retardware wrote on 2021-09-15, 16:55:

Interesting.
Looking up, I found that ATX 2.4 seems to have added a minimum load requirement to -12V, which might cause problems on older stuff.

So... anybody wanna hook up a case fan -12V to 0 and see if that lets ATX 2.4 supplies start?

retardware wrote on 2021-09-15, 16:55:

Because, only reading names like "Cörsäïr", "ÖCZ", "Thërmältäkë" makes me want to skip over, this stuff imho belongs into the trash, not worthy to put on the test bench.

Certain PSU from other brands may also be built by Seasonic, in case you didn't know.

I have an Antec EarthWatts PSU made by that company.

BitWrangler wrote on 2021-09-15, 17:12:

retardware wrote on 2021-09-15, 16:55:

Interesting.
Looking up, I found that ATX 2.4 seems to have added a minimum load requirement to -12V, which might cause problems on older stuff.

So... anybody wanna hook up a case fan -12V to 0 and see if that lets ATX 2.4 supplies start?

If that 's the only difference and reason for them psus not bootin' It's worth a try.

Now. do you hook the fan +12v to ground and the fan gnd to -12v or was it the other way around ? 😉
What's the minimum requirement anyway? (too lazy to look it up)
Would a couple of LED's be sufficient since a nice high power fan would be more than the 0.3A that some psu's are spec'ed for on their -12V line.

Black to most negative and red to most positive, so black or normal 0V connection to -12V and red or normal 12V connection to 0v.

The ATX 2.4 PSU's seemed to boot fine on the P4S533 and P4S533-E, it was the motherboard that didn't want to boot with them, so I doubt putting a fan on the -12v will do anything in this situation.

Limpem wrote on 2021-09-16, 08:38:

The ATX 2.4 PSU's seemed to boot fine on the P4S533 and P4S533-E, it was the motherboard that didn't want to boot with them, so I doubt putting a fan on the -12v will do anything in this situation.

Did you measure all the voltages to be present and accurate, including the PG signal, on the powered up psu and non booting motherboard combo ?

To be honest, I didn't measure those voltages.
Although I have a multimeter, I wouldn't know how to measure it when the ATX cable is connected.

I understand Seasonic are really impressive PSUs but as far as I understand it seems like modern version of lower end brands seems to improve at every new models. I know they might not have a warranty lifetime or perfect voltages like the mentioned brand but compared to the past common PSUs things got better (or not?) in the last years at least in the protections layout design. Then I always see the usual chinese capacitors inside these low wattage PSU (even the supposed good ones sometimes) but I also see generally they improved the internal component choices and design for what they costs.
I don't know if voltage ripples can damage over the years the components but at least shouldn't we expect to at least work enough good inside the warranty period?
Lately I bought a Coolermaster MWE500 v2.0 and far from being an high end PSU it seems like they much improved the internal design beside the capacitor brand and temperature limit. With the price of the lower Seasonic you can buy almost another one brand new at the end of the warranty period. If I could afford it I would have bough the Seasonic without even thinking about it of course but for a Socket 775 mainboard is more than enough to make it last longer.

If you recap your motherboards with super low ESR capacitors to survive ripple currents and recap your PSUs to super high quality caps to avoid ripple currents, you're wasting money at one end or the other 🤣

Limpem wrote on 2021-09-18, 11:11:

To be honest, I didn't measure those voltages.
Although I have a multimeter, I wouldn't know how to measure it when the ATX cable is connected.

The connectors are open on both ends.
Just stick in the multimeter probe "nails" into the appropriate holes; they will contact with the crimped-on connector plugs of the cables coming from the PSU.
This is very easy and the risk of damage is minuscule, as the probes cannot slide away and short out things.

386SX wrote on 2021-09-18, 11:59:

I don't know if voltage ripples can damage over the years the components but at least shouldn't we expect to at least work enough good inside the warranty period?

You only know the qualities of a PSU if you put it on the test bench and meter it in various load configurations, measuring the output on voltmeter and oscilloscope.

The damage mechanism of ripple coming from the PSU is this:
The caps on the mobo/cards/peripherals are only intended to even out the ripple caused by load spikes.
This causes only minor currents, limited to very short times with long recovery times inbetween.
They are not designed to even out ripple coming from the PSU.
For this reason the ATX spec defines a maximum of 50mV PSU ripple (1%) on 5V, and 120mV on 12V.

Excessive PSU ripple causes the capacitors on boards/devices to buffer the PSU ripple also.
The consequence of this is that they are permanently loaded with continuous high currents they are not intended/designed for.
In turn they deteriorate, get hot, life expectancy is massively reduced, and early failure (bloating, leaking, explosions,...) is common.

And now think a step further:
Why do you, for example, see series of tantalum explosions on some particular boards, while the majority of the same type just behaves well and inconspiciously?
My personal guess is that very often the blame for symptomatic damage done by bad power supplies is being put onto their innocent victims.

retardware wrote on 2021-09-18, 12:44:

You only know the qualities of a PSU if you put it on the test bench and meter it in various load configurations, measuring the […]
Show full quote

386SX wrote on 2021-09-18, 11:59:

I don't know if voltage ripples can damage over the years the components but at least shouldn't we expect to at least work enough good inside the warranty period?

You only know the qualities of a PSU if you put it on the test bench and meter it in various load configurations, measuring the output on voltmeter and oscilloscope.

The damage mechanism of ripple coming from the PSU is this:
The caps on the mobo/cards/peripherals are only intended to even out the ripple caused by load spikes.
This causes only minor currents, limited to very short times with long recovery times inbetween.
They are not designed to even out ripple coming from the PSU.
For this reason the ATX spec defines a maximum of 50mV PSU ripple (1%) on 5V, and 120mV on 12V.

Excessive PSU ripple causes the capacitors on boards/devices to buffer the PSU ripple also.
The consequence of this is that they are permanently loaded with continuous high currents they are not intended/designed for.
In turn they deteriorate, get hot, life expectancy is massively reduced, and early failure (bloating, leaking, explosions,...) is common.

And now think a step further:
Why do you, for example, see series of tantalum explosions on some particular boards, while the majority of the same type just behaves well and inconspiciously?
My personal guess is that very often the blame for symptomatic damage done by bad power supplies is being put onto their innocent victims.

I understand it and agree. Usually some decades ago most common people didn't even know (me also) a lot about PSU and how much they are important, not to mention powering some high end config where some spent a lot on many components but with the cheapest PSU. But I understand its importance in the early 2000's cause a brand new cheap PSU once powered did break immedately an IDE disk and almost burned the connected vga. And it was brand new one! Also happened that brand new cheap PSUs stopped working (who knows in which way with which voltage peaks or whatever) after a couple of weeks with just 50W of load..
But lately PSU tests and reviews are more common than ever and I suppose even manufacturer begin to invest a lot more into the design of the PSU compared to the awful late 90's paper light PSUs. I know we are still far from having such quality into low end brand ones anyway. Are those high end PSU still a better choice? Of course! But on the other side modern PSU situation "feels" like being better than in the past.
What I wonder is that beside having a good PSU, maybe also the mainboard component choices/design quality make me ask how much sometimes too many modern ones seems to stop working without visible problems maybe for some micro IC that stop working, some mosfet gone, or maybe an almost invisible passive component detached somewhere.

Interesting problem you've had there , I have a very similar 470w Thermaltake PSU in my in my p4/gf4 98se build and it works perfectly with the Asus p4P800 board I have.
Glad you've got it sorted , there are some strange problems that come along with these retro builds aren't there.
Lets see some pics when you get it all finished , also , I didn't realise there were P4 boards that supported the V3 agp cards , got me thinking about my next build now. 🙄