Rohde & Schwarz CMU200

[_Wim_]

(but if this is the way the user is supposed to work, it would have been better of R&S to prevent a crash if the user did something different).

Yes, the software could certainly be much more robustly implemented and be more intuitive in use! I also believed for a very long period the normalization was not working.

• Using ‘Level sweep’ (great for detecting linearity issues!), the program sometimes performs the measurement as expected; at other times, it does not execute the expected sweeps.

When getting "min level error", this is sometimes caused by the user specifying a level outside of the limits of the selected output. But apart from that, I also have had many crashes I could not explain.

• “Power Measurement Error” appearing at random moments. Repeating the exact same measurement later usually results in no errors. (It might be the CMU itself, but this is hard to tell)

My CMU does this to. Not sure what is causing this.



With good remote software the CMU could indeed be a even more useful, but the amount of effort is quite big. I once started writing my own software, but gave up after realizing how much work it would be to achieve the features I wanted.

[Bicurico]

As the author of the "VMA Simple Spectrum Analyser Software" I would like to thank Rudi for his review.

The particular version of my software for the CMU200/CRTU has never been popular, only few people use it. I want to believe that the reasons for that are:

1) There are not that many owners of a CMU200 or CRTU
2) The ones owning such a device, don't have a Keysight USB GPIB interface (note: I have a clone from China, which cost under 100 Euro and works without any problem)
3) People don't know about my software

Anyway, the absence of many users has refrained me from correcting bugs and continuing the development beyond my personal needs.

Hopefully I can iron out some of the issues Rudi mentioned and implement Wim's tip about the Regional Settings - boy has this caused me trouble in the past and i did not know there was this simple fix. This should be presented by Visual Studio as a message box on each new project!

And yes, I do believe that any old spectrum analyser capable of outputting the swept samples in a fast way (GPIB or network) can get a new life by means of a PC software doing all the data processing. This is what I did with my software - it was not designed as a remote GUI for the device and it's particular functions but rather a new implementation of a spectrum analyser that only communicates with the hardware spectrum analyser to set fundamental parameters like START/STOP frequencies and then receives the samples.

Ideally such a software should be universal and use specific DLL's to deal with the hardware communication. I failed to build the software like this because I never expected to grow it to this state. It might be simple and a professional programmer might replicate this in a better way in little time, but it took me several years (about 7 years, if I am not mistaken) to get to the current state, where I have versions for the R&S CMU200/CRTU, HP/Agilent 859x Series, Siglent SSA/SVA, SMA/NWT/D6/LTDZ, TinySA/TinySA Ultra, Anrist SSA.

A final note about the R&S CMU200/CRTU: they are fine test equipment that offer a great spectrum analyser. However, they are still sold at a price that defies the purchase of a brand new Siglent SSA3021-P, which can be crossgraded to a SVA1032. Compared to that, the CMU200/CRTU offer a better resolution and higher sweep rates (IMHO), but they lack measurement functionality. The main concern, however, is that they suffer from several issues, which need to be named. First off, many might have a broken or out of spec RF frontend. As a consequence, the specific selftest will fail and measurements will be out of spec. This is the RF input board. But the RX/TX boards can be faulty, too. Finally the RX/TX boards come with 1x (CMU200) or 2x (CRTU) Digital Boards. These are small daughter boards. When they fail, you won't be able to repair them and they are really hard to find. Additional problems come from the PSU and the PC board. I own 5x such devices. 4 are fully working, albeit the selftest does not pass 100% on all of them. One is not fully working (bad RF input board), but usable as a SA. Then I have a case with just the main board, that has some bent pins from the previous owner. Was purchased to remove needed parts for the other 4. This is to say that I started with two broken ones to rebuild one and ended with all of these...

Kind regards,
Vitor

[RBBVNL9]

_Wim_

With good remote software the CMU could indeed be a even more useful, but the amount of effort is quite big. I once started writing my own software, but gave up after realizing how much work it would be to achieve the features I wanted.

I'm not a programmer myself, but I certainly believe writing such code, debugging, making sure it works in different environments, and maintain it over time is a significant effort. Especially for a single person doing that in his/her own time. Also for that reason, I applaud Vitor's VMA program, which is useful but also very stable.

Concerning FreRes: I'd like to believe (expect?) that a leading M&E equipment provider as R&S does have the knowledge, professionals and resources to write good software. (This program was squarely aimed at its core audience, and already in version 5.30...) 

As for using the normalisation function: both my CMU and my cables are fairly flat, and to make sure the normalisation was doing what I would expect, I did a little joke: I took a bandpass filter, normalized on that, and then did a measurement with input and output simply connected. That way, I should get the _inverse_ of the bandpass filter.

So I got what I asked for ;-)

[_Wim_]

Also for that reason, I applaud Vitor's VMA program, which is useful but also very stable.

Fully agree!

I do think one of the reasons it was not more adopted is the fact some payment was needed. We live an age where people (incorrectly) think software should be free, because they are used to that when installing apps on their phone. However, app developers typically have others ways of making revenue (typically selling your data to others), which is not viable for software like the VMA program.

This however has totally killed the market for selling software.

Concerning FreRes: I'd like to believe (expect?) that a leading M&E equipment provider as R&S does have the knowledge, professionals and resources to write good software. (This program was squarely aimed at its core audience, and already in version 5.30...) 

Fully agree here also. I guess because they offered it for free, they did not care as much as they do for their payed software. Maybe this was written & updated by thesis students? 

[RBBVNL9]

I’m having a conversation with someone whether the CMU200 actually uses FFT for its Spectrum Analyzer functionality. If it uses FFT in a ‘hybrid’ mode you would expect that its LO would jump in discrete amounts, where the jump length would be around the FFT span used at each step. Now, the relevant boards in the CMU (RXTX in particular) are very well sealed and it may be hard to get a probe there that could test that.

However, looking at the block diagram (service manual, page 3.2 , can be found here https://gotroot.ca/cmu200/CMU_Service.pdf) I see that the RXTX board has three IF frequencies (IF1 = 843.085MHz /1816.115 MHz, IF2 = 486.515MHz and IF3 = 10.7MHz), but ALSO features a bypass switch via which the entire front-end frequency band (e.g., 10MHz to 2.7 GHz) is connected straight away to the A/D converter!

Any thoughts?

[_Wim_]

Hi RBBVNL9,

Just watched your video about adding the B17 IQ IF option. As my unit has a working B17 option (it was installed when I bought the unit), I had a quick check in the setup menu. As you can see from the attached pictures, you should see a part number and modification index, so something is not correct. I would try to reinstall a firmware version with the B17 option installed.

I also attached an old pictures I had of the internal RF wiring. My unit is not loaded with as much modules as yours, but at least it gives you an idea (but the issues you are experiencing are most probably not related to the RF cables)

Hope you can get it to work, if you need additional info, please ask. 

[_Wim_]

Before doing a full firmware update, have you done this?

[RBBVNL9]

Dear Wim,

Thanks a lot! Since I finished the video session, I made considerable progress, and now I know that the problems I was experiencing were caused by at least three (!) different causes. A teaser: my problem was partially caused by wrong cabling, but I will later explain why that happened in the first place ;-) I am getting closer to resolving all and will make a video trying to summarize it in a good way so other can benefit from it.

The photo you sent on how your card shows up in the Setup menu is very interesting, however. Even after a 'firmware upgrade after board change', I still have an empty space in the field where you do have information. One possible explanation might be that I, in fact, installed the CRTU-B7 unit instead of a CMU-B17. I was not aware of this at that time, as the identification stickers were not present on my board (it seems such identification stickers are there only on units that were fitted later in time, not in units that were installed in the machine at the time of purchase)... 

Anyway, I hope to follow up soon, but thanks a lot for sharing!

[RBBVNL9]

Dear Wim,

As you can see from the attached pictures, you should see a part number and modification index,

One more follow-up... Your Setup screen reports a part number "1135.6290.03" for the IFIQ card.   

I got a bit confused there, as I remembered seeing different numbers for the CMU-B17 and the CRTU-B7 units. I quickly looked through the documentation I have here and found that for both boards, there are multiple numbers associated with them (see attached overview). R&S equipment is not only complex in terms of technology but also sometimes in terms of administration

[_Wim_]

R&S equipment is not only complex in terms of technology but also sometimes in terms of administration

That is quite surprising seeing so many different numbers for what is essentially the same thing! Quite an interesting find!

[RBBVNL9]

All,

Getting the IQIF card working in my CMU200 turned out to be no less than a quest. Basically, three things went initially wrong:
1: Errors in R&S documentation, leading me to faulty wiring,
2: CRTU-B7 is NOT equal to CMU-B17,
3: Need for “firmware update after board change”. 

I also learned quite some other things that may be relevant for CMU users, like that in specific circumstances, the instrument simply swapped the I and Q lines (!)

But now I got all up and running, and made a video documenting it.

https://youtu.be/VUljs46ViIk

[F4IUJ]

Hello,

I got myself a CMU200 for not much money. I use it mainly as a generator and a spectrum analyser. Unfortunately, I had a bit of an accident as I was using it as a signal generator to calibrate a small qrp rig VFO and I sent by error a 37dBm pulse into the front end of the CMU200 RF3 which is limited at 13dBm in my case 

I suppose that will teach me to put an attenuator even I don't think that I need one! But beyond this, is there a chance that I can repair the front end ?

Kind regards

Yannig

PS : thanks Rudy for the videos, I watch them!

[Bicurico]

If you search this forum you will find posts regarding the repair.

It involves opening up the front end module and replacing some IC's - can't remember which ones.  There are a total of 4, one for each input port.

Definitely doable, but quite some work and not totally noob-proof. For one, you are dealing with SMD components, then you need to desolder one component containing Berlium (if I recall correctly). This is why you have a safety hazard sticker on the front end module (or should have).

The white powder inside the module should just be aluminium oxide - not that dangerous. Try to vacuum clean it, but do not (as I did) blow on it...

The other option is try to buy a front-end module or a broken CMU200/CRTU. Chances are, though, that the purchased replacement is broken, also.

[Bicurico]

This is the thread I was referring to: https://www.eevblog.com/forum/repair/rohde-schwarz-cmu200-repair-(analyzer-levels)/

Check this link of one of the posts: http://wunderkis.de/cmu200/rawpic/

It is the "C4" component you need to replace.

The RFP 50-17 AMZ component is the one you need to dessolder to get to the back of the PCB. This is the one containing berryllium: https://nj.gov/health/eoh/rtkweb/documents/fs/0222.pdf

[F4IUJ]

This is the thread I was referring to: https://www.eevblog.com/forum/repair/rohde-schwarz-cmu200-repair-(analyzer-levels)/

Check this link of one of the posts: http://wunderkis.de/cmu200/rawpic/

It is the "C4" component you need to replace.

The RFP 50-17 AMZ component is the one you need to dessolder to get to the back of the PCB. This is the one containing berryllium: https://nj.gov/health/eoh/rtkweb/documents/fs/0222.pdf

Thank you, that is really helpful. There is one thing that puzzles me, if I see properly there is just one "C4" component but I seem to have damaged just one channel. Shouldn't it be a part specific to RF3 in my case ?

[Bicurico]

I might be wrong (long time since I looked at this), but there should be 4 such C4 components, one for each input. Two on each side of the PCB, which is why you need to dessolder the component I mentioned before.

[switchabl]

IIRC there are only two. The one for RF3 is on the top side. The one on the bottom is for RF4 and you need to desolder the attenuator and the connectors to get to that one. RF1 and RF2 don't have dedicated amplifiers.

[F4IUJ]

OK, it makes sense now...

[luc-on7kz]

Hello,
at the hamradio event I have buy a CM300, I'm not the specialist from these equipment's and has not tested completely before I buy 
ant there are issues 
first is the generator has a 3dB to low output on all the outputs and frequenties will these be a hard or software problem?? who can help me with these ??
A other fault is the message LO0Tunediag must be giving a value from 26.5v but must be between 0 and 20v I suppose these is someware a local oscillator
but don't find it direct into the (poo) service manual.
The generator and spectrum analyzer is only functioned to 2.2Ghz, readed spectrum levels are correct but not higher than 2.2Ghz
So who can help with tips/trick enz...
Greetings Luc ON7KZ

[RBBVNL9]

Luc,

Your observations suggest some problems indeed. But I'd suggest you first carry out all the self-tests (including loop 1/4 and 2/3) and relevant calibration procedures. This video may help you with some of this.

Then there is the service manual.

. rudi

[luc-on7kz]

Thanks Rudi,
having watching the video but it has notting telling me what i not know,
The service manual is poor no detailed schematics or even a more detailed blockdiagram or the place on the PCB is not to find
A lookup to the RX/TX board has also no visual issues except a small corrosion on one of the SWA filters, but where to find these parts and second problem placing these parts, I have some experience with SMD but these are just a little to high, I will not destroy the working parts.   
Any tip will be appreciated
thanks into advance
Luc On7KZ

[_Wim_]

The generator and spectrum analyzer is only functioned to 2.2Ghz, readed spectrum levels are correct but not higher than 2.2Ghz

Are you sure it is both the generator AND spectrum analyzer that are limited to 2.2GHz? I would expect it is ONLY the analyzer part. If that is the case, I would not expect it is the same corrosion issue with the filters as I had, as the filter used for >2.2GHz is also used below 1.2GHz which is working fine.

My first suspect would be a damaged RF switch (SW-239 from Maccom). Try to measure these in circuit (there a quite a few of them) and check if one of them is measuring very different with a multimeter. This could be an indication of a damaged switch.

A better (but much more work) approach is to check for each switch if a signal is seen on port A and port B when sweeping over the entire range. This however requires to solder test wires to each of the RF-switch A&B ports to be able to measure them when the unit is powered on. So I would first try to measure with a multimeter on the board itself to see if no obvious suspect is detected.

These boards have big copper planes which make it hard to remove components. So do be careful with a hot air gun as you have to add a lot of heat. As easier way in my opinion is to cut the legs of the part you want to remove with a fine cutter, and then remove leg by leg with a soldering iron. This is off course only possible if you are already sure the component in broken.

[_Wim_]

My first suspect would be a damaged RF switch (SW-239 from Maccom). Try to measure these in circuit (there a quite a few of them) and check if one of them is measuring very different with a multimeter. This could be an indication of a damaged switch.

Have a look at this post: https://www.eevblog.com/forum/repair/cmu200-rx-level-issue-failed-loop-test/msg3916070/#msg3916070

If you board looks similar, I would expect one of the two RF switches above mixer 1A & 1B might be broken. If your board looks different, it might be a good idea to try and reverse engineer like I did based on the block diagram. These boards are fairly easy to follow where the signals go, but it still takes several hours to complete...

[luc-on7kz]

Hello _Wim,
Reading and learning a lot here on these forum, your tips are great
Its so that both RX an TX fail above 2.2Ghz
The spectrum is working perfect till these frequentie, even better than my siglent SAA3021 (31) and these have i tested whit a 1Ghz signal from my R&D SMG
In the TX part i have 3dB less than it must be and these on all the outputs RF1, RF2 and RF3 i suppose somewhere a attenuator fault or ??
The search is on, I hope that it will be possible to make the CMU300 be working 100%
If that is OK than I will looking for the B41 module
greetings Luc

[_Wim_]

Its so that both RX an TX fail above 2.2Ghz

Hi Luc, in that case it is more likely that something common is broken (and thus not an RF-switch). It could be that "LO0" is used both for the RX and TX part. In my diagram I labeled LO1A and LO1B, but probably R&S labels them LO0 and LO1. As you get a tune error on LO0, this would make sense. So I would check with your SAA3021 if you get an LO signal on both mixers first.

Edit: it is very well possible the LO frequency is out of range for the SAA3021, a power meter might be a better option to check if an LO is present

Edit2: just reread my old post, the LO of mixer 1A was shared between RX and TX, so I would check this one first