@bloodem

Could you give your 1400 Thunderbird a short run in Prime95 (in place large FFT's), and measure the 5V rail with a multimeter?

I find it really odd how Segotep rated this cheaper model, with a higher output on the 5v+3.3v rails 🤔 I'm very curious how these two models would fair against each other.
_________
Much later EDIT:

For only 15euro more, I just bought the "Segotep ATX-500WH".
I just finished testing it in the exact same circumstances, with the GF3Ti500 GPU installed:

The Segotep ATX-500WH delivers exactly 0.1V more than the Segotep ATX-500W12 , and the +5V ATX wires aren't hot any more, and it also seems to have an extra transformer inside.

So I very much agree with @bloodem :
For the price, I fully recommend the Segotep ATX-500WH !!

B24Fox wrote on 2023-09-22, 23:25:

@bloodem

Could you give your 1400 Thunderbird a short run in Prime95 (in place large FFT's), and measure the 5V rail with a multimeter?

Oh, sorry, missed your message (i.e.: forgot about it 😁 ).
Anyway, I can't test the Thunderbird right now, because it's in storage (in a different city).

Glad you like the Segotep ATX-500WH. It's a decent PSU (definitely much better compared to some of the PSUs I was using 2 decades ago).

bloodem wrote on 2023-09-26, 19:16:

Oh, sorry, missed your message (i.e.: forgot about it :-D ). Anyway, I can't test the Thunderbird right now, because it's in sto […]
Show full quote

B24Fox wrote on 2023-09-22, 23:25:

@bloodem

Could you give your 1400 Thunderbird a short run in Prime95 (in place large FFT's), and measure the 5V rail with a multimeter?

Oh, sorry, missed your message (i.e.: forgot about it 😁 ).
Anyway, I can't test the Thunderbird right now, because it's in storage (in a different city).

Glad you like the Segotep ATX-500WH. It's a decent PSU (definitely much better compared to some of the PSUs I was using 2 decades ago).

It's ok; no harm done, as after today's test, I'm DEFFINITELY keeping this psu 😁

Done some more testing with the Segotep ATX-500WH, and found out that if you populate all MOLEX + FDD connectors (in my case: 2x HDDs + 1x CDrom + 1x FDD)... and measure at the end of the red/black (+5v)cable, against the unpopulated SATA rail; it has a severe voltage drop (almost 0.2V) on the rail/wire with everything populated.

So I highly suggest cutting off the entire SATA rail, and soldering a MOLEX rail instead, so the load can at least be split in half.
(getting thicker gauge wires, and soldering them directly to the PSU's PCB, would be the ideal route)

Don't even wanna know what a Voodoo 5 will do to that thin cable.
AFAIK, Voodoo5 takes almost 30W from the +5V rail .. and probably, most of it from the molex connector.

I keep looking at this PSU, https://www.amazon.co.uk/gp/product/B0CBXBSTHY, claiming a 35A 5V rail. I know that it's no-name junk, but it looks a lot nicer than the no-name junk that came bundled with cheap cases that I used to use back in the Athlon XP days. I have an socket A board I'm considering reviving into a working machine so perhaps I'll try it.

WildW wrote on 2023-09-28, 20:36:

I keep looking at this PSU, https://www.amazon.co.uk/gp/product/B0CBXBSTHY, claiming a 35A 5V rail. I know that it's no-name junk, but it looks a lot nicer than the no-name junk that came bundled with cheap cases that I used to use back in the Athlon XP days. I have an socket A board I'm considering reviving into a working machine so perhaps I'll try it.

Seems to have pretty good reviews on eBay:
https://www.ebay.co.uk/urw/ACE-A-500BR-ATX-Po … -lastEditedDate

(nothing mentioned about the 5V rail, though..)

When looking into PSUs for old 5V systems its good to have reviews that look into cross-loads, which I really doubt any ebay or amazon reviews will have. It's always just "it works so its 5*".

Some time ago I bought a PSU (Akyga brand I think) that had 25+A for 5V,. but after I put it in an old Athlon system I saw that the 12V rail was actually giving over 13 Volts due to the 5V load. And this was just in the Bios without the CPU being stressed.

My PSU has only 20A on the 5V line and it can handle the most powerful systems I throw at it (Athlon 1400 @ 1500 MHz, Barton XP @ 2400 MHz), paired with a Radeon 9800 Pro. It makes sense if you do the math (75W CPU, 15W graphics card, 5W hard disk = 95W). The only problem would be graphics cards that draw a lot more than that from the 5V line, but I believe those are rare - I've read that the Geforce 6000 series already uses 12V.

Next time I use such a system I'll remember to measure the current on the 5V line, would be interesting how close it is to maxing out.

also you can probably further draw some more on the cpu side with a 2700mhz oc mobile barton of the nicest samples, or perhaps a dual socket A board with two XPs, and I guess with a 9800XT instead of Pro

I tested my corsair ax1600i that is on my main pc with psu tested and it could handle 200W load on the 5V rail easily.

Munx wrote on 2023-09-29, 05:52:

When looking into PSUs for old 5V systems its good to have reviews that look into cross-loads, which I really doubt any ebay or amazon reviews will have. It's always just "it works so its 5*".

Some time ago I bought a PSU (Akyga brand I think) that had 25+A for 5V,. but after I put it in an old Athlon system I saw that the 12V rail was actually giving over 13 Volts due to the 5V load. And this was just in the Bios without the CPU being stressed.

Yes. Very good point. Avoid both cheap as well as group-regulated power supplies!

B24Fox wrote on 2023-09-28, 21:52:

Seems to have pretty good reviews on eBay: https://www.ebay.co.uk/urw/ACE-A-500BR-ATX-Po … -lastEditedDate […]
Show full quote

WildW wrote on 2023-09-28, 20:36:

I keep looking at this PSU, https://www.amazon.co.uk/gp/product/B0CBXBSTHY, claiming a 35A 5V rail. I know that it's no-name junk, but it looks a lot nicer than the no-name junk that came bundled with cheap cases that I used to use back in the Athlon XP days. I have an socket A board I'm considering reviving into a working machine so perhaps I'll try it.

Seems to have pretty good reviews on eBay:
https://www.ebay.co.uk/urw/ACE-A-500BR-ATX-Po … -lastEditedDate

(nothing mentioned about the 5V rail, though..)

This PSU's manufacturer

1) does not seem to have website (at least I could not find it)
2) has no visible company coordinates (address and/or phone number) that could be discerned on any online photos of it or its packaging (maybe there is something in the manual, if there is one)

This brings about 2 questions :

a) to what degree can one trust any of the specs provided ?
b) if something goes wrong with the PSU (it breaks and/or kills other components and/or catches fire, etc ) who is liable and/or can actually be made to face liability ?

A001 wrote on 2023-09-29, 14:18:

Munx wrote on 2023-09-29, 05:52:

When looking into PSUs for old 5V systems its good to have reviews that look into cross-loads, which I really doubt any ebay or amazon reviews will have. It's always just "it works so its 5*".

Some time ago I bought a PSU (Akyga brand I think) that had 25+A for 5V,. but after I put it in an old Athlon system I saw that the 12V rail was actually giving over 13 Volts due to the 5V load. And this was just in the Bios without the CPU being stressed.

Yes. Very good point. Avoid both cheap as well as group-regulated power supplies!

A 12V hungry HDD could be used as "ballast" in this situation. Some models can draw almost 1Amp from the 12V rail (you need to check on the label).
Obviously not ideal; but that should lower the voltage on +12V Rail in some cases by ~0.3V

TBH, I'm still wondering how bad is an extra 10% on the 12V rail, seeing as most (if not all?) hardware that use that rail, step down that voltage to whet they need.
The HDD motors might not be to happy though; as I presume they're the only thing running directly off of 12V.

B24Fox wrote on 2023-09-29, 19:20:

TBH, I'm still wondering how bad is an extra 10% on the 12V rail, seeing as most (if not all?) hardware that use that rail, step down that voltage to whet they need.

In most cases that's actually better, because more voltage equals less current and less heat.

The Serpent Rider wrote on 2023-09-29, 19:30:

B24Fox wrote on 2023-09-29, 19:20:

TBH, I'm still wondering how bad is an extra 10% on the 12V rail, seeing as most (if not all?) hardware that use that rail, step down that voltage to whet they need.

In most cases that's actually better, because more voltage equals less current and less heat.

I'm no expert; but isn't heat (W) = Volts x Amps ??
So even if the amount of Amps that are drawn, would go down (*which I don't think it does) ; you'd still have the same Wattage, i.e. the same heat output.

*because, for example, if you undervolt a CPU, you are guaranteed to have a smaller heat output

B24Fox wrote on 2023-09-29, 19:51:

I'm no expert; but isn't heat (W) = Volts x Amps ?? So even if the amount of Amps that are drawn, would go down (*which I don […]
Show full quote

The Serpent Rider wrote on 2023-09-29, 19:30:

B24Fox wrote on 2023-09-29, 19:20:

TBH, I'm still wondering how bad is an extra 10% on the 12V rail, seeing as most (if not all?) hardware that use that rail, step down that voltage to whet they need.

In most cases that's actually better, because more voltage equals less current and less heat.

I'm no expert; but isn't heat (W) = Volts x Amps ??
So even if the amount of Amps that are drawn, would go down (*which I don't think it does) ; you'd still have the same Wattage, i.e. the same heat output.

*because, for example, if you undervolt a CPU, you are guaranteed to have a smaller heat output

Same wattage using higher voltage and lower amperage does indeed produce less heat from the wires that the current is going through, but I don't think the same goes for power regulation components.

While ~+10% voltage is still within ATX spec (as outdated as it is), personally I don't feel comfortable subjecting old budget Socket A boards that were known for all sorts of corner cutting to it.

Same wattage using higher voltage and lower amperage does indeed produce less heat from the wires that the current is going through, but I don't think the same goes for power regulation components.

It also affects VRM efficiency, with higher voltage providing better results. That's why it's easier to get decent PSU 80+ ratings from 220-230V source.

well... "the more you know... " 😄

thanks guys for the little lesson 😁

In case someone stumbled across this thread searching for a modern PSU to power his retro machine, I can fully recommend EVGA models as other posters mentioned before.

I have this EVGA 450 BR in my Socket 7 system:
https://www.evga.com/products/product.aspx?pn=100-BR-0450-K1

My rig:
- Asus P5A Board
- Pentium MMX 233MHz
- 64MB SDRAM
- Trio64 PCI graphics card
- SB16 clone soundcard + WP32 McCake daughterboard with external front panel (needs floppy power)
- GUS PnP clone soundcard
- 1x "HDD" (via IDE->CF Adapter)
- 1x 3.5" HxC floppy drive emulator
- 1x CD-ROM drive

I have no issues with the EVGA 450 BR.
It is absolutely silently. Very impressing!

It has 20A on 3.3V + 5V rail, combined 120W which should be more than enough.
What I like about the EVGAs is that they still have the 20-pin mainboard plug so you don't need an adapter.
It comes with 3x Molex and 1x Floppy connector which can be too little for some systems.
In this case just use Molex splitters and Molex->Floppy adapters. These are cheap.
The EVGAs are still in production and so built of new components.
I think I will put a few aside just to be prepared.

If you need more power you can look for higher wattage EVGA models.
But you have to look carefully on the specs. Some have more 3.3V/5V power but some have less. Also watch out for the combined 3.3V+5V wattage.