What's new
What's new

Desktop mill for optical parts

Optical

Plastic
Joined
Jan 8, 2021
I'm looking for a quite specific machine and was hoping some of you might have insights. The purpose of the machine is to produce optical surfaces with freeform shapes (think of the water surface of a pond, smooth and wavy). The surface finish should be close to mirror, but a small amount of polishing is acceptable as a post-process. Productivity is not the issue, only the precision/surface finish is important.

Typical tool: ball nose of diameter ~2mm (0.08in)
Axis: 3
Axis resolution: 0.0001mm / 0.1µm (0.000004in)
Axis travel: ~ 100x100x100mm (4in)
Material: soft metals
Typical feed rate: 200mm/min (8in/min)
Typical surface size: 20x20mm (0.8in)
Typical stopover: 0.02 (0.0008in)
Spindle: up to 20k rpm, very low runout
Budget: ~30k US$
Machine weight: ~100kg (220lb)

No need for tool changer or flood coolant. Material remove rate is not important.

I know well the mono-crystaline diamond SPDT machines like Precitech, Moore Nanotechnology, etc... but those are overkill for our needs (and too big and too expensive).

I have found some interesting machines like
Minitech GS
MDA Precision V9-TC8
NS CNC Mira X7

Any experience with some of those?
Do you know of other manufacturers? Ideally well connected to Europe.
I'm based in Switzerland.
 
Hi Optical:
The two contenders that immediately spring to mind are in Switzerland, they are Kern and Hermle.
There are others who make high end claims too, but these are the two that most machinists are aware of.

These are breathtakingly expensive machines but when you're trying for sub micron resolution you cannot get what you need any other way.
This is way way beyond Haas territory and you need to be aware of that from the start.
The three machines you referenced are toys...thy are probably OK for a cosmetically acceptable part but I cannot bring myself to believe they will be even in the ballpark for the kinds of work you are trying to do.

There is a little mill that I saw in a Youtube video that would be about perfect for your needs too.
Sadly I can't remember the brand...I believe it was a Mori, but it featured very small travels and very high resolution.
It was in a garage somewhere in the USA, and I saw it as party of a shop tour.
The guy who owned it had set up a small tool and die business in his garage and I believe the shop tour was conducted by John Saunders, but I just did a quick search and couldn't find it again.
He was using it to hard mill mold components and carbide parts.

Super nice machine if you can live with its tiny size.
It was a real industrial machine; probably weights 4000 lb but was the size of a phone booth.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining

I found it; the guy is Adam Demuth and his shop is in Ohio.
Here's a link:
CNC Shop Tour | Adam Started His Own CNC Machining Business! - YouTube.

The machine is a Mori Seiki NVD 1500 DCG.

MC
 
Axis resolution at 1/10 micron and price at under 30K.
I should note to the OP that axis resolution is not machine capability when making parts. Lots of lies, false beliefs and dreams that do not work out here.
Resolution is just gearing and/or encoder counts. Actual accuracy way different. One can gear down to 1/100 of a micron resolution easily if you give up speed but very hard to make a part there.
Even if given accuracy numbers done by the accepted machine tool standards rember that these are "unloaded numbers" so your mileage will vary.
Ideally you want this number to 1/10 of what you want to hold when making parts. I know this gets into crazy land once the tolerances get small but it is so very real.

Those in micromaching wish this price tag possible in a true sub-micron milling machine for small or tiny parts.
You need to add a extra zero in the price tag and then lots of options.
The good side is that being able to do this type stuff pays well as most shops and conventional machining centers can't do this type work.
Being able to inspect it final or to make size adjustments in a run can get expensive also.
Bob
 
Thanks for the replies. I should have specified that we already use a machine for making prototypes that comes quite close in terms of precision. It a Roland MDX 540 (costs about 30k) that would definitely be considered as a "toy" machine. We have optimized every parameters for our process to be able to achieve a good surface finish. However, the spindle has too many vibrations and the resolution is 0.001mm.

I strongly believe that you can get very good precision from small desktop machines if you take the time to generate perfect tool paths.
 
Thanks for the replies. I should have specified that we already use a machine for making prototypes that comes quite close in terms of precision. It a Roland MDX 540 (costs about 30k) that would definitely be considered as a "toy" machine. We have optimized every parameters for our process to be able to achieve a good surface finish. However, the spindle has too many vibrations and the resolution is 0.001mm.

I strongly believe that you can get very good precision from small desktop machines if you take the time to generate perfect tool paths.

Bob already said as much but I'll try to rephrase.

You have conflicting specifications and some overoptimistic expectations. 0.1μm resolution is meaningless on platforms that are physically capable of accuracy barely better than two orders of magnitude greater than that.

The fact that you are almost satisfied with the MDX tells me that you absolutely do not need 0.1μm resolution. The specification of that machine states +/- 100μm positional accuracy, 50μm positional repeatability.

Using the Mira X7 that you linked as an example, since it's the only one to state it, it claims 0.3μm resolution, with 5μm positional repeatability. Assuming that was even remotely true, which it almost certainly isn't, you can see that the sub-micron resolution is completely pointless.
 
I'm going to pile on that your budget and expectations from a benchtop machine are unrealistic. Since you're in Switzerland you should take a look at the Willemin 301S2 301S2 - Willemin-Macodel It's 5-6x your budget but the most affordable machine that approaches your desired spec.
 
Hi again Optical:
You wrote:
"I strongly believe that you can get very good precision from small desktop machines if you take the time to generate perfect tool paths."

Of course, all depends on how you define "very good precision"
The influence of spindle bounce and bearing irregularities and motion control limitations and thermal effects and etc etc, are all not trivial influences.
Since you said "optical" I assumed you need to get the kind of precision where these influences become show stoppers...is that not true?

Resolution of 0.001mm is hard to believe from such a lightly built machine as the Roland you referenced.
For those not familiar with metric units, that's 40 millionths...precision lapping territory, not conventional milling territory.
That may be the smallest commanded increment in the motion controller, but I find it impossible to believe your system error is any better than 0.01 mm and I'd be super impressed if it was that good.
I wouldn't be surprised to find it's more like 0.02 mm or 0.03 mm when you mill something and measure it

I don't say this to be an obnoxious prick, but I do know something of working to these levels of precision, and the machine tool takes on an outsize role as you get below the 0.01mm range.
There's a reason why a Hermle is built so heavily and with such attention to detail...everything begins to matter in a whole new way when you have to guarantee the tool tip is exactly where you think it is within ten microns at high feedrates after it's been running for an hour or a day.

So let me ask you this...have you ever been to Kern or to Hermle and asked for them to demo machine a part for you?
I bet you'd be shocked if you took a part off the Roland, set it alongside a part made on one of these machines and compared the two.
I think it would be highly instructive, and you have the advantage of living right next door, so I encourage you to take advantage of that.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining
 
Thank you again for taking the time to reply. Let's assume a resolution of 0.001mm is enough (we can angle the parts to lift this requirement). Our application is very sensitive to resolution but less to the global positional accuracy as we will machine the whole surface in one operation very slowly. (We actually have the MDX 540S model with higher accuracy, +/-20μm)

We tested the Willemin 701S, great resolution but made for industrial production which is overkill for a lab environment.
And Kern machines also. A bit better than what we achieve with the Roland.
Those machines are expensive for other reasons than their resolution.
On the Roland, we clearly can measure the tool moving 1μm, if you take very light cuts, very slowly on soft metals you will keep this precision.
 
Hi again Optical:
If the Roland is almost good enough have you considered having it rebuilt to make it the best it can be, rather than investing in a whole new machine?
I don't know of anybody who would take on a project like this, but there must be SOMEONE out there who could do it.

You might be able to upgrade the spindle and the ballscrews, map and compensate the motion errors, stick it in a climate controlled room and do your magic.

Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
Vancouver Wire EDM -- Wire EDM Machining
 
Thank you again for taking the time to reply. Let's assume a resolution of 0.001mm is enough (we can angle the parts to lift this requirement). Our application is very sensitive to resolution but less to the global positional accuracy as we will machine the whole surface in one operation very slowly. (We actually have the MDX 540S model with higher accuracy, +/-20μm)

We tested the Willemin 701S, great resolution but made for industrial production which is overkill for a lab environment.
And Kern machines also. A bit better than what we achieve with the Roland.
Those machines are expensive for other reasons than their resolution.
On the Roland, we clearly can measure the tool moving 1μm, if you take very light cuts, very slowly on soft metals you will keep this precision.

Couple points. If you are doing this now, you aren't getting better IMO, until you approach the million dollar mark

I happen to have a 1um Mitutoyo drop indicator (just mentioned this morning in another thread :D). It's practically unusable in a 'normal' shop environment. You said lab however, I assume full climate control (temp +/-1/2F, humidy control, air flow, so no hot/cold spots on your equipment) How are you measuring to 1um? Is your stuff calibrated by a reputable company? Are you soaking your parts if you are machining and inspecting in 2 different places?
 
@Mike1974
I just measured the machine servo resolution, not the part accuracy which I don't care so much about. Our optical application requires very good local precision as we care about the angle of the surface, not the global precision.
Climate control is not important, if the tool length increases by a few microns from the start of the machining until the end, it doesn't matter as it doesn't impacts the part locally. Imaging that we only care about the surface finish of the part but not its precision (both are linked but not one to one).
 
Thanks for the replies. I should have specified that we already use a machine for making prototypes that comes quite close in terms of precision. It a Roland MDX 540 (costs about 30k) that would definitely be considered as a "toy" machine. We have optimized every parameters for our process to be able to achieve a good surface finish. However, the spindle has too many vibrations and the resolution is 0.001mm.

I strongly believe that you can get very good precision from small desktop machines if you take the time to generate perfect tool paths.

If that hobby-grade machine is really almost good enough, you might try a Haas CM-1 in metric mode. It still won't hold a candle to a Hermle or such, but it will do .001mm resolution (not necessarily accuracy), and will cost a fraction of the price. You also get a 30,000RPM ISO 20 spindle, or optional 50,000RPM, which while not nearly as rigid as a 30 taper, will be much more rigid than that of the hobby machine. If I were you I'd contact the local dealer and have them cut a test part to see if it meets your needs.
 
If that hobby-grade machine is really almost good enough, you might try a Haas CM-1 in metric mode. It still won't hold a candle to a Hermle or such, but it will do .001mm resolution (not necessarily accuracy), and will cost a fraction of the price. You also get a 30,000RPM ISO 20 spindle, or optional 50,000RPM, which while not nearly as rigid as a 30 taper, will be much more rigid than that of the hobby machine. If I were you I'd contact the local dealer and have them cut a test part to see if it meets your needs.

Thanks, making a test part could be interesting, but seeing that the spindle is 3.7 kW (5hp) makes me think this is a machine meant for rough work and steel machining. There will likely be a lot of vibration and big runout. A Nakanishi spindle has a power of 300W but is optimized for precision which I think is more adapted for what I'm looking for.
 
Thanks, making a test part could be interesting, but seeing that the spindle is 3.7 kW (5hp) makes me think this is a machine meant for rough work and steel machining. There will likely be a lot of vibration and big runout. A Nakanishi spindle has a power of 300W but is optimized for precision which I think is more adapted for what I'm looking for.

I've seen people who have one claim that they can't even hear the spindle while it's running. A more massive spindle column makes for improved rigidity and accuracy. Just as a point of reference, my 2015 Haas VF-3SS, with a 15,000RPM CAT-40 spindle with 30HP (claimed anyway), has less than .0001" runout. Have the test part cut, I'm really curious to see if it would meet your needs. I'm thinking of getting one for small medical device parts.
 
Couple points. If you are doing this now, you aren't getting better IMO, until you approach the million dollar mark

I happen to have a 1um Mitutoyo drop indicator (just mentioned this morning in another thread :D). It's practically unusable in a 'normal' shop environment. You said lab however, I assume full climate control (temp +/-1/2F, humidy control, air flow, so no hot/cold spots on your equipment) How are you measuring to 1um? Is your stuff calibrated by a reputable company? Are you soaking your parts if you are machining and inspecting in 2 different places?

I thought I had replied but my message seems to have disappeared, sorry.
I only measured the machine servo resolution, not any kind feature on the part. My application requires good angular direction of the surface normals which means I need a good local precision but don't require a great global precision. If the tool length expands of a few microns from the start of the machining and the end, it doesn't matter as it won't impact the local precision. Climate control is therefore not important.
If you will, I only care about the surface finish and not about the precision.
 
If that hobby-grade machine is really almost good enough, you might try a Haas CM-1 in metric mode.
I wuz gonna say, we've had a couple people get Office Mills for tiny work, definitely not a Hermle but better than a Roland. They were happy with the machine.

But he seems to have a bunch of preconceived notions so ... what the heck.

(P.S., had a south bend with resolution of fifty millionths, too. Whoopee, that made it real accurate :D)
 
Thanks, making a test part could be interesting, but seeing that the spindle is 3.7 kW (5hp) makes me think this is a machine meant for rough work and steel machining. There will likely be a lot of vibration and big runout. A Nakanishi spindle has a power of 300W but is optimized for precision which I think is more adapted for what I'm looking for.

I don't know where you've been getting some of your notions from, but 3.7kW is nothing as far as machining. Our standard mill here is 30kW our smallest is 19kW. 3.7kW would only be able to take very small light cuts in steel and moderate cuts in aluminum. The vibration and runout is a matter of money. The higher end machines will have better bearings and be better balanced for running at the RPM's needed for mold work.

To try and put it into an optical term; you are wanting a pair of Swarovski quality binoculars for Bushnell prices, while arguing that you can get them as clear as the Swarovski if you just fiddle with them for a bit.
 








 
Back
Top