[evlatests] More on Switched Power 'Saturation'

[Rick Perley]

    Bob Hayward and I journeyed to antenna 24, 2 weeks ago, to do some 
specific tests, hoping to clarify the mystery surrounding the apparent 
compression (and sometimes expansion!) of the PDif values when observing 
sources of different flux densities in situations where we know the 
system gains do not change. 

    The experiment reported on here is very simple:

    1) We observed 3C84, at C-band, in basic OSRO mode, with all antennas. 
    2) Antenna 24 (which shows the strongest compression effect in its 
RCP  channels) was stowed (but remained in the array). 
    3) Extra noise power was inserted into the RF in antenna 24.  We 
utilized a variable attenuator so that the added noise increased in 1dB 
steps.  Twenty-two steps were utilized.  The maximum added noise was a 
factor of 5 greater than the 'cold-sky' noise.   This extra noise should 
replicate the effect on the front end of (say) observing a strong source 
like Cygnus A.
    4) The station boards monitored the total powers, and produced, as 
normal, values of PSum and PDif.  (Unfortunately, because we have to 
adjust the attenuators by hand, our presence in the antenna was 
required, thus preventing it from tracking the source.  Hence, no 
visibility measurements are available to compare to the changes in 
switched power). 
    5)  'Set-and-Remember' was turned off, so that the Executor adjusted 
the T304 attenuators to keep the power to the samplers in the 'safe 
zone'.  This change in power caused by this is easy to see in both PSum 
and PDif, and I've adjusted the results to remove the attenuation. 

    Expectation:

    Following correction of the powers induced by the attenuator changes 
in the T304, we would hope to see *no change* in PDif, since the 
injected switched noise power was kept constant.   However, from 
previous observations on/off Cyg A and 3C273, we know that the values of 
PDif on the RCP side become smaller as the strength of the source 
increases.  Thus, if our experiment indeed replicates an observation of 
strong sources, we expect to see a reduction in PDif as the injected 
noise power increases.   This effect should be nearly absent on the LCP 
side. 

    Result:

    Sadly (perhaps, depending on your point of view), the expected 
effect was clearly seen.  On the RCP side, the values of PDif 
monotonically decreased as the injected power rose (following correction 
for T304 attenuation).  On the LCP side, virtually no effect was seen.  
The RCP PDif diminution was ~15% at the maximum power injection -- a 
factor of 5 over the 'cold sky' value.  IFs 'A' and 'B' (tuned to 4900 
and 5000 MHz, respectively) were essentially identical. 
    I attach a plot, showing the PDif rolloff.  All values of PSum and 
PDif are normalized to 1.0 for the 'cold sky' setting, allowing direct 
comparison. 

    Discussion:

    Although the expected effect was seen, its magnitude is quite a bit 
less than expected -- the 'on/off  Cyg A' experiment -- which triples 
the cold-sky system temperature --  showed a considerably larger 
reduction in PDif than the ~8% reduction seen in the experiment.   No 
simple explanation for this comes to mind. 



   
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 24NonLin.ps
Type: application/postscript
Size: 17426 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20110527/796115dc/attachment.ps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 24NonLin.jpg
Type: image/jpeg
Size: 41874 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20110527/796115dc/attachment.jpg>