Guider Plots 2¶
The night of Dec 15 was a full SN night as we were coming
off the community time. This means that we were taking relatively
long exposures and were working predominately near the equator.
The first thing to look at is the QR summary. The long sawtooth rise
starts a little after we went to the equator. Oscillations occurred all the
way up the ramp. At the top- I don't know; it sure looks like something
snapped. I don't believe atmospheric seeing had anything to do with this.
This next plot show what I mean by "oscillation". This guider image
shows the guider offsets sent to the telescope during one of the images
taken during the rise in the QR seeing summary. Most of the guider
images on the Guider Plots page are from this ramp. The period of the
oscillation is roughly 10 seconds. Oscilliations either don't occur or happen
rarely when we are working the southern fields, SPT-W and SPT-E.
After the snap, the oscillations disappeared. Somewhat after that
we moved off the equator. Now the problem with image quality was with
what I call "jumps" in the guider plots, which produce the extreme e images:
These are characterized by a jump in one direction, always y and usually positive,
followed by the guider slowly bringing the telescope back to the original pointing,
followed by a second jump in the opposite direction, here negative y, again followed
by the guider slowly bringing the telescope back. It's plausible that these are always
paired but that sometimes the second of the pair occurs when the shutter is closed.
These jumps were happening quite clearly on Dan Kelson's community nights on the
12/13 and 12/14. He noticed that the guider jumps always were associated with
a donut analysis dody spike.
This plot shows the donut analysis dody trace, and spikes, for 12/15.
The data are the calculated decenter in y deduced from the donut analysis
in the exposure taken while the guider shows a jump. The donut analysis
then calculates a dody to send to the hexapod, but at the moment does not
The question now is: what is the cause of the donut analysis dody? It could be
because the image jumped and returned, smearing the image and causing
the donut analysis to interpret it as a shift in dy.
Then Dave pointed out that the BCAM data shows the same spikes, at least
some of the time.
The BCAMs are mounted on the side of the primary itself and point at BCAMS
mounted on the filter changer, so on the camera barrel.
As the BCAMS see a shift we
know two things: the jump occurs somewhere in the primary mirror-camera barrel
system. If the jump had been due to a tracking glitch, say, then the guider and
donuts would have registered it, but the BCAMs wouldn't have. Second, since the BCAMS
register a shift, that means that the shift positive in y and then negative in y, the
dual jumps, in the guilder plots do not bring the system back to its original
The oscillations remain a mystery, but the jumps are in the primary mirror-camera barrel system.
Some comments on Jim's results by Gary:¶
- We see a 0.1 Hz period, but this could be the guider sampling frequency beating with a faster telescope motion. Mike Warner should be able to look at the TCS telemetry and tell what the real frequency is.
- A tilt of the primary will be seen as a displacement (and a small tilt) by donuts and BCAMs, because the donuts/BCAM report the postion of the focal plane relative to the primary assuming the primary mirror is fixed. The is a 10m lever arm converting primary tilts into focal plane displacements. Recall that CTIO has sensors recording the lateral position of the primary.
- I think a more exact statement of you conclusion would be "some or all of the jumps are in the primary mirror - camera barrel system" - there could be more than one source of jump, indeed both the primary mirror support and the cable wraps are suspected. Maybe recording the directions of jumps could help classify them. As astronomers, we will have to call them Type I and Type II jumps :)