CMU200 Rx level issue / failed loop test

[Sas]

Yes, the whole chain is used for the factory calibration. The unit I repaired was within 0.1dB so there was little difference in the filters I swapped.
Has anyonw cracked the code on the calibration file? (where is it stored and what is it's structure?)

Michael

The difference of 0.1 dB is acceptable in the two new filters. I was unable to decode the table structure. I stopped while trying to run VERM and CMU on a virtual machine. There are traces in the configuration files that the software can run under dos itself, without the entire device (for programmers). I think the calibration software is already installed on the device. I tried to figure it out two years ago, now I don't remember much. Re-calibration after the factory is probably done by connecting the standards and sending a few commands via GPIB. I think there is one general table for tx and one for rx that can be calibrated outside of the R&S factory e.g. in LGE Korea. It seems to me that the correction of other tables is only possible at the R&S factory (e.g. temperature corrections).

Where did you buy DFC3R836P025HHD ?

[Sas]

Then I removed the soldered copper tape again (which probably overstressed the 836 filter, see below), and solder a tiny copper plate on top  (plate pressed flat on the surface filter, only soldered the sides while pressing down). This improved the result on the 1842 filter to only -3dB (was -10dB before), my best result so far. Unfortunately no more output on the 836...

Soldering surfaces is a risky idea. The filter is not soldered from the bottom to the PCB. PCB for RF doesn't like high temperatures. After all, the result of -3.5dB is quite a good result. It is not known how linear the generator is, in my CMU200 it has a deviation of about +/- 0.5 dB for -30dBm in the entire frequency range.
Yesterday I checked the first channel of my CRTU. The worst loss on an internal loop 1->1 is -1.5dB and 2->2 is -2dB. External loop 1->4 is -2.1dB and 3->2 is -2.4dB. The loss of -2.1dB is repeated for -35dBm in loops 1->4 and 3->2, which excludes problems with RF inputs. Probably the 836 MHz filter has too much loss, there is some fault or the module requires calibration. I don't have new 836MHz filters to verify this.

I add photos of filters before and after cleaning from the first channel of my CRTU. I cleaned the upper surface of the filters with a cotton swab and isopropylene alcohol, then a mixture of methyl alcohol, water and detergent (dishwashing concentrate), neutralized with isopropylene alcohol, then cleaned with ketchup (it cleans copper perfectly) and neutralized again with isopropylene alcohol. I isolated the filters from the RF absorber with laminating pouches (0.15mm). I did not want to stick anything to the surface of the filters so as not to damage the copper layer.
I will check the results of the loop measurements after the initial repair in the CMU after the return from the buyer (he replaced the rx / tx board and more ).

[Sas]

FreRes CMU loop 3-> 2 after replacement of BPF 1842 MHz with a new one and BPF 836 MHz tightly covered with copper tape. The tape is soldered on the corners of the pcb filter. The 836 MHz filter repaired in this way has too much loss of about -4.4dB. If the filter factory has a loss of -2.6dB, it is about -7dB after repair. I note that this filter was in a very bad condition (photo 31_CMU_B_836_return_buyer _cheat.jpg).
Internal loop 1-> 1 and 2-> 2 test in the range of 10-1200MHz are from + 0.3dB to + 3.1dB and in the range of 2205-2700MHz they are from -1.4dB to + 1dB. I guess it should be like that, in the internal loop of the same input there is too little isolation in the signal paths. Someone has already written about it.

Is there any way (apart from printing a screenshot onto a CF card) of logging these results (any text format)?
Do you switch the RF4 IN input correctly in FreRes (the analyzer remains on the RF2 input instead of switching to RF4)?

[_Wim_]

Is there any way (apart from printing a screenshot onto a CF card) of logging these results (any text format)?
Do you switch the RF4 IN input correctly in FreRes (the analyzer remains on the RF2 input instead of switching to RF4)?

The failed results are stored in a file cst.err when running in continuous loop test mode. The good results unfortunately are not. A workaround is installing an attenuator in the loop path so all results are bad...

To get the cst.err file, you have to have a keyboard attached to the CMU. When all the tests have run (in continious test mode), you have to hit alt+F4 to exit in dos modus. The cst.err file is found in the firmware folder (if I remember correctly). This file has to be renamed/copied in order to prevent it from being deleted automatically.

After renaming, you can reboot the CMU and transfer the file to your hard drive using the "remote service tool" software.

There might be an easier way, but that is the only way I know of.

[Sas]

At CMU (buyer's cheat back), after several hours of warm-up, the RF3 OUT output died    . Yes, I thought the screw heads were damaged for a reason. He also replaced the RF IN / OUT module  .
And I wanted to try the log method with the cst.err file. The generator works on RF1 and RF2 (loop 1 i 2 is ok). It may RF3 OUT work when it cools down, but finding such thermal faults is not easy. We'll see tomorrow. I'm fed up, I'm going to sleep .

Edit:
The MMIC amplifier at the RF3 OUT output is damaged. The cool one worked as normal, but as it warmed up it limited the gain to the band above 2.2 GHz, and then the full band. Cool it worked, warm no, and over and over again. After disassembling the in / out module and heating the amplifier with hot air, it got a short circuit on the output (and power supply) of 60 ohms. It is now completely dead.

[Michael-VK2BEA]

Where did you buy DFC3R836P025HHD ?

The replacement 836 filter I canablalised from another board bought on eBay (from the TX side .. it also has corroded RX side filters)


I placed an order for the BHD version on the dubious site I mentioned earlier (ecTransistors.com) but have not heard anything from them despite several emails. The 'order' still shows pending and the PayPal payment is the 'approved' state. I'm not optimistic though.

After seeing Wim's message the other day, I did another search and found some (BHD) on eBay which I ordered. (I had looked on eBay previously but not found this)
The really curious thing is that the price on the dodgy site and the eBay seller are identical (except they added a bunch of fees and taxes). Maybe its someone scraping eBay and then charging high shipping and fees.

BTW .. If you need just a couple, send me a message.

Michael

[Sas]

After seeing Wim's message the other day, I did another search and found some (BHD) on eBay which I ordered. (I had looked on eBay previously but not found this)

I ordered BHD filters on 23 Dec 2021 from the same ebay source. Theoretically, the package is flying to me. I wonder who will get first
 (USA vs Europe send 29 Dec 2021). If I did not get it within a month then I will send a message.
According to the IEEE MTT (thanks a lot) that you posted, BHD filters can be older type filters, with an electric wall. I still have to wait for the MMIC amplifiers to the RF3 output, I can't do anything else until the parts are delivered.

[Michael-VK2BEA]

According to the IEEE MTT (thanks a lot) that you posted, BHD filters can be older type filters, with an electric wall. I still have to wait for the MMIC amplifiers to the RF3 output, I can't do anything else until the parts are delivered.

Unfortunately it's going to be difficult to get it to fit. I'm thinking that it might need an intermediate PWB (with castelated edge 1/2 holes) to mount the BHD and lift the end of the (bigger) filter over some other smd components. 

Michael

[_Wim_]

I can happily report that the BHD type filter can be bodged on with good results 

Probably not ideal as not all the filter castelations are grounded, but seems to work ok. I first covered the small components above the filter with capton tape to prevent shorts, then moved the small cap at the input of the filter a little to the side (but still on its original pad), aligned the filter such that the cap fitted just inside the castelation input of the filter, and the solder the ground and the output of the filter.

Covered the new filters with capton tape to prevent future corrosion from the RF absorbers and reassembled. All tests past, even the FM Calibration that failed before!

Remark: removing the 836MHz original filter was not easy. I do not know if this was related to my soldering abuse from before, but I had a hard time getting the filter of the board. Luckily nothing got damaged in the process.

[Michael-VK2BEA]

I didn't notice before but your layout is different than my RXTX board. In mine the 936 MHz filter is at the edge of the board.

[_Wim_]

I didn't notice before but your layout is different than my RXTX board. In mine the 936 MHz filter is at the edge of the board.

Ach indeed, you board does look quite different. Something to watch out for when attempting this repair. It was already very tight on this layout, but luckaly it just fitted.

[_Wim_]

Now that I look at the pictures again, I did also install the filter in reverse (output used as input). I presume these filters are fairly symmetrical, but should maybe check if no additional spurs are seen by doing it this way....

[Michael-VK2BEA]

Now that I look at the pictures again, I did also install the filter in reverse (output used as input). I presume these filters are fairly symmetrical, but should maybe check if no additional spurs are seen by doing it this way....

I think you are right ... they should be palindromic (50 ohms in and out, passive and linear). I'll see if I can confirm on the VNA. I think there should be a way to fab an intermediate board that gives a better ground (and lifting above the other components).

BTW It seems R&S used these filters in other devices. I spotted them the other day in Shahriar's (Signal Path) repair video of the FSL spectrum analyzer. (see them here in the bottom left  https://youtu.be/60hasxvl5KU?t=1044)

[_Wim_]

I think you are right ... they should be palindromic (50 ohms in and out, passive and linear). I'll see if I can confirm on the VNA.

 

BTW It seems R&S used these filters in other devices. I spotted them the other day in Shahriar's (Signal Path) repair video of the FSL spectrum analyzer. (see them here in the bottom left  https://youtu.be/60hasxvl5KU?t=1044)

Good find. Probably not a "cheap" donor though... 

[Sas]

All tests past, even the FM Calibration that failed before!

Congratulations on the repair . Can you do the 1-> 4 and 3-> 2 loop test with a positive result?

BTW It seems R&S used these filters in other devices. I spotted them the other day in Shahriar's (Signal Path) repair video of the FSL spectrum analyzer. (see them here in the bottom left  https://youtu.be/60hasxvl5KU?t=1044)

SA R&S FSH3 also has the same filters.

I don't have MMIC amplifiers yet and work stopped.
I got the same filters from the same seller. The interesting thing is, each of the filters is different, one of the filters has large offsets between the sections. The entire 3 pole BPF consists of three separate filters.

[_Wim_]

Congratulations on the repair . Can you do the 1-> 4 and 3-> 2 loop test with a positive result?

A single frequency fails on the TX1 => RX4. But this is related to CH1 of the frontend, which was a separate issue as reported in the first post (and as far as I remember, CH1 has always been a less good performer since I have the CMU). I typically do not use CH1, so I never looked into this.

There does seem to be a small jump (+-0.6dbm) when switching IF-frequencies, but will try to fix that with the frequency dependent attenuation settings

[_Wim_]

There does seem to be a small jump (+-0.6dbm) when switching IF-frequencies, but will try to fix that with the frequency dependent attenuation settings

A little better

[_Wim_]

Meaured spectrum of a BG7TBL RF noise source (https://www.epanorama.net/blog/2017/08/29/bg7tbl-rf-noise-source/)

No strange jumps or peaks, so I expect most is working fine, certainly sufficient for my modest needs in RF (my low cost China RF cables and connectors are probably creating at least as much deviations...)

[Sas]

I fixed the CMU.. The CMU passes all tests without errors, including the 1-> 4 / 3-> 2 loop test. Adds the RF3 OUT -> RF2 loop test image from FreRes after repairing the RF3 output and replacing the 836 MHz BHD filter 

[_Wim_]

I fixed the CMU.. The CMU passes all tests without errors, including the 1-> 4 / 3-> 2 loop test. Adds the RF3 OUT -> RF2 loop test image from FreRes after repairing the RF3 output and replacing the 836 MHz BHD filter 

Congratulations. Looks very good! Can you post a picture how you installed the BHD filter?

[Sas]

Congratulations. Looks very good! Can you post a picture how you installed the BHD filter?

I soldered the same, isolated with Kapton tape. I removed the solder mask and added a GND solder pad (PB tin) under the filter on pcb (i marked the size and location in the photo). I soldered the filter with hot air, then soldered the top of the filter to ground and the in / out connection with a soldering iron (new toy on board JBC CDE-2BQA  ).

The waveforms look similar, my deviation from +0.7 dB to -1.2 dB. My gain drop is more linear. If you have a problem, it may be possible to correct it with an attenuator (I think 6 dB, I marked in the photo).
It will not change the fact that you may have too low gain (from -0.1 dB to -1.4 dB) in the whole spectrum analyzer operating range or too low signal generator level. I don't know why. Spectrum analyzer gain too low or generator RF level too low. Missing +0.8 dB gain (or generator RF level) causes errors from loop test 1-> 4 / 3-> 2. May require calibration or some repair.

[_Wim_]

I soldered the same, isolated with Kapton tape. I removed the solder mask and added a GND solder pad (PB tin) under the filter on pcb (i marked the size and location in the photo). I soldered the filter with hot air, then soldered the top of the filter to ground and the in / out connection with a soldering iron (new toy on board JBC CDE-2BQA  ).

The waveforms look similar, my deviation from +0.7 dB to -1.2 dB. My gain drop is more linear. If you have a problem, it may be possible to correct it with an attenuator (I think 6 dB, I marked in the photo).
It will not change the fact that you may have too low gain (from -0.1 dB to -1.4 dB) in the whole spectrum analyzer operating range or too low signal generator level. I don't know why. Spectrum analyzer gain too low or generator RF level too low. Missing +0.8 dB gain (or generator RF level) causes errors from loop test 1-> 4 / 3-> 2. May require calibration or some repair.

Thanks for the additional info and pictures. Maybe I will experiment a bit more with the filter placement in the future, but for now I am quite satisfied how it works. Nice JBC you have there, makes my JBC BT-2BWA look "old fashioned" 

[Bravo]

I found some filters on Arrow electronics website. They are smaller than the originals but have the correct spec.
So some kapton tape & thin wire will be required.
They are discontinued & very cheap. They have large stocks ............until throw them out.
The 836.5 Mhz one has a  part No B39841B4147U410  @ $0.45 for one.
The 1842.5 Mhz one has a part No SF1192B  @ $0.73 for one.

[Bravo]

I ordered some filters from 
https://www.ectransistors.com/10-PCS-INDUCTOR-TRANSFORMER-MMCL-DFC3R836P025BHD-3R836P025-DFC3R836P025BH.html

After paying for them & courier shipping, they told me that the price had gone up 500%, so I payed the extra. They then shipped the filters by economy post. Assuming the packet is not empty, they are still cheaper than the ones on Ebay, but you have to buy 10.   

[Bravo]

I fitted the small 836Mhz filters from Arrow,  part No B39841B4147U410. They work fine.