[evlatests] More on Power (non-)Linearities

[Rick Perley]

    I reported this morning on the comparisons between various power 
meters for 4744 MHz.  Since then, I've reviewed another frequency, 6744 
MHz, where we took the same range of data (cold sky, sky dips, Cygnus A, 
and hot loads).  The significant difference is that the absolute powers 
at this frequency heading into the meters is 2.2 dB lower -- and this 
affects some of the results. 

    For both frequencies, excellent linearity is seen over the range of 
power spanned between cold sky and Cygnus A ( a factor of about two).  
So I calibrated the FFT system and the TPD-B power meter by matching 
them to the Milhouse values over this range.  I then solve for offset 
and look for non-linearities visible at the hot load level (a factor of 
about 12 above the cold sky). 

    Summary:

    4744 MHz. 
   
    *  The raw power levels are -30.56,  -24.98 , and -27.20 dBm for the 
Milhouse, TPD-B, and TPD-A power meters.  (The FFT method used IF A). 
    *  The offsets are -1.0 and -0.5 K for the FFT and TPD-B meters -- 
in the sense that these meters find Tsys is lower than Milhouse.
    *  With these relations, the non-linearity seen on the hot load is 
-26K and +9K for the FFT and TPD-B meters.  (in the sense that the FFT 
method shows a deficit (compression), and the TPD-B an excess (expansion). 

    6792 MHz

    *  The raw power levels were -32.8,  -27.13, and -29.65 dBm for 
Milhouse, TPD-B, and TPD-A. 
    * The offsets are now -1.6 and -0.8 K for the FFT and TPD-B meters. 
    * The non-linearity on the hot load is now -12K and +8 K for the FFT 
and TPD-B meters. 

    Hence, reducing the power by 2.3 dB seems to have reduced the FFT 
method non-linearity by a factor of about two, but didn't significantly 
change the TPD-B results.