My main PC is down, and I’m pretty sure I’ve got either a bad power supply or a bad motherboard. Are there any basic tests I can run with my multi-tester that will rule one or the other out? It’s an HP m7167c, with an ASUS P5LP-LE mobo and I’m pretty sure a relatively standard ATX power supply (300W).
A PSU can spin up its fan and still be bad. To check it slightly more thoroughly put a load onto it - plug a hard drive in, for example - fire it up using the paper clip method and then use the multimeter to check the voltage on the 12V and 5V rails.
I’ve got a multimeter down in my electronics graveyard, but strangely never thought of adding a load to check the rails. :smack: I’m definitely adding this to my standard operating procedures when I suspect a flaky PSU. Thanks a bunch!
One of my “ongoing projects” (I bought the parts, but Lord knows when I’ll build it) is a computer power supply load tester. I mention it here, because if others take it up or make suggestions, I may actually finish it this winter. It seems like almost everyone I know hits a possibly dying PS every year or two – and guess who they come to? There are go/no-go testers for $15 or less, but its hard to trust them when you can’t confirm them, and they offer precious little in the way of specs or docs, even if you call the manufacturer.
Phase I is easy: a series of mobo power sockets scavenged from dead mobos, wired in parallel to variable high current resistors (nichrome wire strung between ceramic posts). Resistance is calibrated by marking various points with measured hot ohms @3.3/5/12V. Hot vs Cold ohms (high initial draw) turned out to be an issue. I stopped development to think of a more elegant way to deal the change in resistance than sliding an alligator clip from an initial “warm up” position to a final “test” position on the nichrome
Test: Measure the voltages with the resistances set to draw, say, 75% of rated current on each rail. I think this alone should detect almost all bad PSs.
Phase II: add some fast high current transistors (or similar) to short out part of the nichrome to simulate transient load jumps/drops of 1-4A (e.g HDD spinups)
Phase III: since I don’t own a multiGHz scope at the moment, build a “trigger box” with a cheap high speed logic chip to latch an LED if hi/lo transients are present. By adjusting the sensitivity, I could compare the transient levels of a “known good” PS to the one under test. I’m not sure how much this would add in detecting normal failure modes, but this might tell me the answer, and it’s a LOT cheaper than a GHz scope
I estimate the total parts cost of ~$20 (I used a lot of salvaged/junkbox parts, but the nichrome and ceramic posts were just a few bucks), so pretty much any hobbyist should be able to build it, using basic safety precautions. The output side is low voltage. A short to line voltage is extremely unlikely, but should immediately blow the PS fuse on testing.
Alas, I’ve decided that discretion is the better part of valor where my main box is concerned, so I’ll be taking mine to a shop this week to let a pro deal with it. It’s getting too close to Christmas for me to spend any more time without access to Quicken.
Try swiping a friends power supply and hooking it up to your box, no need to install it. Would take around ten minutes to verify if it is the power supply or not.