700 replies to this topic

[mark cowan]

0.02 mm thick brass  (AKA shim stock) is less than 0.001".  I can predict fairly confidently that will not work. 

 

Reason is two-fold - you'll not be able to keep it flat in practice so it will act as if it were thicker, and second is that it will try to fold under tension and/or loading forces.  But try away.

 

The mechanics of building a wire spider are dead simple for the most part and they work very well in practice.  I appreciate the time you're obviously taking to discover this.

[BGRE]

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.

Young's modulus for Brass ~100GPa

Young's modulus for steel ~ 200GPa

Hardly similar. 

[Oberon]

0.02 mm thick brass  (AKA shim stock) is less than 0.001".  I can predict fairly confidently that will not work. 

 

Reason is two-fold - you'll not be able to keep it flat in practice so it will act as if it were thicker, and second is that it will try to fold under tension and/or loading forces.  But try away.

 

The mechanics of building a wire spider are dead simple for the most part and they work very well in practice.  I appreciate the time you're obviously taking to discover this.

I thought I had ordered 0.2mm but the foil that arrived was...well, obviously thinner. I’m not taking it very seriously as I don’t think it will prove practical for the reasons you give, but what the heck, one might as well play games while having fun! 

 

I’m confident I can make a practical 0.2mm thin spider easily enough. Thats 0.008”. Built from SS cable ties I won’t need to cross them, so hypothetically 1/4 the diffraction of 0.4mm wire, and at worst 1/2, plus be stronger than wire.

 

But I’m also curious as to how thin a practical solid vane can be. I’m thinking of laser cut 0.1mm steel, so if anyone has experience working with thin sheet like this before I place an order please pipe up.

Edited by Oberon, 28 April 2019 - 08:06 PM.

[jtsenghas]

Just a note about cutting such thin steel and maintaining a good square edge:

 

In my industry we often cut thin steel shim stock of a few thousandths of an inch for precision adjustments on machinery. Scoring the material with multiple passes of a sharp utility knife alongside a straightedge while it sits on steel plate will nearly cut through it without bending it much.  Slight flexing it back and forth at the line will snap it off neatly. Sanding the cut edge with fine sandpaper will both make the edge less dangerous and remove any significant hook on the edge that is similar to the desired edge on a cabinetmaker's hand scraper.

 

Remember though, as far as diffraction goes, off-axis light will see a considerably thicker vane. For objects even 1/4 of a degree from the center of the field of view, a 0.004" vane 2" wide will appear approximately three times as thick. (0.004" + sin(. 25 degrees) * 2" is approximately 0.013").

Edited by jtsenghas, 28 April 2019 - 09:20 PM.

[mark cowan]

I thought I had ordered 0.2mm but the foil that arrived was...well, obviously thinner. I’m not taking it very seriously as I don’t think it will prove practical for the reasons you give, but what the heck, one might as well play games while having fun! 

 

I’m confident I can make a practical 0.2mm thin spider easily enough. Thats 0.008”. Built from SS cable ties I won’t need to cross them, so hypothetically 1/4 the diffraction of 0.4mm wire, and at worst 1/2, plus be stronger than wire.

 

But I’m also curious as to how thin a practical solid vane can be. I’m thinking of laser cut 0.1mm steel, so if anyone has experience working with thin sheet like this before I place an order please pipe up.

You get to 0.2mm you're in pretty standard territory for vanes. 0.1mm at .004" or so appears doable also from what I've seen posted.  Cutting it flat, lasers make sense.

 

As J.T. notes you don't get any sort of pass on off-axis sources with vanes though.

Edited by mark cowan, 28 April 2019 - 09:28 PM.

[tommm]

I looked at wire stretching a bit. I started with the extreme case of a wire spider without crossed wires, i.e. the wires go straight out to the UTA normal to the optical axis, like the edges of a rectangular shaped vane; and a simple longitudinal (along the optical axis to the primary mirror) displacement dz of the secondary holder.  I call it extreme because the load is normal to the wires.

 

The only force to resist a motion of the secondary along the z axis is increased tension in the wires due to a change in the wire angle, phi, and length, L, as the secondary moves up or down since the wires, unlike vanes, can rotate about their connection to the secondary. Using fig.1 below, for a movement, dz, the length of the wire increases by dl = (dz^2 + L^2)^1/2 - L = 1.25 x 10^-6 inch for L = 10", dz = 0.005", and the increase in force, or tension, in the wire required to effect this deformation is e*A*dl/L = 0.0012 lb, where e is the modulus of elasticity, equal 30 x 10^6 psi, A is the cross sectional area of the wire (0.020" diameter assumed), and dl is the elastic deformation of the wire (stretch). Four wires must stretch to permit this movement, so a total force of 0.0048 lb. Very small as expected for this geometry.

 

 

Now cross the wires. Say the angle, theta, of a wire wrt a UTA radius is 15 degree as shown in fig.2. In this case for a displacement dz along the z axis, dl = dz*sin(15) = 0.005*0.26 = 0.0013", and the increase in tension in the wire required for this deformation is 1.2 lb. Since 4 wires must stretch, about 4.8 lb of force is required to move the secondary 0.005" along the z axis in this case, roughly 1 lb/mil displacement.

 

 

The increase in tension for a given movement in pitch or yaw can be estimated in a similar way, since for small displacements, the displacement can be treated as linear even though it results from a rotation. It seems the angle of the wires in the plane they are crossed in doesn't matter here since it has no component in the plane of the displacement of the holder for pitch or yaw.  For both pitch and yaw the wire angle is 45, so a given displacement of the secondary holder results in more elongation of the wire than in the prior case.  For a 0.005" displacement, dy, dl = dy*sin(45) = 0.0035", and the increase in tension in the wire required for this deformation is  3.3 lb.  Again 4 wires must stretch so about 13.2 lb force is required, or about 2.6 lb/mil displacement. 

 

Of course the loads are actually torques, so relative moment arms they and the wires act through will change things depending on holder design, but for the moment arms I've seen on offset holders such as Jonathan's (not the central bolt, "cross" type) it seems the load and wire moment arms aren't all that different so I'd expect that to be a significant, but not large effect.  Solid vanes with at least two bolts at each end cannot rotate about their connection points, so pitch and yaw loads are counteracted by shear stress in the vanes resulting in greater stiffness.

 

I had started making a wire spider primarily for lighter weight than my present holder, but also because of decreased diffraction (though I don't know how big this effect is) and, I thought, greater stiffness as stated e.g by Reiner here .  I have 0.020" piano wire, but based on the above estimates I think I will get 0.030", and maybe wait to see what Jonathan finds before replacing my existing spider and holder. 

Edited by tommm, 30 April 2019 - 09:25 AM.

[mark cowan]

Jonathon's design (relying on differential tension) is not the best way to do it though IMHO. 

 

Geometry can be used to much greater advantage, leveraging resistance to stretching by crossing wires across the hub itself, which puts them into an active role resisting both vibration and twisting forces by increasing the stretch movements, compared to the usual approach. The design shown below was produced empirically, and has also been built by several other people, in various versions, with uniform good results:

 

 

 

And beyond that approach reducing the offset helps a great deal as well, from basic principles.   

 

The following prototype (with dental floss strings ) is also an empirical design and incorporates most of the above (crossing wires across the hub) as well as reducing offset torque to minimum.  It is a very simple construction by comparison (one flat plate if in a hexapod), but requires a extended upper cage to mount it in.  J.T. has been building one for a 12.5" and I'll be using it for a new 20" thin meniscus hexapod project I'm starting. 

 

 

 

 

Stereo cross pairs for those who dare:

 

 

Edited by mark cowan, 30 April 2019 - 02:37 PM.

[Jeff Morgan]

I also changed the Heim joint mounting method, but I'll discuss that in more detail in the Hexapod thread.

 

Looking forward to that update.

 

The return of warm weather has restarted my Hexapod project. After the failure of my carbon fiber/balsa core UTA rings I restarted with more conventional materials and I'll be ready for the Heim joint attachments is a couple of weeks.

Attached Thumbnails

• 

[Oberon]

Jonathon's design (relying on differential tension) is not the best way to do it though IMHO. 

 

Geometry can be used to much greater advantage, leveraging resistance to stretching by crossing wires across the hub itself, which puts them into an active role resisting both vibration and twisting forces by increasing the stretch movements, compared to the usual approach. The design shown below was produced empirically, and has also been built by several other people, in various versions, with uniform good results:

 

 

 

We've discussed this before, and I disagree that your geometry is better, but rather than spoil the thread with an argument perhaps you would be so kind as to post me one of your machined supports pictured above (which I will return) so that we can do a direct performance comparison and see what emerges. I'm always happy to be proven wrong.

[tommm]

 

 

...Geometry can be used to much greater advantage, leveraging resistance to stretching by crossing wires across the hub itself, which puts them into an active role resisting both vibration and twisting forces by increasing the stretch movements, compared to the usual approach...

 

It is not clear to me that this is so. For one thing, the wires will likely bind somewhat at their 90 deg bends through the holes in the hub, so tension will not be uniform throughout their entire length.  But more than that, in order for the hub to move the resisting wires must rotate about their connection points and stretch a bit, and what matters for this is the elongation of the wire along the portion of its length between the tuner and the hub (elongation is per unit length), and that is determined by tension in the wire between them.  I don't think the part of the wire across the hub enters into it. I actually plan to route the wires across the hub similar to this and bratislav's design (which I am somewhat copying), but I don't expect it offers any advantage in stiffness.

 

At your link you state: "After working with the secondary a bit I thought that if the wires that converged on the focuser side of the hub went from top-bottom and bottom-top instead of describing a trapezoid the resistance to bending along that side would improve. It does."

 

Yes, you basically went from Fig.1 to Fig.2 in my post#206.  I actually would expect this design to be less stiff than the more conventional hub/wire connection used by Jonathan and others due to the small offset of the wires at the focuser side of the hub.  It seems to have a bit greater moment arm to resist pitch than conventional, but it should be less stiff in resisting yaw due to the small wire offset on the focuser side.  The conventional hub as wider offset on all wires to resist yaw.  It still may work fine - and I don't doubt you that it does, the proof is in the pudding as they say. There is only one way to know for sure if it is more stiff, send one to Jonathan to test with all other things equal - just swap out one holder for another for same size secondary mirror. But practically, what matters is that it is stiff enough.

 

My expectation is that wire stretching for adequate sized music wire is not a significant contributor to lack of stiffness, and the variables that effect it lie elsewhere such as UTA, truss, truss connection, and wire adjuster assembly stiffness.  As far as "adequate size" I think 0.020" is sufficient for 4" minor axis mirrors and smaller, maybe for 5", when coupled with sufficient offset on the hub, but that is just based on rough calculations. Mine are an upper bound for expected wire elongation - expect it is actually a bit more stiff because of some simplifications I made, so somewhat smaller diameter wire would be ok. Will see what Jonathan finds - unless it turns out to be too tedious to separate all the effects and he tires of it. Time consuming task.

[tommm]

Maybe the figure below illustrates why I think the "double offset" holder is less resistant to yaw.

A more conventional hub is shown, with wide offset between all 4 spider wire pairs. On the double offset design (Mark) two pair of wires have much narrower offset, so seem less resistant to yaw. 

 

 

 

Edited by tommm, 04 May 2019 - 06:50 PM.

[mark cowan]

We've discussed this before, and I disagree that your geometry is better, but rather than spoil the thread with an argument perhaps you would be so kind as to post me one of your machined supports pictured above (which I will return) so that we can do a direct performance comparison and see what emerges. I'm always happy to be proven wrong.

Certainly I would hate to "spoil" your thread with an argument, Jonathon. 

 

PM me your address and I'll loan you the presently available spider (top pic) that is intended for a 14.7" application at around f/4.5 (I think that's a 2.6"), used with 0.010 or 0.008" wires.  Then we'll see.  I'm always happy to be proven right.

 

I could also send you the piece of wood in the second version but probably since I'll be machining it out of aluminum I'll just make two later on, it's dead simple to construct especially for the hexapod.  J.T. might have a spare though.

 

NB I should however point out that the main design criteria was to reduce damping times for vibration, not necessarily improve resistance to loading forces other than by reducing torque moments.  You might want to test for vibration damping times as long undamped modes are extremely annoying in practice, but you'll need a calibrated hammer of some sort...
 

Edited by mark cowan, 04 May 2019 - 01:49 PM.

[mark cowan]

Maybe the figure below illustrates why I think the "double offset" holder is less resistant to yaw.

A more conventional hub is shown, with wide offset between all 4 spider wire pairs. On the double offset design (Mark) two pair of wires have much narrower offset, so seem less resistant to yaw. 

 

Pitch and Yaw.JPG

I tested it in a mocked up truss as a wooden block with c-clamps providing exaggerated torque loads and it worked very well.  Plus the design in practice has lived up to it's claims.  But this is an "argument", mostly theoretical, and Jonathon requests there not be one so please stop discussing it...

[jtsenghas]

I could also send you the piece of wood in the second version but probably since I'll be machining it out of aluminum I'll just make two later on, it's dead simple to construct especially for the hexapod.  J.T. might have a spare though.

 

NB I should however point out that the main design criteria was to reduce damping times for vibration, not necessarily improve resistance to loading forces other than by reducing torque moments. 

This is a great thread  and I compliment you on your rigorous testing, Jonathan.

 

I don't in fact have a spare flat wooden elliptical holder. My first attempt, which had the wires attached to slits within the board did not adequately keep the wires from touching the edges of the glass for the relatively steep angles involved, so that prototype holder was converted to BTUs in my wood stove.

 

My major design criterion for my work in progress, hopefully to be done in time for the Cherry Springs Star party in four weeks, was also fast damping times.

 

This slight variation on Marks latest prototype doesn't quite have the attach points on the UTA coincident, but it does have all ends of the eight wires attached with twisted loops.

 

I'm trying for ADEQUATELY stiff for collimation tolerances with wide angles and offset attach points and minimal cantilevering of the secondary center of gravity from the wire lines of force.

 

No doubt the almost instantaneous damping time is helped by the asymmetry of the wire lengths and angles.  The different resonant frequencies of the wires almost certainly is an important part of the design.

 

I want to caution, though, that this layout is VERY tricky to tune in for mirror centering and angles, with some adjustments requiring all eight wires to be tweaked. Marks suggestion to have a way of moving pairs of wires parallel to the UTA axis would help almost certainly in my mind.  It might not be the last spider on this scope, but it will be the first  

Edited by jtsenghas, 05 May 2019 - 07:07 AM.

[jtsenghas]

Jonathan  I suspect that what will emerge from your experiments is that many designs of wire spiders will be adequately stiff provided they have an adequate wire diameter, sufficient offsets and wide enough angles between them.

 

The "best", depending on one's criteria may be those that don't put extreme stiffness demands on the UTA in multiple directions (as Mark's latest version and my variation of it clearly do) . The weakest link probably is more often stiffness issues in the rest of the scope SYSTEM, as Tommm mentions above. 

Edited by jtsenghas, 05 May 2019 - 07:44 AM.

[Earthbound1]

Great thread Jon! Thanks ALL!

[Oberon]

Meanwhile back at the lab, a clutter of solid straight vane spiders is emerging.

This set is being built using identical techniques, identical dimensions and identical vanes - 50mm x 0.5mm aluminium - but in 4 variants. Look closely you should see that 2 are offset and 2 are centred, 2 have a 50mm section and 2 have a 20mm section. I'll follow this with similarly dimensioned curved vanes to get as direct a comparison as possible.

 

[Ed Jones]

 

NB I should however point out that the main design criteria was to reduce damping times for vibration, not necessarily improve resistance to loading forces other than by reducing torque moments.

Yes what about damping time?  Any info?  Wire spiders tend to vibrate.

[Dale Eason]

Yes what about damping time?  Any info?  Wire spiders tend to vibrate.

If my wire spider vibrated on my 10 F3 3D printed scope I did not notice it.  Any vibration stopped in less than a second.  So fast I did not really notice it.  However I only use that scope to 150x.

Dale

Edited by Dale Eason, 19 May 2019 - 01:53 PM.

[Oberon]

Yes what about damping time?  Any info?  Wire spiders tend to vibrate.

Not forgotten, but not yet implemented. Need to make a standard hammer...

[ckh]

Wires can easily be damped. Lightly touch a vibrating guitar string.

[mark cowan]

That oversimplifies that particular problem.  Especially for longitudinal mode vibrations, which are the killer.

[ckh]

Longitudinal waves propagate along the wire at the speed of sound in the wire. For SS wire that's roughly 6000 meters per second.  Assuming the ends don't move, for a 10 inch wire the lowest longitudinal frequency possible is about 12 kHz.  Such a high frequency vibration with be damped almost instantly by the UTA and connections.  Longitudinal vibrations are independent of tension. 

 

Traverse vibrations depend on the density and size of the wire, the length and tension. Those frequencies are far lower and take longer to damp without a damper.  How much the vibration modes affect the motion of the diagonal is another issue. but I don't think the longitudinal mode will last long enough to notice.

[mark cowan]

Oh OK.   However...I said longitudinal mode vibrations (i.e. mechanical oscillations) not sound waves.

 

Experimentation shows that implementing a dampening device (even a finger on the wires) is not very successful at suppressing vibratory modes of the spider. 

 

These are not expressed as simple string vibrations, which are also at higher frequencies than what the spider moves at when shocked.

 

Perhaps they should be called something else here, like "primary hub oscillation", so this kind of confusion doesn't arise.

 

The support wires are obviously stretching and relaxing along their lengths when this happens as the hub moves back and forth in some direction of freedom.  It's at low frequencies, a few cycles per second at best.  These are longitudinal vibrations of the wire (but not sonic waves).

 

To sum it up: neither transverse nor longitudinal waves at high frequencies are involved.  It's all relatively slow and hard to suppress because coupling to any viscous sort of snubbed is inefficient.

Edited by mark cowan, 21 May 2019 - 07:15 PM.

[BGRE]

Merely a mass (secondary assembly) suspended on a set of springs (spider wires). The frequencies of the fundamental resonances are easily calculated if the mass and the various spring constants are known. In principle one could perhaps use eddy current damping but the suspension system for the magnets used would need to be stiff and well damped. Various active damping schemes are also possible in principle but present similar issues with the supports for the required actuators (pieozoelectric, etc) and sensors.  .