I've a question for electronics engineers

[k-four-nine]

Short version:
What's the 'cheapest' way to measure high frequency (~1MHz) ripple voltage (fluctuating DC) of a switching power supply?

Long version:
My cousin's HP Z400 server is acting weird as of late. And since I'm the 'go-to-geek' of the family, it's my 'moral duty' to fix his problem. Anyhow, after spending 30 minutes on all sort of diagnosis, I'm convinced that it's something to do with the PSU. Naturally, I picked-up my al-cheapo multimeter and tested all 3 voltages, but everything was well within tolerance of ±5%. In fact, the main 12V rails are all nice and dandy; sitting at around 12.07-12.10V which is fantastic. Now, here's the problem:

The PSU feels quite hot to the touch on load, which means the ripple voltage is probably spiking beyond the tolerance of 1% and causing system instability. Problem is that I don't have access to an oscilloscope, or even a high speed multimeter with True-RMS for that matter. I checked AliExpress and good meters cost just under 10,000 which is a tad too much for something I'd only use once or twice a year, tops!

Now this is probably a VERY stupid question but... Is there a way to somehow "transform" the ripple voltage into DC current or at least reduce ripple frequency from ~1MHz to 50-60Hz so my cheap as chips multimeter could read it? I won't mind soldering together some capacitors and stuff on perfboards as I've no plans for this weekend!

Thanks in advance!

[Hassan1991]

First there is no Frequency in DC voltages.
And you need oscilloscope without that you can say anything about ripples. Just in case if there are then voltages shouldn't be accurate and over heating causes resistance to increase which disturb values. There are cheap small oscilloscopes available cost like 2k.

[k-four-nine]

First there is no Frequency in DC voltages.

Sure DC has no frequency 'cause it's direct but that's not what I meant. I'm trying to measure the unfiltered AC ripple (usually in few dozen millivolts) that's present on the output of all AC>DC converters and power supplies.

[Hassan1991]

Search for the oscilloscopes i told you.
You're talking about adding a filter. Better to repair as it's already made to serve the purpose.

[shafqat35]

Measuring (without scope) or transforming ripple is not easy. If you suspect it is a ripple, that is creating problems, then try to bypass it with a parallel cap. Remember the cap can make the PSU unstable. on a side note I dont think it is a ripple. Check if heat dissipation is working correctly ie fans etc

[capsat]

HP Z400

it is a work station. in Karachi at Regal Computer market you can get a fresh supply for under Rs. 2000. In lahore Hafeez centre is the right place.

before that remove the power supply from machine.

open power supply cover blow clean the unit check the cooling fan free run. TEST. If the CAPCONs are not bubbled out due to stuck up cooling fan, problem will be resolved.

Otherwise get a fresh clean supply.

[k-four-nine]

Yes, fan is spinning just fine. In fact, I've lubed its bearing with some motorcycle engine oil to make sure it spin at full RPMs. But the PSU still feels warm, even in winter.

I think a capacitor is blown up somewhere inside which is causing the extra ripple but the crazy thing is voltages. They remain stable and don't sag below the recommended 5% tolerance even when I fully stress the CPU and GPU. The PSU just shuts-off by itself after 15-20 minutes, probably due to overheating as a safety measure.

lahore Hafeez centre is the right place.

I went to Hafeez Center but couldn't find anyone selling Z400 workstations, or their PSUs for that matter. Asked an electronics shop there to fix the PSU and they asked 2,500 for the repair with no guarantee!

Otherwise get a fresh clean supply.

And that's another major problem. The PSU on the Z400 is proprietary and has a different, non-ATX pin layout for the 24-pin mobo connector. Sure, adapters exist but they go for $25 + shipping on ebay. Couldn't find anything on AliExpress.

[capsat]

you may ask some friend of yours in Karachi to spare some time and visit Naz Plaza or Regal Computer market such HP stations are dumped in these markets. you may share the PSU photo with him. All taken care of including fuel & time compensation to your contact it will be under 4000 tested removed from active unit. ( shipping handling included)

[Ijaz_Ahmed]

1) Put the meter to measure AC Voltage and check the AC content of each O/P to see if there is ripple present.

2) Add a big electrolytic capacitor to each suspected output (parallel connection)

3) Try adding a series inductor to the suspected output, make sure inductor wire is sufficient to handle the current.

Lastly, you can visit me if you are in Islamabad. Lots of oscilloscopes present

Can you show few clear pictures of the power supply ?

[k-four-nine]

Lastly, you can visit me if you are in Islamabad. Lots of oscilloscopes present

Can you show few clear pictures of the power supply ?

Thanks a lot for the offer but I'm from Lahore, unfortunately!

Here's how it looks like:

And here's the voltage distribution chart of a typical ATX standard PSU:

12V @ yellow, 5V @ red and 3.3V at orange. I'm only interested in the 12V rail as CPU and GPU run almost exclusively on 12V. 5V and 3.3V don't matter all that much.

Anyhow, I was thinking about adding a ~200W dummy load at two molex connectors (each Molex is good for around 9A @ 12V) and then testing the voltage at the 6-pin PCI Express connector, which goes into the graphics card.

At first I was considering two 100W 1.5Ω power resistors but now I'm thinking about buying a pair of cheap 12V 100W halogen lamps used in cars, as I don't want to build a heatsink for the resistors! As a bonus, 12V bulbs are MUCH cheaper and not as dangerous as resistors!

So with this in mind, how do you think I should proceed? Any advice would be greatly appreciated!

Again, thanks a lot, brother!

[Ijaz_Ahmed]

Unfortunately a low resolution picture, so I can't read model number.

CPU runs at 3.3V, GPUs need lots of 12V power.

Multiple output PSU don't regulate every output. Instead, the output with highest A current is ususally regulated. Some PUSs use a cross regulation scheme. So I think if you load 12V with 200W, and leave 5V open then the PSU may not be able to maintain regulation and you wouldn't get the desired results.

What problems exactly you are having on the server ? If you have added the GPU later, does the PSU meets the GPU requirements ?

[capsat]

take this along pay visit to Hafeez Centre. if you can not find it there

visit hall road ask for spectrum electronics the shop is inside the market guy name is saqlain he is known champion to repair audio amplifiers he can fix this supply for you.