[evlatests] Update on Linearity Tests

[Rick Perley]

    The observing file on Cygnus A which revealed the apparent 
non-linearity in the PDif power measurement as a function of the input 
power was repeated today, with the monitor data logging turned on to a 
faster rate (1 Hz).  This allows a detailed comparison of the power 
levels as seen at the station board to those seen in the T304. 

    It was quickly found that the output power meter on the T304 
reproduces very well the powers seen on the correlator station board.  
More to the point, the characteristics of the non-linearity discovered 
in the digital power are replicated well in the T304 power meter.  This 
points to the T304 as a possible origin of the problem. 

    However, as the non-linearity found on Cygnus A is nearly absent at 
X-band, but is strongly present at C-band (and at lower bands), 
suspicion is focused not on the T304 itself, but rather the input power 
spectrum from the LSC converter (T302).  It may be the harmonics of the 
LO power in the T302 (which are known to exist) are getting into the 
T304 (despite an input filter which is designed to prevent this), and 
upsetting the linearity of the T304.  The engineering group is 
investigating. 

    Other tests are also proceeding.  If the T304 is in a significantly 
non-linear state for C-band observing, the effect should also be noted 
in the correlator visibility data.  The effect should be notable on 
3C273, data for which I have in hand.  Vivek is looking carefully at  
the stability of the switched power from experiments (like pointing) 
where there are no band changes and no large range of input powers -- he 
reports some instabilities which should not be present. 

    There will be more to come on this.