For sure I'd run the error signal through an FFT to see what your primary frequencies are. After that, if the oscillations are present in the control loop then a low-pass or notch filter in the feedback might help suppress the noise. I'm super excited to see what you come up with!
Running a different gas like argon or changing the temperature of the gas feed through the bearing might tell you if its a cavity resonance issue. Or it could be due to the air supply itself - such as if your air supply is vibrating. Hitting the resonance of a valve can also cause feed issues
Air bearings are highly progressive by nature, so that may be a tough one! Only thing that comes to mind is more massive bearing parts to smooth the micro-oscillations.
A few ideas: 1. You mentioned in your paper that this oscitation motion is in Θy axis, is there any way you can mount your spindle so that it can reach a tool mounted at the center of this rotation of the z-stage? 2. Tuned mass damper - since you seem to have a nearly constant resonant frequency this may help to reduce the amplitude significantly. 3.Mount additional mass as far from the center of rotation of the z-stage as feasible to reduce the resonant frequency of the carriage to try and get the air bearing to react quickly enough to compensate 4. Modify the z-air stage to have either better tuning of the orifice on the x-axis constraint or add further constraints or damping in this axis either further along the length of the z-axis or on the sides of the y-axis constraints.
Do you have an accumulator in your pneumatic line, or is fed directly from a compressor? Accumulators can help dampen pressure fluctuations if that's contributing.
Flow rate might be one for me, either too high or too low but next to the sweet spot unless it's getting a flux. I'd be curious if there's also vibration in the floor. Next to that I'd be looking into how the part is held.
It would be interesting to measure the stiffness of the air bearing. Also floating height would be of interest. If you think you have an oscillation of 1 uin, how small do you suppose the footing height should be? Similar or less? Is there vacuum preloading on these bearings (can’t remember if you covered this before).
Can you post a link to the paper you wrote about the lathe? As for the problem you’re experiencing, would it be possible to cancel out the vibrations with acoustics?
Also try an ultra-precision diamond turning tool with either a positive or negative rake, plus a smaller radius than the .500 micron (.020"), that would give you a much smaller surface area contact point and stop possible chatter
I'm not bashing your tool edge or controlled waviness Kieran, I am very familiar with Edge Technologies and the manufacturing of ultra-precision diamond turning tools. I'm suggesting to Cyrus to try a smaller radius with a better controlled waviness which is more easily achievable with a conical radius on a mono crystalline synthetic diamond.
@@JimThompson-t8f I see, the best I am able to reliably measure a tool is .04um across any meaningful sweep of the tool arc, we could certainly try to make a tool to those specs. But the machine dynamics are well outside of my wheel house.
Agreed, the best waviness that I can measure is an .02-.04 um on extremely small radii, with windows no larger than 80°-100°. I think Cyrus was trying to find a correlation between your waviness chart and his readings. But your window was 82.1° which he clearly wasn't using the whole window. That said, if he's turning a true flat surface, the highest peak on your radius would be all that's touching the part, assuming the radius was small enough and did not have a large surface area point of contact. That's why I suggested a smaller size radius, to reduce the surface area contact point and going with a conical ovet a cylindrical waviness radius. Now, if he was plunge turning, the peaks and troughs would definitely come into play. Also if he had a steeper positive or negative rake angle, the shearing or tearing of the aluminum mirror could be reduced and give a smoother, chatter free cut. I assume you went with a 110 plane over a 100 plane synthetic crystal for hardness reasons on the front cutting edge too. Personally, I think it's a machine / mechanicall, electrical or air-bearing induced error that needs to be addressed.
@@JimThompson-t8f We did orient cube-cube for that tool, we generally only put conical radii on our tooling. One of the other phenomenon we see sometimes in aluminum is tool break in. We expect see some rainbow in the part until the tool edge wears in enough to get through the subsurface damage in the diamond/nano radius grows to a point where the edge "burnishes" (not sure what the correct verbage is)the hard particulate rather than tears them out. I also do not know what wavelength he is targeting for is optical surface, so "rainbow" may be ok depending on the wavelength he wants to look at.
Not related to the vid - What happens if you sharpen a standard tungsten carbide or CBN tool to the sharpness of your diamond tool and used that instead? is the only concern with using these softer materials that they just don't hold a super sharp edge for as long so they can't be used to achieve these optical finishes? I guess grain size of the tungsten carbide composite could effect the practical sharpness limit.
You can’t get WC or CBN as sharp as diamond, and hence you can’t get the same optical finish, or realize the same near-zero cutting forces, which contributes to part accuracy
@@cylosgarage Makes sense. I'm curious what's the minimum depth of cut you can effectively take with your setup on aluminum? I remember seeing some paper on this lathe. I didn't bookmark it and I can't remember where you posted it. Could you please link it? Thanks!
Can you characterize the air bearing run out as a function of air bearing pressure? I found that higher pressures can result in less runout for my rheometer....
That spindle walked so this lathe could run. I’m really proud of the x axis, the hydrostatic bearings of which represent the cumulation of all the things I learned doing the air bearings. Completely designed and machined in house. The project has a finite timeline associated with it, otherwise we would’ve made everything ourselves. But when offered a PICo spindle, you use a PICo spindle :)
@@cylosgarage yeah it is kind of unfortunate how difficult these machines are to build from scratch. Tens if not hundreds of years of combined experience and research go into them,
If you can't eliminate the cause of those vibrations, you need to filter them out, right ? I don't know how feasible this is, but increasing the weight of the axis would act as a mechanical low pass filter because of the added inertia. I've used this trick to stabilize the reading of a scale on a "vibrating" table. It's a different situation but the method should apply, right ?
@@cylosgarage Do you know if the servo electronics are reporting notable difficulty keeping the Z-axis motor in position? As in, if you can decompose the error motion at the cutting point into polar coordinates, ie. into the sum of (1) rotation-about-a-fullcrum (with the fulcrum defined by your friction-bar coupling position) and (2) a back-and-forth component colinear with the friction bar. Is force 2 negligible? (I can't recall if modelling error this way, as a rotation of a rigid body, is a sane decision with your design... I suppose at this level of precision there really is no such thing as a rigid body...) Anyway, ignoring that, might be time to increase the budget a little and pick up some more graphite and epoxy, lol.
Need more dumping. I vote for that famous strong magnet and thick copper trick (ua-cam.com/video/sENgdSF8ppA/v-deo.html), as any move induce Eddy current and tend to keep part stay still, and as copper is not ideal condictor energy is converted to heat in copper. No idea how big must be magnet and copper parts, what to attach to base and what to part with dimond tool, but i think is can decrease vibrations.
My 2 cents. Is it possible to lock the air slide (shut down air) at some point and use a mechanism to move just the cutter independently? James Webb Web telescope has a mechanism with very interesting specs (details in this video ua-cam.com/video/5MxH1sfJLBQ/v-deo.html) that has a resolution < 10nm and if I am not mistaken range of 20mm that could inspire you to some solution. So the air slide would be like a "coarse" adjustment and the mechanism a fine positioning.
The JWST compliant mechanisms are cool, and maybe in the fine range they'd work. The coarse range wobbles though, not sure it's suitable for machining..?
For sure I'd run the error signal through an FFT to see what your primary frequencies are. After that, if the oscillations are present in the control loop then a low-pass or notch filter in the feedback might help suppress the noise. I'm super excited to see what you come up with!
This would be the first step for sure. I would imagine he has already checked this however.
Running a different gas like argon or changing the temperature of the gas feed through the bearing might tell you if its a cavity resonance issue. Or it could be due to the air supply itself - such as if your air supply is vibrating. Hitting the resonance of a valve can also cause feed issues
Air bearings are highly progressive by nature, so that may be a tough one! Only thing that comes to mind is more massive bearing parts to smooth the micro-oscillations.
A few ideas:
1. You mentioned in your paper that this oscitation motion is in Θy axis, is there any way you can mount your spindle so that it can reach a tool mounted at the center of this rotation of the z-stage?
2. Tuned mass damper - since you seem to have a nearly constant resonant frequency this may help to reduce the amplitude significantly.
3.Mount additional mass as far from the center of rotation of the z-stage as feasible to reduce the resonant frequency of the carriage to try and get the air bearing to react quickly enough to compensate
4. Modify the z-air stage to have either better tuning of the orifice on the x-axis constraint or add further constraints or damping in this axis either further along the length of the z-axis or on the sides of the y-axis constraints.
Ah, the old powerline dampening trick of mass damper.
I fucking love this comment section. Never change, guys.
I just comment here to come across smarter than I am, LOL.
How stable is the air supply? Could adding a accumulator (airtank) help. Sort of how your hydraulic pump reduces oscilations?
i think the vibrations are on a much smaller scale than something like that, probably more on the scale of turbulent air flow in a pipe or something
Ballast air tank (accumulator tank)is easy to try.
Do you have an accumulator in your pneumatic line, or is fed directly from a compressor? Accumulators can help dampen pressure fluctuations if that's contributing.
We have very a stable supply that feeds the entire building as well as an accumulator before the bearings
Flow rate might be one for me, either too high or too low but next to the sweet spot unless it's getting a flux. I'd be curious if there's also vibration in the floor.
Next to that I'd be looking into how the part is held.
It would be interesting to measure the stiffness of the air bearing. Also floating height would be of interest. If you think you have an oscillation of 1 uin, how small do you suppose the footing height should be? Similar or less? Is there vacuum preloading on these bearings (can’t remember if you covered this before).
Compliance is around 1-2lb/microinch. Fly height is .0001”. Fixed opposed pad with no additional preload
May be improving the stiffness of the air bearing with proper constraints. I would also test for hydrostatic (oil) bearing.
I think your flexture tool post may not be stiff enough and is giving you too much deflection during the cutting cycle
Very cool 👍
Can you post a link to the paper you wrote about the lathe? As for the problem you’re experiencing, would it be possible to cancel out the vibrations with acoustics?
He has the link to his paper under his community tab.
Also try an ultra-precision diamond turning tool with either a positive or negative rake, plus a smaller radius than the .500 micron (.020"), that would give you a much smaller surface area contact point and stop possible chatter
It is an ultra-precision tool, the plot is showing the tool edge (in red) is round to .15um peak to valley.
I'm not bashing your tool edge or controlled waviness Kieran, I am very familiar with Edge Technologies and the manufacturing of ultra-precision diamond turning tools. I'm suggesting to Cyrus to try a smaller radius with a better controlled waviness which is more easily achievable with a conical radius on a mono crystalline synthetic diamond.
@@JimThompson-t8f I see, the best I am able to reliably measure a tool is .04um across any meaningful sweep of the tool arc, we could certainly try to make a tool to those specs. But the machine dynamics are well outside of my wheel house.
Agreed, the best waviness that I can measure is an .02-.04 um on extremely small radii, with windows no larger than 80°-100°.
I think Cyrus was trying to find a correlation between your waviness chart and his readings. But your window was 82.1° which he clearly wasn't using the whole window. That said, if he's turning a true flat surface, the highest peak on your radius would be all that's touching the part, assuming the radius was small enough and did not have a large surface area point of contact. That's why I suggested a smaller size radius, to reduce the surface area contact point and going with a conical ovet a cylindrical waviness radius. Now, if he was plunge turning, the peaks and troughs would definitely come into play.
Also if he had a steeper positive or negative rake angle, the shearing or tearing of the aluminum mirror could be reduced and give a smoother, chatter free cut.
I assume you went with a 110 plane over a 100 plane synthetic crystal for hardness reasons on the front cutting edge too.
Personally, I think it's a machine / mechanicall, electrical or air-bearing induced error that needs to be addressed.
@@JimThompson-t8f We did orient cube-cube for that tool, we generally only put conical radii on our tooling. One of the other phenomenon we see sometimes in aluminum is tool break in. We expect see some rainbow in the part until the tool edge wears in enough to get through the subsurface damage in the diamond/nano radius grows to a point where the edge "burnishes" (not sure what the correct verbage is)the hard particulate rather than tears them out. I also do not know what wavelength he is targeting for is optical surface, so "rainbow" may be ok depending on the wavelength he wants to look at.
I'm wondering if increased mass might not reduce the jitter of the air bearing?
Not related to the vid - What happens if you sharpen a standard tungsten carbide or CBN tool to the sharpness of your diamond tool and used that instead? is the only concern with using these softer materials that they just don't hold a super sharp edge for as long so they can't be used to achieve these optical finishes? I guess grain size of the tungsten carbide composite could effect the practical sharpness limit.
You can’t get WC or CBN as sharp as diamond, and hence you can’t get the same optical finish, or realize the same near-zero cutting forces, which contributes to part accuracy
@@cylosgarage Makes sense. I'm curious what's the minimum depth of cut you can effectively take with your setup on aluminum? I remember seeing some paper on this lathe. I didn't bookmark it and I can't remember where you posted it. Could you please link it? Thanks!
Can you characterize the air bearing run out as a function of air bearing pressure? I found that higher pressures can result in less runout for my rheometer....
I think you need more viscous air. 😃
Air so thick you could cut it with a knife, LOL. I see a new business venture.
So basically the lathe right now is made up of a salvaged Z axis and a "brand name" air bearing spindle? What happened to your creations?
That spindle walked so this lathe could run. I’m really proud of the x axis, the hydrostatic bearings of which represent the cumulation of all the things I learned doing the air bearings. Completely designed and machined in house. The project has a finite timeline associated with it, otherwise we would’ve made everything ourselves. But when offered a PICo spindle, you use a PICo spindle :)
@@cylosgarage yeah it is kind of unfortunate how difficult these machines are to build from scratch. Tens if not hundreds of years of combined experience and research go into them,
maybe you can make small enough chanels in your air bearing in order to make the flow laminar
different gas than air thats denser could dampen it? or is that not feasible?
If you can't eliminate the cause of those vibrations, you need to filter them out, right ?
I don't know how feasible this is, but increasing the weight of the axis would act as a mechanical low pass filter because of the added inertia.
I've used this trick to stabilize the reading of a scale on a "vibrating" table.
It's a different situation but the method should apply, right ?
Hi sir , what grade aluminium is used here
6061
@@cylosgarage
Thank you for the reply sir, what kind of insert is used for turning sir.
@@sandeeps7979 monocrystalline diamond, .5mm nose radius
Is the Z-axis stage using porous or orifice air bearings? (Sorry if I'm making you repeat yourself. :P)
Orifice, which I hear is part of the problem
@@cylosgarage Do you know if the servo electronics are reporting notable difficulty keeping the Z-axis motor in position?
As in, if you can decompose the error motion at the cutting point into polar coordinates, ie. into the sum of (1) rotation-about-a-fullcrum (with the fulcrum defined by your friction-bar coupling position) and (2) a back-and-forth component colinear with the friction bar. Is force 2 negligible?
(I can't recall if modelling error this way, as a rotation of a rigid body, is a sane decision with your design... I suppose at this level of precision there really is no such thing as a rigid body...)
Anyway, ignoring that, might be time to increase the budget a little and pick up some more graphite and epoxy, lol.
Personally I just go straight for oil
Need more dumping. I vote for that famous strong magnet and thick copper trick (ua-cam.com/video/sENgdSF8ppA/v-deo.html), as any move induce Eddy current and tend to keep part stay still, and as copper is not ideal condictor energy is converted to heat in copper. No idea how big must be magnet and copper parts, what to attach to base and what to part with dimond tool, but i think is can decrease vibrations.
My 2 cents. Is it possible to lock the air slide (shut down air) at some point and use a mechanism to move just the cutter independently? James Webb Web telescope has a mechanism with very interesting specs (details in this video ua-cam.com/video/5MxH1sfJLBQ/v-deo.html) that has a resolution < 10nm and if I am not mistaken range of 20mm that could inspire you to some solution. So the air slide would be like a "coarse" adjustment and the mechanism a fine positioning.
The JWST compliant mechanisms are cool, and maybe in the fine range they'd work. The coarse range wobbles though, not sure it's suitable for machining..?
*Promosm*