Spikes in PSU are BAD Right?

[VERTIGGO]

I'm replacing all my rig's guts at the moment, because I've had data corruption for 4 months and several drives have de-formatted themselves during normal use (my c drive is "uninitialized" as I write; can't reboot). I've contemplated replacing the PSU as well but I'm not sure what is the best way to be confident that it's working.

Regardless, I have been running SpeedFan and SensorsView, and noticed a few spikes on the 12V, which if I'm interpreting this correctly is approximately 13.5V or 12.5% (BAD):

Would you say based on this graph that the PSU is dangerous to my hardware? I have a new PSU in the mail and I have to decide which one to use.

 

[Zap]

You cannot trust software readings. Use a digital multimeter.

 

[PreferLinux]

Multimeter wouldn't pick up the spikes. You'd need an oscilloscope to do that.

 

[VERTIGGO]

My father is actually working on an oscilloscope/logging apparatus (as his hobby, independent of my issues if you can believe), so I might be in some real luck.

 

[Modelworks]

Multimeter wouldn't pick up the spikes. You'd need an oscilloscope to do that.

You just need a decent meter. Many, mine included have a max and min setting. They save the highest and lowest values for the readings until reset.
Don't cost much either, about $75 for one that can do that.

 

[BTRY B 529th FA BN]

Interesting :hmm:

 

[WhoBeDaPlaya]

You just need a decent meter. Many, mine included have a max and min setting. They save the highest and lowest values for the readings until reset.
Don't cost much either, about $75 for one that can do that.

Fluke?

 

[Modelworks]

Fluke?

Not fluke , though you can get them from fluke if you want to pay the price. This is similar to the one I use. One i use is no longer sold.
http://mcmelectronics.com/product/TRIPLETT-9045-/74-340

 

[VERTIGGO]

Yeah I've been using a Fluke 87 III on just a mobo, cpu and random stick of 2GB, but without load it hasn't fluctuated more than 12-12.34V... it's actually frustrating that it won't fail, because now I have no idea what caused the data corruption, and everything from the mobo, cpu and SATA cables are still suspect. I am using a new sandy bridge setup so I'll see in a while if the issues arise again, and I'll know to rule out the PSU mobo and cpu, but as always, you never really know for sure when your system is working, only when it isn't.

 

[bryanl]

If the graph truly represents 53 minutes of data, then each spike is several seconds wide, or far too long in duration to let the computer continue operating and recording voltages. A computer typically can't operate if a dropout lasting more than 1 AC cycle, or 1/60 second (16ms), occurs.

Most multimeters take no more than about 100 measurements per second and update only their bar graphs at that rate. The digits are more likely to be updated only a few times a second, but even the higher rate is too slow for analyzing spikes.