SGP Focus Settings

[Atmos]

I've been trying different software packages for semi-automation so that I can at least get some sleep at night! Over the last two weeks I've tried CCDAP which constantly crashes on my new cheap imaging laptop so that's out of the question.
ACP does a pretty reasonable job but a bit more than I'd like to spend at this stage.
I've got SGP so I figured I'd give it another shot. Previously I'd tried using the same focus technique as MaximDL which would give a nice U curve and was really accurate. Because SGP uses the HFD of stars all over the place it seemed to get a bit confused with stars of various sizes came into play which would make a jaggered curve.

Last night I decided to do the same kind of thing that FocusMax when calculating its V curve, work with a FWHM of 40-50 on either side of focus and it was very accurate. So I did 3x3 binned (saves a LOT of time on a slower computer and a 16mp camera), 15 points and large steps. The seeing was shocking last night (best I got was 4.4") so I'm not sure whether it was any better or worse than any other method.

What do others find works well?

[DJT]

I don't bin for focussing. focus max suggest binning for coarse focus but 1*1 for the final touch. If you are trying to setup so you can sleep whilst you are imaging then what are you trying to save time for?

[Atmos]

SGP is the only focusing software I've ever considered binning on, FocusMax does a great job but I cannot use it while using SGP.

At the moment in SGP I am doing a 13 point focus, doing it unbinned each 1s exposure takes a good 10s for SGP to calculate the HFD of the hundreds (sometimes a few thousand in the case of ETA Carina) of stars in the image. 3x3 binned it takes 2s.

[Spookyer]

I have used SGP for a couple of years now. I have had the occasional issue but generally focus results have been good with both refractors and now a RC scope as well. I generally use 7 or 9 focus points and 2x2 binning. You have to ensure your step size covers enough ground to get a decent curve. They have suggested ways of calculating that and then you may want to tweak it. I use the Luminance filter for focusing and have calculated offsets for the other filters. This is much quicker and more reliable in the long run especially when using NB filters.

[Slawomir]

I also use 9 points and 2x2 binning, but focus on each NB filter instead of using offsets. Exposures are longish - 15s, so the process takes a few minutes, but I do not refocus often, so it does not bother me. One day will certainly look at calculating filter offsets.

[Atmos]

If you want to calculate filter offsets, get trial version of FocusMax 4 and CCDAP or ACP.

[Slawomir]

Thank you Colin for your advice.

I might first try in SGP writing focuser position for each filter when it is in focus and then subtracting values to get offsets. Should have done it ages ago but was too lazy...if that does not work then I will look at other options. But I will wait for the arrival of my new scope before I experiment with offsets :D

[Shiraz]

same as Suavi - SGP, 9 point with 2x2 binning and restrict the size of the focus field for broadband (who needs 700 stars?). at f4 expose for 4 sec broadband and 10s through NB filters (may need to use full field for NB - not many stars). focus is real quick, so redo every 12 minutes (generally works OK, but may decide to use shorter updates if focus is likely to be drifting in rapidly varying temps)

use fairly small steps to get a u curve around focus, rather than a sharp v - seems to work best and the symmetry of the curve is a good visual check on collimation (which needs occasional tweaking).

Astronomik filters seem to be true parfocal - no offsets needed on this system.

[codemonkey]

I use 7-9 points (think I'm back to 7 now), with reasonably large step sizes. Making the steps too small gave me inconsistent results--it's easier for the line to bounce around a bit that way.

Key thing is making sure the backlash is dialed in. If the top of the curves don't match up, you've got backlash and should adjust it one way or the other until they level up.

I've also found that it gets confused when stars enter or exit the frame during the focus routine. It also frequently confuses nebulae with stars and that can mess it up too. I have a lot more success with it on galaxy shots where the galaxy rarely confuses the star detection algorithm.

[Slawomir]

Lee - I had the same occasional issues in the past. The new version of SGP has an improved algorithm for focusing and it works for me every time, and I only shoot nebulae in narrowband :-)

Brett- if the chip is tilted, even slightly so, then checking with a mask on a bright star that is on axis may give a slightly different focus point than SGP that averages HDR for many stars across the chip and thus finds optimal position that will/should yield on average the sharpest stars in the entire image.

[Atmos]

Well I just downloaded the latest version of SGP this afternoon so I'll hopefully have better luck with it :)
Might drop the number of steps to 9 and increase the step size to 150. Will go WAY out of focus on each side.

[Slawomir]

With 2.5" Moonlite I had step size about 30 at f/5.6 and @1.33 arcseconds per pixel. With new algorithm and 9 steps it always worked.

[Atmos]

Each step on mine is 1.27 microns so 1000 steps is 1.27mm.
What is your step size?

[Slawomir]

Moonlite's full step size is 0.00016" so approx 4 microns.

So perhaps step size of 100 would be adequate in your case? Maybe a bit less since your critical focus zone is narrower than mine?

[codemonkey]

Interesting... I did upgrade SGP recently, but I can't remember if I had the newer AF stuff prior to that upgrade. I think I did... I know I had issues with NGC 2070 recently with SGP detecting nebulae as stars. Could be user error too!

Colin I think they recommend going out to about 3-4x the best HFR when doing AF. i.e. if you find the best focus you can achieve reports a HFR of 1, then you want to make it so that the top of the v-curve is between 3 and 4. Step size of 100 is probably a reasonable starting point, just play with it until you get a ax HFR within that 3-4x range.

[Octane]

Psst, CCD Commander.

H

[barx1963]

So it is
Not just me! SGP stubbornly refuses to do AF on NGC2070. I ended up slowing away to run focus on each filter change then getting it to slew back each time. Tonight I am doing the crook and AF is just purring. To be honest I probably don't need to focus each filter change, just do it to make sure it is always as good as it can be. Must look into working out the offsets though!