Taking another pass- diamond turning a spherical mirror

[@Puke0010]

Former professional diamond turner here with 12 years optics manufacturing experience. Very cool to see someone making their own DT machine. A few tips. If you can, get a larger radius tool. 0.030in -0.060in radius tool would help smooth out the surface tool lines. Spray your coolant form inside (X0) towards the outside (OD) of the part. That blows the chips away from the surface about to be cut preventing those chips from dragging and scratching the surface. Ring at center could be something in your X axis like a bump or debris. It doesn't take much for anything wrong in the machine showing up as a feature on the surface of your optic. Good luck!

[@jonbeno9926]

Extreme care mounting an optic to a hydrostatic diamond turning lathe with a hacksaw going in the background... 😂 Love it

[@R.Daneel]

That was my question. Why wasn't that hammering and machining noise (and people walking around) a concern?

[@koharaisevo3666]

@@R.Daneel The entire machine is sitting on a vibration isolation table, maybe that level of vibration isn't a concern?

[@cylosgarage]

vibrational isolation table is doing a great job, but yeah, should waited till late at night ideally.

[@jackflash6377]

As a machinist my thoughts are: 1. Taking too light of a cut can result in the tool just rubbing. 2. Did you calculate SFM and use the correct value and are you aware you need to keep it constant across the turned surface? I mean speed up at the center. 3. All tools have a minimum cut depth. Less than that and they are just rubbing like I mentioned in 1. 4. I use diamond tipped cutters all the time and I've found that I need to run them slower than what is recommended (SFM wise anyway) 5. Make absolutely sure the tool is not deflecting the stock. It needs a very strong backing to keep from being deflected while "cutting" as your not really cutting anything, your scraping which generates forces. I'm taking 0.15mm depth of cut on a 6mm thick copper bar being supported by a 50mm thick fixture and it deflects with even that small of cut. You may know all that already, just running my mouth. Cool project.

[@Puke0010]

I agree. Your finish pass should be minimum 0.0002in . When using a diamond too shallow a cut does result in a rubbing effect. Agree that everything needs to be as stiff as possible. Constant RPM is best for DT optics. Something this small 500-600 RPM throughout the cut will be fine.

[@capncharlie7894]

the sounds in the background make what Cylo is doing so much more remarkable. Hacksaws, grinders and chips flying ... then a few feet away Cylo's directly machining mirrors. Great video!

[@Bob_Adkins]

What's your occupation: Cylo: Brain surgeon and diesel mechanic.

[@cylosgarage]

ha, thanks. Doing this in a makerspace is a super non ideal environment lol.

[@daleeason9687]

Interesting project. If you want more info on DFTFringe. Let me know. I'm the author. I hope you can change the 3D plot parameters so that we can see the surface in 3D. You just need to increase the vertical scale using the controls on that display.

[@cylosgarage]

Ha! Thanks for stopping by Dale. I'm absolutely loving DFTFringe! It's honestly helped a lot with my understanding of optics. I completely missed the vertical scale settings though, d'oh...

[@daleeason9687]

@@cylosgarage The astigmatism of the second test is lower but still way to big. But if you did not average several results then it could be from air currents or even other sources. you can know for sure if it rotates with the mirror.

[@animus3d663]

Inner circle is most likely caused by an inconsistent surface speed. As you get closer to the center of the part, your radius decreases,so you would want to speed your spindle or slow your tool path in order to keep a consistent surface speed. Many lathes have a code for it, G96 Also from experience it tends to show up sharply like that, as you move out of resonance you start to rub/chatter

[@solarguy6043]

Came here to say that. It's a very common problem if the goal is high precision and/or an excellent and consistent surface finish. There is some optimum sfpm to get the best performance. It's not (often) a precise rpm with no wiggle room. It's a modest range. But once you leave that range/sweet spot.....you will get a discontinuity in the surface finish as it leaves the happy frequency and enters the unhappy and inefficient frequency.

[@maxwellvos1578]

Does the low ring line up with the work holding interface? The force from the draw bar may be flexing the plate, with the interface acting as a fulcrum, then measuring in the relaxed state creates these small errors?

[@adamtanasiewicz8119]

That's a good point

[@dainius4168]

Exactly what I came here to say

[@DTOptics]

Agreed

[@waterlife.1905]

Almost needs negative compensation for the forces induced to offset any obtrusion?

[@cylosgarage]

Video on this point soon. Spoiler alert tested it today and it has a small effect but its not nearly the main source of error

[@kuglepen64]

Time to call in the OG (Original Gelbart)

[@cylosgarage]

😂😂😂

[@mrjacob8836]

At these scales my first thought is i wonder if you're running into floating point, or other numerical precision issues in linuxcnc. You may be able to experimentally determine if that's what is going on by changing the radius of the sphere you're cutting and seeing if you can corelate the errors back to a specific maybe Z relative to the sphere geometry?

[@cylosgarage]

Not my personal top contender, but very plausible. I have a plan to investigate this soon.

[@cameronskinner6197]

Very interesting project, you have there!! It is difficult to isolate the source of your surface error since you have several potential variables in play simultaneously (as noted in the many replies/comments). I might suggest trying to machine a flat before trying to machine a figure (spherical surface). This would allow you to test your interferometry setup and potential eliminate/minimize the question of your interferometry optics and axial alignment. Then consider machining a series of channels/grooves/rings (rectangular cross section) into the surface of differing depths so as to assess the centering/axial alignment issue and accuracy of the cut. You might also want to intentionally perturb the system as it is cutting by 1- getting up/sitting down 2- dropping something on the floor next to, a few feet away, quite far away etc. so that you can isolate the source of the surface ripple. It might be the bearings or it might be telegraphed from the environment. In my research work I can detect when the busses/trucks go by our building as they cause the laser beam to bounce around despite being on an isolated optical table. Only an air cushioned table would solve that problem for me and it might be required for you. In the early days of AFM they used to suspend the whole AFM on bungee cords to isolate them from the environment - that might be your fate as well….

[@erniecamhan]

The ring happens because the speed needs to be accelerated the closer to the center the tool is, the workpiece is rotating faster closer to the edge, and slower the closer to the middle

[@Puke0010]

Surface speed isn't a factor when diamond turning. Constant RPM works better actually.

[@WaveofThought]

I don't have any first-hand experience with precision machining, but my first thought is thermal expansion. At these scales wouldn't even small changes in temperature of the part and machine over the duration of a pass cause the ~100nm scale errors you are seeing?

[@juslitor]

Aye, even when hand polishing a mirror, the thermal expansion can get you.

[@DomManInT1]

Trying to hold millionths of an inch profile tolerances: #1 A temperature variation of even one-tenth of a degree will cause variation in material and machine dimensions. #2 Any outside force applied can change machine alignment. (Like if you place your hand on the machine or even if you place your feet on the floor differently.) #3 Any external vibrations that are not compensated for will change tool to work piece orientation by millionths.

[@cylosgarage]

yep. its a fool's errand. luckily im a fool!

[@tetrabromobisphenol]

@@cylosgarage Well, at the very least, you COULD make your errand a bit less foolish by at least addressing the temperature concern. You can make a nearly constant temperature environment by use of a PID loop heater and a box as an environmental chamber. Most of them will hold to within a tenth of a degree if you use an RTD for temp measurement.

[@jonbeno9926]

Two thoughts.. first on machine performance, is the any reason to suspect the linear encoder to account for some of the error? The fact that some amount of it repeated seems to point towards more towards the machine. As a suggestion, try bumping the diamond tool a bit in Z and re-running and comparing the result, now using a different portion of the linear axes. Alternatively, take more material off and see if the errors change in radius on the optic. Another alternative would be to compare to a flat part with the Z locked. As a second thought, did the material come out of round stock? I wonder if the material properties might vary radially and the core defect you are seeing might be related to hardness/grain size variation or something like that. Perhaps try a different piece of material?

[@user-qq6vd8bz9o]

I'd look into a much steeper negative rake on the monocrystalline diamond tool. That may help with a smoother cut and less chatter = a better waviness / surface fltness.....try 15°-25° negative rake. Good luck Cyrus

[@cylosgarage]

But in aluminum? I’ve never heard of using negative rake mcd tools in aluminum before

[@Mister_G]

@@cylosgarage Zero rake for Al 👍

[@Puke0010]

@@cylosgarage You do not want negative rake for aluminum. That's for crystalline materials. I would suggest a larger radius tool. 0.030in-0.060in on a shallow part like this will help smooth tooling lines. A good speed to RPM is 500RPM with about 1mm/min feed.

[@MrYrodz]

I found out about your channel thanks to this video. Like and subscribe immediately. Optics, metalworking and metrology are my favorite topics and all in one video.

[@aeryn_mobil]

Lots of good tips in here: larger tool tip, optimizing air coolant to clear tool & just worked surface, optimizing air speed to minimize material tear away, checking x axis drive mount for debris, making sure you're actually cutting deep enough so your tool is cutting and not rubbing... have any of these helped yet? Your mirror mount seemed rock solid. When you started off with oiling the mirror, and showing us the flat line you had achieved when you ran the tool tip near the OD circumference, I was impressed... have you considered doing the same thing, feeding it inward on the X just as the tool tip would see? Maybe any machine introduced errors would be highlighted in this way. Regarding the tool holder/mount, rotational speed in relation to center, heat, and vibration: worst case scenario thinking here; tool is rubbing some, heat is transferred into material, vibrations picked up in the mirror resonate, material vibrations interfere at harmonics highlighted by the rotational speed at given circumferences. (What was the rotational speed at the break points we see in the interference scan? Are these always repeated?) Can you put a feeler Guage on the tool holder, independent of the tool mount, while making your next cut. (With another camera to record it.) The mirror mount. Is it mass dampened? Is the center of the mirror supported from within, or is it like a drum? You went to some effort to check your X dimensions and center. Check your Y. Account for tool tip size, and experiment with its cutting position. Experiment with its cutting angle.

[@ramous5182]

Could that inner circle be a change in material properties? like a heat affected zone from drilling a hole in the back? Or maybe you are crossing a threshold in surface speed where the chip breaking changes? Does the circle change size with rpm?

[@jrucker2004]

fun fact: your crosshair joke is more accurate than you might think. The origin of the word comes from Robert Hooke used hairs just like you did on the lens of a telescope.

[@danielstjean3476]

Some optics used spider webs as well up to the 1940's

[@cylosgarage]

Ha! Awesome!

[@AleksMakesStuff]

Hi from a fellow telescope builder! In my experience, the mirror like this is a little sensitive even to its own weight. Looks like your error may come from the bar pull, then compensated by defocus in DFTFringe. You may check it by varying bar tension in interferometer. Also I would recommend starting with making a flat, it makes errors better visible

[@Fgekc]

Increase spindle rpm as you get closer to the center. You see that pattern on a lathe when you don't use spindle compensated speeds.

[@walter67435]

Gosh, I know very little about this stuff but thought immediately "speed of the cutting edge over the surface." If something could be done to maintain a very precise force of the tool across the entire surface perhaps something really perfect would emerge. Certainly the tool reacts a little bit to the force on it... ?

[@graealex]

Continuing the thought, right in the center you can't really cut anything, and you need a damn good machine to keep up RPM with the loss in surface speed near it.

[@connordavis5086]

It looks like it's repeating the same form errors to me. Could you try running a corrective pass with your measurements on top of your final pass to see if it goes away? Also you could try running again off center to see if the same errors happen to determine if it's because of speed or programming.

[@eackerw85]

Just my two cents. I think the center might be from not having enough material at the smaller diameters. On or lathes we always use constant surface footage. When we get close to the center the lathe often maxes rpms out trying feed enough material. Another thing is the banging in the background could cause issues in finish quality at these levels. Also the Heat from the sun hitting the machine could be something to think about. You're playing on a level most people never even think about.

[@dylanpowell3854]

If the machine is repeatable would running a spring pass possibly help halve the errors in the profile, if the profile repeats then youd know that would be the inaccuracies is the machines abilty to correctly follow the profile that your aiming for. Since your working on such a macro scale i cant promise thatd itd be useful. It could also be noted thats the tool could be being pulled down and preloaded like a spring whilst you are cutting then as the force and pressure builds it returns ,to some degree ,back to the set height of the tool. Itd explain the wavyness of the finish because as itd pull down the distance to the tip of the tool and the part would increase, when its springs back that distance would decrease again. As you said its also likely thats its also just vibration from the air bearing By no means do i expect it to help but could be another angle

[@ShevillMathers]

Have you used a Foucault Test on it to see the surface figure? This really show the surface figure. Fascinating work, I have hand ground many glass optics over thee years for my astronomy work.

[@jrobi47]

I would try increasing the spun mass to increase stability. The more mass the harder it is to move

[@CATANOVA]

Your profile error at around 8mins in is likely due to tool pressure and part stiffness where the stiffness of the part is variable to match the reverse side features. Lapidary with diamond slurry is probably the best method to produce the best flat mirror surfaces.

[@cylosgarage]

@@CATANOVA this is not flat. What?

[@movax20h]

Fascinating. I am not a machinist, but maybe linear speed is too low when close to the center, and need to do variable rpm to compensate? Also thermal expansion might be an issue, need to pre-warm everything before doing any measurements, and maybe even model thermal expansion, because once cut, and then cool down, it will assume slightly different shape. Or probably easier pre-warm bearing by letting it run a bit, but make sure part is cooled down to room temperature.

[@demianmoody148]

i used to measure parts on cmms and i've seen those dips in the center of turned surfaces before, the turners back then explained it with the missing cutting speed at the center of the part, dont know how well that expierience translates to this though

[@truegret7778]

Is your linear cutting speed constant or do you increase as you cut closer to the center? I would think you want to maintain a linear cutting rate (or surface speed). Try to find the optimal "surface speed" for your tool and material combination, and try to maintain that "surface speed".

[@justRD1]

this is amazing! I feel like at these scales something as simple as your air mist being too powerful can affect the cut.

[@cylosgarage]

it's not off the table!

[@justRD1]

@@cylosgarage Would be neat to find out! If your nozzle is stationary, the waving could be the tooling passing through the air stream.

[@iverpossehl3316]

I wonder what the function of this mirror will be when you get the surface the way you want it? Once you get your sphere, you could tweak it a bit to get a parabola for a telescope objective. Maybe?

[@andreicheptea8020]

Maybe you have to speed it down when its approaching the center? The ring might be because of the speed of the plate is increasing in that structure and somehow resonates or thermal dilates the material? It might even be the stick-slip effect...

[@robertmckelvey3091]

Your artifacts match up with the amount of sharpie you got on that when you're turning it

[@cylosgarage]

that's a theory too, the rehbinder effect could be at play here

[@gags730]

I think he needs something like a borescope but with magnification. You could probably add an OTC magnification lens for a digital camera or the like.

[@ashpalmer1365]

Is it possible that the pulse of the air mister from the compressor is impacting the surface finish?

[@turtlemann14]

was the spindle and work holding balanced? could this be micro chatter? the center being dull looks like a velocity problem from my machining background(keep in mind that i have never tried to cut a mirror finish nor used diamond tools). sometimes, depending on the sharpness of the tool, when velocity of cut diminished forces can build because it becomes less plastic and or gulling can happen. for this reason we use constant velocity when turning. it might be worthwhile to implement an accelerometer on the spindle and tool to keep track of harmonics. also what kind of coolant are you using? there are no metallic deposits forming on top of your cutting edge are there? being from bar stock, could the rings also be from the tempering on the material as it cooled at the mill? it would be interesting to see what would happen with a piece of plate vs bar stock.

[@adamtanasiewicz8119]

Can you adjust toolpath using measured error values?

[@cylosgarage]

Yes, once I figure out how to write the code to do it...

[@MrYrodz]

@@cylosgarage If you cannot find out and eliminate the cause of the errors, they can be compensated as long as they are repeatable, as in your case. It will be interesting to see how it turns out, good luck.

[@anemac9]

Do you think chips may cause some scratching?

[@cylosgarage]

yeah chip evacuation is super important. I've been battling that consistently

[@sodium.carbide]

analyzing the frequency response on the lathe would be interesting..

[@damienmiller]

Could the ring artifact be related to the clamping force from the drawbolt?

[@marcfaulk]

Depending on the radius you're turning, could you mount the diamond tool on a pivot, so that it moves in an arc across the work piece?

[@robertfontaine3650]

There is something distinct happening in the same place in both runs. Looks more like a programmer error

[@krisjames3984]

Is it possible for the program to have a rounding error at that point? I know almost nothing about machining but at the scale he is using it seems like it would be easy to not use enough digits of pi or something along those lines...

[@Hogla287]

float32’s loose precision at the 7th decimal place or so, float64’s at 16ish. Depending on how the implementation in this program and controllers were done, that definitely could be contributing to something funky

[@DTOptics]

There are many factors in play at this level. Tool wear, tool height, ogive error, work-holding deformation, spindle balance and vibration, specific AL alloy being machined, etc. It takes very little to deform a mirror surface.

[@robertfontaine3650]

@@DTOptics absolutely true but there is a very consistent error in a very consistent place. This

[@JLK89]

Try moving the toolholder along the x-axis a few mm. If the pattern shifts, its non-linearity in the x-axis. If it doesn't, its the mounting system distorting the mirror.

[@aeromethod419]

What is the z axis riding on? Air bearings? Could stiction be causing the profile error?

[@cylosgarage]

Yep, it's on air bearings. It better not be stiction, lol

[@larsbkurin1740]

Do you have increasing speed on the mirror so that there is a constant cutting speed? Next thought - as the radius starts to approach zero the impact increases of the inserts profile on the surface.

[@staryduren]

Internal stress in material? ( if starting from cylinder, i would match somewhat geometry )

[@cylosgarage]

possibly...

[@capncharlie7894]

7:00 I have seen work hardening artifacts from prior passes and/or the original material condition showing through machining passes. you're taking so little off some of the stress in the material may be there from the section that was not cleaned up so deeply. The chip geometry may have changed abruptly at the spot of the ring. the air nozzle scares me w.r.t. surfaces finish as well. it has noise. the noise is pressure over the surface of the workpiece ... so you have a variable load on the end-play bearing. Is the variation in this end play smaller than your tolerance?

[@cylosgarage]

That's sort of what I was thinking about the ring too. With regards to the air though, that effect is on the order of single digit nanometers, and very high frequency. I'll investigate it when I really chase down surface roughness, but for now im focusing on form a little.

[@fearlyenrage]

just WOW what can be achieved with a insert those days!

[@chronokoks]

Using diamonds for cutting on lathes in aluminium has been used for mirror finishes for like 100 years now.

[@fearlyenrage]

Thy for sharing that info.

[@BLUYES422]

try using OBS on your computer next time you do this, you can add the webcam as a source to the scene then you can add images (like a crosshair) to the scene itll be more accurate

[@arthurjohnson3438]

You should set up some macro to calculate your feed rate and match to surface speed as you traverse X

[@1kreature]

Could the ring in center be caused by cutting speed crossing a critical speed boundary? I've noticed my surface finish changes on my lathe when facing if I do not slow down the feed as I close up on center.

[@stumccabe]

Could the ring in the centre be due to the lower surface speed of the tool not being as it approaches the centre? I assume there is an optimal surface speed range for the cutter and the material.

[@GHOSTGXZ]

what is the purpose of the mirror? I mean is it for a telescope or?

[@cylosgarage]

Ideally yeah

[@realcygnus]

👍

[@felixaudet5860]

Just curious about what you want to use this for..

[@cylosgarage]

It'd be fun to make functional optics for telescopes and stuff. but it was mostly just about building the thing

[@flikflak24]

you need some better thermal and wind isolating/ protection m8 man ( most diamond turning lathe's have a insulated sheet metal encluser on it to protect agens quick temp change and then over that they have a plastic greenhouse tent over it so when a door opens the wind dont hit the machine and makes it change/twist do the temp change. since we are in nanometer the temp changes is extremely important after all. if the whole machine is 20 degree C and one side siddently gets cooled down to like 18.5 degree C well cutting the machine will shrink and twist witch can be seen and measured down in those measurement's )

[@dhoodlum4129]

You should just do the whitworth lapping plate method on three pieces. It would take very few passes to make some very nice fringes.

[@chronokoks]

the surface is curved so you would have to not do 3 plate method but a regular optics lapping method :) But it's aluminium, polishing aluminium to nano /sub-nano roughness is a science

[@jonedmonds1681]

Diamond cut the negative and hand lap? Or pitch mould the negative?

[@dhoodlum4129]

@@jonedmonds1681 you would need to make three of them for super flatness, but the are already really flat. I didn't realize it is aluminum, so polishing actually not that easy I imagine, I have not tried.

[@chronokoks]

@@jonedmonds1681 he didn't understand you buddy :D optics lapping/polishing is outside his scope.. he just learned a fancy word like the "wittworth 3 plate method".. I don't blame him.. just a bit of ridicule so he starts learning up on other things too

[@dhoodlum4129]

@@chronokoks You may be right. I've only had my hobby machine shop for 5 years while working a full time job as a software engineer without a college degree for 10+ years. I've only applied whitworth three plate method 4 times. I don't have a degree past highschool. You might be right. However I do believe the only way to make a precision flat surface without any tools at all is to use the "whitworth three plate method." I know it seems like a big scary buzz word, but it's not. It was documented by Joseph Whitworth who also devised the British Standard Whitworth (BSW) screw thread standard. Being the first to document machining fundamentals of precision flatness and screw thread standard, he laid out the foundation of everything related to machining tech today. Does this make sense to you? Although I have my doubts, your feedback is always important. Otherwise, can you just sit down and watch the fucking video?

[@adam207321]

what is the mirror made of ? What material? From experience it is easier to achieve a desireablae surface finish on harder metals like steel stainless. Much more problematic with aluminium

[@chronokoks]

Folks who polish clear coats on cars call these micro scratches "holograms". I think it's an awesome name. And what's the black stuff on the aluminium? Is that india ink?? Wink wink

[@CATANOVA]

At 2mins 38 you can see them belt sanding to get the surface perfect. kzbin.info/www/bejne/hYHNioWajNKNgKM

[@xrfa7422]

Get a clean room when you can afford it.

[@cylosgarage]

ok let me graduate first tho

[@jukkapekkaylitalo]

I think your cutting speed in center of the part is too slow. If y can try slowly speeding up rpm's closer y are of center. Also try turning from inside out. your part warms more when y turn smaller radius. If y have first turned whole outer rim your heat has slowly creeped trough part towards the center. Heat is one reason why getting mirror finish Lapping is usually used. Even the it is done many passes because longer y do it in one time more the part warps.

[@user-qq6vd8bz9o]

Second

[@poetac15]

First.

[@critical_always]

Precision machining? How the fuck are you going to do that when people are banging away in the same room?