SGPro Autofocus help needed

[iapa]

26 minutes ago, steviemac500 said:

Correct me if I’m wrong, but won’t there always be issues with the hitec DC focus as it’s not a stepper motor?

I have used the HitecAstro controller with the skywatcher motor to good effect for a couple of years.

When I say that, I mean I managed good focus, to my aged eyes, with my  200PDS and 10" quattro, not that I got good images LOL.

[RayD]

37 minutes ago, Pompey Monkey said:

I've just run a sequence of auto-focusing: the entire spread of solutions was 28 steps for all my filters (LRGB Baader, Ha, OIII and SII, 3 nm Chromas).

It seems that if the run starts too close to the bottom of the curve, i.e. what would give the classic "hockey stick" profile, then the data fitting algorithm that the software uses doe not recognise the "V".

It then does its best to please and moves the focuser out in order to find the correct profile. Unfortunately, it goes too far, with insufficient overlap of the previous sequence to relocate the inflexion point on the curve. It then tries another move out before retrying and after that there is, of course, no hope of recovery.

Longer exposures and finding just the right step size (12) seems to help, but one swallow etc....

I think I'll message the devs to request widening the the search range rather than just shifting it to the right.

Sure, it might take longer to auto focus that way, but at least it won't shouldn't bomb out!

Ah I see you're using Baader and Chroma.  It may be that you will need to focus using the Lum filter and then apply offsets as switching between them could result in your starting focus point being quite a way out.

I would say just check you settings and step size (I have a few videos with a couple on SGP auto focus here) as step sizes need to be right and can make a huge difference.

[Pompey Monkey]

1 hour ago, RayD said:

Ah I see you're using Baader and Chroma.  It may be that you will need to focus using the Lum filter and then apply offsets as switching between them could result in your starting focus point being quite a way out.

I would say just check you settings and step size (I have a few videos with a couple on SGP auto focus here) as step sizes need to be right and can make a huge difference.

Good video on determining step size. I guess that the 3-5 HFR value comes from the limits of still getting measurable stars?

 

[RayD]

2 minutes ago, Pompey Monkey said:

Good video on determining step size. I guess that the 3-5 HFR value comes from the limits of still getting measurable stars?

 

Thanks Paul.  I think it is just enough to allow the algorithm to calculate sufficient change from the in-focus point to determine the lowest value.  Often people don't get this right but it needs to be done to get a consistently good 'V' curve.

[Jbro1985]

On 02/11/2018 at 17:26, RayD said:

That's odd.  Do you pause your guiding during the flip as that can cause issues here?  Also I suppose if your NB filter is a totally different type to you Lum, then this also could be a problem.  In the latter case I suspect you would need to work out and apply your filter offsets so you focus in LUM and SGP then just applies the correct offset for the imaging filter.  It takes a while to set up but can help in these instances.

I can't remember exactly but I think it is only about 40 or 50 odd steps.  The backlash on the Lakeside I've found to be pretty low, but it won't do any harm to go bigger, so long as you can confirm it is taking out the play.  I did mine with a long cable tie around the focus knob so I could see it moving easily, and then added some.

I had an issue with plate solve not working great after the flip recently and I sorted that by increasing mount settling time setting then the issue on focus arose.  If you haven't, use the settings to pause guiding as above.  Flips (for me) are the biggest reason for failure (I automate and go to bed - perhaps that's the underlying biggest issue !).

Anything to help the sequence post flip routine is a win.

[RolandKol]

Hi guys, 
I have a similar question in relation to SGP autofocus, - I did not want to create a new topic and hope you will advice.

I recently finished my DIY autofocuser with Nema 17 PG27 motor. My calculations and measurements show that one motor step lifts focuser only by 2.42 microns (I able to do it even smaller). 

My telescope SW 130PDS (with SW CC) has a critical focus range in amount of 49 microns. So my motor is able to do 49/2.42 = 20.24 (rounding to 20) steps in the critical focus range.

SGP uses Best Median focus calculation, so it needs 3 best focus points to determine the best one. Initially, this fact made me think SPG step size setting should be 20/3 = rounding to 7.

At the moment I am keeping 9 Autofocus data points. However... with SGP step size 7, - autofocus routine produces quite flat graph and SGP is not able to calculate the mean value.

If I do as SPG suggests, - lets say I have 1 HFR value after bahtinov mask focusing routine (also using motor, but controlling it manually and using SharpCap focus aid).

So my best focus position equals  1 HFR and motor position at 5000, later I get value of 4 HFR while motor is in the position 6000.

SPG step size calc: (6000-5000'motor_position) * 2 / (9'datapoints-1)=250.

I tried this value and here the fun begins...

The first data point SGP creates by adjusting motor out of focus adding to 4*250 to starting position 5000, and reduced it 4 times back to 5000. At 6000, it gets expected HFR value in around 4.

After, it slews focuser back by 250, repeats the sub, - gets a better value and the Graph starts nicely falling looking like one side of the "V" graph, but...

Once SGP sets motor to 4500 (on the opposite side of the ideal focus), HFR value spikes enormously! it was something around 4HFR, next step 5HFR and 3rd step loses focus completely, but SGP calcs HFR something around 2, so it thinks focus is getting better, and finally SGP fails...

Very strange. 

I reduced step SPG step site to 150, it produces very nice V form graph, but... it never hits the focus spot on value... sometimes it gets 1.2 HFR sometimes Even 1.4 and SGP even pops up warning telling that best focus was not achieved (or similar wording).

plus, - SGP step size 250 or 150 in comparison to 7 (which would produce 3 actual motor steps within critical focus range) is a very Very big difference!

Anyway...

I am bit confused... I am not sure if I do it right and HFR results makes me think I have some king of mechanical problem in the focuser.... I looked at it while it was going below 5000,

it did not look like it was slipping in or something... Plus I made it quite tight...

Cloudy nights do not permit to continue with experiments...

Any advice how to solve this puzzle, - would be very handy!

P.S.

My apologies for such a Novel, - I did my best to write it as short as possible.

[RayD]

17 minutes ago, RolandKol said:

Hi guys, 
I have a similar question in relation to SGP autofocus, - I did not want to create a new topic and hope you will advice.

I recently finished my DIY autofocuser with Nema 17 PG27 motor. My calculations and measurements show that one motor step lifts focuser only by 2.42 microns (I able to do it even smaller). 

My telescope SW 130PDS (with SW CC) has a critical focus range in amount of 49 microns. So my motor is able to do 49/2.42 = 20.24 (rounding to 20) steps in the critical focus range.

SGP uses Best Median focus calculation, so it needs 3 best focus points to determine the best one. Initially, this fact made me think SPG step size setting should be 20/3 = rounding to 7.

At the moment I am keeping 9 Autofocus data points. However... with SGP step size 7, - autofocus routine produces quite flat graph and SGP is not able to calculate the mean value.

If I do as SPG suggests, - lets say I have 1 HFR value after bahtinov mask focusing routine (also using motor, but controlling it manually and using SharpCap focus aid).

So my best focus position equals  1 HFR and motor position at 5000, later I get value of 4 HFR while motor is in the position 6000.

SPG step size calc: (6000-5000'motor_position) * 2 / (9'datapoints-1)=250.

I tried this value and here the fun begins...

The first data point SGP creates by adjusting motor out of focus adding to 4*250 to starting position 5000, and reduced it 4 times back to 5000. At 6000, it gets expected HFR value in around 4.

After, it slews focuser back by 250, repeats the sub, - gets a better value and the Graph starts nicely falling looking like one side of the "V" graph, but...

Once SGP sets motor to 4600 (on the opposite side of the ideal focus), HFR value spikes enormously! it was something around 4HFR, next step 5HFR and 3rd step loses focus completely, but SGP calcs HFR something around 2, so it thinks focus is getting better, and finally SGP fails...

Very strange. 

I reduced step SPG step site to 150, it produces very nice V form graph, but... it never hits the focus spot on value... sometimes it gets 1.2 HFR sometimes Even 1.4 and SGP even pops up warning telling that best focus was not achieved (or similar wording).

plus, - SGP step size 250 or 150 in comparison to 7 (which would produce 3 actual motor steps within critical focus range) is a very Very big difference!

Anyway...

I am bit confused... I am not sure if I do it right and HFR results makes me think I have some king of mechanical problem in the focuser.... I looked at it while it was going below 5000,

it did not look like it was slipping in or something... Plus I made it quite tight...

Cloudy nights do not permit to continue with experiments...

Any advice how to solve this puzzle, - would be very handy!

P.S.

My apologies for such a Novel, - I did my best to write it as short as possible.

Have you measured, set up and implemented backlash compensation?

[RolandKol]

11 minutes ago, RayD said:

Have you measured, set up and implemented backlash compensation?

Not yet... was busy learning SGP and clouds later on....

But nice idea, it would probably prevent strange HFR values under motor position 5000, with SGP step size 250.

However, that SPG 250 size in comparison to 7 looks so BIG... I just simply do not feel it's correct at all.

[RayD]

1 minute ago, RolandKol said:

Not yet... was busy learning SGP and clouds later on....

But nice idea, it would probably prevent strange HFR values under motor position 5000, with SGP step size 250.

However, that SPG 250 size in comparison to 7 looks so BIG... I just simply do not feel it's correct at all.

Yes backlash compensation is essential, it can throw things way out.

The actual numbers are only relative to the resolution of your focuser and focus motor, so largely irrelevant.  The important thing is that you get the 4-6 x HFR differential.

[RolandKol]

Thanks a lot,

lets hope I will not need to wait next month to test...

as I will forget everything!

[Laurin Dave]

2 hours ago, RayD said:

Yes backlash compensation is essential, it can throw things way out.

The actual numbers are only relative to the resolution of your focuser and focus motor, so largely irrelevant.  The important thing is that you get the 4-6 x HFR differential.

Indeed it does Ray! Forgot completely about it when setting up new equipment profiles for the Mesu so wasted the first clear night for a month with soft focus.  Guiding performance was great though..  luckily, enough clear sky last night to set backlash and step size using the 4-6x HFR and it worked a treat.  I'd been wondering why the first two points on the focus graph were flat ... 

Dave

[RayD]

2 hours ago, Laurin Dave said:

Indeed it does Ray! Forgot completely about it when setting up new equipment profiles for the Mesu so wasted the first clear night for a month with soft focus.  Guiding performance was great though..  luckily, enough clear sky last night to set backlash and step size using the 4-6x HFR and it worked a treat.  I'd been wondering why the first two points on the focus graph were flat ... 

Dave

Great news Dave.  Glad you got it sorted and the Mesu is performing