[evlatests] L-band 'Set and Forget' Mode?

[rperley at nrao.edu]

Numerous reports attest to the fact that the L-band attenuators are often
set to the wrong level, often by many dB.  The L-band data from my 3C273
data taken a couple weeks ago gives support to the cause being the RFI
environment.

The data were taken in 16 subbands, each 64 MHz wide.  There is of course
switched power available for each.  A review of the PDif and PSum for each
of the 16 subbands is very interesting.

For only about 6 of these subbands is there 'normal and correct' switched
power -- the band between 1370 and 1500 MHz, and between 1800 and 2000
MHz.  All the others show large disturbances, of varying nature.

For most of these the variation in total power is typically a factor of 2.
 For some of the subbands, it's clear the power fluctuations are
quasi-random (clearly the case in the aeronautical bands), in others it's
obvious the switched power is being greatly manipulated by pulsing (as in
radars) emission.

But the worst cases are those subbands covering the satellite transmission
bands, where the range of power seen exceeds a factor of 50.  There are
two of these, so even if we divide these by 16, the effective power range
over the whole 1 GHz subbands is at least a factor of 6 or so (4 dB). 
Such variations are the rule, not the exception.

The receivers themselves are far more stable than this.  It is both risky
and foolish to use the current 'set and remember' algorithm for setting
the T304 attenuators -- from the distribution of the output power one
easily sees that errors of many dB are possible -- even likely.

I advocate a 'set and forget' strategy, as in 'set it once, and remember
it forever'.  We can establish, for each IF on each antenna, the typical
power level on cold sky, and enter the appropriate T304 attenuator levels
in a lookup table, which is occasionally updated when receivers are
changed out, or other changes occur.  Although I realize that managing a
table system is unpopular, I think the risks of the current strategy
outweigh the inconveniences of a table lookup.

Probably we'll need this at S-band also, given the digital satellite radio
bands.