User metadata
Rank Member
Rank
Member

Do you mean the transistors in this PSU may fail prematurely despite the new caps ?

I also have two Delta OEM power supplies (200W but it says the total output can't exceed 110w), they are a bit newer than the FGI since it has a SATA connector

Reply 416 of 472, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

The Delta probably has a 110w limit on the 5V and 3.3V rails combined. This would be fine for a basic P3 without a huge stack of drives or peripherals.

The biggest problem will be if the transistors don't fail. An out of control PSU can potentially deliver huge amounts of energy to your components in the form of voltage in excess of breakdown levels or high current in the event of a short on the motherboard, for instance. One of the most important functions of a power supply is to gracefully shut down in the event of such abnormal operation.

A two transistor standby circuit is ultimately one of the cheapest ways of implementing 5V SB. The problem is that two transistor standby circuits are not robust at all, often featuring little or no protection. Hell, switching control can even be generated via an LC tank circuit, a Zener reference, and a handful of other generic caps and resisters from the Shenzhen market.

All hail the Great Capacitor Brand Finder

Reply 417 of 472, by retardware

User metadata
Rank Oldbie
Rank
Oldbie

Now I have a tale of two AT power supplies also.

According to the date codes, both are manufactured ~1996.
I´ll begin with the Enermax E251BT-V, and continue with the Seasonic SSG-200G after the Enermax issues got fixed.

I got the PSU with an AT case that I bought for my upcoming 486 freeclocker build. It also contained a Pentium mobo which worked fine at my initial tests, so the PSU was proved not to be dead.

To make sure that the PSU is in good shape I decided to clean it and replace the fan.

However, after the work done, I did measure and the PSU shows 10.9V on the 12V line, and 5.32V on the 5V line, which are out of specs.
A crude ripple measurement using multimeters' AC mode showed (approximately) 124mV on 12V and 19mV on 5V.
Sadly it is sunday and I found only some 12V 0.8A fans I could use as temporary load. Will have to get some big load resistors.

In addition I found some strange things on the PSU, which I cannot judge, as my knowledge of electronics is very limited.
So I hope you guys know what is wrong (or what I did wrong) when I show you the photos.

DSCN9575.jpg

The PSU label

DSCN9573.jpg

PSU quite dirty outside, grimy smoker homes' dust

DSCN9576.jpg

The inside after opening looked better than I expected.

DSCN9582.jpg

No Schaffer filter can, as usual. Will have to order one.

DSCN9581.jpg

Decided to replace the old fan (right), which stopped quite soon when tipped with finger indicating worn bearing, with a more powerful fan I took from a server (left).

DSCN9583.jpg

When examining the drive power connectors I found some oil residues, apparently "Kontaktspray" which seems to be popular with some people even in areas where it does not belong.
This obviously did not resolve the underlying cause, low quality, easily worn out connectors. Look how loose they must fit! A true HDD killer plug! Off with that crap!

DSCN9615.jpg

Replaced these with drive power cables I snipped off from an ancient Fujitsu server. Enterprise quality stuff definitely looks better than consumer grade 😀

DSCN9588.jpg

Lots of colophonium drip residues on the components side, which remained at factory cleaning. Removed because they have the potential of becoming conductive.

DSCN9589.jpg

This confuses me a bit. Do diodes work better when well-heated by cramping under that coil?

DSCN9590.jpg

Another strange view.
D11 and transformer battling for space, with the latter having to retreat a little bit.
D14 as heater to bring C17 to boiling temperature?

DSCN9599.jpg

Anyway, 12V have 2000uF, 5V has 2200uF, didn't look at values for -5V and -12V.
All outputs, even the negative ones, seem to have pi coils.
Caps rated 105C and have a logo "CTC".
There are two chips, one LM339N and KA7500B (?).
Mounted at the secondary cooler are BYQ28X 100 (I guess some kind of transistor?) and NDS40-30 (dual diode).

DSCN9600.jpg

The two primary cans have 470uF and are Japanese brand. However, they are rated 85C only.
Rectifier is big block mounted in good height above PCB. Couldn't read full type, obscured by caps.

DSCN9598.jpg

Switching transistors are NT2625N.

DSCN9605.jpg

PCB solder side looks acceptable to me, just cleaned dust and colophonium residues.

DSCN9606.jpg

As I prefer optimal cooling, I connected the replacement fan directly to 12V, not using the fan thermo controller circuit.
(I remember having read somewhere this would be bad - is this true? If so, for which reasons?)

Now my questions:
-What could be the reason for the voltages out of tolerance margin?
Was it only because of wrong minimum load?
Or are the "CTC" caps dried out or otherwise bad and should be replaced proactively?
Or other defects?

I also would appreciate if people who know about power supplies could comment on the technology and quality of that PSU.
In particular important to me is to know how big the risk is that this PSU blows up the retro rig, respective which kinds of protection circuits it has.

Reply 418 of 472, by retardware

User metadata
Rank Oldbie
Rank
Oldbie
retardware wrote:

-What could be the reason for the voltages out of tolerance margin?
Was it only because of wrong minimum load?

Decided to use that Pentium mobo as test dummy.
With it connected, the voltages are fine: 5.12V and 12.10V, with a ripple of 4mV and 30mV, respective. Very good!

So just a fan as dummy load was way too small.

The other questions still remain...
Have used many Seasonic and these seem fairly good.
Not sure what to think of Enermax, as this is the first power supply of them I looked into.
At least I have the feeling this PSU might be a honest 250 watter...

Comments welcome 😀

Reply 419 of 472, by wave

User metadata
Rank Newbie
Rank
Newbie

You can't really measure ripple with multi-meter, as the multimeter is too slow to catch and show you any kind of ripple.

Ripple in smps is high frequency (20KHZ and more) and the multimeter would barely be able to show some fluctuations in voltage. Btw trying to measure ripple both with multimeter and also at V AC is 2 times wrong.