[evlatests] R/L phase differences.

[Bob Hayward]

I'm not quite sure I agree with Vivek's contention that this time 
variation is being caused by not having enough isolation in the LO 
splitters inside the various receivers. The reason for my thinking this 
is two-fold...

1) At Q-Band, the VLA is fine but the EVLA isn't. The VLA and EVLA 
Q-Band receivers have exactly the same LO splitters and have similar 
cabling between the splitters and the tripler/mixer assemblies. If the 
LO splitter isolation was the cause of the problem, the VLA should be 
just as bad.

2) Both the VLA and EVLA receivers use the same LO splitter (although 
the EVLA receiver has a doubler added to provide high-side LO 
injection). The splitters in the K-Band receivers are in a (warm) 
thermalized environment and have short fairly rigid cabling. The K-Band 
receiver should be much less sensitive to thermal changes or flexure 
from elevation changes. If the LO splitter was the problem, K-Band 
should show less of an effect than Q-Band, where the splitter and 
interconnection cables sit relatively unprotected in the cold, cruel 
world of the Vextex Cabin (waveguide systems can't be packaged as easily 
or as compact as coaxial systems).

But we can do tests by adding isolators to a couple of EVLA Q-Band 
receivers to see if this changes things. The EVLA Ka-Band receivers, by 
the way, have extra isolators already (we had a bigger budget) so if it 
is an LO splitter problem, they should look much better.

-Bob


Vivek Dhawan wrote:
>              1. Summary of previous tests:
> 
> 3MHz continuum, IF1 = IF2 in frequency, source= Orion maser.
> 
> Q and K band showed short-term changes in (RR-LL) phase of ~5 deg.
> 
> Apart from the polarizer D-terms, such effects may be caused by cross-
> talk between various paths in the RF chain (unlikely, since these are
> well padded with isolators); or in the LO output fanout of the L301 or
> L302 (neither of which are as well isolated).
> 
> Similar variations are seen in the (IF1-IF2) phases, K & Q. This is
> (R1-R2) phase, so the D-terms cancel exactly as IF1=IF2 in frequency,
> implicating the LO fanout path.
> 
> Most antennas show something, not just a few bad apples. Looking at
> VLA vs. EVLA differences, a puzzle (or clue?) emerges:
> VLA antennas are just as bad as the EVLA at K band. But
> VLA antennas are good (in the noise, <1deg) at Q band.
> 
> 
> 
>             2. New tests:  6.25MHz continuum, 3C84, L, C, X bands.
> 
> L,C,X all show the same variations as above, at about 1-2deg. VLA ants
> are a bit better, but the SNR is getting low, I can't really tell.
> 
> There are a couple really bad outliers, at all bands. Possibly, the
> T304's are old models or something like that: EA16 IF D; EA17 IF B.
> Perhaps also EA14 B,D (intermittent?)
> 
> 
> 
>             3. Next step: in consultation with Hayward & Jackson,
>                the plan is, in rough sequence,
> 
> A. Do Ka band for completeness. (Test scheduled for tomorrow)
> 
> B. Put in isolators on the L302 outputs of a select few antennas, see
>     if that cleans up X band. I pick EA24 and one (or more) of EA1,2,3,4.
>     as hardware permits. Most useful if both L302's are done on the same
>     antenna. Isolators can be left in place for normal use.
> 
> C. Put in isolators after the L301 power splitter at Q band. Also can
>     be left in for normal use. See if this cleans up Q band. Same antennas
>     as in A, to ensure good L301/T304 isolation.
> 
> D. More invasive diagnostics - Feed same RF input to several T304's to
>     narrow down the location of trouble, e.g. cables vs. T304's. This must
>     be undone for normal astronomy.
> 
> Vivek.
> 
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests