The purpose of these tests is to see whether we can correct for the rotation
of the ALFA receiver while the telescope position varies in azimuth during
the "modified" version of basketweave observing, wherein the target region
is too close to the zenith to observe in normal basketweave mode, so the
telescope is steered off the meridian to observe at nonzero hour angles.
To facilitate these tests, a new version of Josh Goldston's program BW_fm
was created, called BWT_mod, which sets up the observations and calculates
important parameters, such as the expected azimuth and parallactic angles
of the telescope at the start of the basketweave, and by how much ALFA
should be rotated before the run to ensure good beam spacings.
First test was in gear 6, ALFA rotated to -22 degrees. Immediately after
start of test, telescope rotated ALFA to 77 degrees. This is because my
position angle was also set to -22 degrees, which is not right. The
parallactic angle was about 80.7 degrees, so a position angle of -22
means CIMA tries to achieve a rotation angle of -80.7 + (-22) = -102.7
degrees, or about +77 degrees in an "allowable" rotation sense.
The lesson learned is that I want position angle = parallactic angle, so
the rotation is very little and the initial rotation of -22 has nothing
to do with position angle.
A more severe problem is that the script does not calculate a proper
parallactic angle during the run. It only takes the initial RA and DEC
and calculates a parallactic angle from that. Arranged with M. Lerner to
modify his program to calculate parallactic angle from current AZ and ZA,
which should give a good enough value of the parallactic angle.
Beam tracks look like this.
Good start to test. Seems to calculate parallactic angle properly this
time. Handled crossing 90 degrees parallactic angle with no problems.
Went from DEC=15:47 up to 21:00, found that feed was rotated -2.3 every
30 seconds in order to maintain constant position angle.
Beam spacings on upward leg of scan were excellent. On downward leg, not
so much. Seems we need a second position angle, since attack angle is
different (by 22 degrees) on the way down. How interesting!
Plots for the first and the third fits file show that one position angle will not do.
Today was the day with lots of RFI, but the test went great. Using two
position angles, got really nice beam tracks both upward and downward.
I gave 20 seconds turnaround time at top and bottom, which might be a bit
more than is really needed. Here are the beam tracks.
The rotation angle of ALFA is plotted here for the first fits
file of this day. The quick change of 37 degrees down to zero has me puzzled.
I didn't really think about it until the next day's test...
Based on success of yesterday's scan, tried a gear 5 test doing the same
thing. Since attack angle is different, need to rotate by -22 degrees
at top, rather than +22 degrees for gear 6. Found that ALFA needed to be
rotated by about -1.3 degrees every 30 seconds during legs of basketweave.
Looked harder than before for signs of RFI due to motor triggering every
30 seconds to rotate ALFA. I saw a fair amount of RFI in general (these
observations were after 7 am) in the wideband, including two spikes which
showed up consistently, but not with a period of 30 seconds (it wasn't
periodic as far as I could tell).
Beam tracks for October 14 for first fits file and second fits file.
ALFA rotation angle plots for first and second fits files.
According to the rotation angle plots, my initial setting of 7 degrees was quickly
reset to 0 degrees when the first parallactic angle measurement was made.
Beam spacings were a bit off, which I thought arose because CIMA had to wait a
while for telescope to slew to target position, and ALFA was rotated a
few times BEFORE the basketweave started. In order to preserve the ALFA
rotation angle specified in the pt1 command file, I requested to ML to
remove any rotations by CIMA until AFTER the basketweave starts.
I was given more test time just as I arrived to do my regular observations, so
I figured I'd re-run my October 14 test, to see whether refraining from rotating
ALFA until AFTER a basketweave leg started made any difference on the beam tracks.
Take a look at the first fits file and the
second fits file.
Here are the corresponding first ALFA_angle and
second ALFA_angle plots.
Very similar to Oct. 14...
The first fits file has some weird tracks that go the wrong way in RA, because the
telescope slewed to the wrong starting position (I have no idea why!). But after
that nonsense, the tracks look very similar to the Oct. 14 tracks. I conclude
that the inital 7-degree rotation angle setting is lost at the beginning of the
basketweave run, and this results in the less-than-ideal beam spacings. I guess
this means that the position angle and parallactic angle should not be equal, but
rather they should differ by 7 degrees.
Which begs the question - why did my gear 6 tracks turn out so great, when the
initial 37-degree angle was also quickly reset to zero? This is mysterious.