PEC traning question.

[MilwaukeeLion]

Is there an optimal exposure you should train PEC at or is it based on seeing conditions similar to normal guiding, setting exp so as not to chase?  Or should you use different exposures for doing a trend?

Thank you for any help..

 

[Oddsocks]

My experience of doing this on a Paramount was that it made little difference what the actual exposure time was only that I had very good seeing.

The first attempt at PEC data collection I chose the brightest star available at Dec zero, binned the image to improve sensitivity (4x4), set a subframe 64x64 pixels to reduce download time and grabbed as many shots as possible, approx 30 data points per minute. The seeing was only middling.

The PEC algorithm in TheSkyX reduced the data to a smooth correction curve and ignored several outliers anyway but good seeing proved more important than data points as I found on this attempt which made PEC worse with 5 arcsec P-P before correction and 14 arcsec P-P when the correction curve was applied.

A few weeks later I ran the PEC data collection again but this time on a dimmer star so a longer exposure was necessary and fewer data points were collected, only 20 data points per minute, with good steady seeing on this attempt the correction curve computed was perfect and gave me a P-P under 1 arcsec and unguided imaging at 1200mm fl, this lasted for a few years right up until the RA worm block failed with a faulty motor encoder.

With the new RA block installed a new PEC curve was created but this time I had a different camera mounted that could not sub frame and even binned 4x4 the download time was around 10 seconds per image, only 6 data samples per minute were collected where before I had 20 data samples per minute but the resulting PEC curve computed (under almost perfect seeing) was just as effective as the prior one at less that 1 arcsec P-P, I even managed some unguided slit spectroscopy at 1800mm fl.

After a worm gear clean and regrease last September I made a new PEC curve with the same camera in not brilliant seeing and the correction curve was useless at ~8 arcsec P-P so as an experiment I reloaded the previous correction curve made under better seeing conditions and the residual PEC was back down below 1 arcsec P-P.

So I would say seeing is the number one factor in creating a good PEC curve. Short exposures in bad seeing is just going to create a load of misplaced data points and lengthening the exposure in such conditions would create distorted star centroids which would also be poor indicators of star position.

As always, YMMV as they say.....

[ollypenrice]

Oddsocks' reply accords exactly with what I would have expected so, if you don't have a night of superb seeing to help you out, using long guide subs to average out the seeing. This ought to give the best possible result. What PEC doesn't need is a mass of very short term corrections on a scale unlikely to arise from repeatable mechanical sources.

Olly