VOGONS


PSU - bust the myth

Topic actions

Reply 340 of 382, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Didn't have time to order the PSU today so that means the hilarity of this comparison will probably be postponed to next week, my apologies. In the meantime I will open up a 15 year old Forton FSP 300W ATX PSU this weekend so that we may see how good it really was for a PSU that was basically the benchmark for quality in my circles circa 2002.

I'm also considering opening up a 25 year old 200W SPI AT PSU but I really fear what that may reveal 😁

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 341 of 382, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
TELVM wrote:
Found some pics of its guts here. […]
Show full quote
PhilsComputerLab wrote:
Any thoughts on this Pentium 4 era PSU? ... […]
Show full quote

Any thoughts on this Pentium 4 era PSU? ...

QsTEE6yl.jpg

Found some pics of its guts here.

05.jpg

Old style half-bridge with BJTs, passive PFC and group regulation. All or most caps are Teapo (could be much worse). Looks decently built though the heatsinks are a bit skinny.

If the caps and fan are OK should be fine to power retrocomps.

appiah4 wrote:

How much power would a Pentium MMX + PCI Graphics Card + Voodoo 2 12MB draw?

Probably not even 100W DC.

This kinda kept in the back of my mind ever since I read it.
My guess is that it's part of how FSP penny-pinched: For some reason they seem to somehow know where to penny-pinch without it compromising like most of the other PSU manufacturers sometimes do or did.

They never used pretty-pained LED-enhanced exterior casings till much later (making some savings there), for example.

The reason they may have used a slightly smaller heatsink might be due to the PSU being a P4-era unit. During this time, PSUs tended to either load lots on the 5v line (Athlon) or the 12v line (Netburst), so it was not expected that many people would load both 5v and 12v. Perhaps that's the reason they made the heatsink in this unit a bit on the smaller side?

But this is mostly just details, there's a good reason my most used PSU is FSP by far! 😁

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 342 of 382, by appiah4

User metadata
Rank l33t++
Rank
l33t++
Tetrium wrote:
This kinda kept in the back of my mind ever since I read it. My guess is that it's part of how FSP penny-pinched: For some reaso […]
Show full quote

This kinda kept in the back of my mind ever since I read it.
My guess is that it's part of how FSP penny-pinched: For some reason they seem to somehow know where to penny-pinch without it compromising like most of the other PSU manufacturers sometimes do or did.

They never used pretty-pained LED-enhanced exterior casings till much later (making some savings there), for example.

The reason they may have used a slightly smaller heatsink might be due to the PSU being a P4-era unit. During this time, PSUs tended to either load lots on the 5v line (Athlon) or the 12v line (Netburst), so it was not expected that many people would load both 5v and 12v. Perhaps that's the reason they made the heatsink in this unit a bit on the smaller side?

But this is mostly just details, there's a good reason my most used PSU is FSP by far! 😁

Agreed to all of that but man the old units with the 80mm fans at the exhaust instead of the 120mm fans at the bottom can make you deaf with prolonged use 🙁

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 343 of 382, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
appiah4 wrote:
Tetrium wrote:
This kinda kept in the back of my mind ever since I read it. My guess is that it's part of how FSP penny-pinched: For some reaso […]
Show full quote

This kinda kept in the back of my mind ever since I read it.
My guess is that it's part of how FSP penny-pinched: For some reason they seem to somehow know where to penny-pinch without it compromising like most of the other PSU manufacturers sometimes do or did.

They never used pretty-pained LED-enhanced exterior casings till much later (making some savings there), for example.

The reason they may have used a slightly smaller heatsink might be due to the PSU being a P4-era unit. During this time, PSUs tended to either load lots on the 5v line (Athlon) or the 12v line (Netburst), so it was not expected that many people would load both 5v and 12v. Perhaps that's the reason they made the heatsink in this unit a bit on the smaller side?

But this is mostly just details, there's a good reason my most used PSU is FSP by far! 😁

Agreed to all of that but man the old units with the 80mm fans at the exhaust instead of the 120mm fans at the bottom can make you deaf with prolonged use 🙁

I agree the 80mm fans are louder compared to 120mm fans in general use, but the FSP units were alright imo in this regard. The nature of sound being annoying can vary from person to person, I'm not really bothered by sounds that are sounding like wind or air moving. It's mostly the whining that drives me nuts 🤣! (like certain harddrives or CPU fans or coil whine).

And it shouldn't be hard to replace a PSU fan for whatever reason. Heck, even I did this once or twice and I didn't even need to solder it 🤣!
Some PSUs of that era had the fans soldered directly to the PCB of the PSU, but others used some kind of plug that was a bit wider compared to the standard 2-pin fan plugs that case fans used before 3-pin fan plugs became the standard.

If you decide to replace a fan with a slower moving part (which I'd only consider to advice someone if the person it far from a total idiot), make sure you load the PSU (way) less as overheating will cause the PSU to become more stressed. You could opt to increase case cooling to move as much of the hot air out before it enters the PSU.
It's still not recommend it though, but if you know what you're doing, this may sometimes be a better option than to not use the PSU at all.
Any part may be fantastic, but if it's too loud to enjoy, I'd never use it anyway. But I also don't want stuff to blow up, so good luck finding a good compromise 😜

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 344 of 382, by Skyscraper

User metadata
Rank l33t
Rank
l33t

Buying an old cheap used unit is perhaps not always a bad idea.

Here is an exampel of a decent CWT made PSU I bought late 2007 or early 2008. Today this unit would probably would fetch less than $10 on Epay. This PSU did not see very heavy use the first year or year and a half of it's life but I sold it february 2009 with a Socket 1366 system running an overclocked i7 920 @ 4GHz and a HD 4870 SLI setup. The friend who I built the system for is the kind of gamer that game 16 hours a day 7 days a week the whole year round, your stereotypical basement dweller! 😁

The system got upgraded with a GTX 580 and later a GTX 780 Ti, while these aren't as power hungry as the original HD4870 SLI the system was still rather power hungy as the overclocked i7 920 C1 CPU has a power draw over over 200 watts...

I traded in the system when I helped the owner build a new system some months ago and today I needed a PSU for testing the Asus NCCH-DL I just bought and thought of this PSU. Lets see how well this unit has aged in a very unfriendly environment.

This is the PSU, the very first generation Corsair TX750. Corsair model CMPSU-750TX. Review http://www.jonnyguru.com/modules.php?name=NDR … p=Story&reid=73

Corsair  model CMPSU-750TX.JPG
Filename
Corsair model CMPSU-750TX.JPG
File size
3.79 MiB
Views
1740 views
File license
Fair use/fair dealing exception

The PSU in the picture above looks rather clean, when I removed it from the system it wasn't as clean, this picture is taken after I got rid of the worst with a vacuum cleaner...

Corsair model CMPSU-750TX filth.JPG
Filename
Corsair model CMPSU-750TX filth.JPG
File size
4.13 MiB
Views
1740 views
File license
Fair use/fair dealing exception

After much huffing and puffing.

Corsair  model CMPSU-750TX inside 1.JPG
Filename
Corsair model CMPSU-750TX inside 1.JPG
File size
3.76 MiB
Views
1740 views
File license
Fair use/fair dealing exception

I took more images of the inside but this PSU is very cluttered and the linked review does better job when it comes to showing the details. I opted for an image showing the volages read by the Asus NCCH-DLs hardware monitor.

Corsair  model CMPSU-750TX voltages.JPG
Filename
Corsair model CMPSU-750TX voltages.JPG
File size
3.84 MiB
Views
1740 views
File license
Fair use/fair dealing exception

There are no visually bad caps and I will test the unit thoroughly during the weekend. The PSU has worked flawlessly for 9 and a half years, 8 of them with heavy load in an unfriendly environment so I think the ~100 euro I payed for it was well spent.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 345 of 382, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie
Skyscraper wrote:

The PSU has worked flawlessly for 9 and a half years, 8 of them with heavy load in an unfriendly environment so I think the ~100 euro I payed for it was well spent.

I've been using a Corsair TX-850 in my main PC since 2008 and it is still going strong to this day. It has endured different builds, gaming sessions, software development and other demanding tasks on a daily basis working almost 24/7. I see no reason to replace it yet, this thing seems to outlive everything else. 😁

Reply 346 of 382, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie
Tetrium wrote:

... The reason they may have used a slightly smaller heatsink ...

Is to cheap out on aluminium. To compensate they have to up the fan revs, hence:

appiah4 wrote:

... They are not quiet though. Don't believe the Noise Killer lie ...

The TX750 was solidly built (Chemicons, ball bearing fan, etc). I'd rather trust a 10 year old TX750 than a brand new $10 PoS.

Let the air flow!

Reply 347 of 382, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie

Hey guys. I haven't read through the whole thread... however based on some recent comments, I wanted to add the following:

1: You can't tell if caps are good by looking. You can sometimes tell if they are bad thanks to leakage/venting.
2: Voltage monitored via a multimeter or BIOS screen can tell you if you are generally running high or low, but won't show you the amount of ripple and noise vs an oscilloscope.

TELVM wrote:

I'd rather trust a 10 year old TX750 than a brand new $10 PoS.

Skyscraper wrote:

Buying an old cheap used unit is perhaps not always a bad idea.

Word.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 348 of 382, by appiah4

User metadata
Rank l33t++
Rank
l33t++
TELVM wrote:

The TX750 was solidly built (Chemicons, ball bearing fan, etc). I'd rather trust a 10 year old TX750 than a brand new $10 PoS.

Well, obviously. I'm not arguing the other way around. But would you really use a 750W PSU on a P1 MMX system? I mean, I also have an 8 year old quality ATX PSU in my current FX-8300 box, and it just keeps soldiering on (although not as efficiently as more modern PSUs I'd think) but I wouldn't waste such a unit on a system that would draw no more than 150W and that'd be a generous marging of comfort..

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 349 of 382, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Outside of problems with crossloading, the only thing using an over-specced PSU will cost you is a bit of efficiency. Certainly you have more options with a lighter demand system, but it doesn't fix fundamental design problems and make truly awful PSUs safe to use.

All hail the Great Capacitor Brand Finder

Reply 351 of 382, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Input filter is minimal, but it has what's needed. Can't comment on the size of the transistors or transformers as I don't know the company designators off by heart, but they're likely sized appropriately. Teapos are unsurprising, but they could be a lot worse. They're also using 105deg rated caps where it's crowded, which is smart.

Are those three lower ESR units on the output filter Teapo as well, or are they Tayeh's put there to frighten children? Either way, I may see a bit of venting which would make me want to check if they're still in spec.

All hail the Great Capacitor Brand Finder

Reply 352 of 382, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie
PhilsComputerLab wrote:

So I finally got around to opening that FSP PSU ...

Nice pics. Save for the oozing/bulging Teapos, it looks fairly decent.

A couple extra Y-caps on input filtering wouldn't harm, however there are two MOVs (between the bulk caps) and one '472' Y-cap across rectified primary ground and earth (good for ripple suppression). Can't read switchers or rectifiers but the trafos look decently sized. Skinny heatsinks but there is a fan regulation board (on the secondary heatsink).

Notice how in many FSPs the secondary heatsink touches and is secured to the case metal with a screw, for heat transfer and structural integrity.

Would be OK for retrocomps after a recap .

Let the air flow!

Reply 353 of 382, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie

For the scientific interest here's a review of a top-building-quality PSU made in Japan by Murata: http://www.tomshardware.com/reviews/cooler-ma … psu,4963-3.html

It has three fuses. Will probably be the only PSU still working after a martian attack with tripods. 🤣

Let the air flow!

Reply 354 of 382, by keropi

User metadata
Rank l33t++
Rank
l33t++

^ that's only 333,33$ per fuse, what a bargain 🤣 🤣 🤣

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 356 of 382, by Skyscraper

User metadata
Rank l33t
Rank
l33t

After messing with hardware for 25 years I killed my first PSU yesterday!

Well I have of course experienced PSUs failing because of bad caps and such but I have never managed to really kill one with fire, smoke and BANGs even though I often push them to the limit and then some.

Or did I really kill it? Was it really my fault? Can I find something or someone else to blame? I sure will at least try. Read PSU Inferno to find out if I succeed.

PSU inferno.

Yesterday I was messing with my "new" Dell Precision 530 MT dual Xeon 2.4 i860 system. The system had an Geforce4 MX440 video card when I bought it and I temporally replaced it with a nice Leadtek Winfast Geforce4 ti 4200. Because the case got beat up and skewed during the shipping the thingy meant for pushing down against the video card securing it when you close the "clam shell" instead pushed down against a capacitor with the result that the card now has one cap less.

The Leadtek ti 4200 with an amputated cap still worked fine but just to be safe I switched it for another ti 4200 and of course removed the offending thingy that caused the amputation. The system was much faster than expected and pushed the poor ti 4200 just as hard as a Socket A system with an overclocked Barton and as I planned to upgrarde the CPUs to 3GHz ones I realized that I needed a much faster video card than the GF4 ti 4600 I planned to use in the finished build.

I found one of my Gainward Geforce 7800 GS AGP cards and did a quick search on the net to see if they were known to work in the Dell Precision 530 MT. I found many forum posts written by people running a Geforce 7800 GS in the 530 MT as this combination always results in an "AGP voltage out of spec" warning. No one had seen any voltages actually being out of spec or experienced any other issues. Some even claimed to have been running a 7800 GS overclocked to 500+ MHz for years without seeing any negative effects other than the warning.

As the "AGP voltage out of spec" warning only seemed to occur with the Geforce 7800 GS and not with other even more power hungry AGP cards the general consensus seemed to be that this was a BIOS bug and not an actual issue.

The power supply is rated for 40 amps on the 3.3V rail, 35 amps on the 5V rail and 17 + 17 amps on the two 12V rails, the total max continuous load is specced at 460W. It's hard to see that power could be an issue with any AGP video card regardless of CPUs unless one is also trying to run a SCSI RAID setup on the 12V rail feeding the video card and one of the CPUs.

A vanilla 7800 GS AGP uses ~50W total, my factory overclocked Gainward 512MB card probably uses a lot more but still not more than at most 100W from 3.3V, 5V and 12V combined. A Prestonia Xeon 2.4 is rated at 65W or 77W depending on stepping and with my single ATA HDD, a couple of fans and some memory the total ends up at 300W DC as a worst case scenario. In reality I never saw more than 250W pulled from the wall during my testing and that's probably only ~200W DC.

With the above in mind I continued my benchmarking without any worries at all and later I played for an hour or two. I experienced no issues and the system performed much better than I though possible with two 2.4 GHz Prestonia Xeons.

No issues until I was going to shut the system off for the night that is. Powering off seemed to progress normally until the PSU was supposed to cut the power. Instead if silence there was an explosion followed by smoke and the nasty smell of burning electronics.

Well this warrants further investigation. Did my power hungry Gainward 7800 GS manage to pull too much power resulting in the death of the PSU? If so why did the protection circuit not get triggered? Why did the PSU never rev it's fans if it actually was overloaded? Why did the PSU die at the moment it was supposed to power down and not during benching or gaming?

Many questions, few answers. We need to see if we can find any clues inside.

At first I could not see anything wrong, all major rails seemed fine, all caps seemed fine, I could not see any burnt components. Explosions, fire and smoke always leaves traces but it took me a few moments to find them. When the PSU powers down the all famous 5VSB circuit should take over the watch. It did not do a very good job this time... I do not really think my Geforce 7800 GS is to blame?

The first thing I did when I got the system was to tear it apart as it had been damaged during shipping. I opened the PSU to clean it and I also took a few pictures for this thread. I thought it would be nice to post some pictures of a high quality PSU in this thread filled with gutless wonders.

Now I got good use for those pictures as I could not find a single image showing the inside of this PSU when I Googled at work today. I could only hope I had taken a picture good enough to identify the now unidentifiable IC in the 5VSB circuit. I was in luck and could easily see that it's a TOP223P TOPSwitch-II "Three-terminal Off-line PWM Switch".

Filename
power integrations_top221-227-327916.pdf
File size
473.69 KiB
Downloads
71 downloads
File license
Fair use/fair dealing exception

Here is an image showing the whole inside of the PSU.

Dell NPS-460AB A REV 02.JPG
Filename
Dell NPS-460AB A REV 02.JPG
File size
4.2 MiB
Views
1574 views
File license
Fair use/fair dealing exception

This is the affected area before the TOP223P blew.

Two days before Armageddon.JPG
Filename
Two days before Armageddon.JPG
File size
3.64 MiB
Views
1574 views
File license
Fair use/fair dealing exception

This is how the same area looks after the "incident".

The aftermatch.JPG
Filename
The aftermatch.JPG
File size
3.54 MiB
Views
1574 views
File license
Fair use/fair dealing exception

I really hope the Geforce 7800GS is alright! I have a few more of these Gainward cards but they are somewhat rare and also really nice so it would be a shame if it got damaged. I care less about the rest of the system as I kind of have lost trust in it but with luck nothing except for the PSU got damaged.

The main selling point when it comes to these Dell Precision 530 MT systems for me is their beefy PSUs making it possible to run dual 3GHz Gallatins and for example a HD3850, something I suspect many other dual socket 603 OEM systems from year 2001 can't handle. Using non standard PSUs is something all first generation dual Xeon OEM systems seem to have in common.

Now I think the best option might be to pay through the nose for a retail i860 motherboard. Using an i860 chipset motherboard is at least a given as I want to see exactly what the first generetion chipsets for Netbust CPUs with Rambus memory can handle. On the single CPU side of the project I use an i850 chipset motherboard from year 2000.

I just have to wonder why did this IC go up in flames? Did it die because of old age or is there an issue in the 5VSB circuit? I can not see any bad caps (one has been a bit discoloured by the fire nearby...) so I am tempted to just replace the IC and do a smoke test. Perhaps it's best to replace all caps nearby.

I have ordered 5x TOP223PN from China costing me 2 dollar for all inluding shipping. There are some cheap higher rated TOP224P ICs in Germany, perhaps the TOP224P is a better choice if I want to avoid another failure in the future?

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 357 of 382, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie
Skyscraper wrote:

... I just have to wonder why did this IC go up in flames? ...

If I had to bet:

"Control Voltage Supply
CONTROL pin voltage VC is the supply or bias voltage for the controller and driver circuitry. An external bypass capacitor closely connected between the CONTROL and SOURCE
pins is required to supply the gate drive current. The total amount of capacitance connected to this pin (CT) also sets the auto-restart timing as well as control loop compensation ..."

^ That critical cap went south and screwed the TOPswitch.

And/or the heat inside that PSU just cooked it.

TOPswitches are usually reliable, but Murphy never rests.

Let the air flow!

Reply 358 of 382, by Skyscraper

User metadata
Rank l33t
Rank
l33t

Yes the critical cap...

If you know or could guess which cap we are talking about I can investigate further otherwise I have to actually read the PDF, measure and follow traces which I probably will not do until I have the replacement TOP223P(N) from China in my hand. 😀

edit +s

Last edited by Skyscraper on 2017-04-29, 15:10. Edited 1 time in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 359 of 382, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie

NDGj6Jv6.jpg

^ My guess from a distance is that the caps circled in red are probably related to +5VSB circuit, at least two of them.

I'd replace with lowish ESR, same uF, and 50V rated (regardless of the originals being 25V or 35V) from some quality long-life series, like Chemicon KYA, Panny FR or Rubycon YXJ.

Let the air flow!