700 replies to this topic

[tommm]

And now for the first results...
 

 

...As anticipated, this spider proved very stiff. The errors measured were smaller than the laser beam spot, and for the most part lost in the noise. Virtually all motion detected was due to the structure, and hence cancelled out by the green laser...

Surprised you say "as anticipated", considering your remark in "An offset wire spider" thread:

 

"What I find interesting about this offset spider method is that it is not only geometrically and structurally superior, but simpler and cheaper to make (and make well) than centered vanes."

 

Maybe that comment was regarding a non-offset vane spider, not the offset one you replaced?

 

So your reasons for going to a wire spider were lower mass (easier OTA balance) and decreased diffraction, not improved stiffness? 

Edited by tommm, 27 April 2019 - 10:11 AM.

[Oberon]

Very much; the driver for wire was reduced mass, benefit was reduced diffraction. The question with wire is always whether or not it is stiff enough, hence the attention given to geometry. In my case the wire spider certainly “feels” stiff enough, and a significant improvement on my classic X Novak spider on my 8 inch. However I never deluded myself it was anywhere near as rigid as the solid vane spider tested here; I recall saying somewhere else that I reckoned I could lift Merope with it it was so strong. Hence my demonstration a few posts back using the hoist.

 

Anyway I haven’t written wire off yet; Merope’s wire spider always had room for improvement (reduced mass and greater angles) both of which have been noted previously. In time I will build a range of wire spiders as part of this project to quantify performance changes with design changes; that is exactly what my UTA is configured to do. I am also looking forward to a spider built with SS cable ties instead of wires, and my tests on curved spiders. I suspect we’ll see many ugly performers.

 

Finally...its worth noting that this spider is so stiff that the limitations of the test structure dominated. In the real world that would mean the telescope truss or tube structure.

[Oberon]

Oberon! Wow thanks for this awesome post and using your time and money for all this research. I've been following from the beginning.

So from looking at the latest results, the offset steel spider seems unbeatable as compared to the wire spider?

I have been going back and forth as to which design to use on the 16" f4.5 I'm building. Had practically made up my mind to go with a wire spider to minimize diffraction and it also allows collimation making the holder simple and solid to build.

Now with these new results I'm kinda doubting on the wire spider. Any recommendations or advise? Please note that I don't have and probably won't get a paracor...

Thanks!!

An offset steel spider is probably unbeatable. I cannot envisage anything improving on a well made offset steel spider for rigidity. However there may be other considerations such as mass and radiation to be mindful of.

 

The next question is how thin can the vanes be? I reckon they can be vanishingly thin, and am planning to try a brass foil that is a ridiculous 0.02mm thin. Just to see what happens.

 

As for wire...watch this space, and be mindful that my experiment is more severe than a Dob in real use. 

[Starman1]

Coma correction is compromised if the optical axis doesn't intercept the center of the coma correcting lens.

If it's even 1mm off the center of the lens, coma will not only not be corrected, but additional coma will be added to the axial image.

Hence, centering the optical axis on the lens is critical.

 

With the Paracorr 1, the math showed a tightening of tolerances by a factor of 6:1 compared to using no Paracorr.

I have not seen a similar math done for the Paracorr II, but the likelihood is that because the correction extends to a wider field and to lower f/ratios,

it is tighter.  Vice Menard may know exactly, but I've read that tolerances go from 0.03D on the focuser axis error without a Paracorr II to around 0.002D with it, which is about

a 15:1 tightening of focuser axial error tolerances.

 

I have not read where collimation tolerances for the primary tighten from the already tight 0.005 f/R³ 

 

To put that in perspective, with a typical 400mm f/4.5, that would be:

FAE tolerances: 0.8mm

PAE tolerances: 0.46mm

The impact of such tolerances are that the scope must be exceedingly stiff to maintain the tolerances in use

and that even reading such tolerances requires tools more sensitive than most.

For such a scope, I would contend only a laser with an aperture stop might be adequate for intitial FAE set up, a calibrated Cheshire for the PAE and

an autocollimator with dual pupils to get close to the actual tolerances necessary.

A star test could refine the PAE, but I don't see how a star test could duplicate the FAE reduction of the autocollimator.

 

This also shows how important it is that however the Paracorr II is shoved off-center by thumb screws, the collimation tools must be so as well.

Centering does not get the tolerances tight enough unless everything is centered, and I don't see that being done in 2" focusers.

Edited by Starman1, 27 April 2019 - 12:24 PM.

[mark cowan]

Very much; the driver for wire was reduced mass, benefit was reduced diffraction. The question with wire is always whether or not it is stiff enough, hence the attention given to geometry. In my case the wire spider certainly “feels” stiff enough, and a significant improvement on my classic X Novak spider on my 8 inch. However I never deluded myself it was anywhere near as rigid as the solid vane spider tested here; I recall saying somewhere else that I reckoned I could lift Merope with it it was so strong. Hence my demonstration a few posts back using the hoist.

 

Anyway I haven’t written wire off yet; Merope’s wire spider always had room for improvement (reduced mass and greater angles) both of which have been noted previously. In time I will build a range of wire spiders as part of this project to quantify performance changes with design changes; that is exactly what my UTA is configured to do. I am also looking forward to a spider built with SS cable ties instead of wires, and my tests on curved spiders. I suspect we’ll see many ugly performers.

 

Finally...its worth noting that this spider is so stiff that the limitations of the test structure dominated. In the real world that would mean the telescope truss or tube structure.

 

Just in general, minimizing the moment arm of the secondary support on a wire spider works wonders.  If you don't do that, the solid vanes always win because they can resist the torque better.  Think of the vane as being a big collection of wires pasted together edge-to-edge. 

[mark cowan]

Coma correction is compromised if the optical axis doesn't intercept the center of the coma correcting lens.

If it's even 1mm off the center of the lens, coma will not only not be corrected, but additional coma will be added to the axial image.

Hence, centering the optical axis on the lens is critical.

 

With the Paracorr 1, the math showed a tightening of tolerances by a factor of 6:1 compared to using no Paracorr.

I have not seen a similar math done for the Paracorr II, but the likelihood is that because the correction extends to a wider field and to lower f/ratios,

it is tighter.  Vice Menard may know exactly, but I've read that tolerances go from 0.03D on the focuser axis error without a Paracorr II to around 0.002D with it, which is about

a 15:1 tightening of focuser axial error tolerances.

 

I have not read where collimation tolerances for the primary tighten from the already tight 0.005 f/R³ 

 

To put that in perspective, with a typical 400mm f/4.5, that would be:

FAE tolerances: 0.8mm

PAE tolerances: 0.46mm

The impact of such tolerances are that the scope must be exceedingly stiff to maintain the tolerances in use

and that even reading such tolerances requires tools more sensitive than most.

For such a scope, I would contend only a laser with an aperture stop might be adequate for intitial FAE set up, a calibrated Cheshire for the PAE and

an autocollimator with dual pupils to get close to the actual tolerances necessary.

A star test could refine the PAE, but I don't see how a star test could duplicate the FAE reduction of the autocollimator.

 

This also shows how important it is that however the Paracorr II is shoved off-center by thumb screws, the collimation tools must be so as well.

Centering does not get the tolerances tight enough unless everything is centered, and I don't see that being done in 2" focusers.

This raises issues of knowing exactly where the optical axis of the mirror is relative to the physical center - or perhaps how closely the optical axis agrees with the latter.  Not for this thread though... except in the sense of minimizing spider misalignment.

[Oberon]

 

Just in general, minimizing the moment arm of the secondary support on a wire spider works wonders.  If you don't do that, the solid vanes always win because they can resist the torque better.  Think of the vane as being a big collection of wires pasted together edge-to-edge. 

Well yes. That is one test that is easy to do and I will demonstrate.

Solid vanes utilising the same geometry will always win over wires because there is more matter to resist stretching (flexure). Wire spiders use so little material we need to apply every trick in the book just to make them viable (good geometry, minimising moment, minimising mass etc). As such they are a great education tool for identifying what matters. Apply the lessons learned in making a viable wire spider to a solid vane spider and you should end up with excellent support.

 

[tommm]

 

Just in general, minimizing the moment arm of the secondary support on a wire spider works wonders.  If you don't do that, the solid vanes always win because they can resist the torque better.  Think of the vane as being a big collection of wires pasted together edge-to-edge. 

That reduces pitch, but not yaw, (edit: I obviously assumed here that you meant moment arm for pitch, but maybe you didn't) and it seems most secondary holders that reduce the moment arm for pitch trade-off an increased moment arm for yaw.  Consider Oberon's holder (vane or wire spider). The moment arm for pitch can be reduced to zero by placing the plane of attachment of the vanes or wires directly over the CoM of the mirror, but there will be greater moment arm for yaw in that case compared to if he places that plane further back from the CoM which permits shifting it down to reduce yaw. Hence my question sometime back if one should seek to reduce one moment arm more than the other.  I would guess that a spider resists yaw much better than pitch, so one should do as you suggest.

Edited by tommm, 27 April 2019 - 05:02 PM.

[ckh]

Jonathan,

 

What is the wire diameter on your spider?

[tommm]

Well yes. That is one test that is easy to do and I will demonstrate.

Solid vanes utilising the same geometry will always win over wires because there is more matter to resist stretching (flexure). Wire spiders use so little material we need to apply every trick in the book just to make them viable (good geometry, minimising moment, minimising mass etc). As such they are a great education tool for identifying what matters. Apply the lessons learned in making a viable wire spider to a solid vane spider and you should end up with excellent support.

 

There is more matter, but most of the tension occurs in an X pattern in the plane of the vane, which is the geometry of the wire spider.  Then if one assumes the wires do not stretch, what permits more movement in the wire spider?  It seems it must be that they they are "simply supported", i.e. their connection permits rotation of the wire about the connection point, which can permit movement of the secondary holder up/down along the z axis (optical axis), and also pitch, whereas a vane is bolted to the holder at two points (or welded) making it far more resistant to such movement of the holder.  I suppose that is the reason for crossing the wires, to get a larger z axis force component to resist such motion.

Edited by tommm, 27 April 2019 - 05:00 PM.

[tommm]

Coma correction is compromised if the optical axis doesn't intercept the center of the coma correcting lens.

If it's even 1mm off the center of the lens, coma will not only not be corrected, but additional coma will be added to the axial image.

Hence, centering the optical axis on the lens is critical.

 

With the Paracorr 1, the math showed a tightening of tolerances by a factor of 6:1 compared to using no Paracorr.

I have not seen a similar math done for the Paracorr II, but the likelihood is that because the correction extends to a wider field and to lower f/ratios,

it is tighter.  Vice Menard may know exactly, but I've read that tolerances go from 0.03D on the focuser axis error without a Paracorr II to around 0.002D with it, which is about

a 15:1 tightening of focuser axial error tolerances.

 

I have not read where collimation tolerances for the primary tighten from the already tight 0.005 f/R³ 

 

To put that in perspective, with a typical 400mm f/4.5, that would be:

FAE tolerances: 0.8mm

PAE tolerances: 0.46mm

The impact of such tolerances are that the scope must be exceedingly stiff to maintain the tolerances in use

and that even reading such tolerances requires tools more sensitive than most.

For such a scope, I would contend only a laser with an aperture stop might be adequate for intitial FAE set up, a calibrated Cheshire for the PAE and

an autocollimator with dual pupils to get close to the actual tolerances necessary.

A star test could refine the PAE, but I don't see how a star test could duplicate the FAE reduction of the autocollimator.

 

This also shows how important it is that however the Paracorr II is shoved off-center by thumb screws, the collimation tools must be so as well.

Centering does not get the tolerances tight enough unless everything is centered, and I don't see that being done in 2" focusers.

It seems highly unlikely to me that most scopes would hold these tight of tolerances, let alone what is required for an f/3 scope, but most everyone sees a dramatic decrease in coma with a Paracorr.  I've struggled to make sense of this for some time.  How is that?

[mark cowan]

Well yes. That is one test that is easy to do and I will demonstrate.

Solid vanes utilising the same geometry will always win over wires because there is more matter to resist stretching (flexure). Wire spiders use so little material we need to apply every trick in the book just to make them viable (good geometry, minimising moment, minimising mass etc). As such they are a great education tool for identifying what matters. Apply the lessons learned in making a viable wire spider to a solid vane spider and you should end up with excellent support.

 

A futile progression, in some ways.  You lost me at "the same geometry" FWIW.

 

 

But hey that's just my opinion.  The logic is flawed.  The way a solid vane spider works is not the same as the way a wire spider works since the vane doesn't really experience much in the way of differential tension.  And you can do things with wires you can't do with vanes at all, like run them through the same plane.  

Edited by mark cowan, 27 April 2019 - 05:35 PM.

[Starman1]

It's simple. A 95% reduction in coma is very noticeable. And it IS possible to build a scope to hold those tolerances. I'm not sure you can do it and make the scope light.

[mark cowan]

It seems highly unlikely to me that most scopes would hold these tight of tolerances, let alone what is required for an f/3 scope, but most everyone sees a dramatic decrease in coma with a Paracorr.  I've struggled to make sense of this for some time.  How is that?

I think because it's not actually correct. 

[Oberon]

There is more matter, but most of the tension occurs in an X pattern in the plane of the vane, which is the geometry of the wire spider.  Then if one assumes the wires do not stretch, what permits more movement in the wire spider?  It seems it must be that they they are "simply supported", i.e. their connection permits rotation of the wire about the connection point, which can permit movement of the secondary holder up/down along the z axis (optical axis), and also pitch, whereas a vane is bolted to the holder at two points (or welded) making it far more resistant to such movement of the holder.  I suppose that is the reason for crossing the wires, to get a larger z axis force component to resist such motion.

Wires can and do stretch. That is their fundamental limitation. That is why it is important to optimise the geometry so that the load exercises minimum leverage. Thicker wires would perform the structural function better but would lose the benefits of minimal diffraction. 

[Oberon]

Jonathan,

 

What is the wire diameter on your spider?

0.4mm.

 

[Oberon]

I know that the performance of the wire spider looks bad, but remember that my test rotates the spider horizontally on its axis, and is much more severe than what a Dob would ever experience. As a reminder, here is what Merope's wire spider looked like when I tested it in the way a Dob works (altitude), and it just scrapes in to meet the extremely tight Paracorr constraints. I didn't include this yesterday because I'm trying to illustrate side by side comparison of the same thing.

Perhaps this will provide assurance that a wire spider can be adequate, noting that Merope's spider should improve with a reduction in mass (remove Delrin block) and improved depth of wire angle (deeper UTA).

[tommm]

The modulus of elasticity of music wire is 30*10^6 psi. 

ref: https://www.acxesspr...properties.html
Hard to see how the small forces the secondary mirror and holder exert on it would result in measurable stretch of say 0.020" wire.

Edited by tommm, 27 April 2019 - 08:57 PM.

[Oberon]

Here is the load. 619 grams. And I'm sure it isn't a bunch of elves playing with the laser to cause its deflection.

[tommm]

Probably not, but that doesn't mean it is due to wire stretching.  Looks like it requires a bit over 9 lb force to stretch 0.020" diameter music wire 1 mil (0.0254mm)

Edited by tommm, 28 April 2019 - 09:19 AM.

[Oberon]

Time will tell. It is hard to pin down sources of flexure. When I build a new wire spider using the same UTA frame as I just used for the solid vane spider then we might be able to separate flexure of the UTA. Right now we're not quite comparing apples with apples.

Edited by Oberon, 27 April 2019 - 09:56 PM.

[Jeff Morgan]

It's simple. A 95% reduction in coma is very noticeable. And it IS possible to build a scope to hold those tolerances. I'm not sure you can do it and make the scope light.

 

My Takahashi Epsilon holds tolerances.

 

It's a 7.1" aperture f/2.8.

 

It weighs 30 pounds and has 1/8" thick vanes.

 

How many people want that?

 

The central issue seems to be triangulating on Coma/Collimation vs. Thermal Effects.

 

AFAIK, Oberon's test rig is the first real effort to quantify these issues. It will be interesting to see where the data - and the trade offs - fall.

 

We've all got skin in the Newtonian game - but it's a game and design with a lot of trade-offs.

[Jeff Morgan]

Wires can and do stretch. That is their fundamental limitation. That is why it is important to optimise the geometry so that the load exercises minimum leverage. Thicker wires would perform the structural function better but would lose the benefits of minimal diffraction. 

 

But not the thermal artifacts of solids.

[careysub]

The cross: A cross cannot resist the beginnings of axial movement, which is why bicycle spokes are always offset.

Collimation: the wires support the mirror and define its location. The wires are adjustable. Thus the mirror is adjustable, and can be collimated.

The spider: and now...just for you...concept diagrams of a fully adjustable internal wire spider with no external protrusions leaving a clean external surface whilst providing a high degree of rigidity permitting a lightweight thin tube. No machine tools beyond a drill press and a router. Every girls dream...

First the overview showing the tube, optimised wire spider and a pair of internal split rings.

 

The top view shows the layout, including the angle points on the ring required to deliver parallel and perpendicular diffraction spikes for an offset spider. Dimensions are all metric, the scale is 100mm/5mm, ring dimensions drawn to 330mm (very close to 13"), and is only 15mm deep leaving a 300mm aperture. You could make do with 280mm if that 15mm is too tight.

Being a split ring it has a wedge to hold it tight. Your first adjustment.

 

 

 

OK here we get serious. See if you can work out how it works while I go and have dinner.

Hint: those special little purple bits move up and down.

 

Note: I lied with my count of threaded rods. You'll need 9 x not 8 x as the drawing shows, the extra one being for the wedge.

Finally; this spider should easily outperform Merope's spider. Its lighter, and the wires are at optimum angles. You could probably get away with the thinnest guitar wire.

 

Sorry not getting back to you sooner to thank you for your work here, partly it was because I did take some time to examine and (try to) fully grasp the design, but then got side-tracked by Real Life.

 

It looks like this whole system is assembled out side of the tube and is then lowered into place (as must be the case, building a "ship in a bottle" is really hard).

 

Since the base of the spider cage is below the focuser there is no problem putting in stop screws through the side to arrest its descent, and then using screw fasteners into the lower ring to hold it securely, so the wedge method is not actually need there. In fact even the upper ring may not need the wedge since I can but in low-profile, or recessed flat-head screws, in the side to hold it without interfering with the sliding sleeve system. But the upper ring wedge might be a nice additional feature. The key innovation here is the adjustments which are internal to the tube. A deep-reach t-handle hex key tool will be needed to adjustment.

 

20 cm (8 inch) fully threaded screws of any narrow type do not exist, but could be assembled from a threaded rod, a coupling nut and a socket (or other headed) screw. If I ordered from McMaster-Carr since they stock Customary unit goods predominately I would use an 8-32 screw (4.166mm) instead of the M4 for price and selection. For example a 7" threaded rod, a 1.5" socket screw, and a coupling nut and permanent loctite (or some other combination to put the nut in the most convenient location) would work. McM has 8-23 socket screws as long as 3".

 

Those are double nut pairs at the lower end of the threaded rod?

[MitchAlsup]

An offset steel spider is probably unbeatable. I cannot envisage anything improving on a well made offset steel spider for rigidity. However there may be other considerations such as mass and radiation to be mindful of.

 

The next question is how thin can the vanes be? I reckon they can be vanishingly thin, and am planning to try a brass foil that is a ridiculous 0.02mm thin. Just to see what happens.

 

As for wire...watch this space, and be mindful that my experiment is more severe than a Dob in real use. 

I have been using 0.008 inch galvanized roof flashing steel for 20 years on my 20" dob

 

Brass has a similar modulus as steel.