Saturday, May 9, 2020

#334 - Wednesday, May 6, 2020 - IT'S WORKINGGGGG

As with many complex mechanical and digital systems, fixing one thing often causes two more problems. (Or more!)  Thus has been the case with all of my recent equipment upgrades.  Luckily, I love troubleshooting (big reason why I became and experimental physicist), so I've spent the last couple nights attacking some new problems.

Focus offsets

I got my new PrimaLuce Lab Esatto focuser installed on Saturday!  The SCT-threaded connection on it slides into the focuser, and you tighten three set screws into a groove on the adapter.  That way, I could set it to the exact angle I wanted, and it's a nice stable connection.  On the camera side, I got the 2-inch eyepeice-type adapter because that's the only way to connect my Lumicon off-axis guider, but this one, unlike my previous JMI EV-1 focuser, has a compression ring with three set screws rather than one.  So hopefully that will hold things in place much more firmly.

From the scope: Meade 0.63x focal reducer, PrimaLuce Lab Esatto focuser, Lumicon off-axis guider (set for ZWO camera backfocus), QHY5L-II guide camera, ZWO 7-position 2-inch electronic filter wheel, and ZWO ASI1600MM Pro camera.

I am now using two different brands of filters -- Astromonik CLS-CCD (light pollution) and RGB filters, and Chroma narrowband filters.  They have different thicknesses, so due to the difference indices of refraction, the focus points are shifted between the two filter brands.  Since I have an electronic focuser, I can measure the different focus points, program them into Sequence Generator Pro, and it will automatically move the focuser to the difference between the two when I switch filters.

The first step was to measure those offsets.  I set the CLS-CCD filter, and then ran the autofocus utility in Sequence Generator Pro four times.  (For the new PrimaLuce Lab Esatto focuser I have, I set the step size to 7,000 steps -- this is how far it will move between focus test points.  I didn't want it to move too much, or the more out-of-focus points would be too out of focus for SGP to get accurate star size measurements).  I recorded the focus position and HFR (half-flux radius, or measurement of the average size of the stars across the image), and then had it autofocus again.  I repeated this for the Ha and OIII filters.  

Then I took the average and standard deviation of those measurements, and set those as the focus positions in Sequence Generator Pro.

I don't actually have an SII filter -- it's just a filler until I get one later on.  Right now, that slot is empty.

Lastly, I needed to get the order of things ironed out in SGP.  This has been kind of tough, since I can't just say "do this, then that, then that;" SGP has its methods, and I kind of have to mess with different combinations of settings to bend it to my will.

I tried first to have it adjust the focus position per filter, and then focus with that filter (by turning off the "Auto focus with filter" option in the auto focus options).  With the variances in the offset measurements, I wasn't sure just setting the offset would be enough. 

Note: SGP doesn't use these offsets as absolute positions; rather, it will look at the last time you focused and which filter you used, and calculate the difference between that focus point and what you have set in the offset.  For example, if I have the CLS-CCD filter set at 200,000 in the filter wheel data, and the Ha filter at 210,000, and I just ran autofocus with the CLS-CCD filter and it found focus at 190,000, then it will take the difference between the actual focus (190,000) and the value in the filter list (200,000), and adjust accordingly, so that the Ha focus position is set to 200,000.  Make sense?

So I tried having it autofocus with the CLS-CCD filter, then change to the Ha filter and offset the focuser to get it close, and then autofocus with Ha to make sure it's in focus.  However, autofocusing with 3nm-wide narrowband filters doesn't work very well.  Even with 20s exposure times, I had several stars when it was near focus, but not enough for it to make accurate measurements when it was far from focus.  Plus, it took forever to run that way -- 9 focus test points, times 20s exposure time each, plus the time it takes to download and measure the stars in the image.  

So what I did instead tonight was to have it focus always with the CLS-CCD filter (a setting in the Autofocus options), and then it sets the filter offset when it changes filters.  Hopefully, this focuser would prove to be good enough that focus would be close enough just using the offsets.

Guide camera

The focus offsets caused another problem: my guide camera wasn't really in focus anymore.  I'm using an off-axis guider, and it's before the filter wheel in the optics train because trying to get a guide star through a narrowband filter would be basically impossible.  However, its focus is fixed, and moves with the camera, so when I drop in the narrowband filter, and the main camera focuses with the index of refraction difference of that filter, my guide camera is then out-of-focus.  Enough so that when I tried running things on Sunday, PHD couldn't hold onto any guide stars, and I didn't get any frames that night.  

Then I got a piece of advice from my astro-buddy Tolga: put the guide camera in focus with the main camera when the focuser is set for halfway in between the Astronomik and Chroma filters.  That way, it'll be a little out of focus for both, but not super out of focus for one or the other.  Smart!  (Of course, the real solution would be to get Chroma filters for my LRGB as well, but that's going to run me like another thousand dollars, and I'm not ready for that yet).  So on Monday night, I set the focus for halfway between the two, focused the guide camera, and then PHD was better able to hold onto the guide stars.  So that worked!

Dialing in polar alignment

I was curious about how well the mount performs unguided, so on Monday night, I opened up PHD and unticked the "Enable guide output" option in the brain dialog box.  This allowed me to see how the guide star moved around without actually sending guide commands to the mount.

It didn't exactly go running off, but it wasn't very good either.  Doing this reminded me of a tool in PHD that I've only tried one other time: drift align.  It's a way to fine-tune polar alignment by looking at how a star drifts due south and due east or west near the horizon, and then you adjust the altitude and azimuth knobs in an iterative process.  I don't have screenshots, but basically you first slew to a star at the meridian and 0 declination for the azimuth adjustment, hit Drift and watch the direction of a plotted trendline for enough samples to even out the noise, and then adjust the azimuth knobs until that trendline is mostly flat.  Then you do the same for altitude, but for a star that is due east or west, near the horizon.  There's a helpful document for this here.  I found that instead of moving the star all the way to the magenta circle, I needed to move it only like halfway to the circle -- all the way would far overshoot.  Drift alignment is a real pain and a slow process, but I've been told that it can have a real benefit for permanently-mounted scopes.  I'm not mounted on a pier, but I also don't take my gear down, so it stays pretty steady.  (Until the next earthquake, of course...)

After this, I checked whether I needed to update my T-Point model or not after this adjustment -- I slewed to a star, and while it didn't land in the middle, it wasn't super far off, and they already weren't landing in the middle before.  So I didn't touch the model.  I did re-calibrate PHD, and it looked pretty good.

A different autoguider?

Because my guide stars are so ugly -- partly because the off-axis guider is, well, off-axis, and deep in the coma of my Schmidt-Cassgrain, and partly because the stars aren't quite focused -- I decided I'd give another piece of guiding software a try, just for yucks: Metaguide.  I've used it recently for collimation, and I thought it'd be interesting to try out guiding with it.  It uses a different principle than every other method of autoguiding out there: Instead of doing long integrations (3-10s) to avoid "chasing the seeing," it sends corrections many times per second to the mount, sort of acting like a slow adpative optics-type thing.  It totally turns the thinking on its head, but the author (who spoke about it on TAIC recently) swears by it.  My hope was that it would be better able to grab oddly-shaped stars.  The interface looks a bit messy at first, but it has a lot of options and tools built in that are really handy.  

I had previously got it set up on my mount and guide camera when I was collimating my C11, so the tricky bits with doing that were already taken care of.  Calibrating ran quickly and without issue on a bright star.  Then I just needed to play with gain and exposure settings to see how short of an exposure time I could get away with.  Unfortunately, I had to do 2s exposures in order to get the guide star some of the time, so it more or less defeated the purpose.  It also didn't talk to SGP as smoothly as PHD2 does: it would frequently not "hear" the pause or stop commands, so that when SGP tried to resume, it was still already running, and SGP just sat there until its timeout ended and it's like "sure I'll just go ahead and continue the sequence anyway, assuming it's running."  

So the first frame came down, and oof.  It wasn't great.  It didn't like run away or do anything crazy, but the guiding was kind of all over the place.

Cropped image of some stars near M100, 10m exposure in OIII

The next frame was no better.

Are there some settings I could may adjust?  Perhaps.  But with 2s exposure times, when M100 has a pretty bright star right inside my guide camera, I don't think it's going to work out.  Not with an off-axis guider at least!  Might work better with a guide scope with brighter stars. 

So I switched back to PHD.

At last, success

After all these tweaks over the past couple of nights, I finally, finally had a great night.  The forecast called for an exceptionally clear sky.  The Moon was full, so I swapped out some of my wideband targets for narrowband ones.  The sequence had Ghost of Jupiter Nebula, M100, M51, and M57 Ring Nebula.  I did H-alpha on the two galaxies, and Ha and OIII on the two nebulae.  I ran out of time on the Ghost of Jupiter, so I didn't get any frames there (it goes behind my lemon tree at about 10 PM, leaving little time after astro-dark), but the rest of my M100 images and my M51 and M57 images came out really, really well.  Nice, tight, well-focused and well-guided stars, I could see signal in all the frames, no clouds, everything was great.  I didn't get a great number of frames because all my narrowband frames are 10 minutes long, but it did run all night without issue.  Yayyyyy!!!!

I settled on the following autofocus options:

Note: Step size depends entirely on your focuser.  My new focuser has a full range of some 400,000 steps.  My last focuser had a full range of only 8,000 steps, so the step size was very different.

I am also doing all plate solving with the CLS-CCD filter as well.

There's not much to see on the galaxy narrowband images, but here's a zoomed-in image of the Ring Nebula, a single 10-minute frame in Ha:

I can't see the dim halo around the Ring that I've seen in other images, but it may yet appear on stacking.  We shall see :)

Seeing in the morning that the sequence had run flawlessly the whole night and got a bunch of great images put me in a really, really good mood for the whole rest of the day.  A little bit of starshine in these difficult days.

No comments:

Post a Comment