[evlatests] Bandpass issues

[Vivek Dhawan]

3-bit tests at K band showed the bandpasses of ~half the antennas to
be variable by 1-2% on timescales of minutes. This is x10 worse than
we typically see (though careful tests were done a while ago, in A-
array).  All current tests were done on 3C84.
Best antennas 4,12,17,27,5RCP, 22RCP; worst 2, 5LCP, 22LCP;
rest in between.

Stacking all the 8 samplers on the same frequency (4R, 4L) shows that
the bandpasses are independent of sampler. 8-bit bandpasses agree with
the 3-bit at the same sky frequency, so its not the samplers (whew!)

C band (8-bit) was done next, and shows curious ~80MHz ripples which
are modulated with a ~5minute period. Again these are x10 worse than
the 0.1% bandpass amplitude stability we have seen before. The delay
steps are a few seconds apart, so I dont think that could be the cause.
Best: 21, 15, 17, 27, 28, 2  Worst: 1, 6, 10, 11, 20, 22, 23.
Is there any correlation with radome material on the C-band horns?

X-band (8-bit) looks much better on most of the antennas. 10, 21 look
just fine. ea1 looks less fine, but the ripples are smaller than at C,K.

I'm concluding the problem is upstream of the T304s (i.e. not including
them). Some issues are clearly front-end related (K5-LCP, K22-LCP) but
are too small to see on the bench. Generally I find it hard to accuse
the front ends and converters, since they are unchanged since previous
tests.

Any other ideas?

Are other people seeing this?