24 replies to this topic

[rasmubur000]

Hi all, I'm working on designing a 12" dob, and have a handful of questions regarding mirror selection and mirror cell design.

 

Being in the US, there seems to not be many options for purchasing mirrors... I've narrowed my choices down to just getting a GSO mirror set for ~$750 after shipping, or going for the hubble optics 12" f5 mirror, which including a secondary and shipping comes to ~$1200. What are your thoughts on if it's worth spending the extra money for the hubble optics mirror? Is the difference that noticeable? 

 

 

Regarding the mirror cell, using the PLOP software, I'm seeing that a 6 point floating support works best, and that a sling is very important. That being said, I'm struggling to understand and decide if it's better to use a cable sling that directly touches the mirror, like used on hubble optics telescopes? Or to do like other designs that just have 2 points with soft plastic 90 degrees apart as the bottom lateral support. I'll try to post a couple of the designs I'm referring to. 

 

[brebisson]

Hello,

 

For 30cm, 6 points is indeed great.

 

simple side supports at 90° would also work well and might be ok.

 

You can use bearings set vertically, but I have been using linear bearings (used a lot in small cnc and 3D printers). which might be easier to use (https://fr.aliexpres...0909626059.html)

 

300 are not heavy and quite forbidding, so don't sweat it...

 

GSO are OK in my opinion, they are at l/4 which is perfectly OK for visual.

I would not go for the HO unless you are an exceptional visual observer.

 

Cyrille

[Pinbout]

I prefer a moving frame to collimate the mirror so I often use a 6 point cell.

 

but on the edge, I always use a wiffletree edge support. 

 

https://www.youtube....h?v=oDKj1ZkqGKI

 

the hubble will be better figure, from my experience. is the cost worth it for you? if you like low mag views, gso is fine.

 

here's a 12" moving frame cell. I would typically use delrin wheels from Home Depot [drawer glides] on the edge support but this one I'm using nylon socket cap head screws... they will bend as the mirror shifts.

 

https://www.youtube....h?v=76LjhPB1EHk

 

a hanging moving cell frame to collimate from the front for an UL scope.

 

https://www.youtube....h?v=ifSMbmCxf1I

Edited by Pinbout, 19 June 2025 - 09:55 AM.

[MeridianStarGazer]

Hubble is soda lime, whereas GSO is BK7. Actually for a bit more they have fused silica. But I will argue that is not important, considering the HO is only 1" thick. It will cool twice as fast.

[MeridianStarGazer]

Please post your mirror thickness and wave error for 6 points and the allowable wave error. I can use those to extrapolate for other mirror sizes, thicknesses, and cells.

[Tangerman]

Please post your mirror thickness and wave error for 6 points and the allowable wave error. I can use those to extrapolate for other mirror sizes, thicknesses, and cells.

Or you could just run Plop and get accurate results for whichever design and mirror you could fancy.

[MeridianStarGazer]

Or you could just run Plop and get accurate results for whichever design and mirror you could fancy.

He already ran the numbers. And I don't need plop once I have one set of numbers.
I could dig through old posts.

Some of us are not software inclined.

[rasmubur000]

My question and concern with the wiffletree edge support vs the sling is that I don't know if it's possible to simulate in Plop how this type of support effects the mirror, and why it's preferred over a sling?

 

Looking at the plop outputs below, the 2 mirror options don't seem to be much different, though obviously the thicker GSO mirror has less deformation, they are both within acceptable range. 

 

Personally, a thicker mirror doesn't bother me, as I always arrive at my observing site with plenty of time before total darkness for setting up the telescope. For mirror selection I'm more concerned if when looking at planets, the moon, or DSOs, how much I'd be sacrificing on image quality purely based on which mirror I get. I'll add, if I got the GSO mirror, I'd be able to afford some nicer eyepieces than the gold line eyepieces I use on my 8" currently. 

 

 

Here's the Plop outcomes for the 24mm thick Hubble Optics 12" mirror

 

And the GSO 38mm thick 12" mirror.

Edited by rasmubur000, 19 June 2025 - 12:23 PM.

[Dale Eason]

This link may help.  https://www.cruxis.c...ecalculator.htm

 

Slings when done well and placed well work well.  However they are more fussy to get right when out in the field than the two edge support.  Slings can have a lot of friction that makes them hard to position correctly unless you have some form of keepers to hold them in the right place.

 

For a 12 inch mirror a wiffletree edge support is not needed.  It too is a little fussier than a single point in that if the axis of the lever support is not parallel to the optical axis then one point of the lever arm will be behind the other on the mirror's edge.  Yet you need both to be within about 1/16 of an inch of the COG of the mirrors edge.

[Pinbout]

 

how this type of support effects the mirror, and why it's preferred over a sling?

the whiffletree edge support is relatively a newer concept in amatuer telescope making. 

 

it keeps the mirror from sloshing around.

 

the sling you still need 2 posts to keep the mirror from moving too far

 

both the whiffletree and sling suppor the mirror at the center of gravity on the edge.

 

form my Hubble i made whiffetrees at the end of my whiffletrees since it has two layers. 

Edited by Pinbout, 19 June 2025 - 12:49 PM.

[Tangerman]

My question and concern with the wiffletree edge support vs the sling is that I don't know if it's possible to simulate in Plop how this type of support effects the mirror, and why it's preferred over a sling?

I'm not sure that Plop is written to simulate the different types of edge support, so I can't offer you anything quantitative. I think much of the drive towards whiffletree support was driven by this article: https://www.loptics....rorsupport.html

 

It's possible that there was prior work that I'm not referencing, but I wasn't interested in astronomy at the time this was developing, and the above is a good article with lots of documentation.

 

For a 12" mirror, I'd suspect that a whiffletree support or a cable sling would work just fine, and it really just depends on what you prefer.

 

I don't have extensive experience with GSO vs Hubble Optics. I might be able to go out this weekend with my new telescope, which has a Hubble Optics mirror, and I have prior experience with my GSO 10", but I don't think I'll be able to really compare the two, the apertures are so different and I won't have both side-by-side. 

Edited by Tangerman, 19 June 2025 - 01:22 PM.

[sixela]

A whiffle tree edge support is complete overkill for a GSO mirror of that size. Two points in the plane of the COG at +-45° will do just fine, and it's a lot less easy to mess it up.

For the Hubble Optics you do need a whiffle tree but along the optical axis (!) to support each mirror slab independently.

Even 6 point cells are almost overkill for the rather thick GSOs of that size -- they do "just fine" on three points, really, unless you really want to worry about fairly irrelevant aberrations.

Edited by sixela, 19 June 2025 - 01:04 PM.

[rasmubur000]

I'm not sure that Plop is written to simulate the different types of edge support, so I can't offer you anything quantitative. I think much of the drive towards whiffletree support was driven by this article: https://www.loptics....orsupport.html 

 

It's possible that there was prior work that I'm not referencing, but I wasn't interested in astronomy at the time this was developing, and the above is a good article with lots of documentation.

 

For a 12" mirror, I'd suspect that a whiffletree support or a cable sling would work just fine, and it really just depends on what you prefer.

 

I don't have extensive experience with GSO vs Hubble Optics. I might be able to go out this weekend with my new telescope, which has a Hubble Optics mirror, and I have prior experience with my GSO 10", but I don't think I'll be able to really compare the two, the apertures are so different and I won't have both side-by-side. 

Tangerman, thanks for the article, though your link was a bit broken. I just had to delete some stuff from the end of the URL. I appreciate being able to see the artificial star tests shown there to really demonstrate the differences and to back up his explanations.

[rasmubur000]

This link may help.  https://www.cruxis.c...ecalculator.htm

 

Slings when done well and placed well work well.  However they are more fussy to get right when out in the field than the two edge support.  Slings can have a lot of friction that makes them hard to position correctly unless you have some form of keepers to hold them in the right place.

 

For a 12 inch mirror a wiffletree edge support is not needed.  It too is a little fussier than a single point in that if the axis of the lever support is not parallel to the optical axis then one point of the lever arm will be behind the other on the mirror's edge.  Yet you need both to be within about 1/16 of an inch of the COG of the mirrors edge.

Yes, that helps a ton. It makes a lot of sense to use this in tandem with Plop, as Plop doesn't have many options for edge support.

 

 

 

To respond to others' comments, I have some machining skills and access to my university's machine shop, so I'd rather just do wiffletree edge support, since I'm already going to be making wiffletree supports for the bottom of the mirror. All the resources y'all have provided will help give me the background and theory to properly support a mirror cell. I'll probably do a design very similar to Pinbout's, making the mirror cell frame out of CNC'd 3/4" plywood, or use sendcutsend to make it out of cut and bent sheet metal.

 

 

I'm still unsure if I'll go with GSO or Hubble Optics for the mirror. It seems like there's so many other factors on what you see through a scope than just the mirror that it'd be hard to isolate without just buying both to test directly. I certainly would appreciate having the reassurance of the actual test results from a US manufacturer rather than getting a "diffraction limited" mirror with no receipts for half the price.

Edited by rasmubur000, 19 June 2025 - 01:33 PM.

[Tangerman]

Tangerman, thanks for the article, though your link was a bit broken. I just had to delete some stuff from the end of the URL. I appreciate being able to see the artificial star tests shown there to really demonstrate the differences and to back up his explanations.

I don't know why it didn't work. I think I've fixed the link now.

 

Anyways, lots of good points being made here. If you're using a Hubble Optics sandwich mirror, you'll need the whiffletree to properly support it. Mine has that — maybe I'll get a picture next time I'm out with it. The GSO would be fine with either support, or even just the 90° posts as the Cruxis site that Dale linked to indicates. 

 

I think the main draws of the Hubble Optics sandwich mirrors are faster cooling and a much lighter mirror. If you live in Utah, faster cooling is important, temperatures keep dropping throughout the night in the desert, unlike where I currently live. Of course, you'll have to decide if those benefits are worth $450 to you. It's possible that the GSO would still cool well enough if you get a good fan setup to cool it.

[rasmubur000]

oh, important to note is that Hubble Optics uses solid plate glass, nonsandwich for the 12" f5 that I'm looking at. It is still 24mm thick compared to the 38mm of the GSO. 

[Starman1]

Hi all, I'm working on designing a 12" dob, and have a handful of questions regarding mirror selection and mirror cell design.

 

Being in the US, there seems to not be many options for purchasing mirrors... I've narrowed my choices down to just getting a GSO mirror set for ~$750 after shipping, or going for the hubble optics 12" f5 mirror, which including a secondary and shipping comes to ~$1200. What are your thoughts on if it's worth spending the extra money for the hubble optics mirror? Is the difference that noticeable? 

 

 

Regarding the mirror cell, using the PLOP software, I'm seeing that a 6 point floating support works best, and that a sling is very important. That being said, I'm struggling to understand and decide if it's better to use a cable sling that directly touches the mirror, like used on hubble optics telescopes? Or to do like other designs that just have 2 points with soft plastic 90 degrees apart as the bottom lateral support. I'll try to post a couple of the designs I'm referring to. 

 

 

OK, for mirror cells you want to minimize wavefront deformation.

For the edge, a cable sling contacting 180° is best, a 4-point Whiffle tree is next best, then two points 90° apart and worst, 2 points 120° apart.

For the back support, an 18 point cell is best, then 6-points.  For come reason, 9 points is worse than either 6 or 18.

These considerations are all about minimizing wavefront distortion to as little as physically possible.  The thinner the mirror, the more important a large number of points of support are.

 

I've run the numbers for wavefront deformation (RMS) for the various options:

GSO (Edge)

Cable 0.2nm

Whiffle 0.3nm

90° 0.6nm

 

Hubble (Edge)

Cable 0.3nm

Whiffle 0.4nm

90° 0.8nm

 

The wire sling wins, but it is easier to remove a mirror from a whiffle tree design and the difference is only 0.1nm RMS (1/500 wave at 500nm)

 

GSO (back) Peak-Valley

6 points 5.9nm (0.01 wave)

18 points 1.9nm  (0.004 wave)

 

Hubble (back) Peak-Valley

6 points 9.1nm (0.02 wave)

18 points 2.2nm (0.0044 wave)

 

The 18 point cell wins by a large margin.

 

To put all this into perspective, a 1/4 wave or 0.25 wave mirror is common at these prices.  6 point cells add a bit too high a % of wavefront deformation to that figure to be considered.

Also, your mirror might be better than 1/4 wave P-V, and then adding the wavefront error from 6 point cells is even worse.

Edited by Starman1, 19 June 2025 - 02:29 PM.

[rasmubur000]

 

To put all this into perspective, a 1/4 wave or 0.25 wave mirror is common at these prices.  6 point cells add a bit too high a % of wavefront deformation to that figure to be considered.

Also, your mirror might be better than 1/4 wave P-V, and then adding the wavefront error from 6 point cells is even worse.

Thanks Don, I think that sums up the mirror cell support analysis quite nicely.

 

It was my understanding that while a 6 point cell is much worse than 18 points, the difference in RMS is within what is considered perceivable. 

This is from the GuiPlop Users Manual:

"A reasonable limit for the RMS error due to the mirror cell is about 1/128 wave, which is about 4.2e-9
meter, or 4.2e-6 mm, for 550nm green light. This means 1/64 wave on the wavefront reflected from the
mirror, and corresponds to about 1/16 wave peak to valley. This is based on Toshimi Taki's suggestion that
about 1/4 of the error budget may be allocated to the mirror cell."

 

I'm wondering if you could provide any insight on this?

 

Another question this brings up for me: How do I determinine what rms error threshold I should be considering?

 

 

Here's the info on the mirrors:

Hubble optics: 1/10 or better in PV, about 1/50 in RMS, and 0.95 or better in Strehl ratio, with the 96% enhanced aluminum coating

GSO: GSO guarantees diffraction limited performance, but their mirrors typically have a mirror surface quality of 1/16 wave RMS at least, with 92-93% coating.

[Starman1]

Thanks Don, I think that sums up the mirror cell support analysis quite nicely.

 

It was my understanding that while a 6 point cell is much worse than 18 points, the difference in RMS is within what is considered perceivable. 

This is from the GuiPlop Users Manual:

"A reasonable limit for the RMS error due to the mirror cell is about 1/128 wave, which is about 4.2e-9
meter, or 4.2e-6 mm, for 550nm green light. This means 1/64 wave on the wavefront reflected from the
mirror, and corresponds to about 1/16 wave peak to valley. This is based on Toshimi Taki's suggestion that
about 1/4 of the error budget may be allocated to the mirror cell."

 

I'm wondering if you could provide any insight on this?

 

Another question this brings up for me: How do I determine what rms error threshold I should be considering?

 

 

Here's the info on the mirrors:

Hubble optics: 1/10 or better in PV, about 1/50 in RMS, and 0.95 or better in Strehl ratio, with the 96% enhanced aluminum coating

GSO: GSO guarantees diffraction limited performance, but their mirrors typically have a mirror surface quality of 1/16 wave RMS at least, with 92-93% coating.

Most after market tests show that these mirrors *barely* meet the Maréchal Condition (0.80 Strehl), and some do not.

So I would think it wise to add as little as possible from errors you can prevent.

 

The eye is most sensitive at night at ~500nm due to the Purkinje Effect, which increases the error in the mirror because of a shorter wavelength.

(one of the reasons I don't like to see interferometric analyses done with a red laser at 650nm.  A 0.05 wave error at 650nm is 0.065 wave error at 500nm).

 

A "system" (mirror cell, sling, primary mirror, secondary mirror, eyepiece) that yields 1/4 wave P-V on the wavefront, in good seeing, will yield good images.

If you start out with a primary mirror that is 1/4 wave P-V and then add to it with errors from other sources, then the final wavefront will be poorer.

It's why I recommend the most accurate secondary you can afford, and to minimize errors from the mirror cell and sling.

 

Admittedly, Seeing conditions often make the best optics mediocre, but we deal with what we can control.

I'm not comfortable with the hardware adding 1/16 wave P-V.  Maybe that's because I've had higher end mirrors for decades.

Maybe it's because the scope builders I used made scopes with low error mirror cells.

 

It's also a matter, obviously, of how thick the mirror is.  The 2 mirrors you contemplated had very different thicknesses: 8% and 12.5% of aperture.

At that aperture, temperature equilibrium is required for the best image quality, so the thinner mirror would have the advantage.

And, since you're the designer, add both a rear fan and some cross-vent fans and you will see what the mirror can deliver a lot sooner in the night.

I just sold a 12.5" scope with a 32mm thick mirror and it cooled very quickly with 3 fans, and stayed at the ambient temperature all night.

My previous 12.5" scope had a 55mm thick mirror and even with a fan, it didn't get to equilibrium until after 2am with the fan started at sunset.

I'd go for the thinner mirror.  It'll be easier to transport if leaving it in the mirror box during transport.

Edited by Starman1, 19 June 2025 - 03:45 PM.

[Pinbout]

Here’s a bit of history, for dobs that just use the collimation bolts attached to the rear support points (not a moving plate) the cable sling on linear bearings is a God send.

 

but with a 12” on a moving plate cell, whiffletree edge support is best. 

[Oberon]

Ditto for hexapod, which assumes a fixed cell and has no collimation mechanisms for the primary mirror, greatly simplifying mirror support (full collimation is achieved with turnbuckle trusses).

[Arjan]

Starman1: the wavelength of the laser used for intererometry has nothing to do with how surface error is reported. DFTFringe scales this to a standard wavelength, 550nm I believe.

[Dale Eason]

Starman1: the wavelength of the laser used for intererometry has nothing to do with how surface error is reported. DFTFringe scales this to a standard wavelength, 550nm I believe.

Yes that is correct.  Not any different really than measuring a distance in inches but reporting it in centimeters.  Most modern interferometry analysis programs usually report the result around the 550nm region.  They usually indicate what wave length they report the wave front at as well.

Edited by Dale Eason, 20 June 2025 - 02:40 AM.

[Arjan]

Another Mirror cell design consideration is the diminishing returns you get with many-point cells.

Sixela already mentioned the risk of overdesign in an earlier post.

 

While you theoretically may get 1/500 or better cell induced deformation according to PLOP, there are many implementation parameters that are not considered by this tool. The increased complexity of many-point cells require very careful engineering to get close to the theoretical value.

 

David Lewis says 1/128 surface error is a good limit, but I think he is also on the safe side. You need very careful interferometric methods to actually verify this, in controlled circumstances (on a bench in a lab).

Also note that both RMS and PV errors don't just add up, so I am not exactly sure what is meant by Taki's statement that max 25% of the total error budget should be reserved to the cell. What are the other sources? How much error do you allow for your typical use-case?

 

IMHO it is better to go for simplicity rather than theory. Then 3 or 6 point cells generally suffice for average sized mirrors. That and two-point edge support (for an AltAz mount). And by all means, use PLOP to assess this.

[brebisson]

Hello

 

>IMHO it is better to go for simplicity rather than theory.

>Then 3 or 6 point cells generally suffice for average sized mirrors.

>That and two-point edge support (for an AltAz mount).

 

Hello, I 100% agree with you here!. 3 is enough up to 250, 6 up to 400.

This applies to any "normal" mirror. super thin mirrors are different, also if you do astrophoto, you might want to make better cells as astrophoto is less forgiving. but we were talking about "standard" here..

 

>And by all means, use PLOP to assess this.

 

For 3 point cells, you don't even need to, assuming a standard sized secondary it will end up being at the 38-40% radius, which is as precise as you are going to get with your supports which will be non punctual (plop assumes punctual supports).

 

cyrille