will such a CNC work for making an imaging lens mold?

[shoujin]

I've been trying to avoid having to use custom lenses and trying to use stock or off-the-shelf components but I may need few custom lenses. A colleague of a friend lately acquired a precision CNC milling machine and makes internal parts of wrist watches like gears and such and says he can make aluminum molds for small lenses very cheaply for me. While it is still a lot of work and time spent to injection mold the lens, make precision housings and get AR coatings on the acrylic lenses, the idea of having access to a CNC that I can make my own lens molds and making the lens assembly from start to finish locally is something refreshing and exciting. But will the CNC machine meet the precision requirements for pmma imaging lenses? It doesn't seem to be made specifically for lens manufacturing and I'm not aware if those are even more accurate or not. What specs should I look at and what values to expect for this to work? Right now all I can say is the machine makes mechanical part of a wrist watch and it's very impressive to the naked eye, but not enough to judge its use for optics, no matter how confident the owner is it will suit the job.

[VDX]

... super-precise "optical grade" CNC milling needs accuracy and resolution below 1 micron -- common precision CNC-mills are more around 5 to 10 microns.

You can get optical surface accuracies too by polishing them, but then the overall precision is more an averaging, than accurate ...

Viktor

[shoujin]

Depends on what we mean by precision. His can move at sub micron steps but the drill bit (is that the term for it? probably not) can cut a hole only 1 micron or more.

[VDX]

... it's no problem to move sub-micron steps - have some linear stages with 0,1 micrometer resolution, which will give a repetability of maybe 1-2 microns over 100mm travel, when assembled in a rigid cartesian setup.

But when milling a 3D-shape with this setup and measure it, the overall result over an area of let's say 20mm (typical lenses diameter) will vary by more than 5 microns with ambient temperature, humidity and eventual bypass of a heavy truck on a neighbouring street in 200 meter distance or a train in a kilometer ... manufacturing with "optical" precision is much more than simply microstepping ...

Viktor

[shoujin]

will vary by more than 5 microns with ambient temperature, humidity and eventual bypass of a heavy truck on a neighbouring street in 200 meter distance or a train in a kilometer ... manufacturing with "optical" precision is much more than simply microstepping ...

Like I said, depends on what you mean by precision. I'd love to know what precision means here as well as what determines it.
From what I know any 3d printing, CNC milling or injection molding can be affected by ambient temp and humidity. As for your other example of vibrations from the earth, well what is the difference then in these optical CNCs that make them have higher precision than a micromachining CNC or lathe with precision sub-micron steps and 5 micron bit and not being operated from inside a bunker?

[VDX]

... but they mostly are operating "inside a bunker"

This sort of machines are placed on really heavy "seismic masses", which are isolated against vibrations and installed in a climate controlled room.

Here in Germany at the IMM (Institute for Microtechnik Mainz) they installed such a "super precision lathe" with some ten Nanometers precision and sometimes got "random" errors with some hundred Nanometers to Micrometers ... needed some months of investigation, until they could point this "errors" to trains entering the main station in 7 Kilometers distance!

There was a rock structure beneath the building, which "connected" the building with the main station.

They had to change to another building and add some more isolation and damping to get the lathe running nominal ...

Viktor

[shoujin]

I don't know what kind of optics those make but I have a feeling they are of a much higher tolerance than what I need.
But still interested what precision means, the diameter of the bit, the step, the error range, or something else/more.

[VDX]

... my smallest mill-bits have a diameter of 0,3mm ... the thinnest drill is 0,1mm.

This was not meant for making optics, but "only" some sensors and parts and tools for micro-assembling them

Viktor

[buffo]

A colleague of a friend lately acquired a precision CNC milling machine and makes internal parts of wrist watches like gears and such and says he can make aluminum molds for small lenses very cheaply for me.

I do not believe he will be able to achieve the necessary surface precision needed to make tooling for quality injection-molded lenses without spending a *lot* more money on the milling machine, the tooling, and the vibration/temperature/air pressure isolation system(s), as others have mentioned.

I looked into something similar to this for a carbon fiber mirror project, and I was astounded at the specs needed for mirror tooling. (Carbon fiber mirrors use metal molds.) We're talking flatness measured at no more than 147 nm! (So .15 microns, or 1/4 the wavelength of sodium light.) And the surface polish was an order of magnitude smoother than that. (Although that didn't need to be done on the mill - the mold could be polished after it was machined.) The molds were going to cost us tens of thousands of dollars to manufacture. In the end we went with off-the-shelf material rather than trying to make something custom. Injection mold tooling for lenses is going to require similar precision, both in terms of overall dimensions (for the quality/uniformity of the lens) and the surface finish (for proper release from the mold).

While it is still a lot of work and time spent to injection mold the lens, make precision housings and get AR coatings on the acrylic lenses...

When you add the cost of machining the tooling to all the above, you'll see that you won't save any money unless you are planning to make a whole bunch of lenses. Far better off to buy something that is already available. And if that isn't an option, requesting a custom run of a specific lens design will still likely be cheaper than trying to do it all yourself.

the machine makes mechanical part of a wrist watch and it's very impressive to the naked eye, but not enough to judge its use for optics, no matter how confident the owner is it will suit the job.

Exactly. The naked eye is a poor judge of surface quality for things like lenses or mirrors. Making good lenses is difficult. Pangolin spent a lot of time and money developing their safety scan lenses, and some of the horror stories from that development process were very similar to what we encountered with our failed mirror project. I wouldn't recommend it to anyone.

Adam

[mixedgas]

I built an interpolating interferometer at work that can see 10 nanometer steps.

With the case off and the vertical baseplate exposed, body heat causes a measurable bias from 6 meters away.

We can detect footsteps 20 meters away.

The laser system in use, a heterodye interferometer by HP, has a baseline cost of about 20,000$ for the stabilized laser and another 15,000$ for the motion controller, and is the same model of interferometer used controlling diamond turning machines for custom lenses.

Given the cost of just the measuring laser, are you really sure you want your own machine for this?

Contract out the work to a professional.

Steve