The Starry Night, 76

:: home ::




              <<   41   42   43   44   45   46   47   48   49   50   51   52   53   54   55   56   57   58   59   60c   60   61   62
              63   64   65   66   67   68   69   70   71   72   73   74   75   76   77   78   79   80   81   82   83   84   85   >>   SRCH


12/10/2011. Time at last to do some finer fine tuning. I downloaded a trial of CCD Inspector to take a look at the state of the AT10RC and its imaging train. Collimation looks great (a few seconds off); field curvature with my ST2000XM is just under 20%; tilt in x and y is good to a few tenths of a degree. I'm very happy with these results, and I will probably register the software in order to keep track of and improve these specs.

I need to correct some previous remarks about the steel-tubed AT10RC not suffering noticeably from focus drift under falling temperatures. While checking overnight sequences of recent images, CCD Inspector causes me to think that I really do see focus shift with changing temperatures after all. In the fairly lax regime of >2 arcsecond star images that I consider reasonably good, this is not immediately obvious. Stacked images look pretty good without worrying about this. But in almost every sequence, there is indeed a trend toward larger stars as the night goes on. These more severe December nights may be more taxing in that respect than were early- and mid-autumn nights. (Focusing at 50°F for a run that ends near 20°F represents almost twice the delta of earlier tries.) Or it may yet be that my mediocre seeing changes that much during the course of the night (but almost always for the worse? not likely). I need to correlate these graphs with meteorological records to see if they do, indeed, match up with changing air temperature. If so, then it's time to get the temperature compensation routines in RoboFocus calibrated and on the job.

I installed A-P's high precision azimuth adjuster on the Mach1. These are now standard on that mount, and A-P made sure the part could be retrofitted easily. I like the last instruction ("step back and admire a thing of beauty!"). Even the price is meticulously fair: new mounts are as much more expensive as this part costs to add on. While fitting the new base, I relocated two spiders who had made a home inside the mount but I offed a third who elected to hide in a hole from which I couldn't coax, blow, or otherwise extract him gently. A lot of mud had been transported into a few of the mount's hollows. Wasps? I cleaned all that out, and used diced foam to stop up as many access paths as I could find.

I remounted the telescope with no drama. But I could not for the life of me get it polar aligned using the bore scope. RA and Dec tattletale plots in PHD were widely seperated; corrections were large. 1800-sec images of M31 in H-a showed gross field rotation. Maybe something shifted while I was cleaning the scope's optics. I had to do something and something turned out to be a really neat trick. It occured to me some weeks back that PHD "knew" and was displaying everything needed to do a dead-nuts polar alignment job. Surely, somewhere out there, someone had documented how to use it for that purpose. Google is your friend.

"PHD Guide polar alignment" (without the quotes naturally) turned up my new favorite alignment routine. This comes courtesy of an amateur named Pete on the Stargazer's Lounge, only slightly edited (not so much for clarity but because I can't help myself):

I spent some time last night between clouds using PHD to do a drift alignment, and ... it works a treat.

Method - set scope to near meridian and celestrial equator; start PHD and do a calibration; start guiding with DEC corrections turned off; turn on the Graph, and change the RA/DEC button to dx,dy.

If you see a steady drift in the dy, adjust azimuth East or West until the dy graph remains flat, apart from the seeing-causing slight random movements.

Repeat the procedure with the scope aimed East or West on the Celestial Equator. Calibrate PHD and again guide only in RA. Watch the dy trace, but this time [change elevation] until the dy trace flattens.

This took me about an hour including longish runs to ensure dy was really flat...

The nice thing is you can quickly see the affect of each mount adjustment within a few seconds, especially if you adjust in the wrong direction!


Thank you, Pete! I like this because it uses only software I already have installed and running at the 'scope and because it doesn't require me to remember specific directions of movement or any rules of thumb ("if the star moves N and you're aimed W of the meridian, then move the azimuth E; reverse these instructions if your star is E of the meridian or the star moves S..." give it a rest!). Nor do I need to fiddle to make sure the detector's x and y axes are aligned with the sky before getting down to business. It fits my "just play with it and do more of whatever works," mindset.



One bright Pleiad in full Moonlight
3x900s L, 1x900s RGB (with mismatched flats)
Note that a cable dragged -- or something --
for a few seconds during the G exposure.
L frames show 2.1" stars, round to better than 0.5%


And it does work a treat! The combination of the new azimuth adjuster and an effective computer-aided alignment method is very satisfying. The difference between a dy trace moving up or down sharply is a tiny adjustment. The slope of the line offers an immediate indication of approximately how far off the alignment is. Some kind of systematic search is needed to find the area of adjustment in which the critical range lies. This first time out, I spent a lot of time locating the right ballparks in both altitude and azimuth, but I bet it gets quicker quickly. My restricted east and west skies insure that I have to iterate at least twice since I can't get very low in the west anywhere near the celestial equator. I need to experiment some to see how critical (if at all) is the declination of the chosen stars. I only watched the dy curves for up to one minute at a time to be sure they were reasonably flat; when not flat, just 10-20 seconds of watching 3-second integrations leaves no doubt that alignment is off. As I type these notes, I am running 15 minute exposures of a star in the Pleiades. Guiding is good to 0.60 RMS which works out to under 1/2 arc second. The plot has only rarely touched full-pixel deviations. This is good news. In the resulting images, stars are round, everywhere, often to less than one percent(!) compared to what I've been calling good enough (sometimes 10-15% or more). If there is still field rotation, there is very little. I begin to think that I may never have made a properly guided astrophoto.

So... Yes, PHD Guide can cope with polar misalignment. But yes, it's really better if it doesn't have to.


12/14/2011: I messed with the polar alignment some more, not because it needed much work but just to get some more experience with the method. I don't think I did the alignment any favors with this lab exercise [right! the "aspect map" in CCD Inspector, see below, confirms that I messed it up nicely, which means I am ready for more practice]. I did three 1800 second frames of NGC 6960 while fixing supper and waiting for Amy to get home from her (last ever?) night class, and afterward slewed to M33 intending to gather H-a until it hit the treeline. Weather had other notions and closed the sky early. I only got a set of 300s color frames, one full H-a subexposure, and 75% of the second one. Tracking was noticeably better than before I started doing computer aided alignment, most of the time, but large excursions again punctuated the run. I shifted the counterweights outward, which answered this insult before, but I had no chance to test the mod. There is little field rotation (it's measurable but not visible); stars in some of these images are as small as 1.8 arc seconds and no larger than 2.4 on any (incidentally: I focussed tonight without reference to the measured PSF's on the chip; rather, I just made the diffraction spikes from Vega as sharp as possible; the result was unambiguous down to 5 RoboFocus steps). The chip would only cool to -33°C at the beginning of the evening, so I used it at -30°C instead of -40°; tonight is a warmish December night. There is far, far too little signal and too much noise to make good photos from tonight's harvest, so I'll make small ones instead:


n6960 and m33



12/14/2011. I registered CCD Inspector and upgraded to the latest version. Good lord. I may never see daylight again. So much data to mine. I believe it will help me tune just about any aspect of the astrographs that I find wanting. The curvature map to guide collimation adjustments, the aspect map to diagnose polar alignment by quantifying field rotation, the graphs of FWHM PSF's to measure focus shift, and who knows what all else. You put >$15k out in the yard, why balk at ~1% more to keep it all in trim?

Likewise, a collimating ring for the focusser is enroute from Teleskop-Services in Ransburg, Germany. You know what they say: in for a penny, in for 131 Euros. The only part of it I mind is that too much of that went for shipping (TS has a sterling reputation for service, though, and I don't mind paying for that if that's where part of the shipping assessment goes). Still: I do want the additional degree of freedom. CCD Inspector convinces me that if I can control the tilt of the sensor, I can get smaller, more consistent stars, especially when I ask CCDI to extrapolate to larger sensors than the KAI2001M detector in my ST2000XM (like, especially, the many-times-larger sensor in the Canon 5D). With this addition, I ought to be able to fully collimate the R-C, end to end, finally. I'm seeing 0.5 arc second differences from side to side as it is. Why is that significant? Well, consider the intensity of light within star images: tightening PSF's from 2.1 to 1.6 arc seconds should let me see 0.6 magnitudes deeper. That's equivalent to increasing the AT10RC's aperture to 13.1 inches. If I could get that much more from eliminating focus-shift and another bit from avoiding field rotation, the next thing you know I'll think I have a 16-inch astrograph.


Except where noted, deep-sky photos are made with an SBIG ST2000XM CCD behind a 10-inch Astro-Tech Ritchey-Chretien carried on an Astro-Physics Mach1GTO. The CCD is equipped with Baader LRGB and 7nm H-a filters. A Meade DSI Pro monochrome camera looking through a modified Orion off-axis guider keeps the OTA pointed in the right direction. The imaging camera is controlled via Nebulosity 2; the guide camera is operated by PHD Guide 1.13, both by Stark Labs. The stock focuser on the AT10RC has been augmented with Robofocus 3.0.9 using adapters turned on the lathe downstairs. Maxim DL5.12 performs image calibration, alignment, and stacking; Photoshop CS4 and FocusMagic 3.0.2 take it from there. Gradient Xterminator by Russell Croman and Astronomy Tools by Noel Carboni see their share of work, too.


:: top ::



                   © 2011, David Cortner