FSP400-GHN. 23A on the 5V rail, under 40Euros. Have several of these and they are great for powering old builds, S462 included.

From the other side of the ocean (Europe), how about the Enermax EG365AX-VD(G) it is rated for 185W of combined power, is it good?

May I get a little cheeky? Any modular power supplies that will fit the bill or at least get reasonably close 😜?

Yes, the fully modular Seasonic M12II-520 Evo.
The 5V rail is rated at 24 amps (130W total for the 3.3V + 5V load). I tested it just for fun on a Thunderbird 1.4 GHz system with a GeForce 3 Ti 200 card and it worked perfectly, did multiple benchmarks without any issues. Now, I'm not sure if it can take this strain for very long periods of time, but I have faith in Seasonic, they usually can handle more than the advertised amperage.

This might sound silly, but has anyone ever directly measured the amperage draw on each of the rails for common setups (430 boards w p1, 440bx with early and late p2, etc ) to see what its actually doing at rest and under a gaming load?
Knowing even a typical load would help a lot.

Cobra42898 wrote on 2020-04-26, 13:29:

This might sound silly, but has anyone ever directly measured the amperage draw on each of the rails for common setups

Yes, of course and I posted the results for my Athlon system

Neat software!
i thought about measuring externally, but that way is much easier.
what systems does it work on?

Cobra42898 wrote on 2020-04-26, 14:44:

what systems does it work on?

There are native SIVs for pretty much any version of Windows 9x/NT/2K/XP and later, but you need a PSU that reports the power usage. See Menu->Tools->Link Limits to check which are currently supported.

Nemo1985 wrote on 2020-04-26, 08:30:

From the other side of the ocean (Europe), how about the Enermax EG365AX-VD(G) it is rated for 185W of combined power, is it good?

Have one on my slot a system, no complaints.

sunaiac wrote on 2020-04-26, 18:45:

Nemo1985 wrote on 2020-04-26, 08:30:

From the other side of the ocean (Europe), how about the Enermax EG365AX-VD(G) it is rated for 185W of combined power, is it good?

Have one on my slot a system, no complaints.

Thank you for the feedback, I think that such psu, if the label is honest, should be used in a socket a system.

the 365 and 465 are good psus, I had one of them explode on me though once ^^

Nemo1985 wrote on 2020-04-26, 19:02:

Thank you for the feedback, I think that such psu, if the label is honest, should be used in a socket a system.

I'm no PSU guru, nor did I put it through an oscilloscope test.
But the brand should be reliable, and the power delivered is good for those systems.
I bought it specifically for slot A and socket A systems (I have two PSU of the 365 model)
P3s are less needy, and P4+ are 12V oriented.

imi wrote on 2020-04-26, 19:13:

the 365 and 465 are good psus, I had one of them explode on me though once ^^

Old Capa ?
I guess I should start consider redoing mine then 😀

probably, that was many many years ago, equipped all the P4 office PCs with them, and it happened while the PC was still in use... so not that old even ...I still have a few of them... and I do not know which one is the exploded one x3 ...so should check all of those before I use them.

Garrett W wrote on 2020-04-25, 19:01:

I seem to remember someone saying that the motherboards that have the 4pin P4 plug are not entirely safe from this issue either and you still need a powerful 5V line. Has anybody ever looked into that?

OP, are you running the original Thoroughbred 2800+ or a Barton model? Phil's Computer Lab tested a couple of PSUs and Duron/Athlon processors some years back and found that even a lowly Corsair VS450 can handle Durons all the way up to ~57W TDP, although I'm not sure if those numbers can be trusted (not to mention TDP is weird).

I did. Tested with three nforce 2 motherboards (Abit AN7, Asrock K7NF2 and a Gigabyte Ga-7N400) that have a 4 pin 12v CPU power connector. Using a modern Seasonic Core GC, 80+ Gold, 650W witch boasts 20A on the 5v rail I encoutered instability running a 3200+ (2.33GHz FSB 333 version) and a Geforce FX 5900 Ultra .

Switching to a much older Mushkin XP-650AP (circa 2007 I think, no efficiency certification but a rock solid PSU) that sports a 30A 5v rail made the system stable.

I ran the PC like that for a while, then decided to run a couple more hard disks in it, and the problems came back. Not as severe as before, but the system will sometimes lock up in demanding games after it's been running for more then 60 minutes. I also noticed that under windows 98, even with 512MB of ram, the PC will hang in 3dmark 2001 in the first half of the test suite, something that does not occur under windows XP. This happenes with all 3 motherboards. Using the 650w seasonic PSU, the system will hang after loading the taskbar with all motherboars except for the abit AN7.

After that, I decided to go nuts and tested the system again using my Corsair HX 1200 platinum, witch according to it's specs boasts a 30A 5v rail. It was perfectly stable under XP and 98, using all motherboards - so at this point I was thinking "what gives?"

I got a hold of a clamp amp-meter, and decided to measure the amount of power going trough the 12v lines of the 4 pin CPU aux connector. Under full load, in windows 98, using the Abit AN7 board, the CPU uses 4 to 6 amps off the 12v rail witch is not quite right, since the 3200+ has a power rating of 60.4 / 76.8 Watt (or at least this is what CPU world claims). Then I put the amp-meter on the 5v lines running to the ATX connector, and noticed it was drawing over 15 amps from that line as well.

I still don't understand what exactly is happening, but it seems the CPU either uses power off both the 5v and the 12v rails, or the 4 pin CPU connector only feeds the PCI and AGP slots on the motherboard. Even so, considering the power draw I measured, any modern PSU with a 20A 5v rail should be enough for a 3200+ right? But in practice that's not the case...

Perhaps the issue is how "clean" the 5v rail is? Personally, I'm stumped.

I posted a small thread on this topic earlier.
The PSU Matchup Thread

You may be able to find new build Startech and Athena power ATX supplies that follow the ATXv1.3 power allocations. Athena are generally junk, but the Startech PSUs are fine.
https://www.startech.com/ca/Computer-Parts/PS … ply~ATXPOWER300

I believe FSP is no longer producing those older design PSUs, although they did produce them quite late compared to most other OEMs so you can probably find NOS samples with not so much shelf time.

gdjacobs wrote on 2020-04-26, 19:50:

I posted a small thread on this topic earlier. The PSU Matchup Thread […]
Show full quote

I posted a small thread on this topic earlier.
The PSU Matchup Thread

You may be able to find new build Startech and Athena power ATX supplies that follow the ATXv1.3 power allocations. Athena are generally junk, but the Startech PSUs are fine.
https://www.startech.com/ca/Computer-Parts/PS … ply~ATXPOWER300

I believe FSP is no longer producing those older design PSUs, although they did produce them quite late compared to most other OEMs so you can probably find NOS samples with not so much shelf time.

Do you know who the OEM is for those Startech units ? I can't imagine they are making them themselves and I find hard to believe that the OEM is not selling them under its own brand or to other distributors .
Unless Startech invested in a warehouse full of them when they were still being made and is now sitting on NOS that it is selling as new to an admittedly niche audience .

darry wrote on 2020-04-26, 20:30:

Do you know who the OEM is for those Startech units ? I can't imagine they are making them themselves and I find hard to believe that the OEM is not selling them under its own brand or to other distributors .
Unless Startech invested in a warehouse full of them when they were still being made and is now sitting on NOS that it is selling as new to an admittedly niche audience .

I believe Startech made the older ones. Newer ones seem to be ATNG.
https://www.tomshardware.com/reviews/power-su … rer,2913-9.html

Socket3 wrote on 2020-04-26, 19:20:

I did. Tested with a three nforce 2 motherboards (Abit AN7, Asrock K7NF2 and a Gigabyte Ga-7N400) that have a 4 pin 12v CPU powe […]
Show full quote

Garrett W wrote on 2020-04-25, 19:01:

I seem to remember someone saying that the motherboards that have the 4pin P4 plug are not entirely safe from this issue either and you still need a powerful 5V line. Has anybody ever looked into that?

OP, are you running the original Thoroughbred 2800+ or a Barton model? Phil's Computer Lab tested a couple of PSUs and Duron/Athlon processors some years back and found that even a lowly Corsair VS450 can handle Durons all the way up to ~57W TDP, although I'm not sure if those numbers can be trusted (not to mention TDP is weird).

I did. Tested with a three nforce 2 motherboards (Abit AN7, Asrock K7NF2 and a Gigabyte Ga-7N400) that have a 4 pin 12v CPU power connector. Using a modern Seasonic Core GC, 80+ Gold, 650W witch boasts 20A on the 5v rail I encoutered instability running a 3200+ (2.33GHz FSB 333 version) and a Geforce FX 5900 Ultra .

Switching to a much older Mushkin XP-650AP (circa 2007 I think, no efficiency certification but a rock solid PSU) that sports a 30A 5v rail made the system stable.

I ran the PC like that for a while, then decided to run a couple more hard disks in it, and the problems came back. Not as severe as before, but the system will sometimes lock up in demanding games after it's been running for more then 60 minutes. I also noticed that under windows 98, even with 512MB of ram, the PC will hang in 3dmark 2001 in the first half of the test suite, something that does not occur under windows XP. This happenes with all 3 motherboards. Using the 650w seasonic PSU, the system will hang after loading the taskbar with all motherboars except for the abit AN7.

After that, I decided to go nuts and tested the system again using my Corsair HX 1200 platinum, witch according to it's specs boasts a 30A 5v rail. It was perfectly stable under XP and 98, using all motherboards - so at this point I was thinking "what gives?"

I got a hold of a clamp amp-meter, and decided to measure the amount of power going trough the 12v lines of the 4 pin CPU aux connector. Under full load, in windows 98, using the Abit AN7 board, the CPU uses 4 to 6 amps off the 12v rail witch is not quite right, since the 3200+ has a power rating of 60.4 / 76.8 Watt (or at least this is what CPU world claims). Then I put the amp-meter on the 5v lines running to the ATX connector, and noticed it was drawing over 15 amps from that line as well.

I still don't understand what exactly is happening, but it seems the CPU either uses power off both the 5v and the 12v rails, or the 4 pin CPU connector only feeds the PCI and AGP slots on the motherboard. Even so, considering the power draw I measured, any modern PSU with a 20A 5v rail should be enough for a 3200+ right? But in practice that's not the case...

Perhaps the issue is how "clean" the 5v rail is? Personally, I'm stumped.

Clamp ampmeter tool is usually intended for AC current measurement not for DC current. Is yours a DC capable type?

Cheers,

pentiumspeed wrote on 2020-04-26, 20:58:

Socket3 wrote on 2020-04-26, 19:20:

I did. Tested with a three nforce 2 motherboards (Abit AN7, Asrock K7NF2 and a Gigabyte Ga-7N400) that have a 4 pin 12v CPU powe […]
Show full quote

Garrett W wrote on 2020-04-25, 19:01:

I seem to remember someone saying that the motherboards that have the 4pin P4 plug are not entirely safe from this issue either and you still need a powerful 5V line. Has anybody ever looked into that?

OP, are you running the original Thoroughbred 2800+ or a Barton model? Phil's Computer Lab tested a couple of PSUs and Duron/Athlon processors some years back and found that even a lowly Corsair VS450 can handle Durons all the way up to ~57W TDP, although I'm not sure if those numbers can be trusted (not to mention TDP is weird).

I did. Tested with a three nforce 2 motherboards (Abit AN7, Asrock K7NF2 and a Gigabyte Ga-7N400) that have a 4 pin 12v CPU power connector. Using a modern Seasonic Core GC, 80+ Gold, 650W witch boasts 20A on the 5v rail I encoutered instability running a 3200+ (2.33GHz FSB 333 version) and a Geforce FX 5900 Ultra .

Switching to a much older Mushkin XP-650AP (circa 2007 I think, no efficiency certification but a rock solid PSU) that sports a 30A 5v rail made the system stable.

I ran the PC like that for a while, then decided to run a couple more hard disks in it, and the problems came back. Not as severe as before, but the system will sometimes lock up in demanding games after it's been running for more then 60 minutes. I also noticed that under windows 98, even with 512MB of ram, the PC will hang in 3dmark 2001 in the first half of the test suite, something that does not occur under windows XP. This happenes with all 3 motherboards. Using the 650w seasonic PSU, the system will hang after loading the taskbar with all motherboars except for the abit AN7.

After that, I decided to go nuts and tested the system again using my Corsair HX 1200 platinum, witch according to it's specs boasts a 30A 5v rail. It was perfectly stable under XP and 98, using all motherboards - so at this point I was thinking "what gives?"

I got a hold of a clamp amp-meter, and decided to measure the amount of power going trough the 12v lines of the 4 pin CPU aux connector. Under full load, in windows 98, using the Abit AN7 board, the CPU uses 4 to 6 amps off the 12v rail witch is not quite right, since the 3200+ has a power rating of 60.4 / 76.8 Watt (or at least this is what CPU world claims). Then I put the amp-meter on the 5v lines running to the ATX connector, and noticed it was drawing over 15 amps from that line as well.

I still don't understand what exactly is happening, but it seems the CPU either uses power off both the 5v and the 12v rails, or the 4 pin CPU connector only feeds the PCI and AGP slots on the motherboard. Even so, considering the power draw I measured, any modern PSU with a 20A 5v rail should be enough for a 3200+ right? But in practice that's not the case...

Perhaps the issue is how "clean" the 5v rail is? Personally, I'm stumped.

Clamp ampmeter tool is usually intended for AC current measurement not for DC current. Is yours a DC capable type?

Cheers,

yup. It's a UT203 made by a company called Uni-T.