So do I actually need a reducer?

[Nikodemuzz]

Hello all!

I have been looking for a refractor in the ~1000mm FL class for some time now (there is a separate thread on the subject). I have sort of taken as a given that I will get a reducer to match the scope. The main benefits that I have thought the reducer would give me are:

- 1. Increased field of view, in other words possibility to image more targets.

-2. Faster imaging, taking the scope from F7-F8 to F5-F6 territory.

I have started to doubt if there is any real benefit for me on either of these points.

Let's address point 1 first. The only other scope I have at the moment (and for the foreseeable future) is the WO SpaceCat. Combined with the APS-C sized chip of my ASI071, I can cover most wide field targets in the sky. When you visualize the fields of view of these options, it becomes quite clear how small difference in FOV the reducer actually makes. In this example the larger refractor is the TSA-120, but the case is similar for others in its class. For those that use a 0,8 reducer instead of a 0,7 the difference is smaller still.

There are a precious few targets that are just the size to fit between the yellow and red rectangles.

To address point 2: Yes the F number decreases. However, if I'm imaging the same target it doesn't make my data collection any faster, I just get more space around it. For smaller targets I would actually lose resolution. Right? So it makes no appreciable difference to imaging at native focal length (with a flattener of course). Nothing becomes any "faster".

Am I missing something? It seems to me that I would get a much larger impact in FOVs by getting a camera with a smaller sensor compared to a reducer. Sampling rates are within acceptable limits on all options for the TSA120. The SpaceCat is its own animal with the short focal length, with that scope I'm undersampling either way.

• 1,10 "/pixel ASI071 + TSA120
• 1,57 "/pixel ASI071 + TSA120 + reducer
• 0,86 "/pixel AS533 + TSA120

Granted, getting a camera is a rather expensive move, but compared to a Tak reducer and adapters, it's not that far off. For most other brands, yes, money lost. 😃 But, I could very well go a season or two with the camera I have now, and get another one from the used market when a good deal emerges. By doing that I could shave a little bit off the telescope investment (in Tak's case, not even that little), and keep things nice and simple with less adapters and bits and bobs to fool around with.

Any thoughts? Am I approaching this somehow backwards?

[vlaiv]

Most people use reducer for point 2 or for point 3.

Point 3 would be - at some point with larger sensor you need field flattener. These come in two flavors - without reduction and with reduction. Most people opt for one with reduction because of point 2.

1 hour ago, Nikodemuzz said:

To address point 2: Yes the F number decreases. However, if I'm imaging the same target it doesn't make my data collection any faster, I just get more space around it. For smaller targets I would actually lose resolution. Right? So it makes no appreciable difference to imaging at native focal length (with a flattener of course). Nothing becomes any "faster".

It does become faster.

We can define speed in simple terms to be: Aperture at resolution.

When you add focal reducer, aperture obviously does not change - but resolution does change as you change the focal length of system. Pixels stay the same size - as you are using the same camera - but due to focal length decrease - each of them covers more of the sky. They become "bigger" in relative terms (cover the same amount of sky as would bigger pixels on the scope without reducer).

That improves the speed as each pixel receives signal from more sky (same aperture) and hence gets more photons which equals signal increase in given time. SNR increases.

1 hour ago, Nikodemuzz said:

Am I missing something? It seems to me that I would get a much larger impact in FOVs by getting a camera with a smaller sensor compared to a reducer.

Nope - it is other way around - you can have the same impact if you get camera with larger sensor not smaller. Larger sensor covers more sky - reducer lets your current camera cover more sky.

Once you get camera with larger sensor - you need to make sure you are also sampling at changed sampling rate - otherwise you will not see speed improvement.

Benefit of larger sensor is that you can use it on larger scope to get the same FOV - and larger scope means larger aperture.

In the end - just be careful. There is something called corrected and illuminated circle for telescope. You can't get good stars past that boundary (and with many scopes quality of stars is questionable near that boundary).

If you wonder why most reducers are in range of 0.75-0.8 and not below - well, for that reason. You design reducer for APS-C sensor size - which has 28mm diagonal and you have x0.8 reducer - then original imaging circle needs to be 28 / 0.8 = 35mm. Majority of scopes won't have good stars and illumination past that diameter.

Same goes for sensor size. It makes sense to use FF sensor only on scope that is capable of illuminating 45mm diagonal.

1 hour ago, Nikodemuzz said:

Sampling rates are within acceptable limits on all options for the TSA120. The SpaceCat is its own animal with the short focal length, with that scope I'm undersampling either way.

• 1,10 "/pixel ASI071 + TSA120
• 1,57 "/pixel ASI071 + TSA120 + reducer
• 0,86 "/pixel AS533 + TSA120

I would argue that

a) you are not under sampling with SpaceCat. If you take sampling rate at face value - it would look like you are under sampling. You are close to 4"/px with ASI071 - that must be under sampling, right? Well - SpaceCat has 50mm of aperture and 50mm of aperture has Airy disk diameter of 5.13" in green light - larger than single pixel. Add a bit of seeing and tracking error to it - and I'd say 4"/px is really not that under sampling

b) without reducer you are over sampling with TSA120.

With that aperture in normal conditions - 1.5" seeing and 1" RMS guiding - you can expect star FWHM to be around 3" and corresponding sampling rate to be ~1.8"/px (FWHM/1.6). With better guiding you'll get down to ~1.5"/px range (FWHM of 2.5").

 

 

Edited February 23, 2021 by vlaiv

[Nikodemuzz]

Thanks for the thorough answer, @vlaiv

I would be using a field flattener anyway, at least I think you need one already with APS-C sized sensors.  The question then is, without reducing, with reducing, or both? I would argue that at least I don't need both because the FOV difference is quite small.

About the speed, I understand what you mean by more of the light from the target hitting single pixels. A drastic example of that would be that the subject is only the size of a single pixel, meaning that all of the light hits that single pixel. I guess my point is, how much does the SNR increase, and is it enough to account for the loss of resolution? I'm sure the F-stop rule doesn't apply here anymore (meaning that one F-stop equals double the light or half the exposure time).

That's a good point about the reducing factor and the original illuminating circle of the scope, I didn't know that!

I'm sure you are also correct about the sampling rate. Taking that into account, would you say that with those sampling rates it would be better to image with the TSA-120 reduced, even smaller targets? Because the additional resolution wouldn't materialize anyway due to oversampling.

Edited February 23, 2021 by Nikodemuzz

[vlaiv]

31 minutes ago, Nikodemuzz said:

About the speed, I understand what you mean by more of the light from the target hitting single pixels. A drastic example of that would be that the subject is only the size of a single pixel, meaning that all of the light hits that single pixel. I guess my point is, how much does the SNR increase, and is it enough to account for the loss of resolution? I'm sure the F-stop rule doesn't apply here anymore (meaning that one F-stop equals double the light or half the exposure time).

Indeed - classical F/stop rule does not apply here as we are deep into - "any noise source is high enough to cause issues" territory , but then again - depends on how you look at it.

Say that you use 0.8 reducer, and that actual SNR improvement is only 1/3 of what F/stop rule would suggest. Increase in pixel area is (1/.8)^2 = 1.25^2 = 2.5 - so signal increases x2.5 times, or additional 150% of original value. Say that we only get 50% of original signal value - that is like imaging for 50% more time.

If you image for 2 hours - than this is not really drastic - you add another 1h to get 50% more imaging time. If you image for two days and need to add another day - well, it is becoming significant now, right? Two vs three days spent on target.

I think that best approach to speed is - set your working resolution and then throw as much aperture at it as you can

By the way, reducers are not the only way to manipulate with working resolution - there is binning also. Problem is that you are using OSC sensor and in the start you are effectively using twice as low sampling rate as you think you are using. This is because of bayer matrix on the sensor and the fact that pixels are in spaced by factor of two (R-G-R-G-... and G-B-G-B-....) - so in fact red is every other pixel in X and Y direction - so is blue, and if you look at green being G1 and G2 - then those are spaced every other pixel too.

In process of reconstruction of the image - missing pixel values are interpolated - and that means that if you bin your image 2x2 - you won't get the same result as when you bin mono data. There will be some improvement but not what one would expect from 2x2 bin on non interpolated data.

43 minutes ago, Nikodemuzz said:

I'm sure you are also correct about the sampling rate. Taking that into account, would you say that with those sampling rates it would be better to image with the TSA-120 reduced, even smaller targets? Because the additional resolution wouldn't materialize anyway due to oversampling.

Indeed. I think that we often over sample without realizing it.

You can check how much you are oversampling by measuring FWHM of stars in your image. Proper sampling rate is about x1.6 less than FWHM. If you FWHM is say 3.2" then good sampling rate for that resolution is 2"/px.

Here is general guide line of scope aperture vs resolution:

100mm aperture in regular seeing with regular mount will be capable of doing 1.6-2"/px (there is quite a bit of variation at this end - depending on how good the mount and seeing is)

scopes in 80-72mm range are going to be wide field instruments with 2-3"/px

150mm of aperture is going to manage 1.4"-1.6"

200mm is needed if you want to attempt resolutions of about 1.2"

I don't really think anyone can properly manage resolutions smaller than 1"/px so I would put limit there. If you are at 1.1"/px or lower - 99% chance you are over sampling.

Here is an example of that - I took this image with my 8" RC - and seeing just did not play ball that evening:

This is at 1"/px - image looks soft there and in fact - actual resolution of the image is closer to 2"/px - and can look like this:

There is nothing in above image that can't be seen here - and here it looks better (higher SNR) and sharper because it is properly sampled.

Most people confuse sampling rate with FOV and zoom level. I advocate that image should look good when viewed 1:1 or 100% zoom rather than screen size. Most people over sample but due to fact that they use cameras that have say 4000x3000px and viewing those images on screens that have 1920x1080 or so pixels - images get resampled down x2 - and resolution seems fine in that case - as you get x2 reduction - but you loose SNR that way.

If one was to sample at proper resolution - it would result in SNR improvement of x2 as well as sharp image.

Hope that makes sense?

[Nikodemuzz]

5 hours ago, vlaiv said:

Hope that makes sense?

Yes, makes sense! Well explained, thank you.

I don't think your math quite checks out (1,25^2=2,5), but I get the point nonetheless. The benefit is there, but it might be a bit smaller.

Binning is something I knew I was giving up using the OSC camera, and those compromises were calculated ones. Mono imaging might very well be in my future, but at the moment I'm very happy gathering experience with OSC. During processing I have already noticed things that would be easier and better if I had mono data, and one can't argue against the shortcomings that the Bayer matrix brings. Should I go mono, I would like to get high quality filters and fully automated equipment to take the fuss out of the experience. That investment will have to wait a while still.

The sampling rate was something I hadn't paid enough attention to prior to starting this thread. My thinking has changed as a result, meaning that I now think going in with a reducer is the correct choice. Flattener, perhaps unnecessary. 😃 

[vlaiv]

1 hour ago, Nikodemuzz said:

I don't think your math quite checks out (1,25^2=2,5), but I get the point nonetheless. The benefit is there, but it might be a bit smaller.

You are absolutely right. I don't know how that happened. I entered digits into calculator and pressed something (maybe *2 instead of to the power of 2?) and I was not paying much attention to the figure I got.