Monday, August 4, 2014

Calibrating the Atlas Focuser

I have to recalibrate the new focuser:
  1. Autofocus parameters in SGPro
  2. Temperature compensation
  3. Filter defaults
  4. Filter offset
Before I could start, I noticed that it sometimes took several attempts to connect to the focuser. And later, SGPro would sometimes hang when it needed to move the focuser out for the first measurement. I asked on the SGPro mailing list - need to find a way to generate debugging information for the Atlas Focuser. Asked FLI about it.

Next, I wanted to calibrate autofocus in SGPro. But the very small resolution of the Atlas focuser means that even with a step size of 1,000, I have to take more then 20 (better 30) data points to create a good V-curve. Unfortunately, SGPro doesn't allow step sizes greater then 1,000. I asked on the mailing list and the maximum will be raised to 10,000 in the next SGPro release. Yei!

So, for now, I have to use a step size of 1,000 and 31(!) data points to get a good V-curve. Which of course takes quite a while :-(

While I was playing with the step sizes, I also played with binning. And I noticed that there isn't much of an accuracy improvement from binning 1x1 to 2x2. But of course, the exposure times are going down significantly. But when I set binning to 3x3 or even 4x4, the accuracy in the focus region gets very bad. So, binning 2x2 is what I will now be using.

With regards to temperature measurement, the Atlas focuser has 2 measurements: internal and external. The internal one looks much more reasonable (the external one reads very high). So, I'm using that one for now, but also asked FLI what these two temperatures are about.

For the next night, I created an SGPro sequence to do the focus measurements. In the autofocus settings I set "Refocus after each frame", so that SGPro would run the autofocus routine every time. As SGPro writes the temperature and the focuser position into the FIT file, I could run this sequence all night and evaluate it the next morning.

1. Sequence
8 Luminance images
3 Red images
3 Green images
3 Blue images
3 Ha images
3 SII images
3 OIII images
(I set this sequence to end at 3:30, so that I would take another series of Luminance images)

2. Sequence (to be started at 3:30)
8 Luminance images.

All images are 0 second exposure time.

With the Luminance images at the beginning and the end of the night, I can determine the temperature coefficient of my focuser. With the other images, I can determine the focus position differences. With the temperature coefficient I can then calibrate them all to the same temperature.

The next morning I noticed a) it wasn't enough time to capture all filters. The narrowband filters take 30 second exposures. With 31 exposures that's 16 minutes per images!

And b) it also got cloudy at the end of the night and I only had 5 luminance images at the end of the night.

Nevertheless, I think I had enough measurements for the temperature coefficient:

Evening MeasurementMorning Measurement
Temperature Difference3.71
Focus Position Difference1673
Focuser Steps per 1 degree451

So, 451 steps per 1 degree (which shows the small resolution of the Atlas Focuser - the Robofocus focuser had 15 steps per 1 degree!!!)

Next, the individual filter measurements.

From the Averages (you can see all measurements in this spreadsheet):

TemperatureFocuser Pos.TemperatureFocuser Pos.TemperatureFocuser Pos.TemperatureFocuser Pos.
Diff to Lum Temp0.53910.72661.1641
Adjusted Pos243822426923874
Diff to Lum Pos413526921
So, these are my filter offsets and filter defaults for the RGB filters. Next, I need to take the defaults and offsets for the narrowband filters. But I'll do that after the Oregon Star Party (where I will use the LRGB filters only).

In the next nights, I will still focus before each frame to compare how well the temperature compensation works.