Rohde & Schwarz CMU200 repair (analyzer levels)

[Velund]

Hello,

I currently trying to fix Rohde & Schwarz CMU200 purchased as a non-op unit from eBay. Most of problems was solved, a lot of black dust was cleaned from everywhere, alignments (Vref and reference frequency) was made, but now I come to a thing that I can't understand (at least at once).

Unit have 2 RF generators, and both work fine, I can see right levels and frequencies on another (calibrated) spectrum analyzer. Signals (both from external and internal RF generators), connected to any of CMU inputs correctly measured by wideband power meter in CMU (0.1-0.2 dB difference, comparable to expected losses in test cable). But CMU analyzers (both spectrum  analyzer and power vs time analyzer) show lower value. At start difference was about 4 dB, it slowly decreased and now "stabilized" at about 2.3 dB @ 100 MHz after running continuously for a few days. Vary between 2-3 dB across whole band.

All self tests in maint mode shows similar difference (about  2-3 dB), both with internal and external loopback.  Any combination of inputs and outputs (if internal RF gen is used) give same result, so it is not a frontend problem.

Unit was reset to factory defaults, to exclude possibility that user attenuation data is loaded. Nothing changed.

Maybe someone can advise where to look next?

[Velund]

Checked again - using generator of CMU and very short test cable (U-like piece of semi-rigid with N connectors). On a dozen of random frequencues from 30 to 2500 MHz deviation was between 2.0 and 2.3 dB. 

I can play in "add 2 dB game" and use it as it is, but would like to find a root cause of this problem.

[Rasz]

At start difference was about 4 dB, it slowly decreased and now "stabilized" at about 2.3 dB

self healing = electrolytic caps somewhere?

[capt bullshot]

It's been mentioned somewhere here:
There's an option to re-calculate all the compensation tables inside the unit, e.g. required after swapping modules. It's somewhere in the starting menu (firmware manager).
For my unit, running this once helped to pass the external loop tests.

[Velund]

There's an option to re-calculate all the compensation tables inside the unit, e.g. required after swapping modules.

It was done many times, both before and after adjusting voltage reference. Does not change anything.

BTW: I looking at RF gen and can't believe that there is no external modulation possibilities at all. Is it for real, or I just look in a wrong places?

[capt bullshot]

AFAIK there's no possibilit to ext. modulate the RF gen. At least not with the standard analyzer / generator mode.

[Velund]

Today tried to use CMU for something useful, to work in 2.3 GHz range, where Rigol is useless. With success, of course (keeping in mind 2+ dB deviation of spectrum analyzer). CMU was on full day, at the end I tried to measure -10 dBm from external sig. gen at 100 MHz with same RBW and span as before - and found that deviation is 2.9 dB, likely it start to drift to higher side now.

[Wilts01]

Did you manage to find out the cause of the problem.  My CMU200 has a similar fault.  What I have found is that
a)  Output levels regards of being RF1, RF2 or RF3 seem ok.
b)  If I use the Wideband Power Mode (i.e. at the front end of the analyser), that is OK.
c)  If I use the "low distortion mode" then that is OK as well.  However, if I use "Normal" or "Low Noise", then the problem occurs.

I have also found that the problem can be level/frequency dependent - when I use FreRes to make a sweep, the problem goes at the higher power levels.

I was wondering whether the problem is likely to be on the RXTX board or on the Front End board.  I think it is more likely to be on the RXTX board but would value other's advice.

[Velund]

Did you manage to find out the cause of the problem.  My CMU200 has a similar fault. 

If question is to me, then answer is negative. I still not managed to find a root cause. And, to be precise, fault is similar but not exactly similar. Mine shows lower level in any mode, no difference between "low distortion", "Normal" and "Low Noise". And delta changes with frequency or amplitude not so much. Changing RBW also not cause severe changes.

With your symptoms I'd look first at switches, while with mine I'd look at analyzer IF amp. But, for now, I need CMU as 2+ GHz capable spectrum analyzer for project and not yet ready to take it apart for a prolonged time.

[Wilts01]

Velund
Thanks for the reply. 

It sounds that are faults are not related then.  The other issue with mine is that it fails the internal loop back tests for RF1 and RF2 at lower signal levels for some frequency bins.  Similarly for external RF loop back tests.

When you say switches, which ones do you mean?

Thanks

Paul

[Velund]

When you say switches, which ones do you mean?

Look in following thread, maybe you'll get more ideas.

https://www.eevblog.com/forum/repair/rohde-and-schwartz-cmu200-rxtx-module-issues/50/

I mean that it may be problem in plugin in DIG/DSP board, that (probably) controls some switches in RXTX.

[Wilts01]

Update.  After some more tests, I came to the conclusion that the problem was probably with the RF FrontEnd module so I bought a replacement via eBay.  Fortunately my theory was correct.  It is now much better.

The only downside was that the new RF FrontEnd module has the 2nd RF Generator attached so now limited to 2W in max.  But at least it is working.

[Velund]

Just for information - described problem was finally tracked to TXRX unit. Replaced whole TXRX unit with the one off eBay (from parted CRTU-RU) - analyzer now fine, within 0.2-0.3 dB compared to my calibrated HP8920A within its freq. range (to 1 GHz). Not yet tested at a higher freqs. But now have 0.7 - 1.7 dB generator level error. Fortunately, not drifting, 0.7 dB is around 50 MHz, while 1.7 dB is on top of CMU frequency range, looks like increasing more or less proportional with frequency.

Update - checked generator output on RF3 out connector using Narda 10 MHz-12 GHz thermocouple power meter head and found a "walley" with 2.6+ dB difference around 1900 MHz. Also, external 1-4 loopback in self-tests have few errors near this frequency.

[Bicurico]

Hi,

I have a kind of similar problem as described here:

My CRTU is working OK, except that RF4 IN will show a -20dBm too low signal. The trace looks exactly as if connected to RF1 or RF2, but the scale is wrong by those -20dBm.

I am pretty sure that the issue is within the RF Frontend module. Has anyone opened such module? Any ideas what to look for?

As I understand it, this RF Frontend has the function of switching between the 4 RF connectors and attenuating the signal according to the input port specs.

Regards,
Vitor

[capt bullshot]

RF4 in is the "high sensitive" input, RF front end module should be the same as CMU200. There's an MMIC amplifier near the N connector inside the RF front end module - this might be the culprit. There's at least one thread here, maybe two about replacing this MMIC (I also did on my CMU200). It's an chip marked "C4", type SCA-14 by Sirenza, former Stanford Microdevices. Replacing this chip may result in degraded calibration of the RF front end, but worked for me.

Edit: Found the thread:
https://www.eevblog.com/forum/testgear/sca-14-(c4)-mmic-amplifier-replacement-for-cmu200/msg1066693/#msg1066693
Including some bad news: to access the RF4 related MMIC you'll have to remove the PCB from its housing, since it's located on the bottom side.

Here's some pictures I've taken during the repair:
http://wunderkis.de/cmu200/rawpic/

[Bicurico]

Cool! Thanks a lot.

I will look into this, but probably not this year...

Just went through your pictures: they help a lot. I understand that probably the C4 on the oposite side of the PCB is broken on my RF frontend, as this seems to be the one responsible for the R4 IN connector. That makes sense... Going to search for a replacement IC.

Regards,
Vitor

[Bicurico]

What do you mean with "Replacing this chip may result in degraded calibration of the RF front end, but worked for me."?

Would running the calibration in diagnostic mode fix this?

As is, the values are wrong by -20dBm, so it couldn't get much worse, anyway, I guess...

Regards,
Vitor

[capt bullshot]

Afaik, all the modules of the CMU200 have some kind of calibration tables stored in NV memory on board. The internal calibration run calculates correction tables for the RF analyzer and generator paths from that. Each module gets its own cal table stored during a special module calibration procedure in the factory. If one replaces a component of the RF path, the corresponding table would have to be re-created. I don't think an individual like you and me has the knowledge and equipment to do so, so we'll have to live with some degradation after replacing a component like the MMIC.

Running the external loopback selftest indicates a few dB off for my unit, that's what I said I'm OK with that.

[Ct1jrz]

I was looking for the pictures from the front end board for a long time, since mine had a problem.
I found some burned resistors. But since I don´t have the schematic I need the pictures. I´m very happy to find this ones.
I´m going to try to see if i can repair it.

Thanks

[Ct1jrz]

Not the same board 
Mine it´s different!
Some resistores are the same. Same are different!
There is two small ones that can´t see the values on your pictures.
Can some one help me on that?
Here is the fotos from mine.

[ernemo]

Hello everyone. I had to open my CMU200 for a problem on the rf3 output, the MMIC SBW-5089 burned out, and I took some photos of the RF-FRONTEND board. I hope they will be useful !!

[HF-Bastler]

No, it's possible. It works with AMPS Analyzer/Generator but only in FM. You need CMU-K29 and B41

AFAIK there's no possibilit to ext. modulate the RF gen. At least not with the standard analyzer / generator mode.