Repairing R3131 Advantest Power Supply / Tracking Generator

[rommac100]

So I noticed that my Advantage R3131 Tracking generator was having its output gradually sag (-40dBm to -80 dBm) and decided to tear down the Spectrum Analyzer.
I noticed that the 12.6V rail on the Power supply (AT120-5224/M Nemic Lambda) had a low output voltage (~10V) with no load except the 120mA 12V fan, so something is definitely wrong with the 12V rail. I removed the 12V fan and it is now sitting at around 12.6V with no load but still drops greatly when a small load is applied (used an E-Load and swept between 10mA and 100mA).

Anyone have suggestions for troubleshooting the 12V rail / have access to the schematic for the above power supply?

[fenugrec]

(AT120-5224/M Nemic Lambda)

That sounds like a nearly off-the-shelf part. While you may not find much info about that particular PSU online, repairing smps is "run of the mill" for the members of this forum and I'm sure with a few clear photos it can be troubleshooted without schems.

Statistically the most frequent cause of SMPS failure is bad capacitors. In your case probably on the secondary side (are the other rails properly regulated under load ?).
Another possibility,  sometimes they have a linear post-regulator for some rails e.g. 7812 that can go bad.

[rommac100]

Fair.  See Pictures of the unit (haven't completely disassembled it since I did some load testing of the other rails). Superficially I didn't see any leaky caps but I haven't bother to measure / remove any yet.


5V rail load tested:
0A = 5.24V
.1A = 5.24V
.5A = 5.24V
.7A = 5.24V


12v rail load tested:
0A = 12.65V
.05A = 11.89V
.1A = 8.47V

-12V rail load tested:
0A = -12.64
.05A = 12.1A
.1A = 8.62A

24V rail load tested:
0A = 24.09V
.02 = 23.25V
.05 = 14.37V

I also measured the voltage of other rails while a particular rail was being loaded and they didn't change. Not sure if I need to load another rail in order for it to function properly given that the 24V rail, -12V and 12V all drop a lot when under an even small CC load (entirely possible I am missing understanding the situation).

[andersthuresson]

I would recommend you to recap this PSU due to the age of this unit. It might even solve your issue. Use a scope and check for ripple during loading condition. Some times electrolytic caps leak from the bottom and it might not even be visible until you remove the cap. Get good electorlytic caps and make sure they are rated for temp +105C or better.

I should perhaps mention that you better use a differential probe for your oscilloscope or perhaps a battery powered scope if you have. Regular scopes are ground referenced and crazy stuff can happens if you put your regular passive probe on the wrong point.

[fenugrec]

Some times electrolytic caps leak from the bottom and it might not even be visible until you remove the cap.

+1 on that.

Hard to read on your pics, but if as I suspect those output caps are LXF series, they are replace-on-sight (ref: https://www.badcaps.net/forum/general-topics/general-capacitor-questions-issues/455-the-good-capacitors-thread/page6?t=414&postcount=105#post533421 ), LXF and TXF are my top 2 causes of PCB damage. Don't even bother testing them, just remove and inspect for damage. Replace with quality high-temp, high-hour parts (modern parts of similar size should have at least similar ripple ratings to these 20+ year old caps).

IME the bulk high voltage caps are usually not a problem. I don't suspect any issue on the primary side on your PSU there since you have good regulation on at least some rails.

[ARF]

The power supply of this spectrum analyzer is common group stabilized. Please note on the label that the minimum load on the 5V rail is 2A. If you disconnect the load from 5V, then the 12V rail will have wrong voltage.
This power supply is working correctly.
R3131 spectrum analyzer has a number of known problems with input sensitivity degradation. A correct solution to the problem was never found, but a workaround was invented. You can find information in the large thread R3131.