kinematic mount surface finish 1

[KENAT]

I’m currently checking the drawing of the base part of a Kelvin Mount, a type of kinematic mount (see figure 3 of this doc to get some idea what I’m talking about

http://www.optics.arizona.edu/optomech/references/papers/Lake 2000.pdf

)

The overall surface finish is only 32 microinches (.8 micrometers).

Based on what I’ve read in Precision Machine Design by Alexander H. Slocum and on a website I found a while back(

http://www.precisionballs.com/tech_papers.html

especially

http://www.precisionballs.com/Serious_Kinematic_Coupling.html

) I believe a much finer surface finish should be applied to the contact surfaces of the groove, cone & flat. Slocum doesn’t say much but appears to use finishes of down to about 12 microinches in his examples. The website says:

the surface quality of really high quality Kinematic components is between 0.1 to 0.2 microinches (0.003 micrometers), (2.5 to 3 nm) Ra.

In my application the item put into the mount isn’t cycled a lot, so it wont really wear itself in.

A precision measuring device hangs off this mount so I want pretty good repeatability from the start on the rare occasion it is removed. That said there is some adjustment/allowance in the system to cope with some misalignment.

The material the designer has selected for the base is 303 SST. It has some fairly complex machining on other features so changing to some of the materials that have better properties for this kind of thing may be problematic, if nothing else they’d probably have to be some kind of insert or similar.

I don’t have the material of the mating balls to hand but can find it if it’s of significance.

The load isn’t particularly high, I’d guess well under 10 lbs, I can find out if need be.

Any guidance from anyone with knowledge of this type of thing would be appreciated.
For instance would it be fair to say there’s no point making the finish of the groove, cone & flat much better than the finish of the balls?

What surface finish can I reasonably achieve without massively increasing manufacturing costs? I have the standard chart that shows finishes for different processes but the finest it has is .5 microinches for Grinding, Electro-polish, Polishing, Lapping and superfinishing. I’ll obviously suggest the designer contacts the machine shop we’ll probably use to confirm but I’d like some idea up front.

Appreciate any help anyone can give. If you need more information ask, I may have to be a little careful due to IP but can probably give some if needed.

KENAT, probably the least qualified checker you'll ever meet...

 

[IRstuff]

How much load? What is the baseline of the mount? What is your repeatability requirement?

In general, if you are only using a single mating object, the surface finish is probably secondary, since the final settled in position is a function of the local irregularities, and so long as they are not distorted or worn away, a removal and replacement should go back into the same orientation.

Usually, what potentially creates problems is if you have multiple objects mating to a single mount, then there might be some variance because each object was aligned against a different reference.

TTFN

FAQ731-376

 

[KENAT]

What do you mean by baseline of the mount? Sorry if this is a really dumb question.

The other 2 I'll find out tomorrow, no one else around right now to ask.

I put above that I think the load is less than 10lb. I have a figure of better than 10 microns for repeatability in the back of my head but I'm not sure so I'll have to check.

An earlier iteration of this design had some fretting or something going on and didn't have the repeatability we desired so this is why I'm looking at this more closely, maybe considering second order factors.

It will normally be a matched pair, same measurement device in same mount. While it could change if the measurement device is exchanged there would be a tolerance stack within the measurement device which would mis align it no matter how good the mount.

Thanks,

KENAT, probably the least qualified checker you'll ever meet...

 

[IRstuff]

Sorry, I missed the loading in the OP. How is the load attached? Is it simply placed on the kinematic mount? Or will there be compression applied, e.g., is the load bolted or held down through the mount? Some optical kinematic mounts use springs to hold the two halves together.

The baseline is the distance between the 3 points. I was thinking angles, hence the change in height divided by the baseline would give you an angle error.

Anyways, most kinematic mounts are simply gravity loaded and the load is simply placed on the 3-point. Under those conditions, wear and fretting is minimal. If you've got fretting, that seems to imply some sort of vibration environment, since you say that removal is seldom. This would then require some additional force to keep the load from bouncing on the mount, and potentially a harder material than what you had originally.

I don't know much about the materials, other than that I assume that it needs to be a relatively hard steel. For optical applications, the surfaces are nearly mirror-like, but that's for applications where microradians of angle repeatability is dictated.

TTFN

FAQ731-376

 

[KENAT]

Thanks IRstuff.

The mass of the load is around 3-4 lbs. However there is also a spring loaded element to help hold it in place, I don't know the extra load from this.

Distance between groove and cup, in the horizontal plane is about 2.1 inch. The flat area is actually in the vertical plain; the load is cantilevered out so the moment with help from the spring gives the load here. Nominal contact point is about equidistant and down about 1.4" it?s not directly under the groove & cup though, off set maybe .55 ish. The combination of vertical plane and using Kelvin rather than 3 groove design has complicated our attempts at analyzing this design.

There should be relatively little vibration as it's mounted on an air table but, there are stages on the table to position the sample so when these move there may be some movement. I think the 'fretting" (really not sure it was fretting or even galling, I never saw it) was caused by it being taken in and out a few times during testing but can't be sure.

Our measurement device is down in the nanometer range, and although we can correct for mis-alignment to some extent the less we have to the better.

I think I?m at the point where I?m going to suggest better than 32 finish for the contact points/areas but leave it up to the designer to determine the value.

Kinematic mounts come up every now and then though here so I?ll appreciate any general input, links to resources etc.

Appreciate the help.


KENAT, probably the least qualified checker you'll ever meet...

 

[KENAT]

(Those ? should be' not sure what happened when I pasted from word)

KENAT, probably the least qualified checker you'll ever meet...

 

[IRstuff]

Well, if the number of removals was very small, then the material was definitely not hard enough. My recollection, and these guys, say you need hardened steel:

http://www.precisionballs.com/All_Kinematic_Balls.html

Hardened steel will take a fair amount of abuse before galling or fretting. Most of our optical products use hardened steel mounting plates.

Neither Newport nor Thorlabs say much about the materials, but this other company has an interesting concept and discussion about their concept and the relative stress difference:

http://www.g2-engineering.com/spherolinder-engineering.html

TTFN

FAQ731-376

 

[KENAT]

Thanks IRstuff.

Turns out this is an interim step. Basically minor adjustments to the current design to keep development going.

He's actually developing a new mount which takes some of these other things into account, I think he said he was using some components from

http://www.precisionballs.com/All_Kinematic_Balls.html

Really appreciate your input and I may come back to this topic when I see the new mount.

Ken

KENAT, probably the least qualified checker you'll ever meet...

 

[sreid]

Maybe a couple of things. I prefer a kinematic mount using three groves at 120 degrees. One advantage is that the sides of the grooves are easily lapped.

Depending on the hold down force, how much the balls can deform the mating surface is more than one might think (Hertzian Stresses). With hardened steel balls and 303 for the grooves. any surface roughness is "Peened" out on the first few insertians.

 

[KENAT]

Thanks Sreid, this is a vertical mount so a simple 3 groove design isn't practical. That said I think a modified 3 groove design may be better, I've already had this conversation when I was asked to look at it with the designer. I'm hoping the new mount maybe uses this, I'll wait to see.

The measuring device actually has adjustment screws that go into the cup/groove so the first time it goes it is adjusted to be level etc. Using the 3 groove design my colleage is concerned about movements in different axis getting coupled.

Thanks for the info on the roughness, the problem is the measurement device gets put in once (maybe a few times), the system set up around it and then it will be removed at least once probably more before being sent to customer and re-installed.

Hence we don't really wear it in before we set everything up around it. I have seen suggestions to do so and to do so with a greater than service load which I may bring up when I see the final design.

Appreciate your input.

KENAT, probably the least qualified checker you'll ever meet...

 

[Auspex]

Since your not allowing the contact points to 'wear in', the initial surface finish is more ciritcal to achieve high levels of repeatability. Keep in mind, the surface roughness must be compressed before the inherent stiffness of the elements develop a repeatable resting position.

Additionally, since the repeatability is also related to the Hertzian contact between ball and flat elements, stiffness of the mating pieces are also crucial. This is a good motivator to utilize hardened steel or even carbide (I have even seen them with pcd diamond spheres). You can get around using a ultra stiff contact with an adequate repeatable preload. I have studied the repeatability of three groove and split groove kinematic couplings in the past and magnetic preloading helps tremendously.

 

[KENAT]

Thanks Auspex. You've confirmed my thinking. As it was a stopgap we just improved it to 4. However when I see the final version I'm going to bear in mind what you and IRStuff said.

KENAT, probably the least qualified checker you'll ever meet...

 

[lumenharold]

You might find some additional information within the technical papers at:

http://www.precisionballs.com/KinematicComponentsDirectory3.htm

Harold
SW2007 SP2.2 OPW2007 SP0.0

http://www.lumenflow.com

 

[KENAT]

Thanks lumenharold, that's part of the same site I mention above but I'm not sure I'd hit that page yet.

KENAT, probably the least qualified checker you'll ever meet...

 

[lumenharold]

Yes, I noticed that you referenced the site in your opening post as did IRstuff in his response. I wasn't sure if you had a chance to read in that part of the site and thought your designer might use the info.

Sound's like a fun project though....

Harold
SW2007 SP2.2 OPW2007 SP0.0

http://www.lumenflow.com

 

[KENAT]

"Sound's like a fun project though.... "

Kind of,

I nearly inherited it when it wasn't working properly with a crazy tight timescale (that's when I did my research, months ago)...

Fortunately the original designer stepped up to the plate and we're muddling through.


I'm looking forward to reviewing the final design, so long as I have time!


KENAT, probably the least qualified checker you'll ever meet...