Back to forums

Good Example of Why I Wouldn't Use HFD Change to Trigger Refocus

Planewave CDK17 Acquisition Auto-Focusing Deep Sky
16 replies•1.1k views
Kevin Morefield avatar
Kevin Morefield avatar
I've seen some choose to trigger a re-focus based on an increase in FWHM or HFD.  This never made sense to me because I want to refocus before focus has shifted outside the CFZ rather than after.  By definition, if you can see the effect of being out of focus it is too late.  Worse yet, I think there is some assumption that seeing is constant.  It is not. 

Below is a good example of how seeing can change over a short period of time.   This is CDK17 data from Obstech and at these seeing levels the CDK is pretty much a seeing monitor (if in focus).  Within one hour the FHWM moved up and back down over 1".  Looking at the background values (mode of the image) it looks to me like there were some thin high clouds passing.  But later, after the background values stabilize, the FWHM is still varying by over 0.5".

The answer is to focus on temperature change.  Expansion and contraction of the OTA, etc. is the cause of focus shift and it is measurable in real time.  Different scopes can have vastly different rates of expansion and contraction and different critical focus zone (CFZ) widths.  So you need to at least roughly calibrate your system to fine tune how much temp change can move your system out of the CFZ.  I did this by plotting the different focus results I saw at different temps.  I settled on a change of 2 degrees C for the CDK17, 1 degree for the DeltaRho and 0.5 degrees for the FSQ.   

Well Written Helpful Insightful Engaging Supportive
Bruce Donzanti avatar
Bruce Donzanti avatar
Interesting analysis, Kevin.  Thanks for presenting these data.  Frankly, I though most folks use both HFR and temperature in some combination and I don't think most folks think seeing is constant all evening- at least I learned that in Florida.  As a NINA user, I have my program set to refocus if temperature changes by 2 degrees or a 5% change in HFR, for either the C11" or FSQ85 in use.   Perhaps I should do a similar run like you did, but if I find out that a 0.5 degree change is best, I'd be refocusing all evening.   I understand your point about the HFR/D/FWHM change as the horse has already left the barn (similar to the depth reader in my boat- you get the depth after you pass over it and too late to avoid the sandbar), but I am trying to strike a balance.  Perhaps my approach is wrong.

Bruce
Helpful Engaging Supportive
Brian Puhl avatar
Brian Puhl avatar
I 100% agree with you, however for somewhat different reasons. 

If HFD bounces around in a backyard system, it's typically because guiding went to hell.  Now this can be linked to seeing conditions, but more often than not it's due to imperfections in gear mesh on cheaper mounts.   That's not a focus issue and therefore the corrective action should not be to refocus.   

Also, the wider the field, and further we get from proper sampling, the less HFD fluxuates.  I can't find any proper scenario in my head where refocusing on HFD would ever be better than refocusing on temp.  

Your points stand strong.  I just think the majority of us are less dependent on seeing conditions.   For my big frac, it's focus after 1 degree.    For the 100s they refocus after 2 degrees.   I never saw the huge changes in seeing with the 100s but the 150 definitely sees it bouncing throughout the night.
Helpful Respectful Engaging
John Hayes avatar
John Hayes avatar
Kevin,
This is an excellent point and it's why I check focus using astigmatic focusing every guide cycle rather than waiting for each exposure to end.  I've attached an image showing just one example of when the guiding is good but the focus signal is unstable.  I have the time constant set to be fairly long so that the system is most responsive to more slowly varying thermal changes.   This data shows the changes in focus that you can get from the atmosphere (due to second order changes in the wavefront caused by atmospheric pressure gradients).   Controlling focus in real time while the shutter is open can, in principle, improve image quality by holding focus to within the depth of focus throughout the exposure.  It also improves throughput since there is no need for any V-curve focusing in between exposures.

The defective focuser on my scope screws this up a little bit by sometimes (not always) causing a bit of image shift in the DEC direction, but I hope to repair the problem next month when I head back down to Chile to install a new focuser.

John



Well Written Helpful Insightful Engaging Supportive
Kevin Morefield avatar
Kevin Morefield avatar
Bruce Donzanti:
Interesting analysis, Kevin.  Thanks for presenting these data.  Frankly, I though most folks use both HFR and temperature in some combination and I don't think most folks think seeing is constant all evening- at least I learned that in Florida.  As a NINA user, I have my program set to refocus if temperature changes by 2 degrees or a 5% change in HFR, for either the C11" or FSQ85 in use.   Perhaps I should do a similar run like you did, but if I find out that a 0.5 degree change is best, I'd be refocusing all evening.   I understand your point about the HFR/D/FWHM change as the horse has already left the barn (similar to the depth reader in my boat- you get the depth after you pass over it and too late to avoid the sandbar), but I am trying to strike a balance.  Perhaps my approach is wrong.

Bruce

If I were to "belt and suspenders" my refocus decision I'd use temp change and some time limit.  Looking at my ~3 hour sample above, the FWHM increased by 5%+ 11 times.  

With my FSQ, which seems to be very temp sensitive, I do see it refocusing very frequently in the early evening at temp falls rapidly.  Often it will refocus after the each of the first two or three subs.  But I'm OK with that because I know I would be out of focus if I didn't.  Keep in mind I'm using Centigrade here.
Helpful
Kevin Morefield avatar
Kevin Morefield avatar
Brian Puhl:
I 100% agree with you, however for somewhat different reasons. 

If HFD bounces around in a backyard system, it's typically because guiding went to hell.  Now this can be linked to seeing conditions, but more often than not it's due to imperfections in gear mesh on cheaper mounts.   That's not a focus issue and therefore the corrective action should not be to refocus.   

Also, the wider the field, and further we get from proper sampling, the less HFD fluxuates.  I can't find any proper scenario in my head where refocusing on HFD would ever be better than refocusing on temp.  

Your points stand strong.  I just think the majority of us are less dependent on seeing conditions.   For my big frac, it's focus after 1 degree.    For the 100s they refocus after 2 degrees.   I never saw the huge changes in seeing with the 100s but the 150 definitely sees it bouncing throughout the night.

Brian,

I agree that seeing fluctuation is unlikely to cause variation in HFD or FWHM in heavily undersampled systems.  That said, temperature will still cause the scope to go out of focus.  So I'd be sure to have a temp based refocus criteria.  In this case it might not hurt to also have an HFD/FWHM refocus too but I would want to know why it is moving up if temp hasn't changed?  This would start to point me to focuser slippage, mirror slop or other mechanical problems.

Kevin
Helpful Insightful Respectful
MaksPower avatar
MaksPower avatar
While I have an ASIAir and EAF on my scope - and use a temp sensor to trigger refocus, I've found there is a way to minimise the issue - stabilise the temperature of the OTA.  For years I have used an insulated wrap around the OTA with a heater strap underneath applying 3W…5W to keep dew off the corrector.

After switching from visual to imaging a year ago I've found there is another benefit - after  30-60 minutes the OTA reaches a thermal equilibrium slightly above the dew point. While I have ASIAir set to refocus on a 1 degree ambient change (using its external sensor) the  changes are minuscule - like 0.1mm - and often the scope can run for an hour or two without the need to refocus.

But with no heat strap and no insulation the focus can shift several mm (scope is f/12).

It should be effective with any scope with a cylindrical metal OTA,  even open ones (cassegrains or Mewlons).
Helpful Engaging
Scott Hall avatar
Scott Hall avatar
With my SVX102T I refocus every 2 Celsius, or every 120 minutes, so I can catch big swings in temperature change and also account for shooting through a different amount of atmosphere over time where critical focus will creep.  NINA Hocus Focus HFR Improvement Tolerance is set to 10% and Autofocus R2 threshold is set to 0.98, to ensure I don't accept a bad focus run.  This is something the ASIAIR can't do and one of the big reasons I ditched it.  I got tired of it autofocusing and having a bad run and wasting hours of integration time.
Well Written Helpful Insightful Concise Engaging
Kevin Morefield avatar
Kevin Morefield avatar
John Hayes:
Kevin,
This is an excellent point and it's why I check focus using astigmatic focusing every guide cycle rather than waiting for each exposure to end.  I've attached an image showing just one example of when the guiding is good but the focus signal is unstable.  I have the time constant set to be fairly long so that the system is most responsive to more slowly varying thermal changes.   This data shows the changes in focus that you can get from the atmosphere (due to second order changes in the wavefront caused by atmospheric pressure gradients).   Controlling focus in real time while the shutter is open can, in principle, improve image quality by holding focus to within the depth of focus throughout the exposure.  It also improves throughput since there is no need for any V-curve focusing in between exposures.

The defective focuser on my scope screws this up a little bit by sometimes (not always) causing a bit of image shift in the DEC direction, but I hope to repair the problem next month when I head back down to Chile to install a new focuser.

John




John, this is really interesting.  I've never thought that atmospheric effects could impact focus.  In the graph above, are the points above and below the blue lines outside the CFZ or is that the threshold for you to make a focus adjustment during the exposure?  One other question regarding the astigmatic focus routine: what is the margin of error in assessing focus?  

Kevin
Well Written Engaging
John Hayes avatar
John Hayes avatar
Kevin Morefield:

John, this is really interesting.  I've never thought that atmospheric effects could impact focus.  In the graph above, are the points above and below the blue lines outside the CFZ or is that the threshold for you to make a focus adjustment during the exposure?  One other question regarding the astigmatic focus routine: what is the margin of error in assessing focus?  

Kevin

Kevin,
The dotted blue upper and lower bounds are the edges of the CFZ.  The units are a little obtuse in SkyWave because it shows the ellipticity percentage (of the auto-covariance function) rather than actual distance along the optical axis.  You have to input the diameter and focal length and then calibrate the system before SKW can compute the limits on ellipticity using the sensitivity of the system and the known first order parameters.  Either way, as you can see, focus varies in real time due to both thermal expansion of the telescope and from the atmosphere itself.   If you put aside the piston term, the dominant wavefront aberration introduced by the atmosphere is tilt; however, second on the list is the defocus term.  On a really good night, I will see excellent guiding numbers along with a really stable focus signal output.  On rare occasions, I can get good guiding but focus is all over the place (as shown above) and that never leads to low FWHM numbers.  The ultimate challenge with the system is to figure out how to set the feedback parameters to avoid chasing seeing induced variations  while continuously compensating for thermal drift.  

In the above example, the autofocus system is actually working pretty hard to hold the focus signal within the tolerance band but the rate of variation is just a bit too steep compared to how quickly the system can respond because I have signal averaging turn up pretty high to better match the response time to my typical thermal variations.  The other problem is that there is a pretty significant phase lag in the response no matter how fast I make it respond so it can't respond to very rapidly changing signals no matter how fast I try to make it.   Regardless, if I had turned off autofocus during this session, I suspect that the focus signal would have peaked at 2x - 5x the amplitude that you see here.  Just looking at how quickly the system achieves initial focus tells me that the time constant is probably around 30s.  The reason that I don't know the exact number is that you set up various parameters for stacking, averaging, and fractional response that all combine to produce the time constant for the system.  


The screen grab below shows the focus signal when the system first starts.  At startup,  the system is way out of focus but it is quickly pulled  into focus by the autofocus system within less than 5 guide cycles, which is determined by the 5 second exposure time.




John
Helpful Insightful
Arnie avatar
Arnie avatar
MaksPower:
While I have an ASIAir and EAF on my scope - and use a temp sensor to trigger refocus, I've found there is a way to minimise the issue - stabilise the temperature of the OTA.  For years I have used an insulated wrap around the OTA with a heater strap underneath applying 3W...5W to keep dew off the corrector.

After switching from visual to imaging a year ago I've found there is another benefit - after  30-60 minutes the OTA reaches a thermal equilibrium slightly above the dew point. While I have ASIAir set to refocus on a 1 degree ambient change (using its external sensor) the  changes are minuscule - like 0.1mm - and often the scope can run for an hour or two without the need to refocus.

But with no heat strap and no insulation the focus can shift several mm (scope is f/12).

It should be effective with any scope with a cylindrical metal OTA,  even open ones (cassegrains or Mewlons).

I also use insulation wrapped around my scopes (8", 9.25" and 14" SCT's). The whole process works much better by placing the temperature probe UNDER the insulation, taped to the OTA. That way one is measuring the temperature of the OTA itself, which is the true determining factor of tube expansion or shrinkage, rather than the temperature of the air outside the insulating wrap.

Arnie
Well Written Helpful Insightful Concise
Paolo avatar
Paolo avatar
I solved a big problem when a switched from "AF after HFR increase" to "AF after time or temperature".
Kevin Morefield avatar
Kevin Morefield avatar
John Hayes:
Kevin Morefield:

John, this is really interesting.  I've never thought that atmospheric effects could impact focus.  In the graph above, are the points above and below the blue lines outside the CFZ or is that the threshold for you to make a focus adjustment during the exposure?  One other question regarding the astigmatic focus routine: what is the margin of error in assessing focus?  

Kevin

 The ultimate challenge with the system is to figure out how to set the feedback parameters to avoid chasing seeing induced variations  while continuously compensating for thermal drift.

John,

My understanding of seeing is that it is made up of two components; displacement and blur.  It sounds like you are describing the blur component here.  Would that be right?   If so, chasing seeing with guiding would mean chasing the displacement component and chasing seeing in focus is chasing the blur component.  Is that right?

Given the short durations you are describing it would seem there is no way to address that atmospheric aspect of focus other than ONAG.

Kevin
Engaging
Kevin Morefield avatar
Kevin Morefield avatar
MaksPower:
While I have an ASIAir and EAF on my scope - and use a temp sensor to trigger refocus, I've found there is a way to minimise the issue - stabilise the temperature of the OTA.  For years I have used an insulated wrap around the OTA with a heater strap underneath applying 3W...5W to keep dew off the corrector.

After switching from visual to imaging a year ago I've found there is another benefit - after  30-60 minutes the OTA reaches a thermal equilibrium slightly above the dew point. While I have ASIAir set to refocus on a 1 degree ambient change (using its external sensor) the  changes are minuscule - like 0.1mm - and often the scope can run for an hour or two without the need to refocus.

But with no heat strap and no insulation the focus can shift several mm (scope is f/12).

It should be effective with any scope with a cylindrical metal OTA,  even open ones (cassegrains or Mewlons).

Seems like a good option with no downside.  I'm trying to thing how I would wrap my FSQ (which is really temp sensitive).  There is so little of the tube exposed what with the rings and other things on it.
John Hayes avatar
John Hayes avatar
Kevin Morefield:
John Hayes:
Kevin Morefield:

John, this is really interesting.  I've never thought that atmospheric effects could impact focus.  In the graph above, are the points above and below the blue lines outside the CFZ or is that the threshold for you to make a focus adjustment during the exposure?  One other question regarding the astigmatic focus routine: what is the margin of error in assessing focus?  

Kevin

 The ultimate challenge with the system is to figure out how to set the feedback parameters to avoid chasing seeing induced variations  while continuously compensating for thermal drift.

John,

My understanding of seeing is that it is made up of two components; displacement and blur.  It sounds like you are describing the blur component here.  Would that be right?   If so, chasing seeing with guiding would mean chasing the displacement component and chasing seeing in focus is chasing the blur component.  Is that right?

Given the short durations you are describing it would seem there is no way to address that atmospheric aspect of focus other than ONAG.

Kevin

Kevin,
That's correct.  Tilt in the wavefront is what causes image displacement (remember that a telescope converts angles into displacement in the image plane) and higher order terms that blur the image.  The biggest component of the higher order terms is defocus.  Guiding  normally involves just trying to keep the image stationary on the sensor.  The expression "chasing seeing" normally means that guiding is overcompensating for guide errors--due to either guiding too aggressively or due to a time lag in the correction (called a "phase error").  We normally only talk about this stuff with regard to system that is only sensitive to guide errors; however it all applies to a system that is sensitive to guide errors AND focus errors.  ONAG allows for more rapid correction of focus errors but it also increases throughput by constantly keeping the system constantly in focus.  

John
Well Written Helpful Insightful
Bill McLaughlin avatar
Bill McLaughlin avatar
Interesting thread and focusing by temp is what I have done for the past 10 years or so for just that reason. I hate the idea of "closing the barn door after the horse is out"smile. 

 I just calculate the CFZ for the scope and gather temp vs focus shift data so I know how many degrees shift make up 1/2 the CFZ and set my focus by temp to just under that.

The astigmatic focus system looks ideal but I suspect I do not have enough back focus for that and my much loved Active Optics unit.

Personally there are times when the wind and/or seeing are so bad that CFZ (for a given long scope)  becomes moot (early last night was a classic example at my remote site :angrysmile. HFD change might be useful in that situation combined with seeing and weather data to force a switch to a secondary short focal length system. I have always thought about doing that but I am not sure I want to deal with the complexity.
Helpful Insightful
Alex avatar
Alex avatar
having a real-time autofocus system surely is best solution.

ONAG of IF is a great piece for this and would love to have one but it has somewhat 101 mm optical length. Plus back focus of the imaging cam incl. filter wheel it comes to somewhat 135-150 mm minimum needed back focus, if I understood the calculations right.  

As for full frame imaging the beam splitter in the ONAG needs a certain minimum size and considering the need for a 45° angle placement of the BS it cannot be shorter.

Therefore it unfortunately cannot be used with most Newtonian or refractors due to their usually shorter back focus of the corrector.


For example, the flattner for my TOA has 83.7, the reducer even only somewhat ~ 65 mm. 

the technology behind the ONAG, using NIR light for the guiding / focus lock, more seeing immune and assessing the focus correction by astigmatism is excellent.


However, could a well setup regular OAG not be used in somewhat same manner?

Following idea behind:
after initial AF, the guiding software calculates an HFD average of the following guiding images of an equivalent of 30s exposure (10 images when exposure is set to 3s). This average is taken as initial reference. 


Then the average of the next 30s is compared to the initial reference and the following 30s again and so on and so on.

this until based on a certain HFD tolerance defined a movement of the focuser is triggered. Then the routine starts from beginning again.

of course it needs an initialisation the guiding software to know how much focuser travel has what impact to the HFD.

secondly it needs the capturing software to speak with the guiding software on AF status, etc. but giving the already existing communication between N.I.N.A and PHD2 as an example this seems doable.

The exact HFD readings of the guiding cam should not matter much as the routine is based on a difference between the averages. So even the guiding cam might be slightly out of focus, as long HFD values can securely be obtained, no issue.

Surely in the end likely not as good as the ONAG but just for the principle of finding a way of real-time focusing with existing setup.




p.s. sorry, I thought of deleting the post but kept it. The hurdle making it to my believe impossible to use a normal OAG when thinking it over is the guiding software to determinate to what direction to move the focuser. The simple increase of the HFD value is telling nothing about the direction. The ONAG does it well with the astigmatism pattern making it clear very well.