All Sky Camera Experiment

At the IMT we were experiencing stability issues with the ZWO ASI120MC USB3 camera on the MAC/NUC so we decided to move it to it’s own Raspberry Pi4 (4GB) as a cheap experiment.

The Raspberry Pi4 has the advantage of improved networking and connectivity including USB3. One of the downsides of the RPi4 is the increased heat generated by the CPU which is mitigated by housing it in a FLiRC passive case to dissipate the heat and prevent thermal throttling.

Additionally in order to prevent SD card wear and improve the I/O throughput we configured the RPi4 just to boot off the SD card but serve the rest of the file systems from a SSD connected via USB3.

AllSky Camera Software

Using the open-source AllSky Camera project for ASI camera support we gain some advantages. The main advantage is that it’s open source (served via Git), provides an admin and public web portal which is addressed locally as http://allsky.local and the software can be set to start on system startup.

Admin Web Portal

The allskycam software can restart on system reboot but it can also be stopped/restarted via the web admin portal.

The configuration is simple and the mode/behaviour of the software can be controlled via the GUI or via a settings.json file from the command line.

Admin Portal – Camera Settings

I recommend you take the time to create a dark reference image, again this is done via the GUI panel as long exposure frame will show extensive hot pixels.

RGB24 image hot pixels with no dark applied

Non Admin Web Portal

This is accessed via the address http://allsky.local/allsky-website and provides the latest captured image, a constellation overlay and the ability to view time-lapses and star trails.

Local network web site for non admin users

One of the nice features of the software is it’s ability to create a time-lapse of acquired images for the previous night. The following video shows the debayer issues as well as the dewing of the dome cover.

Time Lapse of evening showing dew issues/grayscale bayer

Bob is due to connect up the resistor ring he’s placed around the camera which hopefully should fix the dome dewing we encountered on the first night.

So it remains for us to fix and implement the following :

  • Resolve the lack of colour images (RGB24)
  • Image quality breakup
  • Resolve dark daytime images
  • Secure copy latest image/time-lapse to an external public website
  • Scale VirtualSky constellation overlay (180 degrees) down to 150 degrees

Viewing Report 19th/20th October 2019 – IMT3 Observatory

Viewing time period – 17:31 – 01:48

After nearly a month of not imaging from IMT due to a holiday in Tenerife, a week in New York with work and then Manchester and London along with a run of poor weather it was clear on a Saturday night! Another evening commissioning the observatory was needed, so tonight we will again further refine the polar alignment since the last major modification and distribution of weight where we changed the adjustment plate for Bob’s Tank FS 102 OTA. Again we plan to drift align with PHD.

First thing is to find a star near the celestial equator near the Meridian so that it would display the most movement and thus magnify the error of miss polar alignment. I should be able to find HP 95501 @8pm.

Star to drift align for Azimuth

Next I performed an autofocus using the Luminance filter. HIP 95501 is the star to choose for drift aligning the first part, a 1 second image within Frame and Focus in SGPro showed it just off centre which was fine.

Star to guide on

Next I moved the star to place in the Lodestar FoV. Now I can measure the azimuth polar error, ALWAYS ignoring the RA line. Looking at the Dec line I could see I was out by 2.43′ and 39px. I adjusted the thruster knobs on the MEII to move the star to the outset edge of the purple circle showing the error, in this case the right thrust in and left thrust out. I then drifted again and make sure the purple circle gets smaller and the DEC line a much shallower angle.

After first adjustment 0.53′ and 34px out

I adjusted again and got the azimuth error down to a respectable 0.08′ 5px error.

Azimuth error 0.08′ 5px

The graph on PHD2 should started to look fairly flat, and so I then attempted to fix the polar error for altitude. I selected a star in the West and near the celestial equator such as Rasalgethi in Hercules.

Rasalgethi used for Altitude adjustment

I watched the DEC line only and ignored the RA, the DEC line this time reflecting the error in altitude. Then I adjusted the mount using the altitude adjustment spanner moving the star again to the outside of the purple circle and then retested, finally getting the error down to a suitably small number after only 1 turn of 0.12′ 5px error.

Altitude adjusted to 0.12′ 5px error

Unfortunately at the extreme West I could only expose unguided for 1min on the 12″ at 2.5m focal length, the stars otherwise looked trailed.

Trailed stars at extreme due West

I then went to near the meridian and a 4 min exposure produced nice sound stars.

4 minute exposure near Meridian

I then went on to do an automated TPoint run, but the problem seemed to be that a large number of samples could not be solved. The resulting TPoint model of 118 samples of which only 90 were usable, was worse than the 60 point model I had before. I will therefore redo the TPoint model the next time it is clear.

TPoint model not as great at 60 points I did before

The whole 118 model took approximately 1.5 hours to complete which is so much better than a manual model.

Completed model 118 points

The good thing is the TPoint model told me the polar alignment is excellent!

Polar Alignment is excellent 🙂

It is now @23:30 so I went on to start imaging. First I needed to perform a focus run on Luminance which I did.

Good focus on luminance

I then slewed to M76 to start my image run, a 30s exposure showed stars in focus and little dumbbell prominent in OIII.

Focused stars and M76 centred

This time round I decided to set the Gain to 139 and the Offset accordingly to 21. I also decided on a 10min exposure rather than 20mins front the last set.

M76 1 x 10min Gain 139 Offset 21