Skalabala wrote on 2024-03-05, 02:08:
I used the PSU on the P5A-B overclocked t0 660MHz stable. Then I wanted to try the PSU on the MS6260s. I smelled something and it posted and froze in post.
Board never posted again and post card says 5V is missing 🙁
Should have gotten a better PSU since I am trying to have the fastest socket 7 in the world.
Again, you won't see the PSU run if the 5V is missing.
Even the most crude gutless wonders tend to (usually) shut down if a rail goes away (the few I saw that didn't, ended up in a cloud of smoke.)
So I'm more inclined to think that your POST card is likely just not making a proper connection in the PCI slot and thus not seeing the 5V rail.
I've actually ran into this same exact issue with my POST card on numerous machines - sometimes it's the 5V, sometimes it's some other PSU rail, and sometimes it's a stuck RESET... until I wiggle / re-seat the POST card in the slot. Then all of a sudden, all of these "missing" signals / rails were "fixed".
Skalabala wrote on 2024-03-08, 02:42:
I had a look again and the test card says 3.3V is missing 😒 Using another power supply also.
Yeah, probably your POST card being wonky at this point.
Or if this was on a different motherboard, maybe this one has 5V PCI slots only while the other had only 3.3V slots?
shamino wrote on 2024-03-05, 13:22:
As a general policy, I suggest using a multimeter to measure the output of unknown power supplies before you plug parts you care about into them.
This won't show you the ripple, but that's a deeper thing to get into and requires expensive equipment. Checking the nominal voltages is easy and will weed out most dangerous PSUs.
+1
Actually, a MM won't really show you *all* PSUs that are dangerous, but it could at least show you which ones might not be a good fit for the system.
I have several older group-regulated PSUs like this. These either don't like a heavy 5V load or a heavy 12V load, and one of them hates both, making it very hard to pair with any system. What's interesting is that some of these are actually half-decent to decent PSUs. But they just weren't designed for certain loads. The one that hates both 5V and 12V -heavy loads, for example (and older lower-end Sirtech HiPower platform) tends to squeal with a 12V -based system that doesn't put enough load on the 5V rail, and the voltages tend to be all over the place. It has a poorly designed compensation loop (which I will get to someday.) On the other hand, the same PSU is happy with a heavy 5V -based system... but then the 12V rail is sitting at 12.5 to 12.7V most of the time... which is very close to or out of spec. PSU voltages are stable, but I just don't feel comfortable feeding my HDDs and other hardware such a high 12V rail.
shamino wrote on 2024-03-05, 13:22:A PSU needs a load before it will regulate properly, so for that you can plug in some junk hardware. Or you can rig up a dummy load using resistors, but they need to be beefy enough to handle the wattage that will be going through them.
+1 again.
Though instead of using power resistors as a dummy load, I suggest 12V incandescent or halogen bulbs. Should be pretty easy to find them in car/auto parts store or a hardware store that still carries MR16 12V bulbs for specialized lights (typically home lighting that uses a transformer.) A 20W incandescent or halogen will typically pull about 1A on the 5V rail and about 1.67 Amps from the 12V rail... so these are pretty good to use as dummy loads and won't require special cooling like a power resistor might.
shamino wrote on 2024-03-05, 13:22:On boards that have BIOS voltage readings - ignore those and do your own measurements anyway. It's common for the onboard sensors to be inaccurate.
Like +10 for that one. 😁
I also never advise to trust BIOS voltage readings. Some can be accurate, but it's always worth double-checking.
The negative voltage rails are the ones most commonly mis-read by BIOS (not that these are of any importance, really.)
shamino wrote on 2024-03-05, 13:22:
It's a lot more assuring to have some measurements before you decide how much to trust a PSU, even if it's a good brand.
If it's a good brand PSUs and built well inside, I will always trust it.
But again, it just comes down to: is the PSU a good match for the system you're trying to run?
With group-regulated PSUs, it's not always obvious until you do some measurements, like you said.
StriderTR wrote on 2024-03-05, 18:36:
However, when it comes to power supplies, I just don't see the point in taking that risk or having to do a repair when I can use a brand new modern PSU.
The thing is, some older PSUs can be really bullet-proof, especially with new quality Japanese caps (if they didn't come with those already.) They will continue to work many years after most modern (APFC) PSUs have blown their primary caps. While I don't condone APFC, it's a circuit that puts extra strain on the primary cap(s) in the PSU and therefore they will have a much shorter life compared to non-PFC or PPFC PSUs that run their primary caps at line frequency only.
Skalabala wrote on 2024-03-08, 02:42:
I had a look again and the test card says 3.3V is missing 😒 Using another power supply also.
Yeah, probably your POST card being wonky at this point.
Or if this was on a different motherboard, maybe this one has 5V PCI slots only while the other had only 3.3V slots?
PcBytes wrote on 2024-03-04, 20:19:
(btw I'm Dan81 from BCN 😀 )
Hey hey! 😀
Yeah, I figured this I think a few years back when I came across some posts on Vogons (I think it was a rig/PSU that you had posted on BCN too, so I made the connection pretty quickly.) I see some other BCN members here too, which is pretty cool.