How parfocal are my filters? - an experiment

[michaelmorris]

With the bleedin' typical combination of a clear sky and a fullish Moon, I decided to take the opportunity to carry out a little experiment I've been meaning to do for a few weeks. - Just how parfocal are the filters in my filter wheel?

I have a ZWO 8-position electronic filter wheel containing the following filters

1. Baader CCD Luminance
2. Baader CCD Red
3. Baader CCD Green
4. Baader CCD Blue
5. Baader CCD Ha
6. Baader CCD OIII
7. Baader CCD SII
8. Astronomik CLS

So how to assess how parfocal they are?  It struck me that a simple pretty quantitative test would be take pictures of a star using a telescope with a Bahtinov mask fitted.  The deflection from true centre of the intersecting line on the image with the mask fitted repeated for each filter would give a simple indication of the difference in focus point for each filter.

The interation of Neils Noordhoek's Bahtinov grabber algorithms built into APT software allowed me to measure the number of mean number of pixels the mask's centre line was deflected from a reference point (the luminance filter) gave me a numerical measure of how close the focus points were.

I repeated these measurements 5 times for each filter over a 40 minute period.

Bahtinov mask test
Filter						Mean	SD	Median
Lum	-0.15	0	0	-0.07	-0.47	-0.138	0.20	-0.07
R	-1.91	-1.63	-1.8	-1.89	-1.91	-1.828	0.12	-1.89
G	-1.12	-1.25	-1.03	-1.02	-1.04	-1.092	0.10	-1.04
B	-1.18	-1.51	-1.17	-1.36	-1.66	-1.376	0.21	-1.36
Ha	-1.97	-2.11	-2.06	-2.18	-1.97	-2.058	0.09	-2.06
OIII	-1.04	-1.16	-1.16	-1.04	-0.99	-1.078	0.08	-1.04
SII	-2.07	-2.42	-2.01	-2.33	-2.05	-2.176	0.19	-2.07
CLS	-1.42	-1.69	-1.34	-1.19	-1.42	-1.412	0.18	-1.42

Based on these admittedly somewhat limited data it is clear that the only two filters that appeared to be truely parfocal were the Baader CCD green and CCD OIII filters.  All other filters required some adjustment to achieve critical focus.  However, the Baader Ha and SII filters and the Baader Blue and CLS filters did appear to be quite close to parfocal.

 

[Merlin66]

Michael,

Interesting.

How do those figures relate to real life differences in focal position (Micron) and what was the telescope system used?

 

[MarsG76]

I find that I have to re focus each filter, particularly in NB imaging, but also using RGB filters when planetary imaging. 

I imagine that the reason for it is the same reason you get chromatic aberration in refractors by different wavelengths coming to focus at different points, the IR focal point being different from visible spectrum.

After all the filters are there to pass through only one wavelength of light and so the focus will be slightly off filter to filter.

[michaelmorris]

6 hours ago, Merlin66 said:

How do those figures relate to real life differences in focal position (Micron) and what was the telescope system used?

 

I used an Atik 460EX camera attached to a Skywatcher (Orion) 80ED.

How much these differences effect the final star image will be partially dependent upon how good the seeing is. 

From my experience, the difference between the green and OIII filters would not be noticeable. I suspect the same may be the case with the blue and CLS filters. From my (all be it rather limited) experience, the other difference would, to a lesser or greater effect, probably be noticeable.

[ronin]

Were the measurement made by getting a focus with the Bahtinov mask in position then inserting the different filter and adjusting for the best focus/pattern again with the Bahtinov mask continuously in place?

Just thinking that the image/pattern formed from the filter is an interference one and so eash wavelength could be createing a "best" image at a different position. In interference experiments the pattern formed by Red is ata  different position and often size then those formed by Green and again by Blue.

Just there is the possibility that what is measured or is contributing to the variance is the interference pattern that the mask itself creates at the different wavelengths.

[michaelmorris]

14 minutes ago, ronin said:

Were the measurement made by getting a focus with the Bahtinov mask in position then inserting the different filter and adjusting for the best focus/pattern again with the Bahtinov mask continuously in place?

Just thinking that the image/pattern formed from the filter is an interference one and so eash wavelength could be createing a "best" image at a different position. In interference experiments the pattern formed by Red is ata  different position and often size then those formed by Green and again by Blue.

Just there is the possibility that what is measured or is contributing to the variance is the interference pattern that the mask itself creates at the different wavelengths.

The measurements are the difference (in pixels) in the position of the central interference line at a fixed focus position, with the critical focus position for the luminance filter being the reference point.  This eliminated focuser slippage or backlash as a source of error.

Whilst I understand your point, it has always been my experience that the Bahtinov mask performs well at indicating the critical focus point for all wavelengths and band passes, with no 'offsets' needed for different wavelengths.

[Merlin66]

It comes down to the critical focus zone (CFZ) for the telescope being used....

Basically dependent on the focal ratio, the ED80 being f7.5, you have a "dead" zone of almost 135 micron (based on green 550nm light) before the focusing "error" would exceed 1/4 wave.

So, if all the filters focus within this tolerance ( 4*lambda*f^2) then there will be no impact on the final image.

[Tim]

It's an interesting comparison Michael, and one that I do any time I get a new refractor for review.

I should have thought that a reflector or true apo would provide the most robust results, but I've noticed considerable effects from atmospheric distortion. Also focus changes due to temperature of course so the closer in time each filter is compared, the better.

I'm pretty happy that my Baader sets, both 2" and 1.25" are essentially parfocal to all intents and purposes in the telescopes I use.

However, I always focus each filter anyway, as best practice.

Are you coming to Kelling btw? Hope so!

[ollypenrice]

As others have said, it's impossible to know whether you are testing the parfocality of the filters or the colour correction of the optics. But the information you have gathered will be very useful to you.

At coarse pixel scales I find both the Tak 106 and TEC140 to be parfocal within the pixel scale of the system, using Baaders. However, when I take the TEC down to 0.9"PP I do need to focus separately per filter.

In the real world, anyone wanting to scroll LRGB, LRGB, etc., without autofocus or refocus, would do best to focus in L on the basis that slight mis-focus will be of far less consequence in colour. 

Olly

[michaelmorris]

3 hours ago, Tim said:

... I've noticed considerable effects from atmospheric distortion. Also focus changes due to temperature of course so the closer in time each filter is compared, the better.

The whole test was run 5 times over a 40 minute period with the filters selected in different orders each run to try to reduce the influence of confounding variables such as temperature.  The exposures were all set at 10 seconds, so hopefully any effects of poor seeing would also have reduced.  

The target star was Altair, which was around 42 deg altitude at the time.  Could atmospheric dispersion have been a factor?

[DaveS]

This is something I need / want to check myself.

I *think* my Astrodon NB filters are parfocal in my 130 triplet, but I'm no longer sure about the [OIII]. I would expect HII and [NII] to be, after all they are so close you need 3nm filters to separate them, [SII] might be out as well, as it's further into the red, and correction may be drifting. A triplet apo should be corrected for three wavelengths, in the red, green, and blue spectrum, but exactly which wavelengths is open to question.

Fortunately it's something that can be done under non-imaging conditions.

[Tim]

6 hours ago, michaelmorris said:

 

The target star was Altair, which was around 42 deg altitude at the time.  Could atmospheric dispersion have been a factor?

Borderline I should think Michael. If you test lower in the sky you will likely see a larger variance between blue and red.

[ollypenrice]

For NB imagers I'd have to ask whether or not parfocality mattered. You are not going to be scrolling between filters... are you? You need so much data that you'll be doing one night, at least, per filter and checking focus regularly.

Olly

[Tim]

Personally, being able to demonstrate parfocality in my filter set allows  me to quickly asses the CA qualities of any telescope that turns up for review. I always do a visual test too of course but sometimes the variances which render a telescope unsuitable for full colour imaging are hard to quantify under poor skies, the camera doesn't lie though. An interesting case in point is the little Vixen 100. Parfocal in red and blue but considerably out in green. Wasn't easy to demonstrate with an eyepiece at all but it is impossible to focus a colour camera with it.

[Gina]

I checked my Astrodon filters with my Esprit 80ED Pro and found no measurable difference but the Esprit is a particularly good APO scope.  But as has been said, I usually run all night on one filter and changing temperature is more likely to change focus than changing filters.

[Merlin66]

In our Littrow spectroscope we use a 40mm 200 fl achromat doublet from Edmund Optics. The the chromatic variation is significant.

As a telescope example: Based on an Edmund 90mm 850mm fl achromat......

The focus variation from Cak (393nm) to Mg ( 517nm) can be enormous - up to >5.5mm (!)

From Mg to Ha (656.3nm) the change is much smaller only <0.4mm

Obviously each lens design is different, and it's well worthwhile seeking out the Chromatic curves for your objective, if these can be sourced from the manufacturer.

 

[IanL]

On 9/6/2017 at 19:49, ollypenrice said:

For NB imagers I'd have to ask whether or not parfocality mattered. You are not going to be scrolling between filters... are you? You need so much data that you'll be doing one night, at least, per filter and checking focus regularly.

Olly

You are round these parts, given that you will get maybe two non-Moon afflicted, high-altitude clear nights per summer target if you are lucky. Not willing to wait 12 months for the third filter! Not so bad for winter stuff but even so weather is the limiting factor.