M1 / M 31-- micro-focuser test

 11/24/2024 -- Testing out the new (to me) Meade Zero Image-Shift Microfocuser for the venerable Meade LX10 2000mm F10 SCT... connected to the shoestring astronomy controller on the Odroid XU4....and controlled with Indi/Ekos.  Initial focusing test showed promise using an initial starting movement of 1500ms... this collection against M1 will see how the microfocuser fares with an actual collection.

Meade Zero Image-Shift Microfocuser... into which the Starizona Corrector III is inserted.

Weather started good from 18:00 through mid-night, but clouds are expected and the moon rises (at 28% full) at 01:30... so no real expectations of greatness... but a good night to test equipment.  Using only a 1.25" UV/IR cut filter (broadband color imaging).  After collecting a test image, set exposure to 100s at gain 285 with offset 32 at 0 C.  Spent considerable time re-adjusting the polar alignment (swinging back and forth between -800 to +50". Finally gave up around -11" error...enough to leave notable streaks in a 120s light frame collect without guiding.

Seeing forecast by meteoblue for 11/24-25 suggesting conditions will begin to decline around midnight (just as collection commenced, naturally)

Seeing forecast by cleardarksky also suggests clouds around 01:00 and continuing through morning.

The indi allsky backyard camera noted a pseudo-sky quality metric of ~6000-5000 (not great, not terrible).

Indi/Ekos/KStars focus module from the run launched by the scheduler for M1.

10s light frame (binned 2x2) from the Indi/Ekos focus module following auto-focusing launched from the M1 scheduler.

Internal guider error was ~6" RMS ... nearly all the error in the Dec axis.  Clouds started rolling in visible on the guider view after only 6 light frames collected (~12:16am).  Intermittent wifi network problems were a source of frustration all day.  Ultimately 38 light frames collected over night.

11/26/2024 -- No moon (until after 03:30) but clouds expected to roll in after midnight... set M1 up on the scheduler. Sadly, the focus module did not communicate with the camera driver successfully, and so the scheduled collection did not happen. By the time this was corrected (near 23:30), clouds were hick.

Seeing forecast for 11/26-27 from meteoblue showing clouds expected around 01:00.

Seeing forecast for 11/26-27 from cleardarksky with similar indications of sketchy skies after mid-night.

11/29/2024 - Post-Thanksgiving collection on M1 with no moon and no clouds. Seeing forecast is exceptionally poor, but otherwise a great evening with temps in the upper 20's.   Installed a new EQ6 collar on the franken-mount (Meade Standard Field Tripod...adapted for the Celestron CGEM-DX) to try and improve the setup and polar alignment. The initial installation worked... but left too much slop and movement and will require some more work to fix.  It was not possible to converge to a usable polar alignment...with frequent over & under shoot of the desired position. Tried a see if the internal guider could deal with the polar alignment error and recover some collection for the evening, but no luck - calibration failed repeatedly.

New silver EQ6 collar on the frankenmount.

11/30/2024 - After adding two screws to the new EQ6 adapter on the franken-mount, found some improvement to polar alignment... but it remains sensitive (easy to shift too far one way or the other). Was able to iterate to a polar alignment error <10". Found that the internal guider in Indi/Ekos had no problem with calibration and guiding proceeded successfully with RMS errors < 2".  Set up a mosaic on M31 (2x3 panels) with broadband imaging. Autofocusing with the Meade zero-shift focuser worked out well given a close starting point.  Clouds expected around 22:30-23:00 with poor skies expected by mid-night...so a short set of collections this evening.  Temperatures are low (20's) with the Svbony 405cc reading -5.3(C)... so the cooler remained off for this session.  Indi Allsky backyard camera showed a pseudo skyquality metric of ~6,000 through 21:30.

Seeing forecast by cleardarkskies for Saturday, 11/30 - 12/1/2024 showing an early end to observations with clouds rolling in by mid-night.

Seeing forecast by meteoblue for 11/30/2024 which agrees with cleardarkskies with a gloomy outlook.

Ultimately, 37x100s light frames were collected with the scheduler shutting down collection at 22:30.

Indi Allsky backyard camera for 11/30/2024 indicating worsening skies started appearing at 22:03 with significant clouds at 22:15.

12/01/2024 - Clouds early... still no moon.  Continuing to collect with just the 1.25" UV/IR filter (broad and color) with the SV405cc through the Meade LX10 OTA with the Starizona Corrector III inserted into the Meade zero shift micro-focuser.  Due to the clouds, unable to start until 22:22.  Pseudo sky quality metrics stabilized around 5400 from the Indi Allsky backyard camera.  Had the mount covered with some light snow flakes in the afternoon... but decided to see how well guiding worked without double-checking polar alignment from last evening.

cleardarksky seeing forecast for 12/01-12/02/2024 suggesting clouds early followed by clear skies...but poor seeing.

Seeing forecast from meteoblue with similar predictions with light to very light overcast expected after 21:00.

Indi Allsky backyard camera analysis for the early evening of 12/01/2024 showing impacts from clouds up until 20:35 with pseudo sky quality metric dropping gradually to ~5400 by 23:00.

Scheduler problem/fix:   Reloaded the M31 mosaic from 11/30... but in the scheduler all the jobs are marked 'Complete' (given the scheduler was built for yesterday). Found that the individual mosaic collection jobs in the scheduler can be updated with new parameters and saved into the scheduler with the '+' key (then the original job removed).  This refreshed the scheduler with new jobs.  Given that 37-ish frames were collected for part 6 of the mosaic, removed that job from the list (with sections 1-5 remaining).  Put this in for one run each.

Focus module problem / fix:  Found the OTA out of focus... and in playing with the focus module, found that moving the focuser out produced better focus. However, the main autofocus routine starts by focusing outward... and then moves in from this initial point.  It missed the focus point... and kept moving focus in (further and further out of focus). To fix this, shifted the focus point so that it was out of focus slightly where moving the focuser in would result in correction. This allowed the autofocus routine to work.  So in summary... the OTA needs to be in near perfect focus (or slightly out of focus but at a point where moving the focus inward is correct ... in order for autofocus to work). 

Noted that the mount slewed the OTA to the desired target almost immediately (second try) and that the internal guide routine calibrated successfully and showed excellent guide errors (<1.5" RMS) suggesting that the polar alignment has held from yesterday night. Noted the internal Indi/Ekos guider is set for 2" exposures at gain of 60 with no binning (1:1) through the Celestron 9x50mm RACI finder scope.

Screenshot from Indi/Ekos inside KStars showing the M31 six-panel mosaic

Following the M31 run for panel 5 around 12:30, set up a job in the scheduler to collect against M1 (Crab Nebula). Unfortunately, the focus module did not succeed in finding the focus point until the scheduler ended attempts at 05:02.  :<

12/02/2024 -- Another cloud-free evening with no moon. Indi Allsky backyard camera shows a pseudo sky quality metric of ~5800 (dropping gradually).  Additional broadband collection against the M31 mosaic with 20x100s for each panel 1-4.  Adjusted the focus module to use a different algorithm (Iterative) with an initial step size set to 4500 with a maximum step size of 3000 and maximum travel at 100000.  This appeared to work immediately with M31.  Internal guider calibrated succesfully with a classic L pattern and guided with an error <2" RMS. The guider drift pattern was centered symmetrically on the plot with tight 2" pattern.  This is different from last evening where the guiding looked erratic at times...and significant drifts were visible even during the short 10s focuser collections (wind?).  Similar to last night, the Svbony SV405cc camera was steady at -3.5C without the cooler running.

Seeing forecast from cleardarkskies for 12/02 - 03/2024 showing outstanding conditions..with the usual exception of seeing.

Seeing forecast for 12/02-03 from meteoblue also showing enviable skies for the NCR with the exception of seeing.

12/06/2024 -- Additional session for M31 mosaic.  While near 30%, the moon set at 22:07 freeing the sky for broadband imaging with an additional set of captures for the M31 mosaic.  Sadly, clouds are forecast near 1am which may limit opportunity for collection. Temperatures dip in the 20's, so still no need for a cooler.  Uncovered the scope and checked polar alignment (~40"). Adjusted to below 10" in PHD2 running on the odroid XU4.   Focuser (using the LX10 zero shift microfocus module) succeeded after 11 iterations.  Guider internal calibration was immediately successful. Guider errors started around 1.3" RMS.  Indi Allsky background camera pseudo sky quality values started near 5200 and dropped to 4900 by 24:00.  

Noted that the date and time on the Odroid XU4 is badly off (by 2.5 days in the past).  Attempted to reset the NTP client and sychronize the clock -- but oddly initial attempts with timedatectl and reconfiguring and resetting timesyncd failed.  Similar settings in ubuntu on the Indi Allsky camera raspberry pi produce working NTP synchronization. But FITS files saved by the local indiserver instance have the incorrect time stamp and the DATE-OBS FITS parameter is also unfortunately off.

Seeing forecast for 12/06-07 showing clouds near 1am.

Seeing forecast from meteoblue for 12/06-07 with similar prediction.

12/08/2024  -- Skies started out clear this evening, but the moon is at 40% and is up high in the sky... which is less than optimal for broadband imaging.  However, configured the scheduler to start M31 before clouds move in around 22:00-23:00.  Amazingly, the scheduler did run automatically...and the focuser, alignment (plate solving) and guiding all functioned successfully.  Warmer temperatures did mean setting up the cooler before taking off at 17:00...to have the camera down to 0C by the start of the scheduler jobs (18:24-ish).  

Discovered Indi-Ekos busy collecting light frames when double-checked at 22:00 and four of the six panels in the mosaic had been collected already.  The internal guider showed RMS errors of ~2" with reasonably tight patterns nearly equally distributed in DEC and RA.

Clouds started rolling in around 23:00 and the scheduler was stopped and light frames transferred from the odroid XU4.

12/22/24 -- After many rainy days and nights, finally one cloud free (and relatively moon-free) evening. The LX10 is still mounted with the Starizona SCT Corrector III, Meade zero-shift micro-focuser, wide band collection with just a UV/IR cut filter (1.25") installed.  Cold temps (20's).  Restarted M31 mosaic (in reverse order with panels 1-6) after performing a polar alignment with PHD2 running on the odroid XU4 (achieving <1"alignment error).  Calibration for the internal Indo-Ekos guider was successful and guiding errors were <1.5" RM (very tight pattern). 

Seeing forecast for 12/22-23/2024 from cleardarksky showing cloudless viewing with the moon appearing and decreased darkness after 3am or so...but relatively poor seeing all night.

Seeing forecast from meteoblue with a similar outlook.

Initial Indi Allsky backyard camera pseudo-sky quality metric and star counts for the early evening on 12/22/2024 showing a sky quality metric of ~6800 (not great for a cloudless, moonless evening).

Ultimately added additional targets to the scheduler when it was clear that M31 was drifting out of view.  This included M1 (the Crab nebula) and M44 (the beehive cluster).  

M44 (Beehive cluster) with 31x100s light frames calibrated using flats collected with the LCD panel and neutral density filters in Astro Pixel Processor 2.0 beta.

12/23/2024 -- Clouds expected but the scheduler was set up and ran automatically with collection starting around 21:00... and fortunately clouds were delayed later than expected. At 22:00, the Indi Allsky camera showed a pseudo-sky quality metric of ~6300 (higher than the best skies, but still cloud free).  Guiding results were once again looking good with errors < 2.5" (RMS) and the drift plot showing errors centered and symmetric betwene dRA and dDE. The cooler was kept off with the Sv405cc at ~-4.7C at 22:00. Clouds started to appear on the Indi Allsky backyard camera results around 22:45.

A grim seeing forecast for 12/23-24/2024 showing clouds predicted from the European model near midnight. 

Seeing forecast from meteoblue for 12/23-24/2024 with dense high level clouds predicted even earlier -- at 22:00.

One light frame 1x100s at gain of 285 and offset 32 (M_1_Light_0093.fits) from 12/23/2024. Some saturation on blue stars clear in this light frame.

12/27/2024 -- experimented with neutral density filters (several layers) to provide a flat frame with ~8 second exposure times (as needed by the Svbony 405cc for effective flats).  Used a cushion to keep daylight out and rely on the LCD tracing panel for illumination sitting directly on top of the OTA. Cooled the camera to 0C, set gain to 100 with offset of 32.  Mounted the resulting set of neutral denisyt filters into a poster frame for future convenience.  Collected 30 flat, darkflat and bias frames at these settings.  In configuring the filter set and the gain settings, was attempting to match overall light frame background values of ~16K digital units from the Svbony 405cc when collecting against M 1 (as shown in the light frame above).  

12/30/2024 - Clouds drifted through ... but cleared out by midnight... and so after performing a polar alignment, started collection against M 1. After the clouds passed through, the Indi Allsky backyard camera pseudo sky quality metric was steady at ~5660.

Seeing forecast for 12/30-31/2024 by cleardarksky showing clouds expected until 1am.

Initial M1 (crab nebula) stack (average) in ASTAP using old dark calibration frames (at 180s vs 100s) with judicious selection of light frames within ASTAP (having a quality between 2 and 7, background ADU values between 10897 and 16035, stars between 160 and 378, and having a co-registration solution. This resulted in 181x100s light frames from 327 collected.  The aspect of this result from very rudimentary processing in ASTAP is the quality of the flat frame in terms of removing vignetting and artifacts.

Initial M1 stack (average) result from ASTAP using 181x100s light frames with 30x8s flat frames as well as matching 30x8s dark flat frames and mis-matched dark frames (30x180s). Illustrates the apparent success of the long exposure (8s) flat frames collected using neutral density filters with the LCD panel.

M1 (Crab Nebula) completed in Astro Pixel Processor 2.0 beta 26 using light frames from 11/24, 11/30, 12/06, 12/22 and 12/30/2024 with flats (and dark flats) created with the neutral density filters + LCD panel on 12/27 and 20x100s darks and bias frames collected on 12/27/2024. 184x100s (18400 seconds total) integration. 

1/4/2025 -- following snowfall during the day, skies cleared a bit ... to marginal levels after midnight. Pseudo sky quality metric from the Indi Allsky backyard camera of ~6100.  Good enough to enable some testing.  Camera temperature for the Sv405cc showed -5C... no cooler was employed.

Seeing forecast from meteoblue for 01/04-05/2025 showing some low clouds (~13%) around 3am... with some (8%) at 2am as well. No moon, however.

With guide errors of ~1.5-3" RMS routinely, was curious if improvements were to be found. Immediately after starting to investigate... noted that that within KStars/Indi/Ekos the focal length of the guide scope is set incorrectly.  The Meade LX10 OTA uses the Celestron 9x50 RACI as the guide scope.  Initially this was set inside Indi at 147mm.  In reality, this should be 228mm focal length (the "9" in 9x50 == 9 inches == 228mm).

Even worse, however, is the fact that within PHD2 on the Odroid XU4, the focal length of the guide scope was unchanged since the Orion ST80 OTA was last used... which employs the venerable Cosmicar CCTV lens with a tiny focal length of 25mm.  Within PHD2 this causes the polar alignment error to be greatly inflated for every session since last winter.  Actual polar alignment errors are significantly smaller than PHD2 has been reporting because this value was not adjusted.  

With the correct value, PHD2 reports a polar alignment error of ~-9.5"... which would be 10x if left at the original value for the guide scope focal length. Additionally, the RA and DE graph within PHD2 was wildly variable because of this error ... suggesting serious problems with the mount (which were not in fact present). Doh!  Fortunately, an easy configuration change... and voila... much improved results in PHD2. TBD if the mis-configuration led to issues in the internal guider in Indi.

Indi internal guider calibration was successful, but guiding results showed only slightly lower guide rate errors of ~2" RMS.  Better than 3-3.5" RMS seen previously... and perhaps in line with the update from 147mm to 228mm (vs 25mm for the erroneous guide scope focal length within PHD2).

Set up a schedule for M1 (although now fairly little time remains for collection) just to see what the resulting light frames might look like and if any differences in quality were apparent. Also dropped the exposure time per frame down from 100s to 60s to see if star bloat / saturation is improved.

41x60s light frames were collected over night at gain of 285 and offset 24.

1/5/2025 -- Additional night of reasonable skies with diminishing cloud cover and the moon at 33% but setting at 23:31. Performed a polar alignment with the new / fixed configuration setting in PHP2 on the Odroid XU4 and brought the alignment error below 1".  Noted improved internal guider RMS errors <2" ... likely as a result.  Also set binning to 2...which resulted in lower latency...and may have also improved guide results below 2". 

Phdlogview chart for 01/05/2025 starting at 00:08:38 (23 minute guide session) showing total RMS guide error of 1.81" (0.27 px) with 1.15" in RA and 1.4" in DEC.  For 12/31/2024 (for a session lasting 53 minutes starting at 002:23), this was total RMS of 3.07" (1.34" in RA and 2.76" in DEC).

Set up M1 capture in the scheduler with the same 60s exposure (and gain of 285 and offset = 24) as last evening. Left the cooler off on the Sv405cc since temperatures were below freezing (-7.3C at midnight). Pseudo sky quality metric numbers were similar to last evening (~6500 around midnight) which is not great (~4000 would be very good) likely due to the moon and some cloudiness.

Seeing forecast by meteoblue for 01/04-05/2025 showing minimal clouds and the moon set by 23:30.

Seeing forecast from cleardarkskies for 01/04-05/2025 emphasizing poor seeing this evening/morning.

Reviewed light frames in Astro Pixel Processor -- removing many with guiding errors.  Ended up combing a total of 175x100s with 125x60s frames.

M1 (Crab Nebula) completed in Astro Pixel Processor 2.0 beta 26 using light frames from 11/24, 11/30, 12/06, 12/22, 12/30/2024, and 01/05/2025 with flats (and dark flats) created with the neutral density filters + LCD panel on 12/27 and 20x100s darks and bias frames collected on 12/27/2024. 60s darks and bias frames with gain 228 and offset 32 were collected on 01/05/2025. 175x100s and 125x60s light frames (25000 seconds, 6.9 hours total) integration. Calibrated star color in Astro Pixel Processor...which produced a blue hue. Binned 2:1. Astrometry results here show 1.33"/pixel.

Astro Pixel Processor analytical results for each session sorted by quality from lowest to highest illustrating some big relative changes in sky background intensity from the earlier sessions to the last session on 1/5/2025.

M1 (Crab Nebula) completed in Astro Pixel Processor 2.0 beta 26 using light frames from 11/24, 11/30, 12/06, 12/22, 12/30/2024, and 01/05/2025 with flats (and dark flats) created with the neutral density filters + LCD panel on 12/27 and 20x100s darks and bias frames collected on 12/27/2024. 60s darks and bias frames with gain 228 and offset 32 were collected on 01/05/2025. 175x100s and 125x60s light frames (25000 seconds, 6.9 hours total) integration. Did not perform star color calibration in Astro Pixel Processor. Unbinned (1:1). Astrometry results here show 0.668"/pixel.

M1 (Crab Nebula) completed in Astro Pixel Processor 2.0 beta 26 using light frames from 11/24, 11/30, 12/06, 12/22, 12/30/2024, and 01/05/2025 with flats (and dark flats) created with the neutral density filters + LCD panel on 12/27 and 20x100s darks and bias frames collected on 12/27/2024. 60s darks and bias frames with gain 228 and offset 32 were collected on 01/05/2025. 128x100s and 91x60s light frames (18380 seconds, 6.9 hours total) integration. Did not perform star color calibration in Astro Pixel Processor. Unbinned (1:1). Astrometry results here show 0.667"/pixel.

M1 (Crab Nebula) completed in Astro Pixel Processor 2.0 beta 26 using light frames from 11/24, 11/30, 12/06, 12/22, 12/30/2024, and 01/05/2025 with flats (and dark flats) created with the neutral density filters + LCD panel on 12/27 and 20x100s darks and bias frames collected on 12/27/2024. 60s darks and bias frames with gain 228 and offset 32 were collected on 01/05/2025. 128x100s and 91x60s light frames (18380 seconds, 5.1 hours total) integration. Did not perform star color calibration in Astro Pixel Processor. Binned 2:1. Astrometry results here show 1.33"/pixel.

Examples of four processing approaches in Astro Pixel Processor for the M1 collection. Top right: 25000s binned 2:1. Top Left: 25000s unbinned. Bottom right: 18380s binned 2:1. Bottom left: 18380s unbinned.  Illustrates that binning 2:1 does not impact the overall quality (which is expected given 3" of seeing typical...if not 5"). Also suggests that fewer, higher quality light frames are likely better in this case than more light frames with broader quality.  Between the four, some differences like contrast and saturation based on the fact that unique cropping, remap (brightness, saturation, black point, contrast) and light pollution removal is done on each within APP session individually.

Re-processed the M1 frames in Astro Pixel Processor across a range of quality metrics (everything not obviously problematic from a guiding perspective with 25000s integration to cutting the number of frames dramatically to higher quality levels with 18380s). The re-integrated with a variety of binning factors (1, 0.5, 0.3, .1).  Curious if binning in Astro Pixel Processor would clearly result in better signal to noise... not so much, however (even at a factor of 0.1).

01/09/2025 -- moon at 73% leads to switching to the dual narrow band filter to see if Ha/OIII is usable for M1.  Skies are very mediocre (to include the moon) and even after polar alignment <1" in PHD2, guiding is problematic (likely due to wind gusts).  Undaunted, set up a job in the scheduler after experimenting with exposure (tested 180s and 240s...went with 240s) as well as focusing (25s exposure) and plate solving with the narrow band filter in place.  Gain left at 228 with offset 24. Camera temperature for the Sv405cc was ~-6C...so no cooler was employed.

40x240s light frames captured overnight with the third batch of 20 failing to align at 02:45 (obstruction w/ the house).

Seeing forecast from meteoblue for 01/09-10/2025 showing relatively clear skies, poor seeing (as always) and significant moon in the sky.

Seeing forecast from cleardarkskies for 01/09-10/2025 which is basically off-scale wrt seeing due to cloudy conditions (a bug?)

Indi Allsky backyard camera analysis for 01/09-10/2025 in "moon mode" with elevated but steady pseudo sky quality metrics. After the moon set (~04:20...two hours after the end of collection), these numbers dropped to ~8800 (which is not great).

Two hours of Indi/Ekos internal guide data starting near 23:00 through 00:45 illustrating a total RMS of  4.76" (1.10 px) with an assessed polar alignment error of 0.1'. Guiding in RA and DEC was often within +/- 2.5" but large excursions (especially in DEC) were common. 

01/10/2025 - Collected flat frames (since the UV/IR filter was replaced with the narrow dual band Ha/OIII filter) using the LCD light panel on top of the framed stack of neutral density filters. Utilized 9s exposure with gain of 224 and offset of 24 after double-checking levels in ASTAP.  Then covered the mount and collected flat darks to match.  Then collected dark frames at 240s with similar gain and offset with the temperature of the camera set to -5.8C to closer match conditions overnight with ambient temperatures (light frame 001 with temperature of -6.3 and 029 showing temperature of -7.2C).

1/20/25 -- Moon is finally starting to diminish...clouds are gone for the moment... and the snow is cleared off the deck.  Started M1 job with the dual narrowband filter still in place (with a set of calibration files (including flats with this filter) collected earlier on 1/10/25).  Had re-balanced the mount on 01/10 after having added the micro-focuser to the back of the OTA...so hoped for better guiding.  However, because of the cold, only made a few adjustments during polar alignment and ended up at ~4-5' polar error (not great). RMS errors during guiding likely reflect this (nearing 5"...but later settling down <2"). With clouds expected near mid-night, the race was on to align and configure the mount to collection a few frames for M1 at 180s.  Temperatures exceptionally low (reading -12C on the Svbony 405cc).  

Updated hardware on the Odroid C4 with the addition of the venerable Navisys GR-701W u-blox-7 usb pps gps&glonass receiver.  Sadly the C4 kernel shipped by hardkernel does not include the module for pps_ldisc (only ships with the GPIO pin support)...so the kernel would need to be rebuilt to create the needed module. In any event, configured the indi GSPD driver and it works successfully.  On the indi Allsky raspberry pi system configured rtl_433 software-defined radio to receive packets from a new Accurite 3N1txc which measures wind speed, temperature and humidity for hyper-local weather at the OTA. Configuring weewx to provide weather reporting from this using the SDR driver. 

Seeing forecast from cleardarksky for 01/20-21/2025 showing increasing clouds after 1am.

Seeing forecast for 01/20-21/2025 from meteoblue with similar grim forecast for incoming clouds (and moon -- double whammy).

Astro Pixel Processor result on M1 (Crab Nebula) from dual narrow band collection on 01/09 and 01/20 with a total of 7860s integration (mix of 240s and 180s lights).

1/23/25 -- no moon (until 2am) and few / no clouds / wind motivated an additional night of dual narrow-band filter collection against M1 (and later Abell 21, the Medusa nebula). Indi Allsky background camera showed pseudo sky quality values starting at 10k and dropping to 8k (not great).  Updated polar alignment using PHD2 drift alignment running on the Odroid C4. Two adjustments brought this below 1'.   But when capturing M1 light frames, noted significant (+/- 2.5-3") swings back and forth in RA with the internal guider.  Switched collection to using PHD2 on the Odroid through Ekos/Indi.  This resulted in immediate improvement to guide error...likely due to improved latency (exceptionally poor wifi performance this evening?)  Errors went from typically 3" RMS to below 1".

Seeing forecast for 01/23 - 24/2025 from meteoblue suggesting clear skies but poor seeing expected.

Seeing forecast for 01/23-24/2025 from cleardarkskies suggesting poor seeing but dark/transparent/cloud-free skies.

Only 4 light frames from Abell 21 were collected.  And these suffered from severe focus problems as the autofocus routine was not converging with the settings employed. None of the four were useful. Six light frames for M1 were collected...but these suffered the same out of focus problem.

01/24/25 - Repeated last evening's experiment running PHD2 on the Odroid C4 against Abelll 21 (Medusa nebula) rather than using the internal guider within Indi/Ekos running on the main server (Valinor).   Skies started usable (barely) at 19:00 but clouds rolled in by 21:00.  Although light frames are unlikely to be useable, the goal is to examine guiding logs to see if performance using PHD2 directly on the Odroid performs consistently better. 

Seeing forecast for 01/24-25/2025 from meteoblue suggesting reasonably good skies after 02:00.

Seeing forecast for 01/24-25/2025 from cleardarksky showing relatively good skies after 01:00 (European model)

Indi Allsky Backyard camera view for 01/24-25/2025 showing clouds early but then clear slkies 

PHD2 guide log file and summary stats from 01/24/2025 from ~12:45 to 01:18 on the 25th showing dramatically lower guider errors when using PHD2 directly on the Odroid C4 connected to the mount and Zwo ASI120MC-S guide camera through the 9x50 RACI scope. 

Second guide session from the early morning of 01/25/2025 showing much higher errors in declination than the earlier session but overall stats showed RMS error still bellow 2". Polar alignment error for this session was reported ta 2'.

In examining the resulting light frames in AstroPixelProcessor (of which at least 15 were collected), the focus issue was largely fixed, but ultimately only five were kept for integration due to guiding / tracking / star shape issues.

01/25/25 - Repeated last evening's experiment running PHD2 on the Odroid C4 rather than using the internal guider within Indi/Ekos running on the main server (Valinor).   Skies started usable (barely) at 19:00 but clouds rolled in by 21:00.  Although light frames are unlikely to be useable, the goal is to examine guiding logs to see if performance using PHD2 directly on the Odroid performs consistently better. 

Seeing forecast for 01/25-26/2026 from cleardarkskies suggesting grim outlook with clouds past midnight.

Seeing forecast from meteoblue looking grim from the start with high level clouds predicted in the early evening and heavy clouds rolling in after midnight.

In examining the resulting light frames in Astro Pixel Processor, it's clear the settings used by the focuser in Indi/Ekos did not result in successful iteration to a clear focus with donuts visible in brighter stars.  While useful to track guiding performance, sadly these light frames are not useful for the Abell 21 image.  For the dual narrow-band Ha/OIII filter, autofocus performance is a challenge.

01/26/2025 -- With a forecast of no moon until 06:16 and clouds not expected until midnight, started collection on M1 and Abell 21 in the scheduler. Still using the dual narrow band Ha/OIII filter and guiding is configured to run on the Odroid C4 with PHD2.  Did not have a chance to double check polar alignment...but guiding errors remain low (generally less than 1" RMS). Occasional spikes in RA appear in the guide data, however.  Still configured for 180s lights with gain set to 228 and offset 24. Noted that the initial frames in the focuser were binary (black & white)... and odd issue that's cropped up with the first frames from a session with the Sv405cc camera.  This led to poor focus results from 19:00 to 20:42 when the job was restarted and the OTA refocused.

Seeing forecast from meteoblue for 1/26-27/2025 (in this case, actually, starting at 02:00 on 1/27) showing 50% low, near 80% mid and 60% high level clouds expected. 

Seeing forecast from cleardarkskies showing similarly grim skies after midnight (with the European model suggesting clouds all night). 

Indi Allsky backyard camera pseudo sky quality metric and star count for 01/26-27/2025 (from 20:30 through 24:20) showing fairly reasonable skies (jSQM ~8000) with just a couple minor cloud bands passing through.

Indi / Ekos focus module results from 20:42 on 01/26/2025 iterating towards good focus. Prior to this, the focus module had converged at 19:15...but the camera was providing poor frames leading to a bad result for light frames collected before 20:42.

Light frames from the 26th were the best from all the Abell 21 collection with dual narrow band filter with 19 kept for integration.

01/27/2025 - With no moon, switched to broadband capture by removing the dual narrow band filter and inserting the UV/IR cut filter in front of the Svbony 405cc camera.  This will require new flat frames to be collected for calibration.  Cloud free skies expected until ~02:00.  Changed focus and plate solving exposures down and updated the light frame capture settings back to 60s exposure times with broad band imaging.  Set the scheduler for Abell 21 and NGC 4248 (which captures M106).

Seeing forecast for 01/27-28/2025 from meteoblue showing dark, clear skies until ~02:00.

Seeing forecast for 01/27-28/2025 from cleardarksky with a similar prediction.

01/28/2025 - Good skies, no moon, new target (NGC 1931 - Fly nebula as well as M106 (centered erroneously on NGC 4248)).  Set up NGC 1931 in the scheduler and left for Lunar New Years concert at Strathmore. Came back to find 140+ light frames. Excellent guide errors persist... illustrating the apparent impact of running the guider on the embedded computer vs driving guiding from the main computer with added latency.  Errors were < 1" RMS consistently. Given the switch to broad band imaging, new flats frames will be needed for calibration from collection from yesterday and today.  The Indi Allsky backyard camera showed pseudo sky quality metrics ranging from ~6300 at 22:45 down to ~5900 at 23:30. 

Seeing forecast for 01/28-29/2025 from metoblue showing some clouds expected near midnight then clearing out. And no moon.

Seeing forecast for 01/28-29/2025 from cleardarkskies with a similar rosy picture for this evenings views.

1/29/2025 - NGC 1931 and M106 loaded in the scheduler, but a glitch with Kstars having the time reset to view the sky at 03:00 (from yesterday) blocked the schedule from running in Indi/Ekos.  Once restarted at 22:11, took advantage of clear skies (at least before midnight) and no moon for broad band imaging.  Other than the initial glitch, automated collection from the scheduler proceeded normally with focusing, plate solving and guiding. Guide errors followed the familiar trend of just below 1" RMS. Pseudo sky quality values started at ~6800 dropping to 6050 by 23:20. The Weewx weather station monitoring the Acurite Notos 3N1TXC Wireless Weather Sensor (anemometer, temperature and humidity gauge) showed no wind with outdoor temperature 36F.

Seeing forecast from cleardarksky suggesting exceptionally bad seeing through 06:00.

Seeing forecast for 01/29-30/2025 from meteoblue showing ~15% low and mid-level clouds expected from midnight through 06:00 which may impact the quality of light frames.

2/1/2025 - Experiment -- Kstars on the Odroid C4.  Clear-ish skies expected for the evening / early morning and no moon... so capturing Abell 21 (Medusa Nebula) and M106 but this time running the entire capture from the Odroid C4 over remote desktop.  Ran a drift alignment with PHD2 on the Odroid C4 and noted ~-15" of error... with one lucky adjustment reduced this to ~-2" of polar alignment error.  Started Kstars / Indi / Ekos on the C4 and after some work on configuration, was successful in starting a job collecting 60s light frames on Abell 21.  Set the Svbony 405cc to -5C with the cooler enabled (ambient temperatures ~0C). Set the offset to 24 and gain to 228.

Using the internal guider and note disappointing guide error greater than 2" RMS from the start... which is significantly larger than the guide errors running PHD2 on the Odroid C4 with Kstars on the main computer.  Noted on the anemometer on the AcuRite Notos (3-in-1) Weather Sensor (via the weewx JSON feed) showed wind speeds of ~0.7 MPH (not zero).  

Near 01:00, found the polar alignment off when alignment via plate solving repeated struggled...fixed this with a new drift alignment in PHD2 and discovered the dew shield had fallen off at some point.  After realignment, the internal guider was initially failing. Parked and restarted capture for Abell 21.  After 2am, noted obstructions and switched collection to M106 (centered on NGC 4248).  However, Kstars running on the Odroid C4 died when the initial alignment failed...ending the scheduler & observations for the night.

Initial result after culling through light frames from broadband (60s) and narrow band (180s) collection from Abell 21 (Medusa Nebula) in Astro Pixel Processor 2.0 beta (2.7 hours total) with a faint start of the nebula showing. 

2/4/2025 - Running KStars on the Odroid. Switched over to dual narrow-band Ha/OIII filter with the moon high in the sky at 40%.  Made adjustments to exposure time and gain for the focus and alignment modules.  Set light frame capture to 180s at gain 228 and offset 24. Set the cooler for 0C (started at 8C). Clouds expected near mid-night. No wind...outdoor temperatures dropping rapidly from the low 60's during the day to ~40F by 20:00. Pseudo sky quality metric from the Indi Allsky backyard camera was ~18,500 at 20:45 (not good) in moon-mode (limited to 75% camera gain). So definitely a marginal night for imaging.  Good opportunity to experiment with focus parameters, however.   

Started adjusting parameters in the focus module attempting to get a working set up with the narrow band filter in place. Tried 45s exposure, full frame (all stars). Then tried 60s.  Moved back to 45 with a sub-frame (256 pixels in size).  The moved back to full frame, but configured SEP to use "bigstars" for focusing. Initially, this seemed to help ... given the noise (propensity to pick a bazillion non-stars)...by drastically reducing the number of detections to star-like things. But in close inspection, the detections did not look much improved. Switch to binning 4x4. 

Finally, started guiding and set the exposure ot 80s, 2x2 binning and back to all star detection /extraction.  With this kind of autofocus procedure... best to do this just once at the beginning because it's too time consuming to do in between each job. Did note the guiding errors were ~1" RMS (not bad)...especially w/o updating or checking the polar alignment. Ultimately, 13x80s iterations were required for focusing... beginning at 22:05 and ending at 22:35.

Configured for 45s with gain of 285, the alignment module was successful in plate solving...getting to within 28 arc sec of M1 with 3 iterations.

Seeing forecast for 2/4-5 2025 showing clouds expected near 23:00...but later on the European model.

Seeing forecast from meteoblue for 2/4-5 2025 which agrees with cleardark sky by is definitely not using the European model.