UPDATE 2022-05-26
Paul reports fixed on May 16! Loose cable connection in front end/receiver that seemed to be pointing dependent.
------------------
example pipeline products here:
/lustre/aoc/projects/srdp/problem_data/C_band/
-------------------
UPDATE 2022-05-02
It's on the move again, now centered at 5.7 GHz for 8bit.
22A-301.sb41249612.eb41788345.59700.033183321764
Not sure if it's been mentioned, but I don't think I've seen it in a BP/flux cal. In this example it's evident in polleakage and phase cals.
Solution:
BP/flux (3C286):
polleakage:
phase:
target:
-------------------
UPDATE 2022-04-25
Recent data sets show the issue spanning the entire C:A0C0 and C:A1C1 basebands. The effect was previously seen in the last few spw of the lower baseband, and the first couple spw of the upper baseband.
The problem seems to be isolated to L polarization.
8bit
22A-301.sb41249703.eb41784559.59693.05873708334
3bit
22A-059.sb41209756.eb41775497.59690.41143894676
22A-125.sb41222215.eb41775501.59690.49107805555
--------------------
ea17 solutions look like y = -1/x, ranging from 5.0 to 6.5 GHz (centered at ~5.7), crossing the A1C1 to A2C2 boundary. This can be very pronounced to very minor.
plotsummary corrected amp vs freq is usually seen spiking between 5.0-6.5 GHz as well.
preview:
...