Weather
Started warm and humid – but a cool front passed just around sunset. Clouds were a problem early on – after about 9:30 local time mostly clear, very occasional high clouds, low temps in the upper 50s, humidity around 50%, very little or no wind. Seeing is not great – maybe 2.2 arcsec.
Rig #1: WR134,135, & 137, also WR1
I got the declination assembly back this afternoon and installed on the AP1200. The motor is now much quieter, and so far I’ve noticed none of the intermittent power interruptions. Not sure how that could be related, just an observation at this point. The new motor runs much more smoothly. I setup on WR134 and began the sequence. Some clouds were passing through throughout the first cycle – the second time through was better with only very occasional or no clouds. Note that I only just barely had time to finish two cycles before getting into the maple tree. Noel Richardson preferred that I still try to get two cycles if I can – once that is impossible go to longer exposures for a single pass through. I got somewhat of a late start, but really only going to have another week or so where I can do two full cycles. Once finished (at around 12:30 following Beta Aqr image) I decided to try WR1 in Cassiopeia. At V=10.11 it’s likely too faint, but I’ll try a set of three 1-hour exposures to see what I get.
Rig #2: ASAS-SN 13ck
The clouds kept me from starting a sequence on V1101 Aqr – so I skipped it once it cleared and went straight to ASAS-SN 13ck. It is definitely fainter tonight, going with 1-minute integrations.
Some more ratty data!
Last night’s results were interesting to say the least. The data for ASAS-AN 13ck were a total mess:
I have no idea what caused the discontinuities in the light curves. My usual suspicion of scattered light just wouldn’t do that. It almost appears that the curve changes each time I stopped the sequence to focus. But plotting the image fwhm against the magnitude difference (shown in the blue curve) did not show a correlation. The thing is, the V1101 Aql data came out looking great, where the previous evening’s plot showed a 0.04 mag jump at the time of meridian flip.
Note there is no “jump” in the comparison star curve – in fact the standard deviation for the entire run is 0.007 mag. The thing that I noticed was that, for the September 4/5 data, looking at sample images before and after the meridian flip, the objects fell close to the same location on the before and after meridian flip image sets. But that is after having flipped the post-meridian images so that all images were East-Left and North-Up. In the case of the V1101 data from the previous night the objects were shifted by quite a long way, particularly in X. But, in the case of the ASASAN data, the images from Sept 3/4 were shifted by a lot, while from Sept 4/5 they were pretty close. Yet it is the September 4/5 data that is screwy. So exactly the opposite of what I’m seeing for the V1101 data. So what is up????
One thing for sure is that I am not going to figure it out tonight! Fading fast – nodding off. Out at 4:20am
Hi Edie:
I think I’ve narrowed it down to being PSF shape differences. I use a variance-weighted optimal extraction routine for the photometry which makes the assumption that the shape of the PSF is constant across the entire image. In fact I know that NOT to be the case. I purchased and started using a focal reducer/field flattener for that very reason. But the spacing between the reducer lens and the CCD chip is off by a few mm, which still leaves some field curvature. It can be seen when plotting the FWHM as a function of position. The variation might be as much as .3 pixels or so. But it turns out that things can be much exacerbated by not tending to the focus. It turns out that the temperature was falling rather rapidly that night and so I was not keeping up with the focus changes it was causing. So not only was there the usual issue of slight field curvature, but the shape of the field curvature was changing. When in focus in the center of the image the FWHM size of the PSF is smallest there, and plotted as a function of position the curve showing FWHM very gently rises towards the chip edge. But when out of focus it can be the exact opposite – with the edges of the field being in better focus.
I used to use a standard aperture photometry program and noticed then that I would occasionally get some fairly obvious discontinuities in the comparison star light curves at meridian flip. I figured that was due to scattered light differences between the sky flats and the fact the night sky is not the same color. Using essentially no color filter would really accentuate any color-dependent scattered light differences. But what I also noticed was that if I just enlarged the apertures used to measure the stars that the phenomenon was much reduced. That certainly argues against the scattered light theory. In the data shown in this post the large discontinuities occurred when the telescope was re-focussed. Note that in the article I mentioned that the FWHM itself did not change – but that was based on just looking at the FWHM for one star throughout the night. Instead, when I look at individual images before and after each stop for re-focusing and plot FWHM as a function of position for each image the differences are obvious. Just before re-focus the curve of FWHM as a function of, say, X-position, was a convex curve (pointing upward). After focus it was a concave curve (pointing downward). Since this article I’ve been much more careful about keeping things in focus and have greatly reduced the problem. Mind you, I had a long spell where clouds and life events kept me from getting to the observatories – but all the data I’ve gathered since has led me to conclude that the largest problems are, in this order:
1) focus changes
2) variable psf across image (even if all images are focused)
3) scattered light differences between the twilight sky flats (blue sky) and the night sky.
As to #3, above, I’d noticed in the past that I have smaller issues with difference pre- and post-meridian crossover when using, say, a V filter, which at least somewhat limits the color differences between the twilight sky flats and the night sky.
So I apologize I did not see your comment earlier!
Mike Potter
Beverly Hills Observatory