Back to Ganoksin | FAQ | Contact

Buying a milling machines

I work in Rhino, and have been emailing my files to a service bureau
for prototyping. Now I’m considering buying a milling machine, and
I’m not sure what will work best for me. I need a 4-axis machine.

Research has turned up a wide variety of machines and prices.

I can buy a Micro Mill with MeshCam software for around $3,000; a
unit from Rio Grande catalogue for about $10,000; a Roland mill for
$12,000 from Otto Frei; or a Roland unit specifically for jewelry
(I’m not sure of the difference) Re: Roland’s website, for $18,000.

If I go that high I might as well go another $10,000 for a $28,000
Solidscape R66 rapid prototype machine.

The more research I do the more confused I am. I need 4 axis and I
need the highest resolution possible.

The $3,000 Micro Mill is such a good deal that it seems too good to
be true. The programming of tool paths intimidates me. I’m proficient
at Rhino, but not really computer savvy. Just how complicated is it?

Any advice would be greatly appreciated.

You might also look at Artisan CNC and its new line of servo milling
machines. The basic difference between a 3K machine and the ones
costing 20-30K is the amount of vibration during cutting. A heavier,
more solidly built mill will run more quietly, faster, and more
accurately. Servos are better than steppers.

The initial cost should not be a major consideration. Accuracy,
repeatability, lifetime, and ease of use all should be considered.

The better machines will run thousands of hours and can be rebuilt
when they go out of spec. You may get decades of service from a well
built mill, and they seem to hold their resale value in the used
market. The technology has literally been used for hundreds of years
and differs little between these machines. Yes, there are different
flip jigs, tooling and other addons, and the ease of use of software
is a consideration, but a 4 axis mill in concept is a basic machine.

Google on TD Robotics, Rose. I’m holding out for a 3-D wax model
prototyping machine like the R66 with a robotic carver to take the
prototype software from the best wax models and turn it into jade and
argillite carvings. That’s gotta be worth 100K. If anyone comes
across a machine which can do this, please post.

You will also see an IEEE paper on robotic learning in the Google

What would Michaelangelo think of this?


I work in Rhino, and have been emailing my files to a service
bureau for prototyping. Now I'm considering buying a milling
machine, and I'm not sure what will work best for me. I need a
4-axis machine. Research has turned up a wide variety of machines
and prices. 

By Micro Mill I suspect you mean a Taig. No experience with their
package but I seem to recall people finding their motor driver box
being limited. I run a Taig with the toughest driver box I could I
could bolt together. It has been a few years and I am still happy. A
bit more than $3K but not much even if I count my time putting all
the bits together.

I use Rhino and MeshCam ( disclosure time… I am an unpaid
moderator on the MC forum other than a yearly bottle of fine single
malt. ) I get to kill off spamers and do my best to break the latest
beta, and cultivate friends who are even better at breaking. Simple
pleasures in my life. Bugs tend to be rare and are fixed fast. Good
program and support.

MeshCam is not a full 4 axis program, so far just indexed, although
I do trick it when required. Bad me :slight_smile:

No comments on Roland, I have never even seen one.

Demand Designs
Analog/Digital Modelling & Goldsmithing

Hi Rose,

I use Rhino to do the CAD work for the Knew Concepts saws, before we
convert them over to talk to our machining center. There’s a whole
lot of hand coding involved in that, unfortunately. (Our CAM package
spits out spaghetti code. Since we’re doing many of the same parts,
it’s worth my while to go through and optimize the code by hand. A
few seconds per part may not matter much if you’re only making two.
If you’re doing thousands, it matters a lot.)

I looked at the MicroMill, or what I think you mean by “MicroMill”.
Were you talking about the Taig micromill?

Depending on the size of what you’re doing, and your intent, the
Taig might not be enough of a machine. The websites I found showed
some pretty serious people, with serious measuring tools, setting
theirs up by hand in ways that made it look more like a kit of parts
than a turnkey system.

Are you thinking of machining wax for casting, or metal directly? If
metal, than you need a much more serious tool than you would for

I was demoing the KC saws in the Otto Frei booth at JCK in Vegas a
couple of months ago, and I’m friends with the guy they had demoing
the Roland system. If you’d like, contact me directly, and I can put
you in touch. (I didn’t pay much attention to it: it was on the other
side of the booth, and I was demoing our stuff.)

If you need really “high resolution” you’re talking a ballscrew
machine, which immediately puts you in the $10K+ range.

As far as writing toolpaths by hand, it’s not…horribly hard,
provided you know the basics of running a milling machine by hand to
start with. (I sometimes describe it as like trying to run a
Bridgeport with mittens on.) The kinky part is that while GCode may
be more-or-less universal, every controller has their own
implementations of it, so you’ll have to learn to speak with
whatever accent your controller requires, and the smaller machines
tend to have weird controllers.

Writing 3 axis code by hand isn’t all that hard. I’ve never tried 4
or 5 axis code, but for anything beyond simple "rotate and drill"
sorts of things, I think that’s where I’d really want to have a
CAD package talking to a CAM package, spitting out code for the mill,
before I started hand tweaking it.

The biggest issue with writing code by hand (or trusting the CAM
package not to foul it up) is to make sure you don’t crash the head
into anything at any great rate of speed. You can do shockingly
large amounts of damage in terrifyingly little time if you slip a
zero somewhere.

If you’re talking about direct machining of the metal, try hard to
find a machine with a tool changer (and magazine). Of course, that
immediately puts you into the “costs much more than a car” world,
but for production work, there simply isn’t another choice.

For whatever that’s worth.

Try the Revo 540B from Gemvision. If you want at least the axis
stated here and I think they are under 20k


I bought a Taig 2019CR-ER and completed the CNC conversion myself. I
use it to do wax and acrylic and really enjoy using it. I know that
the lead screw that they use will require some maintenance but for
the price its hard to beat. There is a tone of and
modification available on the net for that machine.

If I was to redo it again, I would go with a package and I would end
up with the same for less money.

Deep Groove has the best prices available

For the CAM operation you should look at RhinoCam, Its a Rhino
plugin that enables you to do everything in one interface and it also
supports 4 axis.

There is also a good forum that you can join that has a tone of
on CNC wax milling:



We teach CNC jewelry making on Taig machines. We find them to be
reliable and tough units. The next setp up we believe is the Levil

From there… Levil WL400… 5 axis. You can get into the CAD /CAM
machines from about $3700.00 to… well, the sky is the limit. :wink: I
have seen folks take junk machines and get beautiful pieces from
them… I spoke today with a gent that commented…"I have spent $45K
and never made a model…Thats sad, but I hear it often. It takes
the whole ball… machine, software and a user willing to spend the
time and effort! Just do your homework and ask a zillion + 6

I hope this can help. Dan.

The $3,000 Micro Mill is such a good deal that it seems too good
to be true. 

The Taig Micromill happens to be a genuinely good deal. It’s made in
Arizona by a family-run firm that used to be aerospace contractors
who designed these micro mills and a micro lathe in order to smooth
out their extreme business cycle. They became so successful at it
that they were able to devote themselves to producing them full-time.
They are rugged and heavy-duty for their size, and have a relatively
fast spindle (10,000 rpm) with a lot of torque,and are capable of
cutting wax or cutting steel. Their encoder-based closed loop motor
control system will stop motion if there’s a discrepancy between
commanded and actual postition. The only other mills in the same
price-range that compete with them are the Sherline mills (made in
California), which are also popular with jewelers. (Sherline
actually makes the rotary tables used on both brands.)

These mills both work right out of the box, but of course there are
always people who have to tweak things…

The programming of tool paths intimidates me. I'm proficient at
Rhino, but not really computer savvy. Just how complicated is it? 

Compared to 3D modeling, producing toolpaths from 3D models is
pretty easy. A particularly simple but powerful package that supports
the 4th axis is DeskProto. One nice thing about it is the free
working demo; once you’ve got a mill, you can use it for a month
before it times out, so you can try it out pretty thoroughly. Since
you’re already using Rhino, you might also try RhinoCAM. It plugs
right into the Rhino workspace, which makes on-the-fly modifications
a lot easier than having to switch programs.

Andrew Werby

Prototype -> Stone Product by Robot

How did Michaelangelo do his prototypes for stone carving? Rose
referred to her R66 protyping machine but does she not have sketches
before that? I have discussed with First Nations people here the
prospect of using clay (as well as sketches and wood carvings) as
prototypes for stone work. Anthropologist Peterson’s 1959 book
titled “Ancient Mexico” could as well be a book titled “Jade and
Clay in Ancient Mexico” since the historical record is presented so
much through this artistic medium. The prevalence of clay (a soft
stone) with harder stone (especially jade) figurines, amulets,
bas-reliefs, statues etc. leads me to think the Aztecs and various
other nations (who called themselves collectively the Mexica people
according to Peterson) used clay to prototype. They may have been
the most advanced of the stone age civilizations, depending on your
scales of advancement.

Brian refers to a “CAD package talking to a CAM package”. But the D
in design can come from clay and there are 3D (Dimensional) scanning
machines which can take the X-Y-Z co-ordinates of the clay objects.
I have cc’d an engineer who builds robots and he may be able to
comment on how accurate scanning robots are. If accurate enough the
scanning robot can generate the code for the final object to be
cut/carved in harder stone.

Others have posted on wax prototyping machines (also called 3D wax
"printers" for some reason) and a wax prototype could serve similar
purposes. But the next stage is the milling of a rough stone into a
finished object ranging in size from a large statue to a tiny
figurine. If it can all be done by grinding I suppose that makes it
easier for the milling robot.

Large robotic stone carvers are now a global resource and until they
get super-smart and go on strike we have an excellent resource of
low-cost labour who can in theory work 24/7. For example you can do
a web search on the TD Robotics system in Italy and the Artek QD
stone mill in New Delhi which claims it can carve stone as hard as

Scaling down to machines which can make tiny jewelry figurines and
bas-relief for jewelry box lids, I have a question for Orchid. Why
do we not see the manufacturers of these machines advertising by
proudly showing off all sorts of carved products? Is it because the
industry is still at an early stage?

It takes the whole ball... machine, software and a user willing to
spend the time and effort! Just do your homework and ask a zillion
+ 6 questions. 

That was beautifully said and important to remember. It really takes
commitment to get full use out of a mill. You can spend a
frustratingly large amount of time working with and adjusting your
mill. If it’s off just a little bit your models aren’t quite right.
In that case you need to stop, recalibrate and run it again. It’s
all part of the job, but something not everyone considers before
taking the plunge. It’s a whole new set of skills you need to

It’s a difficult question to answer, should I buy a mill or job out
model production to a service bureau? If you have the mill, you can
produce and check your models right away, then alter your design and
redo if you like, all in the same day. But you spent many thousands
of dollars and many hundreds of hours for that privilege. If you use
the service bureau, you save the thousands of dollars, but pay $50 -
$150 for each model and have to wait days to see the results. If
it’s not right it’s more money and more waiting. It’s a dilemma.


I spoke today with a gent that commented..."I have spent $45K and
never made a model....Thats sad, but I hear it often. It takes the
whole ball... machine, software and a user willing to spend the
time and effort! Just do your homework and ask a zillion + 6

I wouldn’t contract with a home renovator who could not point me to
satisfied customers so before buying I would ask the manufacturer to
do the same. Otherwise one could well throw lots of good money after
bad and spend double the 45K and still have nothing.

My objective is to find a machine which will carve stone figurines
etc. with maximum automation/autonomy.

Can anyone on Orchid point us to someone who has carved anything in
stone with one of these machines and can answer just a few general
and non-technical questions about the step by step process?

(1) What was the object to be carved and how did you calculate its
3D geometry, ie X-Y-Z co-ordinates?

(2) How did you enter those co-ordinates into the machine?

(3) Did the machine then do the carving autonomously or did you
guide it hand over hand?

(4) Was there any manual work other than placing the rough stone in
the machine?

(4) Can you show us the finished product?

PS - Robotics is moving along at an amazing speed. A technology to
watch is under the European FoF (Factories of the Future) initiative
from 2008. COMET seems to be the plug-and-produce analogy to
ubiquitous Windows. Plug it into any robot and it operates the
robotic system and replaces the built in operating system. With its
huge financial input for improvement, COMET could soon be regarded as
the best robotic OS in the world, rendering all others obsolete.

Hi Peter,

I’m not sure you’re ever going to find a rock cutting CNC without
purpose building one. There area couple of reasons: (A) rock dust is
incredibly abrasive. It’ll grind out the precision guides that CNC’s
depend on in very short order unless you build one specifically
armored against it. (B) underpaid grunt labor in developing countries
is vastly cheaper than an up-armored, custom CNC.

It’s simply cheaper to farm out most rock cutting jobs to Lower
Berzerkistan, so I doubt you’ll see what you’re looking for any time
soon. I could be wrong, but I don’t think I am.

You might want to check into the makers of CNC tool grinders, of the
type used for making milling cutters and other similar things. Those
are grinding machines, and they’re designed to deal with abrasive
dusts without grinding themselves apart. You might be able to
frankenstein one into a figurine cutter, but I guarantee it won’t be
cheap. (on the order of $100,000 plus) (Walters and Anca are
companies I’ve heard good things about, for whatever it’s worth.)


It really takes commitment to get full use out of a mill. 

As you noted too, Mark, :“just a little bit and your models aren’t
quite right”. In field work I guestimate that I can see a tiny
crystal about 1/10 mm but no smaller. The jewelry robot cuts in 3
orthogonal axes only and any others are derivations unless we are in
the hyperspace of matrix algebra but if I market n-dimensional moon
stones will I get rich or join Conrad Black and Bernie Madoff in
jail? As long as the X-Y-Z points on the surface of the finished
product are small enough the deviations from optimal value will not
be seen and +/-.005 inches as has been mentioned would seem to do
the job.

Jade Past

In “Ancient Mexico” (Peterson, 1959) we read that the Mexicans
carved in soft stones (clay) before hard stones like jade which they
valued above all others including gold. It seems likely they used
clay models before the more costly task of carving jade although
Peterson notes that the Olmecs who may have taught the original
skills “worked jade with such ease that it appears to have been clay
in their hands” (page 40).

Jade Future

What we need is therefore not a Scan -> CAD(esign) system but a Scan
-> CAM(illing) system because we have the models (designs) in
hand-crafted clay.

IOW we just put the clay model in one place and the uncut stone in
another, press a button for Scan -> Carve and come back in a few
hours to pick up the jewelry piece.

I am not promoting the following system any more than COMET as
robotic OS (Operating System) from Europe but Orchidists may wish to
click on

to see where we are going.

PS - If COMET were installed in a typical home computer, let’s say
post-2012 (Mexican Calendar end time, page 112) would it operate
robotic peripherals as well as keyboard-screen, like 3D scanners, 3D
printers and 3D carvers … making all of us potentially custom
home-jewellers? Would COMET make Windows, Mac and Linux OS obsolete?

If you use the service bureau, you save the thousands of dollars,
but pay $50 - $150 for each model and have to wait days to see the
results. If it's not right it's more money and more waiting. It's a

My problem with the service bureaus is not the money - I know it
would be more cost effective to stay with that setup. My problem is
the time it takes, and their willingness to either mill or grow in
extreme high resolution. Sometimes they say they will but they won’t.
Sometimes they get a big order from a bigger customer and put mine on
the back burner. It can take weeks, and then if the prototype isn’t
perfect I have to start all over again. I need to be able to control
the time element.

I’m good at Rhino so the input is not a problem. I’m down to two
choices, and all the responses here on Orchid have been very helpful
in my decision. I only plan to mill in wax, but I do not want to
coddle and babysit a mill, so I will either order the Roland Mill
(which Otto Frei has a good package deal on right now) or bite the
financial bullet and go with Solidscape.

Thank you all so much for your input. I may toss my coin today and
go for one or the other.

I'm not sure you're ever going to find a rock cutting CNC without
purpose building one. 

We need more discussion on this because other coming in
says automated machines capable of cutting hard stone into sculptings
are already on the market.

Dan for example, showed a picture of a jade amulet which had been
worked into its final shape by a robotic carver and I have cc’d
Artek in India for clarification since their web site says the QD
stone mill is an industrial robot which will carve even granite. (I
have no vested interest in any manufacturer and am not promoting any
particular one). TD Robotics in Italy seems to be saying the same.
Others may wish to look up these web sites and give an opinion too.

Hi Peter,

Your posts are a bit hard to read, but what I took you to mean was
something to cut figurines, which I’d put in the 1-3 inch tall
range. Both of the machines you mention are purpose built for
monumental sculpture. (Working in feet, not inches.) I doubt they’d
have the resolution to handle jewelry scale objects.

I couldn’t find any prices on the sites, but they’re still clearly
in the $50K+ range.

To be honest, I hadn’t found the two machines you mention. Glad
they’re out there. Who knows? I may be wrong. It’s been known to


something to cut figurines, which I'd put in the 1-3 inch tall

You raise a very good Q about size of product (and size of machine

Dan posted well, “ask a zillion questions”. Technical experts should
realize that just because most of us are “not really computer savvy”
(as Rose noted) that does not mean we are stupid. I can write C code
for geometic points until the cows come home but that technical
ability does not help me much even if the machine is programmed in C
because I just want to operate it with SW installed. We want the best
of the best in a marketplace which is worldwide. Caveat emptor
applies every step of the way.

A really well designed machine should be self-instructional out of
the box.

What can we buy for 100K?

Both of the machines you mention are purpose built for monumental
sculpture. (Working in feet, not inches.) 

Livick posted a video on a machine which carved a human head in
marble. The model seemed to be human-sized. All steps starting with
the scan took 8 1/2 hours and the result looked great except for a
"band" near the bottom. Now if we have a MOBILE machine we can take
it onto the quarry site and carve objects like that in situ.
Outcroppings become carvings. We could also do relief carving on
walls as the Lapisytem from TD Robotics in Italy advertises that it
can do reliefs.

I’d like to start with a 100 ton jade boulder shown on our web site
for proof of concept. This is a world-famous steelhead fishing site.
A 100 ton “Fisherman’s Throne” could be quite a tourist draw. Perhaps
diamond edged circular saw, chain saw and wire saw could do the rough
cuts followed by a TD machine to do the relief work. If it works for
a Fisherman’s Throne, why not a “Jade Palace” carved into the rock

The black jade piece shown by Dan seemed to be an amulet a few
inches across. What size is it Dan?

I doubt they'd have the resolution to handle jewelry scale objects. 

Correct me if I am wrong but .005 inches is the standard for all of
these machines. If so, that is below the threhold for human
perception of +/- carving errors.

I couldn't find any prices on the sites, but they're still clearly
in the $50K+ range. 

100 K is fine as long as we don’t get a box of nuts and bolts for
that and then understand that to set it up and operate it will cost
another 100K or _______? What is the old saying about throwing good
money after bad?

As long as the X-Y-Z points on the surface of the finished product
are small enough the deviations from optimal value will not be seen
and +/-.005 inches as has been mentioned would seem to do the job.

;-)… .005" ??? First, just my opinion and some
know what thats worth :wink: (not) you are making this way more
complicated than it really is… Most of us working in jewelry use
a cutting tool that is.003" to.005" tip diameter. Using the .005" as
an example, we step that over a maximum of 40% per linear pass…
SO… that gives us.002"… or as 99.9% of us do it… in MM…
05mm. Thus… .005"…or .127MM is a football field. A machine that
has .005" =.127mm as a minimum movement or repeatability is worthless
in the jewelry world. You can get that close using your teeth and
even closer with 80 grit sandpaper!

The CNC jewelry making process is reasonably simple. It has a
learning curve no less…and possibly a little more complicated in
some ways than hand carving a wax. The thing that makes CNC less
complicated is that there are a set of paremeters that you can and
will want to learn to live in. When carving at the bench by hand…
these parameters are so broad that they get nearly lost. You may use
25 different hand tools and no one knows how many different carving
techniques. Carving using CNC, you learn a set of parameters and to
some degree, pretty much live there. Hand carving… I seldom set
down at the bench and do the same given job twice in the identical

Sorry for the rambling… but don’t overthink the job. Perhaps it
would be good for you to find a local jeweler doing CNC and simply
ask some direct questions…

Just my 2 cents.

Good Luck. Dan
DeArmond Tool

This is an impressive video from

It looks like human carvers are obsolescent at least for larger
carved figures and reliefs. If we can move a machine like this to our
quarry site, it will serve the purpose. Brian raised an important
question as to whether there is a “machine discontinuity” so that
small carvings for figurines etc. require another HW/SW system. I
have no way of knowing. Can others on Orchid comment?

QD Robotics also has some videos at and maybe QD or Torart can comment
on this issue of using big machines for small figurines a few cm in
size. My only comment is that we do not hire big people to carve big
statues and small people to carve figurines.

Also for historical interest (since a 6 year old project is almost
ancient history in robotics) have a look at the carving of the Sphinx
of Naxos in France and the story of “The Sphinx and the Robot”.