338 replies to this topic

[skround]

Drive and tracking - I'd start stating that backlash is the foe.

 

We are adopting a brushed motor system from BBAstrodesign. The overall recommended reduction sits between 3000:1 and 10000:1.

It's a lot of reduction....But this is where the the dobson configuration nicely pairs with the friction drive concept. The set-up couples a stainless steel rail on a 1200mm diameter rocker with a 20mm h6 ground bar which offers a 60:1 no-backlash as a last stage. A compact low-backlash worm gear reducer for another 60:1 coupled with a 2.5:1 belt and pulley stage completes the drivetrain. Total ratio (60x60x2.5) is 9000:1. Please note that the motors come with a built in 10:1 gearbox that has few gear pairs - too much of backlash imho- conversely the belt will have a tensioner so that the main source of backlash is in the 60:1 worm-gear. Ideally I'd adopt an harmonic drive but it's pretty pricey and those I've found, come with a specced max 30' backlash which translate to 0.5arcsec on the scope axis - hopefully well manageable.

 

I don't see adopting clutches. That's a limitation but if the slewing is 3deg/s then we'll use a bit of patience.

 

Pic#1 of the mirror box with two main elements "pacman" - the rails is not present in this model version

 

Pic#2 The actual rails - 2mm stainless steel. 2 pre-rolled arches for the Altitude and 3 sections for the Azimuth bearing. A full circle would have set me off a fortune - and there's a trick for a smooth section transition.

Attached Thumbnails

• 
• 

[m. allan noah]

About the spider vanes I actually tricked me as well. But if you look at that as a shelf the weight is the secondary mirror assy whereas the upper cage is the 'wall'. Hence the higher moment/stress is close to the wall. Conveniently this allow the secondary central hub to be shorter and avoid the overall upper cage to be higher.

The vanes are not independent cantilevered supports for the secondary. They work in pairs with the one opposite. Also, the secondary is not a point load. It has to fixed against rotation and translation in every axis. I'll reiterate, your current design is backwards.

 

allan

[TOMDEY]

OK, then allow me to explain. If a hexapod truss is made with adjustable poles, as a few of are are doing, then there is no need to provide any other collimation mechanism in any of the mirror supports. This allows both your primary and secondary supports to be sturdier, stiffer, simpler and lighter, collimation be be quicker, simpler and more intuitive, and even permits star collimation from the eyepiece. It also greatly relaxes construction tolerances with respect to focal length.

 

Your design so far allows for it. It is the obvious thing to do. Check out my Merope thread in my signature for more detail.

Here's a picture showing the generalized principle. With six adjustable-length struts, you can precisely tune the location and orientation of the slaved platform (with the mirror on it, there) relative to the base --- in all six degrees of freedom (x, y, z, roll, pitch, yaw). Six is also the magic number that assures deterministic, rigid relationship, with neither too many nor too few members, and therefore no induced stress from redundant ones (as one would get with e.g. eight struts). Ummm... the movements of the slave relative to the base is just a transfer matrix... or intuitive works well. So, think of the base as the PM/mirror box and the slave as the upper cage.    Tom

Attached Thumbnails

• 

[ButterFly]

The vanes are not independent cantilevered supports for the secondary. They work in pairs with the one opposite. Also, the secondary is not a point load. It has to fixed against rotation and translation in every axis. I'll reiterate, your current design is backwards.

 

allan

I agree.  Torsion at the walls is much less of a problem than at the secoondary - the wall is realtively fixed and provides support from the sides.  Thinner in the middle lets that secondary dangle away.

[skround]

Here's a picture showing the generalized principle. With six adjustable-length struts, you can precisely tune the location and orientation of the slaved platform (with the mirror on it, there) relative to the base --- in all six degrees of freedom (x, y, z, roll, pitch, yaw). Six is also the magic number that assures deterministic, rigid relationship, with neither too many nor too few members, and therefore no induced stress from redundant ones (as one would get with e.g. eight struts). Ummm... the movements of the slave relative to the base is just a transfer matrix... or intuitive works well. So, think of the base as the PM/mirror box and the slave as the upper cage.    Tom

Thanks Tom, I somehow overlooked Oberon's post. The hexapod concept is pretty interesting. I just wonder how intuitive and time consuming it might be on the field

[Oberon]

Its completely intuitive if you follow 2 simple rules:

 

1. twist any two poles that converge at the top to adjust the secondary

 

2. twist any two poles that converge at the bottom to adjust the primary

 

Those rules work best when the poles converge at points close to the plane of each mirror. If this is not the case, as in a deep mirror box, then the process may be a little more iterative.

[TOMDEY]

Thanks Tom, I somehow overlooked Oberon's post. The hexapod concept is pretty interesting. I just wonder how intuitive and time consuming it might be on the field

Oberon explains that well. Actually, although I like the hexapod concept, and have happily used it at work ad nauseam... I still favor the traditional octopod truss Dob, with the PM and SM tip-tilts properly done! That's because... although a 3-legged stool never rocks... you're far less likely to fall off a 4-legged one! (because cos45/cos60 is root two.) It's got a "bigger footprint" on the floor. And that does carry over to those rings up top and the box down below. A ring held at just three points is far more prone to flex than one held at four. Very nonlinear.   Tom

Attached Thumbnails

• 

[skround]

The vanes are not independent cantilevered supports for the secondary. They work in pairs with the one opposite. Also, the secondary is not a point load. It has to fixed against rotation and translation in every axis. I'll reiterate, your current design is backwards.

 

allan

Let's check the assumptions taking a simplified case where the scope is pointing vertically - still and without any other external force.

 

Assumption#1 The only significant load on this part of the structure is the secondary mirror and its supporting central hub. Given than this is about 2.5kg I'm not considering the weight of the vanes themselves (150g each)

 

Assumption#2 This load can be concentrated in its CoG - somewhere very close to the central axis.

 

Assumption#3 Every vane is taking i.e transferring trough its structure, 1/4 of the load. You can visualize that by splitting the spider in 4 sectors.

 

Assumption#4 The cage side is much more robust than the spider one. To simplify it the cage is fixed and infinitely rigid.

 

Assumption#5 Such single vane with a quarter of mirror assy dangling off the end (with an offset) and the other end being attached to the cage falls in the classic cantilevered beam with load on the tip

 

I trust you would agree that the end of the vane that needs a thicker section is the one that has more bending moment.

 

Not trying to teach anything here - as I said this area tricked me in the early stage of the design so double checking that is always good!

 

Sorry for the poor CAD execution:

Attached Thumbnails

• 
• 

[555aaa]

In looking at a similar endeavor, I think you might want to take an inventory of your instrument payload. I think you mean a coma corrector, rotator, focuser, filter wheel, guider, and camera. It can be a bit of a challenge to support that with the high precision needed and also the length of these components requires an extra large clear aperture for a fast Newtonian. I am mulling over something similar and debating going prime focus only. As an example of some big low cost science telescopes you might check out the asteroid program at ARI the astronomical research institution which is on a farm.

Edited by 555aaa, 30 August 2019 - 01:20 PM.

[skround]

In looking at a similar endeavor, I think you might want to take an inventory of your instrument payload. I think you mean a coma corrector, rotator, focuser, filter wheel, guider, and camera. It can be a bit of a challenge to support that with the high precision needed and also the length of these components requires an extra large clear aperture for a fast Newtonian. I am killing over something similar and debating going prime focus only. As an example of some big low cost science telescopes you might check out the asteroid program at ARI the astronomical research institution which is on a farm.

The work on the direct drive looks awesome.

I've checked the ARI website few time -the tracking precision of the 0.81m cope is a major achievement. 

 

 

The payload is roughly considered as per picture - sorry for the crude rendering.

Attached Thumbnails

• 

[555aaa]

Thanks. I didn't see the moving secondary focuser; that's clever. The photo you show is for a SCT that's probably f/8 or so and you are at f/3.3 with probably a much larger image circle and so it is a much steeper cone, so I'd be a little concerned with vignetting, So that means a rotator with a big hole in it and the rotator consumes a lot of back focus in a Newtonian. It can be done, just a matter of measuring everything out and optimizing which parts go where. The rotator has to be in front of the filter wheel, so by knowing the image circle and the spacing you can calculate the required rotator aperture and from that, pick one that would work and the corresponding weight. Optec's Gemini rotating focuser looks like an interesting option (I have no relationship with them). It does weigh 3kg all by itself. You could think about building the UTA around that however so it isn't hanging off one side.

 

You can msg me if you want to discuss options on direct drive. There was a guy here, Ctables (??), who got partway through building an alt-az direct drive using homemade axial flux motors but I don't know what happened to that.  You could also do a hybrid going direct in azimuth and a friction drive in elevation.

[skround]

Drive and tracking - I'd start stating that backlash is the foe.

 

We are adopting a brushed motor system from BBAstrodesign. The overall recommended reduction sits between 3000:1 and 10000:1.

It's a lot of reduction....But this is where the the dobson configuration nicely pairs with the friction drive concept. The set-up couples a stainless steel rail on a 1200mm diameter rocker with a 20mm h6 ground bar which offers a 60:1 no-backlash as a last stage. A compact low-backlash worm gear reducer for another 60:1 coupled with a 2.5:1 belt and pulley stage completes the drivetrain. Total ratio (60x60x2.5) is 9000:1. Please note that the motors come with a built in 10:1 gearbox that has few gear pairs - too much of backlash imho- conversely the belt will have a tensioner so that the main source of backlash is in the 60:1 worm-gear. Ideally I'd adopt an harmonic drive but it's pretty pricey and those I've found, come with a specced max 30' backlash which translate to 0.5arcsec on the scope axis - hopefully well manageable.

 

I don't see adopting clutches. That's a limitation but if the slewing is 3deg/s then we'll use a bit of patience.

 

Pic#1 of the mirror box with two main elements "pacman" - the rails is not present in this model version

 

Pic#2 The actual rails - 2mm stainless steel. 2 pre-rolled arches for the Altitude and 3 sections for the Azimuth bearing. A full circle would have set me off a fortune - and there's a trick for a smooth section transition.

Addendum: last but not least a quick check on the herztian contact stress - just to make sure that the rail and roller will keep their integrity over time and won't have trail and indentations on the surface.

Rollers are fairly small in diameter (22mm) just like the ground bar (20mm) to achieve an high ratio for the no-backlash last stage of the drivetrain - and the smaller the diameter of rollers the higher the stress. I felt important to check this tricky aspect as it can compromise the accuracy over-time.

The altitude one sees an easier life as the track is at least 50mm wide and there are 4 contact points. The load on azimuth is spread on 3 set of rollers instead so the stresses are higher.

Attached Thumbnails

• 
• 

[m. allan noah]

Assumption#5 Such single vane with a quarter of mirror assy dangling off the end (with an offset) and the other end being attached to the cage falls in the classic cantilevered beam with load on the tip

 

I trust you would agree that the end of the vane that needs a thicker section is the one that has more bending moment.

If assumption 5 was true, I would agree with you. However, you don't have 4 independent chunks of secondary and mount. You have a single one, supported by 4 spokes. The forces and constraints are different for this.

 

allan

Edited by m. allan noah, 30 August 2019 - 07:14 PM.

[555aaa]

Are you using a spring or something to generate the contact force on the azimuth ring? I'm sure that you've thought about that. The contact force sets the breakaway torque as you know, so that there won't be skidding on the rollers. 

 

The LCO (Las Cumbres) horseshoe mount telescopes use a roller drive where the roller is a direct drive using a Kollmorgen (formerly Danaher) cartridge DDR drive. I think they also use a Renishaw tape encoder which augments the DDR encoder. Using a DDR motor on the roller drive will give you a true zero backlash design and the cartridge DDR has a very high resolution absolute encoder (27 bits I think but not with high accuracy). The C041A (the smallest) cartridge DDR is about 5N-m continuous output and 12N-m peak which should be plenty, but they are kind of expensive unless you can find them at your favorite surplus site.

[Oberon]

Allan is right. The vanes are very suboptimal. A spider support operates as a system whereby the secondary is suspended in tension against the UTA ring. The vanes only experience tension, the UTA experiences compression. This permits the vanes to be made of very thin material, which minimizes the diffraction spikes.  From an engineering pov then the optimal vane is a triangle with a single point of connection to the UTA ring and two points of connection to the secondary support. There may be good reasons to diverge from this simple optimum in many cases, but those reasons aren’t apparent here. Your post #33 does not justify your design, it only shows that you don’t appreciate how a classical spider works, and the consequence for observers will be much bigger brighter diffraction spikes.

 

Again, there are many telescopes that for many reasons employ a suboptimal spider, but in your case there is no apparent reason to avoid and nothing to be gained by employing anything other than an optimal spider.

[skround]

The attached pic of CAD model is the result of few months of work and the best of our engineering knowledge. We are currently working on a major update revision. The description below is about the new design so sorry for some mismatch.

I reckon the structure is pretty typical for a Dob in this class. A simplified/integrated rocker box swivels on two ground bars - that's a smart design I've seen on few motorized Dobs. Those bars are supported with heavy-duty bearing on a rotating table that provides the azimuth axis. Under the Az table three 'legs' are housing the roller bearings to allow the azimuth rotation - this is like a big Lazy Susan bearing.
I suppose one could look at the structure as two big 1200mm bearings.

 

The upper cage is again a structure as simple as possible with 2 identical hexagonal elements connected through 3 pillars and a plate that serves as mounting support for the focuser/derotator. Pillars and plate provide the attachment for the spider that holds the housing of the secondary mirror assembly.

 

To connect the upper cage and the rocker box we adopt a truss arrangement. 6 vs 8 beams configurations were compared eventually in favor of 6. The idea is to optimize the number mainly due to the cost of 1.5m poles.

We use FEA to validate the structure. The performance of a telescope hinges on its ability to perform tracking to the specified purpose. That's where the structure itself plays a crucial role.
Ideally, what you are looking for is stiffness, and that's to be insensitive to external perturbations and to be positively reactive to inputs such as tracking, PEC and autoguiding. In other words, a gust of wind shouldn't upset the tracking and if guiding corrections are imparted the reaction shouldn't over or undershoot.

 

If you want to have something stiff but lightweight then your main engineering parameter is called structural natural frequency - or heigenmodes. In practical terms, if you tap on a mount you want the oscillation to be small and dampen out as quickly as possible.

 

And the only way to predict it properly is Finite Element Analysis. We didn't want to spend any significant amount of money before knowing that the mount is fit for purpose. The gifs are showing exaggerated oscillations of structural frequency - the higher the frequency the stiffer the mount.

The target is a bit tricky to set as there not much literature out there. There are definitely papers about big meters class telescopes but not for our purpose although the spirit is the same for any telescope. For our project I set the target to 25Hz.

 

The first iteration showed 19Hz and 20Hz for the first 2 modes. The third mode - not shown- is over 30Hz so it's ok.

 

 

Attached Thumbnails

• 
• 
• 

[TOMDEY]

Saaaaay, skyround... do you have the mirrors yet?! My flow has always been optics in-hand first, then design and build the telescope around that. Just curious. Working them in parallel is generally fastest, provided you know you will eventually have the mirrors. We built the 36-incher using both a surrogate PM and then the real one, once the mount was close to completion.   Tom

Attached Thumbnails

• 
• 
• 

[skround]

Saaaaay, skyround... do you have the mirrors yet?! My flow has always been optics in-hand first, then design and build the telescope around that. Just curious. Working them in parallel is generally fastest, provided you know you will eventually have the mirrors. We built the 36-incher using both a surrogate PM and then the real one, once the mount was close to completion.   Tom

I'd agree for a visual telescope whereas at this kind of diameter and to be an imager we felt to put the accent on a sound mount first.

Is the surrogate for weight balance? Why is it replicating the sag of the actual PM and the back is spherical - just out of curiosity?

[TOMDEY]

I'd agree for a visual telescope whereas at this kind of diameter and to be an imager we felt to put the accent on a sound mount first.

Is the surrogate for weight balance? Why is it replicating the sag of the actual PM and the back is spherical - just out of curiosity?

The back is spherical because the back of the PM is... matching the slumped mirror. Of course, the final front paraboloid and the back sphere are essentially identical, when it comes to mechanical considerations.

 

The 27-point whiffle tree back support and our 30-point side support "Ladder Sling" --- which adaptively supports front and back plates --- Ryan and I came up with / invented that. That support-system was optimized specifically for that mirror, including the slumped back plate. So, we wanted a surrogate for that mirror, to exercise the entire mount, without risking the ($55K?!) mirror set. The sling terminals ride on fore-aft linear bearing rails, so tip-tilting the PM does not introduce cos loads. Ist light was successful... no stress-induced aberrations. No astig, no spherical... no nothin nefarious (whew!)... just pin point stars!   Tom

Attached Thumbnails

• 

[skround]

The back is spherical because the back of the PM is... matching the slumped mirror. Of course, the final front paraboloid and the back sphere are essentially identical, when it comes to mechanical considerations.

 

The 27-point whiffle tree back support and our 30-point side support "Ladder Sling" --- which adaptively supports front and back plates --- Ryan and I came up with / invented that. That support-system was optimized specifically for that mirror, including the slumped back plate. So, we wanted a surrogate for that mirror, to exercise the entire mount, without risking the ($55K?!) mirror set. The sling terminals ride on fore-aft linear bearing rails, so tip-tilting the PM does not introduce cos loads. Ist light was successful... no stress-induced aberrations. No astig, no spherical... no nothin nefarious (whew!)... just pin point stars!   Tom

Oh I see this is a Fullum tecno-mirror - now with the pic is clearer. How much does it weight? 

Also, may I ask you the advantage to put the sling on linear bearings?

A trobule-free first light must have been memorable!

[skround]

Following up on the upper cage hexagonal elements. It needs some detailing now....

 

Attached Thumbnails

• 
• 

[TOMDEY]

Oh I see this is a Fullum tecno-mirror - now with the pic is clearer. How much does it weight? 

Also, may I ask you the advantage to put the sling on linear bearings?

A trobule-free first light must have been memorable!

My recollection is something like 160 pounds.

 

The linear bearings... to prevent over-constraining of the PM. Three knobs adjust the tip/tilt/piston of the PM... so, its position and orientation are tuned to satisfy optical performance vs nominal mechanicals. The sling is in the form of a ladder comprising wire rope  and Delrin steps. That configuration assures that the front and back plates experience equal pure radial loads at all 30 points of contact, provided the end terminals are free to translate longitudinally. Additional Design DOF is the lateral separation of those two terminals. So (not unlike whiffle-trees) the sling on rails adaptively auto-adjusts to its lowest potential energy state aka relaxes/precludes distorting stressors inadvertently being introduced into the mirror... which (most always) would manifest as optical astigmatism.

 

Our Structures guys at work had done a protracted study of alignment and optical manifestations of generic point-loads on telescopes. After all the dust had settled, they found that the biggest performance manifestations were Zernike piston, tip, tilt, power, and astigmatism... regardless of the location or direction of the loads! That agreed with our lab experiences and intuition. Of those five low-order Zernike terms... the first four can be remotely aligned out... but the astigmatism rarely can be, unless decisively provided for in the hardware. My previous hobby-scope (a 29-inch Dob) had significant astig, in its relaxed state. So I came up with my Tweaker Clock, comprising twelve adjustable puller-springs. That allowed me to counter-stress the mirror, tuning out the astig (as my son tweaked the twelve knobs with me looking thru the eyepiece at a star). It worked! Within minutes, the astig went from terribly annoying to... gone... and rarely needed any additional tuning.

 

So, yeah, the linear bearings prevent residual stressors from entering the mirror.    Tom

Attached Thumbnails

• 

[skround]

Thanks for sharing!

 

 

in its relaxed state. So I came up with my Tweaker Clock, comprising twelve adjustable puller-springs. That allowed me to counter-stress the mirror, tuning out the astig (as my son tweaked the twelve knobs with me looking thru the eyepiece at a star). It worked! Within minutes, the astig went from terribly annoying to... gone... and rarely needed any additional tuning.

 

Is that a sort of manual version of active optics? Wouldn't be dependent on the altitude angle?

[skround]

If assumption 5 was true, I would agree with you. However, you don't have 4 independent chunks of secondary and mount. You have a single one, supported by 4 spokes. The forces and constraints are different for this.

 

allan

In a vertical position -as assumed- each spider handles the load transfer of 1/4 of the mirror assy (imagine to place 4 scales at the ends of the vanes in a stand alone condition).

If you are thinking of factoring in the tension on the vanes that doesn't change that much.

 

What if instead of 4 vanes we use one? Would you still make it thicker at the mirror side?   

[ButterFly]

 our 30-point side support "Ladder Sling" --- which adaptively supports front and back plates --- Ryan and I came up with / invented that. That support-system was optimized specifically for that mirror, including the slumped back plate.

Very nice design.  Are those rungs permitted some rotation to help redistribute load or are they locked in place with those clamps between the rungs?