Help needed choosing a new telescope for imaging

[Lee_P]

13 hours ago, vlaiv said:

No, you are quite right - binning does not change FOV.

One of things that focal length impacts is of course FOV, but I commented that which you quoted coming from purely sampling rate perspective under assumption that targets you want to image will fit the FOV.

Sampling rate is function of pixel size and focal length. If we keep focal length the same and increase pixel size by factor of x2 - we get the same thing as halving focal length and keeping pixel size the same - in terms of "/px or sampling rate. Similarly if we increase pixel size by x3 (binning x3) we get the same effect as using 1/3 of FL and keeping pixel size the same.

1600 / 3 is roughly 500mm (or 533.333 to be precise) hence my assertion.

RC will be faster than your current setup even if you want to quadruple FOV (double it in each height and width).

OK thanks, makes sense.

13 hours ago, vlaiv said:

When we did "speed" calculations we concluded that RC + x3 binning is about x6 times faster than your current setup. If you image target for 6h with your current setup - equivalent will be 1h with RC.

Four times faster, I think? RC8 binx3 = 84600; FRA400 binx1 = 19510.

 

OK, let me write my understanding of this binning malarkey and let's see if I've got it:

If I were to use my 2600MC with no binning (i.e. binx1) with an RC8, I'd get an image with a resolution of 6248 x 4176. However, I wouldn't actually be getting useful data across all those pixels because seeing conditions and mount limitations combine to, in effect, "blur" the view. When processing the data, I could binx3. This would make a square of nine pixels into one super-pixel. The plus points of this are that it would increase my SNR by a factor of around four. What takes me 20 hours with my current set-up would instead take five. Or, I could image for 20 hours and get a SNR equivalent to 80 hours with my current set-up. It would also decrease processing time due to the smaller file sizes. The downside is that the image's resolution would only be 2082 x 1392. This isn't that bad though, because the "lost" resolution wasn't useful data. I wouldn't be able to crop in very much with this lower-resolution image, but the RC8's focal length of 1600mm gets me very close in anyway, so that's not a big issue. If I need a wider FoV I could mosaic. 

Is that accurate? The lower final resolution still makes me itch a bit because I'm used to dealing with very high res images (both in astrophotography and regular photography) but I understand that with the proposed set-up it's not really a fair comparison. If I need a higher resolution image, e.g. for printing, I guess I could always use Photoshop's Enhance feature, or Topaz GigaPixel AI, to artificially bump up the pixel count. I just did an experiment doing that with some of my existing data, and it looks quite good to my eye. Enough to somewhat allay my fears about low-res images anyway.

Thanks vlaiv, I definitely owe you a pint! 🍻

[Lee_P]

15 hours ago, Clarkey said:

I think they are generally quite well collimated out of the box. I think many have been made worse by 'tinkering'. My method for collimation is to remove everything xcept the focuser and get the secondary perfectly aligned. I use a TSKOLLI or Reego, but you can use a simple Cheshire for that part. Once the secondary is right I use the DSI method for the primary using a star test.

https://www.google.com/url?sa=t&source=web&rct=j&url=http://www.deepskyinstruments.com/truerc/docs/DSI_Collimation_Procedure_Ver_1.0.pdf&ved=2ahUKEwi62bCmhJX5AhVMPcAKHcvBBws4ChAWegQIKBAB&usg=AOvVaw1TQ8YAhohC9_gr_IRXBo3x

This works for mine but my focuser and primary mirror appear well aligned as I get no tilt issues. If this is not the case it is more tricky - but the DSI method should still work. My main advice would be adjust it SLOWLY and try to remember what you have done to put it back again.

There is not a lot to say about the scope itself. I do have a home-made secondary dew heater, but I have never had a problem with the primary and dew. Cool down is pretty rapid too as it is an open tube. I have only used it with a ASI1600 which gave a good flat field. With a larger sensor you might need a flattener. I still need to try it with my IMX571 camera to determine if it is OK with and APS-C size sensor. I did change the focuser to a Baader ST. The original was OK in terms of tilt, but it did slip slightly with the autofocuser. I hope this helps. Here are a couple of images with the scope. Just remember I am far from an imaging expert!

 

Thanks, those are some good images!

I wonder, can you use the top Vixen dovetail as a carrying handle? It's hard to tell from the pictures if you could fit your fingers underneath it to get a good grip. Maybe with some risers? The collimation does sound like a pain, but it holds well once done?

[raadoo]

21 minutes ago, Lee_P said:

OK thanks, makes sense.

Four times faster, I think? RC8 binx3 = 84600; FRA400 binx1 = 19510.

 

OK, let me write my understanding of this binning malarkey and let's see if I've got it:

If I were to use my 2600MC with no binning (i.e. binx1) with an RC8, I'd get an image with a resolution of 6248 x 4176. However, I wouldn't actually be getting useful data across all those pixels because seeing conditions and mount limitations combine to, in effect, "blur" the view. When processing the data, I could binx3. This would make a square of nine pixels into one super-pixel. The plus points of this are that it would increase my SNR by a factor of around four. What takes me 20 hours with my current set-up would instead take five. Or, I could image for 20 hours and get a SNR equivalent to 80 hours with my current set-up. It would also decrease processing time due to the smaller file sizes. The downside is that the image's resolution would only be 2082 x 1392. This isn't that bad though, because the "lost" resolution wasn't useful data. I wouldn't be able to crop in very much with this lower-resolution image, but the RC8's focal length of 1600mm gets me very close in anyway, so that's not a big issue. If I need a wider FoV I could mosaic. 

Is that accurate? The lower final resolution still makes me itch a bit because I'm used to dealing with very high res images (both in astrophotography and regular photography) but I understand that with the proposed set-up it's not really a fair comparison. If I need a higher resolution image, e.g. for printing, I guess I could always use Photoshop's Enhance feature, or Topaz GigaPixel AI, to artificially bump up the pixel count. I just did an experiment doing that with some of my existing data, and it looks quite good to my eye. Enough to somewhat allay my fears about low-res images anyway.

Thanks vlaiv, I definitely owe you a pint! 🍻

You got it, with the one caveat that if you want to take advantage of the computational speed increase when dealing with the smaller files, you'd have to bin in camera, so the actual .fits output are images of 2082x1392.

That being said, one interesting application of binning an OSC, like your 2600MC, not in camera, but in post, is that you can essentially use the binned data as a Luminance layer, while the unbinned data, undersampled as it may be, would be user for RGB (PI's LRGB Combination to the rescue!). I haven't had a chance to try this myself, but it does sound like it's worth a shot (pun intended).

Edited July 26, 2022 by raadoo

[Clarkey]

34 minutes ago, Lee_P said:

I wonder, can you use the top Vixen dovetail as a carrying handle? It's hard to tell from the pictures if you could fit your fingers underneath it to get a good grip. Maybe with some risers? The collimation does sound like a pain, but it holds well once done?

You can't use the dovetail itself as it is too close to the tube you could probably attach risers or add a handle. I actually use mine to guide with an ST80 mounted on top towards the front. Helps with the balance a bit too as the RC8's are very back-heavy once loaded with photo gear.

WRT the collimation - once set it needs very little, if any adjustment. Mine was pretty good from delivery and only really required a minor 'tweak' to give a slight improvement. In fact the images in my earlier post were taken before I collimated the scope. I wanted to wait until the end of Galaxy Season to start messing with it - just in case!

[vlaiv]

1 hour ago, Lee_P said:

Four times faster, I think? RC8 binx3 = 84600; FRA400 binx1 = 19510.

Yes, sorry, my bad

1 hour ago, Lee_P said:

If I were to use my 2600MC with no binning (i.e. binx1) with an RC8, I'd get an image with a resolution of 6248 x 4176. However, I wouldn't actually be getting useful data across all those pixels because seeing conditions and mount limitations combine to, in effect, "blur" the view. When processing the data, I could binx3. This would make a square of nine pixels into one super-pixel. The plus points of this are that it would increase my SNR by a factor of around four. What takes me 20 hours with my current set-up would instead take five. Or, I could image for 20 hours and get a SNR equivalent to 80 hours with my current set-up. It would also decrease processing time due to the smaller file sizes. The downside is that the image's resolution would only be 2082 x 1392. This isn't that bad though, because the "lost" resolution wasn't useful data. I wouldn't be able to crop in very much with this lower-resolution image, but the RC8's focal length of 1600mm gets me very close in anyway, so that's not a big issue. If I need a wider FoV I could mosaic. 

All correct except one thing - bin x3 will give you x3 improvement in SNR - not x2 which equates to x9 imaging time and not x4.

SNR improvement is related to imaging time by square root function, so x4 imaging time (or x4 number of subs) - gives you x2 SNR improvement, while x9 imaging time gives you x3 SNR improvement.

Since you are binning x3 (joining 9 pixels - which is equal to stacking x9 subs or imaging for x9 longer) - you get only x3 improvement in SNR - or rather, SNR improvement is equal to bin factor.

1 hour ago, Lee_P said:

Is that accurate? The lower final resolution still makes me itch a bit because I'm used to dealing with very high res images (both in astrophotography and regular photography) but I understand that with the proposed set-up it's not really a fair comparison. If I need a higher resolution image, e.g. for printing, I guess I could always use Photoshop's Enhance feature, or Topaz GigaPixel AI, to artificially bump up the pixel count. I just did an experiment doing that with some of my existing data, and it looks quite good to my eye. Enough to somewhat allay my fears about low-res images anyway.

I know this fear exists with people - but that is just because people don't fully understand blur / resolution part of astronomical imaging. Resolution is not there to begin with.

If bin x3 is proper sampling and you wonder what you've lost over having 6000x4000 instead of 2000x1333 - answer is noting. If you want to have 6000x4000 image for say printing purposes or whatever - just take 2000x1333 image and upscale it to 6000x4000. Result will be the same as if you've shot 6000x4000 under original conditions.

 

[Lee_P]

26 minutes ago, vlaiv said:

All correct except one thing - bin x3 will give you x3 improvement in SNR - not x2 which equates to x9 imaging time and not x4.

SNR improvement is related to imaging time by square root function, so x4 imaging time (or x4 number of subs) - gives you x2 SNR improvement, while x9 imaging time gives you x3 SNR improvement.

Since you are binning x3 (joining 9 pixels - which is equal to stacking x9 subs or imaging for x9 longer) - you get only x3 improvement in SNR - or rather, SNR improvement is equal to bin factor.

Ok, does this mean that five hours with RC8 and binx3 is equivalent to 20 hours with FRA400 and binx1 (factor of four); or 15 hours (factor of three)?

 

28 minutes ago, vlaiv said:

I know this fear exists with people - but that is just because people don't fully understand blur / resolution part of astronomical imaging. Resolution is not there to begin with.

If bin x3 is proper sampling and you wonder what you've lost over having 6000x4000 instead of 2000x1333 - answer is noting. If you want to have 6000x4000 image for say printing purposes or whatever - just take 2000x1333 image and upscale it to 6000x4000. Result will be the same as if you've shot 6000x4000 under original conditions.

Totally, this was my (mis)understanding until you talked me through it.

[vlaiv]

33 minutes ago, Lee_P said:

Ok, does this mean that five hours with RC8 and binx3 is equivalent to 20 hours with FRA400 and binx1 (factor of four); or 15 hours (factor of three)?

If you compare two different setups - you should use Aperture^2 * Sampling_Rate^2 formula.

If you compare same setup - native, binned x2, binned x3 and so on - then you can use simplified formula - which is SNR = bin factor, or exposure time decreases by bin factor squared.

These are really the same - in second case, we keep aperture the same - so it can cancel out from equation and we are left with Sampling_Rate^2 part.

Binning increases numerical value of sampling rate and that is why time increase = bin_factor^2.

Hope that part makes sense.

36 minutes ago, Lee_P said:

Totally, this was my (mis)understanding until you talked me through it.

I find that most people best understand it by example that they can replicate, so here is what you can do to get the sense of resolution vs blur.

We can take some image - let's take M51 by Hubble team as our example:

It is very detailed image - very high resolution image. Look what happens if I take that sort of resolution in the image and I reduce it by factor of 3 and then enlarge it back by factor of 3

This is reduced image:

When we scale this back up and compare to original - we get:

Loss of detail and quality is obvious. When you take image that is already high resolution (not by pixel count but by level of detail), and you perform this - you obviously loose resolution.

But what happens if we do that on image that is already blurred?

This is image that much more resembles images taken by amateur astronomers (much lower res than Hubble's one).

Now we do the same - resample it to 1/3 of its original size - we get this:

Now, let's enlarge this small image and compare it to our blurred original:

Look at that - there is simply no difference between the two.

It really does not matter that we are so "close in" with target - if we don't capture the detail. We don't need to "spend" all those pixels capturing the target as there simply is no detail for all those pixels - 1/3 size image is enough to capture all the detail and we can recover "resolution" (or rather pixel count) - simply by enlarging image back - it will be the same as original one at "high resolution" (or rather high pixel count - resolution is really not pixel count but rather sharpness of the data).

You can yourself easily replicate above results - just use some high quality resampling (I used Lanczos) and Gaussian blur.

[Lee_P]

31 minutes ago, vlaiv said:

If you compare two different setups - you should use Aperture^2 * Sampling_Rate^2 formula.

If you compare same setup - native, binned x2, binned x3 and so on - then you can use simplified formula - which is SNR = bin factor, or exposure time decreases by bin factor squared.

These are really the same - in second case, we keep aperture the same - so it can cancel out from equation and we are left with Sampling_Rate^2 part.

Binning increases numerical value of sampling rate and that is why time increase = bin_factor^2.

Hope that part makes sense.

🥴 So am I looking at a reduction in integration time of three or four times..? (Although in reality I'll probably just stick with imaging a target for weeks at a time).

[vlaiv]

1 hour ago, Lee_P said:

🥴 So am I looking at a reduction in integration time of three or four times..? (Although in reality I'll probably just stick with imaging a target for weeks at a time).

If you compare RC8" with your current setup and you bin x3 RC8" one and you fit your target in FOV in both cases - you are looking at reduction of integration time (to reach the same SNR) - that is 1/4 - or 19510 / 86400 = 0.226 of current time.

[Lee_P]

@Clarkey Could you please do me a quick favour and measure the distance from the back of your RC8 to the back of your astrocamera? I'm trying to work out if I'd need an extension pier to stop the camera-end of the telescope thwacking into my pier... Thanks!

[ONIKKINEN]

Have you given thought to the darker skies side of things when it comes to reaching a decent SNR? 

If you visit a bortle 4 area (should be reasonable) you can get a vastly superior image in just one night compared to your current multi-week projects from bortle 8. Actually you might get 2 finished targets per trip, depends on how bad in the B8 end you are now.

This one was quite demoralizing to realize for me. I got 2 hours from a B4 area and compared it to more than 6 from B6-7 and, you guessed it, the 2 hours from dark skies slapped the 6 around like it was a cat toy. More effort though, so not a completely fair comparison but nothing is.

[Clarkey]

29 minutes ago, Lee_P said:

@Clarkey Could you please do me a quick favour and measure the distance from the back of your RC8 to the back of your astrocamera? I'm trying to work out if I'd need an extension pier to stop the camera-end of the telescope thwacking into my pier... Thanks!

From the back of the scope to the back of the camera - making allowances for cables and approximate point of focus - is about 43cm. Bearing in mind the back heavy nature of the scope and the width, I have never had a problem with it hitting the tripod. This is with an AZ-EQ6.

[Lee_P]

2 hours ago, ONIKKINEN said:

Have you given thought to the darker skies side of things when it comes to reaching a decent SNR? 

If you visit a bortle 4 area (should be reasonable) you can get a vastly superior image in just one night compared to your current multi-week projects from bortle 8. Actually you might get 2 finished targets per trip, depends on how bad in the B8 end you are now.

This one was quite demoralizing to realize for me. I got 2 hours from a B4 area and compared it to more than 6 from B6-7 and, you guessed it, the 2 hours from dark skies slapped the 6 around like it was a cat toy. More effort though, so not a completely fair comparison but nothing is.

For sure, but I rarely get the chance to travel, and have a semi-permanent set-up in my garden that makes set-up and pack-down here easy, but conversely would be a right pain to travel with. Plus, astrophotography from a city is kinda my thing 😆

[Lee_P]

1 hour ago, Clarkey said:

From the back of the scope to the back of the camera - making allowances for cables and approximate point of focus - is about 43cm. Bearing in mind the back heavy nature of the scope and the width, I have never had a problem with it hitting the tripod. This is with an AZ-EQ6.

OK, thanks, should be alright with my pier then!

[Lee_P]

@Clarkey, could I please impose another question on you... Could you explain how you get the correct backfocus with a StellaLyra RC8? I asked FLO and got this answer:

"Backfocus is around 255mm from the rear of the tube, a number of extensions are provided to reach this which are threaded in front of the focuser and added or removed as needed. If you set up your current ASI2600MC imaging train, you attach that to the focuser, rack out about halfway and then add tubes to make up the difference needed. You can then fine-tune via the focuser."

What's puzzling me is that if you adjust backfocus using the focusser, then does it not follow that when you use the focusser to focus your image, that will naturally change the backfocus length? I'm sure I'm getting the wrong end of the stick... 

[Len1257]

On 26/07/2022 at 14:13, Lee_P said:

 

I wonder, can you use the top Vixen dovetail as a carrying handle? It's hard to tell from the pictures if you could fit your fingers underneath it to get a good grip. Maybe with some risers? 

I fitted a cut-down handle from WDS Components (as recommended on here!) on a couple of 3D risers at the rear of the top Vixen bar and then a guidescope to the front half of the same bar. I needed a clamp-on counterweight added to the bottom Losmandy bar to achieve balance. I had some pictures somewhere but typically can't find them, I'll take some more later. The short handle is just fine as all the weight is at the rear of the scope. The risers were needed as my lecky boxes sit side saddle in the same location. Picture's worth any number of words.........

[Clarkey]

3 hours ago, Lee_P said:

Could you explain how you get the correct backfocus with a StellaLyra RC8? I asked FLO and got this answer:

I think maybe you are over-thinking this! The 'back-focus' is just the focus point as with any other scope. For normal AP I use the four extensions provided (100mm) and then the focuser and focus normally - but the focus point is a long way back. As the primary and secondary mirrors are not moving you are not changing anything in terms of the optics. Although the supplier focuser is OK, I did change mine to a Baader ST just to make sure everything was rock solid. Not sure it was necessary really but it was nice, new and shiny😁. I don't have my rig set up at the moment, but here is a picture of roughly where my RC8 focuses (there is a tilt adjuster too which adds about 20mm).

As @Len1257 says the weight is far back. I use a combination of guidescope and homemade power box to balance.

[CraigT82]

4 hours ago, Lee_P said:

What's puzzling me is that if you adjust backfocus using the focusser, then does it not follow that when you use the focusser to focus your image, that will naturally change the backfocus length? I'm sure I'm getting the wrong end of the stick... 

Maybe you’re thinking about moving mirror cassegrains here. With this scope the mirrors are fixed and so the focal plane is also fixed at 255mm behind the scope, and the focuser just serves to move the cam/ep to coincide with the focal plane.

With a normal SCT or mak the mirror moves and so does the focal plane. In this design the focuser actually moves the focal plane to coincide with the position of the cam/ep which is fixed. There is an ideal ‘spot’ for the focal plane which corresponds to the point of best correction (spherical) I.e the proper distance between corrector plate and primary).

[Lee_P]

@Len1257 @Clarkey @CraigT82 Thanks for your help! As RC8 owners, can I just check something else... There are lots of RC8 telescopes out there with different branding, but they're all basically the same, right? Reason I'm asking is that the StellaLyra carbon fiber models are out of stock for a long time, but it looks like Teleskop Express and potentially 365 Astronomy have some TS models in stock. I'd like to try mounting the 'scope on my Orion EQ-G mount (similar to HEQ5), so a lower weight would be welcome. Also, these particular ones have Vixen dovetails top and bottom, which would save me having to buy a Losmandy adapter, given that my mount can currently only accept Vixens. That cost saving alone justifies the extra price of the carbon fiber models. So, I just want to check that there isn't something majorly different about the StellaLyras that I'm missing.

[Len1257]

Actually got around to taking some pictures! The handle is a WDS Handle and shortened from the next size up. I lifted it on a couple of 3D risers to clear the side-saddle plate that my electrics modules are velcroed to. I had to drill and tap the Vixen bar to suit M8. The 3D modules are for power distribution and dual dew heaters. The other two images show the 50mm and 25mm extension tubes + tilt Plate + 90mm adapter + Baader SteelTrack focuser + Baader Clicklock + ZWO filter Drawer + ZWO camera. As can be seen the focuser is almost fully in and the auto focus has enough scope with what is showing. I bought A Stellalyra carbon tube so can't comment on any others except that they were more expensive when i bought mine. Ordered in Jan and came in Feb.

[Len1257]

Just for completeness here's the belt drive focuser. Another 3D printed mount.

[Clarkey]

4 hours ago, Lee_P said:

There are lots of RC8 telescopes out there with different branding, but they're all basically the same, right?

As far as I know there is not much between them optically. Focusers and other bits might be slightly different, but as far as I'm aware pretty much the same.

My only question is regarding cool down time with carbon tubes. My 90mm Stella refractor is very slow to cool.

[Len1257]

The RC is an open tube so not like a refractor at all. Cooldown just the same as any open tube jobby. By the time I've set all the rest of the gubbings up, done a polar alignment and platesolved a few targets I reckon it's pretty much there.

[Lee_P]

So I went for the RC8, and have been busy putting it all together. I've got the balance sorted ok, guiding to around 0.8", and I've installed an upgraded focusser that works much better with my ZWO EAF than the stock one.

 

My next hurdle is collimation and / or sensor tilt. I've had a go at collimating, but I'm probably quite wide of the mark. I had about 10 mins of clear skies tonight, so took two test images of Vega: one in focus, one out of focus. (That's useful for testing collimation, right?) I wonder if you kind folks could take a look and offer some feedback on what I need to change? Some of the stars look a bit iffy 🥴 Thanks in advance!

VEGA DEFOCUS__5.0s_Bin1_2600MC_gain100_20220819-232216.fit

VEGA_10.0s_Bin1_2600MC_gain100_20220819-233436.fit

[vlaiv]

10 hours ago, Lee_P said:

My next hurdle is collimation and / or sensor tilt. I've had a go at collimating, but I'm probably quite wide of the mark. I had about 10 mins of clear skies tonight, so took two test images of Vega: one in focus, one out of focus. (That's useful for testing collimation, right?) I wonder if you kind folks could take a look and offer some feedback on what I need to change? Some of the stars look a bit iffy 🥴 Thanks in advance!

Actually you are quite close to good collimation.

It needs to be touched up a bit, but it is pretty good as is. There is some astigmatism and distortion in outer stars that need to be addressed.

Defocused image of Vega is only good for checking secondary collimation. And it is just a bit off as well.

Here is an old trick that shows well if you have defocused image that is not concentric:

It is same image in two frames - one normal, other rotated by 180 degrees. Now you can clearly see that there is a bit of "squish" on one side and that circles are not concentric.