E4438c or any E44xx level adjustment possible?

[Scrapcollector]

Dear fellow Testequipmentadjusters,

Ist there a level correction procedure out there for any of the E44xx series? Normally, when you put the same RF board back in the same device after successful repair, the level is in the specs. But not always.

Also, does anyone know where the M11X1005 (Edit:corrected) sot89 amplifier is obtainable or a working replacement? Seems to be GaN instead of the old shf0189 GaAs(?).

I tried the manual user flatness, which is ok for home use, but not a neat solution. Also, my device did not work with the E4419 power meter for automatic user flatness correction (after selecting the 19b type of course).

The service manual is kind of a joke. It tells you the name of all the procedures you have to do after RF bd repair, but not how they actually work. Clicking them in the Keysight Software I guess.

My 4Ghz E4438c (with several pretest-repair-stickers on the rf-bd) at 0dBm puts out:

250k   0,83dBm
    1M  0,85
100M  0,6
500M  0,44
    1G  0,38
   1,3G 0,25
   1,5G 0,17
   1,8G 0,3
   2,0G 0,05
   2,9G 0,00dBm
   3,0G -0,36
   3,1G -0,58
   ....
   4,0 -0,07dBm

As measured with a Rohde Z-51.
Any Ideas other than typing all this in the user flatness table?

Best regards,
SC
   

[dxl]

I wrote a python script to calibrate flatness on my E4435B (i extended it to 4GHz / E4437B, so i had to calibrate flatness above 2Ghz.). Don't now whether that works without modification on the E4438B, but i would expect the commands to be similar.

https://github.com/svenschnelle/caltools/blob/master/calesg.py

[mankan]

@dxl Thanks for the script, I've been playing around with the commands on my E4436B which I've previously extended the frequency range on.
However something is missing since I cannot get values in file 207 to be applied for frequencies higher than 3GHz.
I have updated the frequency indices in file 65.

Any advice?

[mankan]

A few steps forward. It turns out that the unused part of frequency index has to be updated with the same value (4000 MHz) as the last frequency.
Now the attenuation is applied for all frequencies above 3000MHz except exactly 4000MHz, a bit annoying but I'll survive 

Weirdly enough one cannot write a larger value than 4000MHz in the frequency index even though I increased the upper limit to 4020MHz.

[Scrapcollector]

Thank you very much dxl for sharing your work!

I will get back to you after I have found some kind of "python programming the ESG-Series for dummies" article. Can someone point out to me one or two links to get started?

I a not a programmer, and with my limited programming skills I look at this script, trying to understand what it does before I let it roll on my E4438, with more than one question mark in my eyes. I think I have to understand it enough to make modifications to it, then run it on an older ESG to make shure to not shoot out the wrong calibration tables at my E4438c (I don't know how many of them it has or how to communicate with them yet).

Frequency extending a 3GHz ESG also sounds very interesting. I suppose this is done via a more sophisticated method than copying eproms (and eventually modifying the serial number in the code), right?

Best regards,
SC

[mankan]

I've now solved my mystery, it turns out that you have to have a power offset for each corresponding frequency table entry, even after the end marker. So I had the frequency index ends with multiple 4 GHz for the entire register file but only power offsets for all frequencies up to 4GHz including the two first 4GHz entries. When I wrote the same power offset in the rest of the padded power values I now get correct power at 4GHz as well