New 2.4 Focusing algorithm - ideas


#1

I tried the new focusing algorithm several times last night and it works a charm. I typically use a 7-point array and if the focus point was off to the left side by more than 1 click, it extended by a few more points to be sure.

One idea though - if the focus point is off to the right by more than one click, it repeated the entire array, starting at the highest value. With narrowband filters, this takes an age.

My suggestion is, if the focus module has backlash enabled, why not just add the extra points on the right hand side without repeating all the others too? There should be no detriment to the overall result and be much quicker.

One other thought, now that the focus module will work over a wider field, it can be that the HFD calculation can get confused on the fainter intensity. Is there anyway of flagging that on the initial exposure of a run and increasing the exposure time to accommodate? (I recall FM does something similar)


#2

Thanks for the suggestions. We do not recommend focusing with NB filters. Please go through the process to find your NB filter offsets from your LUM filter and then perform AF with your LUM filter for all filters.


#3

Hi-

I have a related issue with backfocus and the focusing algorithm. I am using the primary focuser on an sct and it works well for me and is very repeatable - but it does require a fair amount of backlash unwinding for it to be repeatable. Unfortunately the focuser I am using is fairly slow and the time it takes to unwind backlash is about 50 seconds. It is inevitable that you need to unwind backlash at least once in a focusing run - but it is very important for me to keep it at a bare minimum. I am finding with the latest changes to the focusing that it seems to unwind backlash over and over as it makes small moves in the “bad” direction.

In addition, even the normal focusing routine will decide on what it thinks is best focus - and then immediately go there - even if I want to re-run the focusing. This means it needs to unwind backlash fully to the new (wrong) focus - and when I start focus again - it needs to unwind it again.

If the focusing works fine the first time then everything is fine - but I agree with the poster here that choosing to expand the focusing range in the “bad” direction may be a very time costly thing to do if there is a lot of backlash.

I’m not sure how unusual it is to have a lot of backlash to unwind - but it is a reality of focusing with the primary focuser and having a somewhat slow focusing system.

For manual focus runs, I would like it to display the value of the estimated new focus position and show it to me first before it goes to that position and then asks me if I want to re-run. For automated runs this doesn’t help - but any way to avoid multiple unwinds of backlash would be good.

Thanks,
Frank


#4

Thanks for the suggestion. That said, we have always been transparent about our roadmap and this, should we decide to do it, is not something we will likely look at for a while (it is pretty invasive to the current implementation).


www.mainsequencesoftware.com