Poster (made for a general audience) can be found here. (Much more technical) paper is in the works: drive.google.com/file/d/1LTucMVWiW7LhGn0QnoJ7--nxAYHHSNTS/view?usp=drivesdk
hi just wanted to say thanks for making this and opensourcing the components and your thoughts throughout the build. I was looking to create my own similar machine for turning camera lenses, and now I have a starting point.
Please post a link to your paper...also, it would be nice to see an Interferometer reading of the surface vs the desired result to see the difference. Nice work!!!
Don't forget to put it on a larger piece of vibration-mounted granite and attach an accelerometer to track vibrations from vehicles outside. You can't really compensate for them in real time but you can learn when there is less traffic and get better results most often in the middle of the night.
The machine base is a 24x18 surface plate, sitting on rubber pads atop of a pneumatic vibration isolation table for optics. The encoders themselves are pretty sensitive accelerometers lol
Awesome to see such a cool project coming together! Thank you for continuing to share your progress, seems like you are not to far off from your form & roughness goals. Congrats! What are your plans to continue working on it? Are you graduating at the end of this term?
Amazing results ! Maybe using ethanol mist cooling might have a positive impact on the finish, but maybe this type of turning is usually without any type of coolant.
Is there any chance you will open source any part of this project? I am trying to construct a submicron-precision X/Y table and would love to be able to learn from your designs and drawings.
Just found your channel. Awesome work. I had worked at Precitech then ABTech about 20 years ago designing lathes just like this. I have a few air bearing spindles and an XY stage that I want to build into a lathe or precision mill eventually. Keep up the great work!
Currently the x axis (linear motor) is performing better. However, the poor performance of the z axis is due to vibrations of the air bearings, not the actuator. It has a noise floor of about 30nm. X axis hold to 2nm
Looks amazing! Always impressed by your projects I'd love to see some other shapes. I was wondering about the temperature of the part during work. It'd be interesting to see the process with a themal camera. I wonder if getting the airflow and temperature distribution just right will help improve the finish even more.
I will attemt to build my own hydrostatic guideway for a similar sized project. What width of carriage and rail do you recommend? And thickness of metal. I am going to make it out of aluminium
I cannot tell you that as I have no idea what your design requirements and constraints are. What loads do you need to support? How stiff does it need to be? My rail is 2 inches by 5 inches, but that doesn’t really help you to tell you that.
I just want to build a hydrostatic bearing for fun. And potentially make it into a cnc, grinder or lathe in the future. Rails will be 50cm long (20inches) I have been looking at aluminium blocks to order to put four making a square rail then four larger ones as the carriage.
A precision air regulator is good practice, but the the Professional Instruments air bearing spindles have a boatload of elastic averaging which makes them quite stiff and arguably one of the lowest error motion spindles available anywhere.
Great work Cyrus, very impressive! Were all the cuts made at constant RPM? Did you attempt cutting at constant surface ft/min? I would love top be able to read the paper you showed at the end, is there a place we can view or download it?
Thanks Carl! In diamond turning, you actually don’t see constant SFM turning almost at all, due to changes in RPM corresponding to changes in balance/vibration and dynamic error motion/ runout. It’s the same reason you always want to dress your grinding wheel at the same RPM you grind at. Diamond tools are relatively SFM insensitive in soft metals, so they don’t mind the drop in surface speed towards the middle. I added the poster in the pinned comment
@@cylosgarage Thanks for the poster. It doesn't appear to be pinned. What is the quantity "L" in the definition of Ra? What is the periodic variation of Z in the surface roughness graph, is this due to a vibration?
Wouldn't you get a better surface finish if you just bought an innolite, installed it in a climate and vibration controlled facility, hired a ultra experienced operator and just told them to do it? Smh ... /s
I achieve a similar appearance in finish just on my mill and lathe using diamond tools. I don't do any optical parts. My customers want the mirror finish for heat transfer reasons. You should do a comparison between the finish you achieve here VS a CNC lathe with a new spindle that hasn't been crashed. Your finish didn't look very good to me BTW. The parts looked scratched up from chips maybe? I get mirror finishes on my lathe, but there is a cusp that can be detected by eye from the right angle. I would be interested in the actual surface roughness of your parts as well as an analysis from equipment made specifically to measure lenses like a "Zygo" brand machine. Otherwise, to me this looks like a lot of effort for something I do every day on my cnc lathe.
That is cool that you are able to get mirror finishes on your conventional CNCs. I think the planar mirror is just a simple demonstration of the diamond turning lathe. As an example, imagine you tried to turn a precise spherical mirror with a P-V of 0.001” on your machine. Then I think you would be disappointed in the result you get from the common CNC lathe, because the granularity of the movements due to insufficient feedback resolution. Besides this, there could maybe be 5 or 10% error due to thermal drift.
Oh, shut the fuck up. If you actually can make parts to the same level you should be able to recognize the amount of work put into this and be glad that anyone would put in the effort. Gatekeeping does no good for any of the trades.
This lathe has a positional resolution of 2 nanometers (.002 micron) and can produce parts with a circular flatness on the order of 1-10 microinches. Andrew explained it well.
@@cylosgarage I don't doubt it. I was commenting on this video that mostly just showcases the finish. I'd be more interested in seeing a video of what makes this really special. A side by side comparison with a workpiece made with a standard CNC lathe with the same program and tools would be even more interesting.
Poster (made for a general audience) can be found here. (Much more technical) paper is in the works:
drive.google.com/file/d/1LTucMVWiW7LhGn0QnoJ7--nxAYHHSNTS/view?usp=drivesdk
As a hobby machinist, this is an obnoxious level of precision. Subbed.
hi just wanted to say thanks for making this and opensourcing the components and your thoughts throughout the build. I was looking to create my own similar machine for turning camera lenses, and now I have a starting point.
Love the contrast between precision air bearings, and then hose clamps around a bolt lol. Nice work, impressive results
A fine example of appropriate application of precision in operation setup.
Please post a link to your paper...also, it would be nice to see an Interferometer reading of the surface vs the desired result to see the difference. Nice work!!!
Working on the paper but I’ve linked the poster
Two years in the making, but you're going to make some awesome rocket motors with this!
Lol
Don't forget to put it on a larger piece of vibration-mounted granite and attach an accelerometer to track vibrations from vehicles outside. You can't really compensate for them in real time but you can learn when there is less traffic and get better results most often in the middle of the night.
The machine base is a 24x18 surface plate, sitting on rubber pads atop of a pneumatic vibration isolation table for optics. The encoders themselves are pretty sensitive accelerometers lol
There is no sound :( Can you reupload this video?
Congrats dude! That is a real accomplishment!
Awesome to see such a cool project coming together! Thank you for continuing to share your progress, seems like you are not to far off from your form & roughness goals. Congrats!
What are your plans to continue working on it? Are you graduating at the end of this term?
Awesome! Congratulations! Nice mirrors!
thank you
You did it man, congrats
Amazing results ! Maybe using ethanol mist cooling might have a positive impact on the finish, but maybe this type of turning is usually without any type of coolant.
Is there any chance you will open source any part of this project? I am trying to construct a submicron-precision X/Y table and would love to be able to learn from your designs and drawings.
Sure, you can have whatever you want. What is it for? What kind of bearings/drives? I can be more of assistance if I know what you need
Wow congrats! Looks fantastic.
Very nice work!
Is that poster or other sort of publication shown in the last few seconds, available to look at?
Pinned comment
awesome project!
Just found your channel. Awesome work. I had worked at Precitech then ABTech about 20 years ago designing lathes just like this. I have a few air bearing spindles and an XY stage that I want to build into a lathe or precision mill eventually. Keep up the great work!
Is the systems linear motion more accurate with the linear motor or the other servo you have and are they similar accuracy?
Currently the x axis (linear motor) is performing better. However, the poor performance of the z axis is due to vibrations of the air bearings, not the actuator. It has a noise floor of about 30nm. X axis hold to 2nm
Looks amazing! Always impressed by your projects I'd love to see some other shapes.
I was wondering about the temperature of the part during work. It'd be interesting to see the process with a themal camera. I wonder if getting the airflow and temperature distribution just right will help improve the finish even more.
Great work!, Awesome finish on the part.
Simply incredible
Well done
I will attemt to build my own hydrostatic guideway for a similar sized project. What width of carriage and rail do you recommend? And thickness of metal. I am going to make it out of aluminium
I cannot tell you that as I have no idea what your design requirements and constraints are. What loads do you need to support? How stiff does it need to be? My rail is 2 inches by 5 inches, but that doesn’t really help you to tell you that.
I just want to build a hydrostatic bearing for fun. And potentially make it into a cnc, grinder or lathe in the future. Rails will be 50cm long (20inches) I have been looking at aluminium blocks to order to put four making a square rail then four larger ones as the carriage.
What are you using for the spindle air regulation? Have you noticed correlation of spindle air pressure and roughness/form error?
A precision air regulator is good practice, but the the Professional Instruments air bearing spindles have a boatload of elastic averaging which makes them quite stiff and arguably one of the lowest error motion spindles available anywhere.
Great work Cyrus, very impressive! Were all the cuts made at constant RPM? Did you attempt cutting at constant surface ft/min? I would love top be able to read the paper you showed at the end, is there a place we can view or download it?
Thanks Carl! In diamond turning, you actually don’t see constant SFM turning almost at all, due to changes in RPM corresponding to changes in balance/vibration and dynamic error motion/ runout. It’s the same reason you always want to dress your grinding wheel at the same RPM you grind at. Diamond tools are relatively SFM insensitive in soft metals, so they don’t mind the drop in surface speed towards the middle. I added the poster in the pinned comment
@@cylosgarage Thanks for the poster. It doesn't appear to be pinned. What is the quantity "L" in the definition of Ra? What is the periodic variation of Z in the surface roughness graph, is this due to a vibration?
can you post your research paper so I can review it?
Pinned comment
WOOOW... Great.... What's the surface finish @0:18 ?
awesome! can you post a link to your paper ? cheers
Pinned comment
Wouldn't you get a finer finish by just lapping the surface with diamond slurry ?
Perhaps, but you can't lap arbitrary profiles. This lathe can do a lot more than make things flat. Flatness is just a useful test of precision.
how is it home made if PIC gave you the spindle?
They told us to do everything else ourselves but don’t bother trying with the spindle 😂
How long does it take you to make a medium size salad bowl from a maple log?
lol
Congrat, thumbs up of course!
that vacuum is so strong
*DANG*
Looks like a metal at a dentist.
Wouldn't you get a better surface finish if you just bought an innolite, installed it in a climate and vibration controlled facility, hired a ultra experienced operator and just told them to do it? Smh ... /s
It’s a pity we can’t hear anything.
Well damn. Lol Excuse my French!
I achieve a similar appearance in finish just on my mill and lathe using diamond tools. I don't do any optical parts. My customers want the mirror finish for heat transfer reasons. You should do a comparison between the finish you achieve here VS a CNC lathe with a new spindle that hasn't been crashed. Your finish didn't look very good to me BTW. The parts looked scratched up from chips maybe? I get mirror finishes on my lathe, but there is a cusp that can be detected by eye from the right angle. I would be interested in the actual surface roughness of your parts as well as an analysis from equipment made specifically to measure lenses like a "Zygo" brand machine. Otherwise, to me this looks like a lot of effort for something I do every day on my cnc lathe.
Let's see your parts, and then we can be the judges of them, too.
That is cool that you are able to get mirror finishes on your conventional CNCs. I think the planar mirror is just a simple demonstration of the diamond turning lathe. As an example, imagine you tried to turn a precise spherical mirror with a P-V of 0.001” on your machine. Then I think you would be disappointed in the result you get from the common CNC lathe, because the granularity of the movements due to insufficient feedback resolution. Besides this, there could maybe be 5 or 10% error due to thermal drift.
Oh, shut the fuck up. If you actually can make parts to the same level you should be able to recognize the amount of work put into this and be glad that anyone would put in the effort. Gatekeeping does no good for any of the trades.
This lathe has a positional resolution of 2 nanometers (.002 micron) and can produce parts with a circular flatness on the order of 1-10 microinches. Andrew explained it well.
@@cylosgarage I don't doubt it. I was commenting on this video that mostly just showcases the finish. I'd be more interested in seeing a video of what makes this really special. A side by side comparison with a workpiece made with a standard CNC lathe with the same program and tools would be even more interesting.