Yesterday morning I saw what looked like halfway decent weather from dusk through about 2200 for capturing some test data in the evening, but I was afraid that I was going to miss the opportunity due to my mother-in-law’s birthday. I managed to get the scope set up, and to find some time to get an imaging session going, at which time it pretty much ran itself.
My objective for this session was to capture some calibration frames and some test data on the Monkey Head Nebula to check a capture change that I made and to check my processing workflow. Overall, I am pleased with the session. The problem areas were all related to focus. Some were bit of Jim silly-ness that I can easily avoid going forward, and I uncovered some technical focusing issues, most of which can be resolved easily.
Set up went very well at about 1650, about 30 minutes before sunset. The temperature was 41 degrees when I went back out to get the session going at 1750, about 30 minutes after sunset. There was no surface breeze at all. The seeing was forecast to be marginal, but I thought that it looked pretty good.
My first task was to shoot some calibration frames. Without too much difficulty I was able to get the Flat panel brightness set to deliver proper Flats, and I shot 30 Flats for each filter in the LRGB set, 120 frames total. I remembered shooting Bias frames last year, but since they are more than six months old, I decided to shoot another set instead of looking for the old ones.
I was able to finish the calibration frames, polar align, get an initial focus, and complete PHD calibration and GA run in just about an hour, which was before the end of astronomical twilight at 1853. I had to wait a few minutes beyond that time for the target to ascend above the trees.
Just a word on sequencing here. I was lucky that I found the main camera through the L filter to be in perfect focus (from the last session), but changing focus after taking Flats is not the right sequence. I need to add to my workflow to capture Flats after achieving an initial focus that is validated by an autofocus run. Doing Flats after polar alignment and initial focus seems to have an added benefit in that the sky is dark enough that I need not worry about light leaks in the imaging train ruining my calibration frames.
I set up NINA to shoot 45 Lum frames at 60s each, and 15 R, G and B frames at 90s each and initiated the capture sequence.
The first focusing issue that I had was self-induced. I set up NINA to do an autofocus run after every filter change, but I forgot to include a line for an auto focus run at the start of the capture run. Since the L filter was already in place when I started the run, there was no autofocus on the L filter. I realized the oversight several minutes into the run, but since I had just obtained a precise focus position with the Bahtinov mask, I decided to let the session run instead of interrupting.
An autofocus did execute after changing to the R filter. This is where the second focuser issue appeared, which was that the new focus position after the filter change was about 100 steps less than the one before it. This is a lot of steps, so the quality of my L frames were now suspect. Just in case, I added a line at the end of the sequence to capture another set of luminance images. This line of the sequence would execute after a filter change, so there would be a fresh autofocus run before shooting the L filter again.
The 100-step problem continued after the change to the G filter, after the change to the B filter, and after the change back to the L filter. As a result of this, the focuser position for the second L filter run at the end was almost 500 steps short of the initial L filter run. The stars for all of the runs looked pretty good, so the different focuser positions didn’t seem to indicate an out of focus position. The autofocus graph did reveal something interesting.
A normal autofocus run starts at a good focus position, backs the focuser out a certain amount, then takes an image and measures the FWHM (star size) and plots it on a graph with the focuser position. It nudges the focuser in a prescribed amount and then measures again. The FWHM plot begins as a downward (negative) sloping line. After about four iterations of this activity and as better focus is achieved, the slope of the FWHM plots will begin to increase from the initial negative slope toward zero, and eventually the slope will become positive on the other side of the best focus position. The finished curve is hyperbolic with maybe five plots on each side of the center, which is the “perfect” focus position.
Again, that description was for a normal run. The runs that I was seeing last night required maybe eight or ten plots to get back to the focus position. I attribute this to a loose clutch between the stepper motor and the draw tube. Because the loose clutch was slipping, each inward nudge was not moving the draw tube as much as it would have moved if it were tight, so more inward steps were required to get to the low point on the hyperbolic curve.
The third focusing issue was that the guide camera was in focus for the initial L filter run. The HFD numbers we bouncing around between the low 3s to low 4s. After the change to the R filter (and subsequent autofocus run, I noticed that the stars in the PHD display were quite bloated, and the HFD numbers were in the low 6s to high 7s. I attributed this to the new focuser position. Guiding was a little degraded, but good enough. At this point I thought in for a penny, in for a pound, so I let it run. This condition persisted for the G and B filter runs, but the HFD numbers got better for the final L run. I concluded from this that the L and the RGB filters are not parfocal
The temp was 34 degrees and RH was 75% when I ended the session at 2230. There was light frost on the grass, but nothing else. I also noticed as I was coming back inside that a light fog had set in.
I was set up to for about 1.9 hours of integration time with the LRGB. If I am able to use the second set of L frames in processing (LRGBL), then I’ll have something closer to 2.5 hours.
An overall conclusion that I have reached is that my guiding is good enough that I want to dedicate sessions when sky conditions are good to capture higher quality calibration and light frames for the purpose tweaking capture and processing. When sky conditions are good enough to work at the scope, but seeing and transparency are marginal, I will dedicate to guiding and focus tweaks.
Thinking through the focuser issues, all of them seem solvable, but I will probably need a night just to work on focusing to resolve them. Not getting an autofocus run before shooting the first L images is easy. I just have to remember to add the line to the sequence. As for the extra 100 steps the focuser was taking to get back to focus in the autofocus runs, I need to check the clutch, and that is easy to tighten. I have a step in my capture workflow to do this, but I was working off script last night. With the clutch operating properly I can measure precise focus positions for all of the filters.
The filters not being parfocal will be tougher to solve that I had imagined last night. Let’s assume that I have the main and guide cameras perfectly and simultaneously focused with the L filter in the optical path. Also assuming the L and the RGB filters are not parfocal, the focuser positions for the filters will be slightly different. The guide camera, which picks off its light in front of the filter, will move with the main camera as it moves from the L focus position to the R focus position. If the guide camera were in focus at the L position, it would have to be out of focus when it arrives at the R position.
Putting the OAG behind the filter wheel is not an option. I hope to transition to narrowband imaging this year. Main camera exposure durations through narrowband filters are on the order of tens of minutes. Guiding could never work with that tiny amount of light falling on the guide camera sensor. Since guiding worked both when well-focused through the L filter, and it worked ok when at maximum out of focus experienced through one of the RGB filters, maybe I should find an average focus that works well for all filters. Another alternative might be to obtain a parfocal filter set. More to come on this one.
I have recently read about a new NINA plugin called Hocus Focus that is an auto focus improvement. Before last night I had decided to put that into the category of new things to try after WSP. Hocus Focus provides better focus by better management of how stars are selected for measuring and how they are measured. Given how important good focus is to good guiding, I am going to load it up and try it.
Guide Log Analysis
Session 1: Calibration run, abandoned because I forgot to nudge north before initiating. I was working off script.
Session 2: A successful calibration run
Session 3: GA unguided run
Session 4: 21m run while I was waiting for the target to ascend above the trees
Session 5: 52m L filter run
Session 6: 25M R filter run
Session 7: 25M G filter run
Session 8: 25M B filter run
Session 9: 52m 2nd L filter run
I was looping 1s exposures. I would have preferred to have been at 3s, but I attribute this error to working off script.
Polar alignment error measurements (Drift tab) varied from .1’ to 2’ through the session. As opposed going from good to bad, or the other way around, the variation as all over the place between the high and low errors reported. As I suspected, PAE cannot be accurately measured, probably due to seeing. Taking the max PAE that I saw in any one session (2’), I can say that my PA was pretty good.
Seeing (RMS noise) and transparency (SNR decrease) seem to have worsened during Session 8 (Blue filter)
Unguided Ra in the GA run was 1.97” due to the worm periodic error. Through the well focused L filter, guided Ra in sessions 4 and 5 improved to about .55”, and maybe got a little worse when the seeing and transparency worsened.
Unguided Dec in the Ga run was .40”. Guiding through the well focused L filter in Sessions 4 and 5 improved RMS to .33”. Dec guiding through the poorly focused RGB filters was .56 or worse. It seems that I would have been better off not guiding on the Dec axis.
The PHD GA set the Dec limit was set at about .35”, but the seeing noise suggests that maybe it should have been set higher. I don’t want to conclude anything at this point because my looping duration was too short. I do want to pay attention to what I see as the possibility of PHD actually making RMS worse.
It is interesting to turn on the Scatter and SNR graphs, and watch how the graphs change through the sessions. Session 3 (unguided) shows Dec as no drift but with seeing noise exceeds the limit. The non-seeing variation in Ra is periodic error, which shows up as the wider left-right scatter. Sessions 4 through 6 show fairly good scatter. Scatter gets worse in 7 and 8 (unfocused, worse seeing), before getting better in 9 with better guide camera focus, but not as good as the other L filter session. I attribute the difference between the to L sessions as worse seeing. I still need to analyze the data in PixInsight. I am interested in seeing how the star size metrics changed through the focusing issues.
Image analysis: While not perfect, because the conditions were not perfect, the data are pretty good. Everything looks to be focused. I saw a little bloat late in the session due to deteriorating conditions. I also saw a little less noise in my RGB (90s exposure) frames than in the L frames (60s per exposure). In short, I think that captures are going well, and there are no further capture issues to work on before WSP.