tayyare wrote:

I think power requirements sometimes over exaggerated. I had a PIII 1000 Coppermine rig that I recently took apart for another b […]
Show full quote

I think power requirements sometimes over exaggerated. I had a PIII 1000 Coppermine rig that I recently took apart for another built. It was my first attempt for a retro build, even long before I started involving with vogons, so it was some kind of a freak without no real practical purpose. The PSU of that machine was a an old and tired Aopen FSP300-60ATV with following specs:

+3.3V 28A
+5V 30A
+12V 15A
-12V 0.8A
-5V 15A
+5Vsb 2A

That is a Fortron Source (FSP) PSU OEMed to AOpen. Good solid period correct PSU. FSPs often need new caps after a while. It's hit and miss though.
.
The problem with using newer PSUs with older systems is by the time you get enough amps/watts on +5v for the older system the PSU has ridiculous amounts of +12v watts.
That makes the total watts higher than what it would be with an older spec PSU for the same amount of +5v amps/watts.
.
I will look at the other you wrote a little later. This time of day I'm only here 5 or 10 minutes at a time.
.

ok. 🤣

The highest ever draw from the wall I've ever seen from my PIII (Tually 1400-S @ 1.6, GF6200 ,768MB of RAM, 1x SSD on Promise s150 TX2+, 1x HDD, 1x DVD, AWE64 Gold, PCI USB 2.0 card, 2x 120mm fans, 2x 92mm fans) is 98W AC.

The PSU is a (restored) Oh Deer with a glorious inefficiency of about 70% at that load, so system real DC draw is 98 * .7 = ~70W DC tops.

If all 70W were drawn from the +5V rail, one 15A rectifier on the +5V rail would suffice.

But that's not the case. The fans and the HDD & DVD motors draw from +12V. Other components draw from +3.3V.

Thus a 12A rectifier on +5V would suffice.

For the record, said Oh Deer PSU previously had spent ~8 years powering a P4 3.0 HT Preshott system without problem. Funny thing is, it managed to do that with only a measly 10A STPR1020CT Superfast (not even Schottky) rectifier on the +12V rail. 🤣

SRQ wrote:

... Dorky lights aside, this looks like a fair quiet PSU- http://www.newegg.ca/Product/Product.aspx?Ite … N82E16817148008

This one is miles better, and cheaper: http://www.newegg.ca/Product/Product.aspx?Ite … 1-086-_-Product

Or this one: http://www.newegg.ca/Product/Product.aspx?Ite … 1-085-_-Product

TELVM wrote:

The highest ever draw from the wall I've ever seen from my PIII (Tually 1400-S @ 1.6, GF6200 ,768MB of RAM, 1x SSD on Promise s150 TX2+, 1x HDD, 1x DVD, AWE64 Gold, PCI USB 2.0 card, 2x 120mm fans, 2x 92mm fans) is 98W AC.

The PSU is a (restored) Oh Deer with a glorious inefficiency of about 70% at that load, so system real DC draw is 98 * .7 = ~70W DC tops.

Frankly I don't believe that. (The highest ever part)
If that were true a 70 watt PSU would be perfectly fine for that system.
Do YOU actually believe that?

A P4 is easier to properly power than a +5v powered CPU.

I'm not much into electronics, but I keep an eye out for typical power issues such as checking the voltages under load, noticing stability or issues or seeing the machine shut down. I've never experienced this with Pentium III gear.

The only time I saw voltages go to the extreme was when I used a 25A 5V PSU to power an Athlon XP 3200+. The 12V went past 13V and the 5V reading was almost down to 4V. THAT is a sure signal that this isn't going to end well 😊

I think I have a photo somewhere...

EDIT: There you go. Now that is a PSU in distress 🤣

PhilsComputerLab wrote:

I'm not much into electronics, but I keep an eye out for typical power issues such as checking the voltages under load, noticing […]
Show full quote

I'm not much into electronics, but I keep an eye out for typical power issues such as checking the voltages under load, noticing stability or issues or seeing the machine shut down. I've never experienced this with Pentium III gear.

The only time I saw voltages go to the extreme was when I used a 25A 5V PSU to power an Athlon XP 3200+. The 12V went past 13V and the 5V reading was almost down to 4V. THAT is a sure signal that this isn't going to end well 😊

I think I have a photo somewhere...

EDIT: There you go. Now that is a PSU in distress 🤣

That looks real bad 🤣! Time to do some hardware swapping before something blows up or ends in a flash of fire!

It was a case of a quick in and out. Booted Windows XP live from a CD, had Prime and sensor software on a USB and off it went 😀

I then used an older PSU which has a 35V rating on the 5V rail. Still, the voltages did also move quite a bit once the XP 3200+ got crunching.

PhilsComputerLab wrote:

It was a case of a quick in and out. Booted Windows XP live from a CD, had Prime and sensor software on a USB and off it went 😀

Where can I get this XP boot CD? Or can I create my own perhaps from XP install media?

PCBONEZ wrote:

Frankly I don't believe that. (The highest ever from part)

The 98W AC peak draw is on system boot, when the HDD & DVD motors draw extra current while they start spinning.

After booting up the most it ever draws under torture is 93W AC, and it idles at 75W AC from the wall.

There's no need to believe me (any moron can post nonsense on the internet after all), get a kill-a-watt and test for yourself.

If you remember back in the day PIII systems like the HP Vectra VEI8 came OEM with a 90W PSU.

http://www.ebay.es/itm/HP-0950-3646-Vectra-VE … %3D230825914925

Tetrium wrote:

... EDIT: There you go. Now that is a PSU in distress :lol: […]
Show full quote

... EDIT: There you go. Now that is a PSU in distress 🤣

^ Pay no attention, that voltage is not real. You just can't trust HWmonitor (or any other software BTW) for voltage readings.

You need a multimeter for that.

TELVM wrote:

Tetrium wrote:

... EDIT: There you go. Now that is a PSU in distress :lol: […]
Show full quote

... EDIT: There you go. Now that is a PSU in distress 🤣

^ Pay no attention, that voltage is not real. You just can't trust HWmonitor (or any other software BTW) for voltage readings.

You need a multimeter for that.

That's true to some extent.
However, if you were to test 10+ PSU using the same setup on a testbench and most have regular voltages and only 1 or 2 show other readings, then this is a good indication that something is amiss.
But then again I tested using the readings from the BIOS and I agree that using a multimeter is the most reliable way to know (and testing while the system is being taxed is even more better).

PhilsComputerLab wrote:

I'm not much into electronics, but I keep an eye out for typical power issues such as checking the voltages under load, noticing stability or issues or seeing the machine shut down. I've never experienced this with Pentium III gear.

The only time I saw voltages go to the extreme was when I used a 25A 5V PSU to power an Athlon XP 3200+. The 12V went past 13V and the 5V reading was almost down to 4V. THAT is a sure signal that this isn't going to end well 😊

EDIT: There you go. Now that is a PSU in distress 🤣

Seen that one. That will happen with PSU faults but also when +5v is loaded too heavy and +12v too light.

Some PSUs use +5v for the feedback to the PSU controller IC and don't monitor 12v at all.
If +5v is heavily loaded the voltage dips so the controller tries to bump up the voltage.
The problem is these controller chips bump up the +5v and +12v at the same time.
Thus if the controller can't keep up with the +5v load the +12v ends up getting bumped too high.

For me working on PSUs (beyond just caps) is a necessary evil. I don't particularly like working on them or I might actually remember which ICs are problematic. Usually if I can't get one right with just caps I toss it into the recycle box.
.

Tetrium wrote:

PhilsComputerLab wrote:

It was a case of a quick in and out. Booted Windows XP live from a CD, had Prime and sensor software on a USB and off it went 😀

Where can I get this XP boot CD? Or can I create my own perhaps from XP install media?

I did build one myself once with Bart PE or something like that, but there is a much easier way. It's called Hiren's Boot CD. A bit dodgy, but has live XP on it 😀

Single threaded only, so you can't use it to stress multi-core machines. For those I just quickly install Windows 7.

TELVM wrote:

If you remember back in the day PIII systems like the HP Vectra VEI8 came OEM with a 90W PSU.

Yes, with on-board video and sound, one HDD and only 2 RAM slots.
Not even comparable to what the OP is putting together.

TELVM wrote:

get a kill-a-watt and test for yourself.

I own two of them, and I already have.

And how much does your PIII system draw from the wall?

PhilsComputerLab wrote:

What are your thoughts on something like this?

http://www.amazon.co.uk/DIGIFLEX-Power-Supply … k/dp/B004IO5CT8

This one only measures voltage, but maybe there is one that does current?

Haven't seen those. Pretty cool.

One for current could be built. It would basically be a packaged version of the test rig I mentioned earlier.
It would need to be much larger for current though.
The required components to measure that many amps are physically larger.
.

TELVM wrote:

And how much does your PIII system draw from the wall?

Mine (only one currently assembled) is dual CPU and has 15 1TB hard drives so that isn't a valid comparison either.
.

Tetrium wrote:

PhilsComputerLab wrote:

I always found Tualatin processors a pain for coolers, as they are slightly higher and put extra force on the clips.

Very much true. I think I even wrote an article about this specific problem in the Vogons wiki and know of a few ways to mount a HSF without putting as much stress to those tabs as would have if I installed the HSF as-is. I mix and match heatsinks, metal clips (sometimes I bend them a bit) and fans and end up with great reliability.

Excellent solution. Easy. Works great. I do that too.