Advice needed on perfect focus

[Padraic M 268]

I spent the full night out last night and got 6 hours of Ha lights on the Bubble and the Horsehead. Reasonably pleased with the results, but even though I followed my usual process and got good focus statistics in APT, I am slightly out of focus with roundy stars and some are even slightly donutty. Samples are attached below.

Problem:
- After getting close to spot-on focus, the APT Bahtinov Aid showed a focus distance oscillating from -0.02 to +0.02. Seeing seemed good to the inexpert eye. Not so sure about transparency as there was some thin, wispy cloud throughout the night. So, I started the night's imaging with focus 'Close' rather than 'On' focus.
- Different subs show different quality stars, ranging from small donuts to circles.

Background information:
- HEQ5 Pro Rowan; SW Esprit 80 with field flattener, SW stock manual Crayford focuser; ZWO EFW Mini; Baader 1.25" 3.5nm Ha filter; ZWO ASI1600MM Pro binned 1x1 @ -20c.
- AA Starwave 50mm guidescope with ZWO ASI290mm Mini guidecam binned 2x2.
- All subs are 300s, gain 139, offset 10.
- Polar alignment with Sharpcap to 17 arcsec ("Excellent"); capture with APT; guiding with Phd2. Focus with Bahtinov mask and APT Bahtinov Aid. Stacked in DSS with Darks, Flats and Dark Flats.
- Mount is well balanced in RA, but is very camera-heavy in Dec.
- PHD2 guiding was around 2"/px. Imaging pixel scale is 1.9"/px.

Questions:
- Do I put the round stars down to seeing, given that the Bahtinov Aid focus distance was bouncing equally above and below zero?
- Can poor seeing cause the donut stars?
- Would an electronic auto-focuser do any better in this situation?
- Would the Seeing Monitor in Sharpcap give useful information? I didn't think to use it last night.
- Could my guiding performance, and possibly the Dec balance, have affected the image quality in this way?
- What are my options in future - abandon imaging for the night? Bin all images in software 2x2 or 4x4 to sharpen the stars at the expense of lower resolution?
- Other suggestions?

Sample 1: Detail from a single 5-min sub of Bubble nebula at 100% showing round stars, and a blurred bubble.

Sample 2: Detail from a different sub of the Bubble nebula at 400% showing donuts

 

Sample 3: Detail from a 5-min sub of the Horsehead nebula at 100%, showing both round and donut stars

 

Finally, both images stacked, calibrated and stretched, scaled to 4x4 in Gimp. 28x300s Ha on bubble, 22*300s Ha on horsehead.

[inFINNity Deck 164]

6 hours ago, Padraic M said:

Questions:

- Do I put the round stars down to seeing, given that the Bahtinov Aid focus distance was bouncing equally above and below zero?
- Can poor seeing cause the donut stars?
- Would an electronic auto-focuser do any better in this situation?
- Would the Seeing Monitor in Sharpcap give useful information? I didn't think to use it last night.
- Could my guiding performance, and possibly the Dec balance, have affected the image quality in this way?
- What are my options in future - abandon imaging for the night? Bin all images in software 2x2 or 4x4 to sharpen the stars at the expense of lower resolution?
- Other suggestions?

- Seeing: unlikely, the stars are very round, so that seems not to be the case.

- Donut stars due to seeing: only in the very rare event that seeing and poor tracking will cause the star to make a circle during your exposure. So, no, not possible.

- Better performance by electronic focuser: Yes, provided it is used in combination with proper auto-focus software (see below).

- SharpCap's seeing monitor: never used it, have no idea how useful it is.

- Guiding performance: no, poor guiding will not result in round stars.

- Future options: get a decent imaging package like SGP or MaximDL. There are some free packages as well that do fairly well, perhaps others can comment on that. I used the focuser routine of SharpCap myself, but was not impressed, especially when compared to the one built into SGP that I normally use.

- Other suggestions: wait for other SGL-members to respond.

As far as I can tell your camera was not in focus. Please note that during the night your scope will cool down and as a result of it will loose focus. The imaging packages mentioned above will allow you to check focus every so many degrees, frames or minutes.

Nicolàs

Edited October 26, 2020 by inFINNity Deck

[ollypenrice 28,425]

I think it's just soft focus. I've never felt the need to use software to interpret the Bahtinov mask image and when I check my visual B-mask focal point against an FWHM measurement I find it's right. I'm not a fan of software for software's sake.

I disagree with Nicolas when he says, '- Guiding performance: no, poor guiding will not result in round stars.' If the tracking errors are of the same magnitude on both axes then you will get round stars but they will be larger than they should be. We saw this when initially testing guide parameters on our Mesu mount. As we optimized the parameters the stars became significantly smaller. This is an unusual situation but not an impossible one. It also shows that round stars are not, in fact, reliable indicators of good tracking. 

Once you have robotic focus working it should give you the best result but I just do it by hand and check regularly.

Olly

[andrew s 4,453]

I noticed you have a manual focuser. Is there any backlash? If so make sure any final adjustment is "in" i.e.  lifting the image train against gravity. Also make sure it does not move once locked down or shift when locking it down.

Regards Andrew 

Edited October 26, 2020 by andrew s

[ollypenrice 28,425]

22 minutes ago, andrew s said:

I noticed you have a manual focuser. Is there any backlash? If so make sure any final adjustment is "in" i.e.  lifting the image train against gravity. Also make sure it does not move once locked down or shift when locking it down.

Regards Andrew 

Very good point.

[inFINNity Deck 164]

1 hour ago, ollypenrice said:

I disagree with Nicolas when he says, '- Guiding performance: no, poor guiding will not result in round stars.' If the tracking errors are of the same magnitude on both axes then you will get round stars but they will be larger than they should be. We saw this when initially testing guide parameters on our Mesu mount. As we optimized the parameters the stars became significantly smaller. This is an unusual situation but not an impossible one. It also shows that round stars are not, in fact, reliable indicators of good tracking.

Hi Olly,

interesting, and you are absolutely right: if guiding fails by the same amount in both directions, you will get bloated round stars. I thought that would be as rare as donuts being created by seeing, but apparently not...

Nicolàs

[Padraic M 268]

@inFINNity Deck, @ollypenrice and @andrew s thanks guys for the very informative commentary. I'm reassured that it's something that can be improved through more rigour and technique, rather than something I have to live with. On the next clear night I might pay more attention to the FWHM reading. Is it safe to say that FWHM should be in a standard range as long as I don't make material changes in the light-path, i.e. with experience I can say that I need to achieve FWHM of 1 or 0.5 or whatever, regardless of what (unsaturated) star I choose? So I can quantitatively say that I am near or in focus in every situation?

5 hours ago, inFINNity Deck said:

during the night your scope will cool down

3 hours ago, ollypenrice said:

I just do it by hand and check regularly

I was happily tucked up in bed for the majority of this imaging session and unfortunately it looks like this may not be possible in future unless I move to automated focusing. That's a definite incentive for me to spend more money!

3 hours ago, andrew s said:

make sure any final adjustment is "in"

Good idea. There is certainly a big difference in tension between "in" and "out", as there's a lot of weight on the camera end. The focus also does change as I lock it down, so I have to adjust the focus lock and the fine focus knob together in very small increments.

I think for now I will treat focus performance and guiding performance as separate issues until I have got the basics right for focusing.

Many thanks all, knowledgeable, helpful and generous as usual.

[ollypenrice 28,425]

7 hours ago, inFINNity Deck said:

Hi Olly,

interesting, and you are absolutely right: if guiding fails by the same amount in both directions, you will get bloated round stars. I thought that would be as rare as donuts being created by seeing, but apparently not...

Nicolàs

I don't think donuts will be created by bad seeing. The stellar image may move around during the exposure but the central part of the star will surely still get the most light.

Olly

[michael8554 1,283]

If I've read your post correctly, you're using a 400mm scope with a 3.92um pixel camera, binned 2X ?

Imaging scale is then 4.04arcsecs/pixel.

Would that give fat round stars........?

Or dew due to the cooling?

Michael

[inFINNity Deck 164]

According to the original post (under "Background information:") the camera is a "ZWO ASI1600MM Pro binned 1x1 @ -20c." The SkyWatcher Esprit 80 is f/5 (400mm focal length, 80mm aperture). This means that it will produce an Airy-disc (at green light) with a radius of 0.66 x 5 = 3.3 micron, so a full Airy-disc of 6.6 micron diameter. This is just under 2 x 2 pixels of the image. The other way around it is recommended to image with a f-number of at least 3 times the pixel-size, so around f/10. The images shown here are therefore slightly undersampled, but that would not cause these fat round stars (contrary even).

It would be great to see the original image at full resolution, but if I look at the noise in the first crop the stars seem to have a diameter of roughly 15 pixels, which seems to indicate a focus issue.

Nicolàs

Edited October 27, 2020 by inFINNity Deck

[michael8554 1,283]

Yes Nicolas, I misread the Binning, my Bad.

Michael

[Padraic M 268]

Hi @michael8554  @inFINNity Deck yes I'm imaging un-binned so the pixel scale is 1.9"/px.

Here's the full Bubble Nebula stacked at full resolution. Only the worst 2 lights were excluded. Stars look approximately 12 pixels wide in the remaining 30 frames and in the stack.

 Bubble Nebula.tif

[inFINNity Deck 164]

Here is the Bubble Nebula on Astrobin, imaged with a SkyWatcher Esprit 80ED and combination of a Canon and ZWO ASI1600MM Cool. Here is another one, imaged with same scope and ZWO ASI1600MM Cool only. So, clearly the scope can perform better at this pixel scale.

Nicolàs

 

Edited October 27, 2020 by inFINNity Deck

[Padraic M 268]

Yes, I've no doubt the scope/camera combination can do better. Those are great photographs. I particularly like Rudiger's final Ha version.

Just waiting now for another clear night to try again, this time with more effort on focus and FWHM watching.