Measuring PSU's

[Frick]

I borrowed a power analyzer from work to see if the PSU (CM G550M) was behind the annoying behaviour (random freezes when the sound "freezes" as well, flash videos crashing with the monitor splashed in colours). I measured the 5V line and got a nice scare until I realized I had the ground probe on the CPU connector and the other probe on the end of a molex chain.

So, I am assuming the proper way to test things is with both probes on the same connector, because that is what that particular component will get. I haven't probed the ATX connector, because frankly I'm not sure how I'd do that.

When monitored on the same connectors all voltages seems ok and within spec. When monitored on the PSU directly they're even better, at least on one connector. I haven't measured the other connectors.

What do I do next? I have measured the CPU and PCIe connectors while under load as well, nothing really happened.

 

[Frick]

Found something else. Might be time to contact the landlord.

That is between the USB shield and an unpainted part of a radiator.

 

[m4gicfour]

106V? Very nice.

You shouldn't let your grounds wear floaties. It seems mean but they're not generally meant to swim.

 

[Swamp Monster]

In the first picture, ripple voltage is about 1.5 to 0.5V, judging by the trace. That is very high, should be below 0.05V on 5V rail.
Unless you have some kind of probe attenuator set to 0x on the probe and 10x on the scope.

Nevermind, in the second picture trace reading does not correspond to V/div, so ripple is not really that high.

 

[OneMoar]

a floating ground creates mud ... .

 

[Frick]

In the first picture, ripple voltage is about 1.5 to 0.5V, judging by the trace. That is very high, should be below 0.05V on 5V rail.
Unless you have some kind of probe attenuator set to 0x on the probe and 10x on the scope.

Nevermind, in the second picture trace reading does not correspond to V/div, so ripple is not really that high.

Again, the first pic is with ground probe on the CPU connector, the other probe on the end of a molex chain. Naturally it'll look bad. The second pic tells me I have a phase shorted or some doofus electrician has done goofed up in the electrical box. Measured between a water bearing radiator and USB ground. Hence the "contact the landlord" bit.

(i could fix it meself unless the problem is a short somewhere, but they wouldn't be happy about it)

 

[OneMoar]

Again, the first pic is with ground probe on the CPU connector, the other probe on the end of a molex chain. Naturally it'll look bad. The second pic tells me I have a phase shorted or some doofus electrician has done goofed up in the electrical box. Measured between a water bearing radiator and USB ground.

sounds like you have a neutral shorted to ground ... and someone writed the plug backwards

 

[Frick]

sounds like you have a neutral shorted to ground ... and someone writed the plug backwards

Aye exactly.

And the computer just froze again, sigh. Only Flash player running. Blergh, I'm getting another motherboard next week, I hope that's the culprit. It's very intermittent, I hate those problems.

 

[OneMoar]

Aye exactly.

And the computer just froze again, sigh. Only Flash player running. Blergh, I'm getting another motherboard next week, I hope that's the culprit. It's very intermittent, I hate those problems.

the ground being like that will certainly cause issues if not fry the board no sense replacing anything until your AC wiring gets sorted

 

[Frick]

the ground being like that will certainly cause issues if not fry the board no sense replacing anything until your AC wiring gets sorted

There's no ground in that outlet. The kitchen has some grounded outlets, but otherwise there's none.

 

[R-T-B]

There's no ground in that outlet. The kitchen has some grounded outlets, but otherwise there's none.

Old wiring sounds like. All the same, I'm pretty sure that's not good. Maybe buy one of those 2 prong to 3 prongs with a ground hookup and ground it to something yourself?

 

[Frick]

I live in a single room apartment on the second floor, not much to hook it up to (especially not if there's a problem).

New stuff. These measurements were done at the CPU connector. When loading the CPU with prime95 and the GPU with Furmark (during which the driver restarted btw) the difference between min/max on the 12V line goes up to 0.65 volts. And because it can only display 3 digits (the Fluke 43B isn't really made specifically for these things ), I have seen it max out at 12.6 volts. I assume some sort of OVP would trip if it exceeds that? Generally it hovers in the 11.7 an upwards area though.

Other than that the voltages seems stable enough, unless they have spikes the scope can't pickup. I have no idea under what circumstances that would occur. I have no idea how to test ripple with it.

Edit: good grief they can set the OVP from 13.4 to 15.6 V! How is that useful?

Edit: Launched WoW and now it varies from 11.7 - 12.5. Within spec but I really wish I had a fourth digit.

Edit: ... What does it mean if I set it to 1ms/d it settles about 0.6V min/max peak, and if I set it to like 500ms/d it shoots up to 0.8V? I have no idea what I'm doing anymore.

 

[Swamp Monster]

1ms/d

1ms/d means if signal waveform's period on screen fits in one division, then frequency of the signal is 1000Hz. 500ms/div means you are seeing signal at hertz range. It is certainly a wrong range to measure PSU ripple.
You need to set it to about 5us range to see classical psu's switching frequency ripple of about 200 kHz. Maybe it is 400kHz nowadays. and set v/div to 20mv/div - , AC coupling mode if it is not already active.
Post a pic if you capture a nice ripple!

 

[Frick]

1ms/d means if signal waveform's period on screen fits in one division, then frequency of the signal is 1000Hz. 500ms/div means you are seeing signal at hertz range. It is certainly a wrong range to measure PSU ripple.
You need to set it to about 5us range to see classical psu's switching frequency ripple of about 200 kHz. Maybe it is 400kHz nowadays. and set v/div to 20mv/div - , AC coupling mode if it is not already active.
Post a pic if you capture a nice ripple!

It's not exactly meant for this, so I had to have it at 2V/d. I did have it on 5us (or thereabouts anyway) for a period and there it behaved very, very well. The min/max difference was about ... up to 0.25V iirc.

I have to thing about the range thing.