CMU200 Options for tracking generator

[DAIRVINE]

Ebay supply:
http://www.ebay.co.uk/itm/CMU200-Radio-Services-Monitor-Tracking-Generator-Duplexer-Measurement-Sources-/252464888662?hash=item3ac8147b56:g:B3EAAOSwbsBXowJ-
This requires 9-12V. You feed in 1816.115MHz and pick up the spectrum analyzer first LO from the middle of the RXTX board. These are mixed to give a tracking generator. It works quite well, though there is some ripple. On the RXTX board there are two rigid SMA leads and a MCX lead which need disconnecting to remove the RXTX board.

You will need to drill a hole in the air guide sheet to access the LO mcx connector. See:
https://www.eevblog.com/forum/testgear/rohde-schwarz-cmu200/msg1007175/#msg1007175

This works up to 1200MHz. For 1200MHz to 2200MHz you use 843.085MHz, for above 2200MHZ you use 1816.115MHz again. The required frequency needs to be at as low a level as works, around -18dBm, and can be generated by the CMU2200. Too low a level gives more ripple on the response.
The local oscillator can use the supplied cable, but could instead use a spare AUX1 BNC and lead if this is unused or not needed. Having all the connectors at the front is more convenient.
Up to 1200MHz you get some ripple. This gets much much worse and attenuated as the frequency goes higher. It still works as a tracking generator, but you have to compensate for the tracking generator level changes.

CMU200 is very sensitive to 1816.115MHz breakthrough to the first IF at 1816.115MHz. Ideally you would use an external oscillator and filter out 1816.115MHz before it reaches the CMU200.

It is possible to use a mixer instead of the Chinese solution. A mixer could output down to DC, which would be useful for looking at signals below 10MHz(also possible is above 2700MHz) with a converter. The problem with a mixer is that you will get 1816.115MHz breakthrough. This breakthrough increases the apparent noise floor, so giving worse results. If you have a separate signal generator (I don't so have not checked this)and are looking at responses with little output at 1816,115MHz, for example a 10.7MHz xtal filter, then the mixer should work quite well. You would probably want to amplify the LO to get a good reference level.

So mixer is possible, and within limits works, and in the end could be better than the Chinese tracking generator. For instant results use a mixer if you have a suitable one. For better results use the Chinese tracking generator, for best results you need to do some work. An improvement to the Chinese mixer would be to generate the required frequencies externally from the ref output. And add a notch filter after the mixer. To keep 1816MHz out of the CMU200. Pick your mixer for the correct frequency range and for good isolation. I got quite good results from a mixer, but it doesn't have the frequency coverage needed.

Block diagram showing filter frequencies:
https://www.eevblog.com/forum/testgear/gotcha-for-cmu200-and-crtu-users/msg694796/#msg694796

Tracking generator review:
https://www.eevblog.com/forum/testgear/rohde-schwarz-cmu200/msg1006966/#msg1006966

Link to some tracking generator install photos I got from the seller:
http://pan.baidu.com/s/1jIi5o8m

The principle here is that the spectrum analyzer uses a fixed filter. It varies the local oscillator to convert the monitored frequency to 1816.115MHz using a mixer. So by mixing the LO with a signal generator set to 1816.115MHz we reverse this process and generate an output at the monitored frequency, so tracking the LO.

Two photos of wide span with the tracking output fed to the RF4 input. One with RF set to 1816.115MHz covering 10-1200MHz and 2200-2700MHz, the other with RF set to 843.085MHz covering 1200MHz to 2200MHz. Both drop by about the correct amount when an (unknown frequency response, 6dB) attenuator is fitted.

So a question. Is it possible to correct the frequency response to level the tracking generator?