Open 5 axis CNC

[hive8]

I wanted to propose something here, that maybe all the members can help out, a open 5 axis compact CNC with the following criteria...

work cube: 200mm to 250mm
high speed spindle: 1000 to 20000RPM
Tool diameters up to 12mm
350 to 475 IPM Rapids
Cutting accuracy 0.015mm to 0.025mm
Position accuracy 0.005mm to 0.01mm
Software: LinuxCNC
Linear Guides
Maybe even Scales for position feed back...
Cost to build: Less then $10k (most machines in this class go for 20k to 40k)
Should be able to cut metal...
Option auto tool changer...

After a prototype put something on kick starter... with different levels
From just the plans and BOM to all the way to a final machine...

I am looking more for a machine that cuts fast, with less deep cuts, more like RMR (Rapid Material Removal). Not something that can take 1inch cuts...

Please let me know if there would be a interest... I do have some funds to get materials and get started...

Get a domain name with hosting... setup a wiki and things (I can do)
Decide on a design (Maybe the heads rotates plus x, y, z movement, or with a trunnion table design or a gantry style design).

Let me know what you guys think and who would be interested... Of course most of the parts for the machine would need to be CNC machined to begin with...

[691175002]

Sorry, but my opinion is that this is completely impractical. The price range puts it in an area where not that many people can build it and your requirements are impossible to meet.

You are severely underestimating the difficulty of maintaining accuracy in rotary stages.
tan( 20 arc seconds) * 100mm = 0.01mm
Which means your entire positional accuracy budget is blown by a single rotary table with 20 arc second error (and the standard single worm tables have error in the 1 arc minute + range).

There are no imports, and there is no way to DIY that kind of precision. Your entire budget (probably several times your budget) is gone in just the rotary tables.

I highly doubt 5 axis machines that can hold 0.01mm position accuracy go for as little as 40K, I'd expect more like 100-200K. I suppose some jewellery machines make such a claim, but they probably don't even know how accuracy is measured.

[hive8]

OK understand let me redo this then how about 45 arc-sec accuracy (0.0125 degrees, 0.000218 radians) with a harmonic drive, i am not talking a worm drive here. In X, Y and Z linear motion it is possible to get that. Would that be possible.

[SCzEngrgGroup]

OK understand let me redo this then how about 45 arc-sec accuracy (0.0125 degrees, 0.000218 radians) In X, Y and Z linear motion it is possible to get that. Would that be possible.

Do some homework on the effects of thermal expansion on machine accuracy, then consider the effects of rigidity (or lack thereof), then look at the accuracy of the ball-screws you could afford to put in such a machine, then price some real spindles capable of doing 20K RPM, then go back and seriously re-think your goals. The best commercial machines can barely get to the tolerances you've outlined. There is *zero* chance of achieving it in a home-built machine. You're off by about an order of magnitude, if not more.

Regards,
Ray L.

[691175002]

Also worth mentioning is that this kind of project isn't something that can be captured in a set of plans and a bill of materials.

Your instructions are going to look something like "Grind this surface flat to 0.002mm and then ensure it is perpendicular to this surface with an error of at most 0.003mm".

There is a point where it is just impossible to work at the accuracies required without a ton of money to begin with.

[hive8]

Ok let me put it different then, what would it take to build something like this for under 10k in a open project... http://www.youtube.com/watch?v=R1DCXe9t3UE&feature=channel&list=UL]PMC Mini 5-axis CNC machining (

And why do companies like this www.mdaprecision.com and Roland offer machines with these advertised specs?

[SCzEngrgGroup]

And why do companies like this CNC Mills, Benchtop Mill, 5 Axis CNC Mills and Lathes - MDA Precision and Roland offer machines with these advertised specs?

The specs are possible but NOT in a $10K home-built machine. You don't get that kind of accuracy without very high-quality components, VERY precise machining, assembly and inspection, a temperature controlled operating environment, a LOT of calibration using very specialized calibration equipment, and *real* closed-loop control on all axes, all of which are VERY expensive, and require a great deal of skill and expertise to pull off.

Regards,
Ray L.

[691175002]

And why do companies like this CNC Mills, Benchtop Mill, 5 Axis CNC Mills and Lathes - MDA Precision and Roland offer machines with these advertised specs?

If the machines accuracy is not reported according to some standard, they can pretty much make numbers up.

If you look at a machine from MDA ( TN5 5-Axis CNC Mill Systems - 5 AXIS CNC System - Milling Machines ) they state the 4th and 5th axis accuracy are both 1 arc minute which means that you already have 0.08mm error in the worst case.

They later state positioning accuracy of 0.01mm which is clearly impossible if they are talking about overall accuracy.

It is likely positional accuracy in this case simply refers to the linear positioning of a single axis which ignores error that will be found in the perpendicularity of axes, flex in the machine and other alignment problems.

If you go through and add everything up:
X 0.01
Y 0.01
Z 0.01
B 0.04
C 0.04
Table Flatness 0.01
Total: 0.12mm = 0.005"

This is a fairly realistic number for a machine of that size and price.

If you assume the total positional accuracy in a 3 axis machine is 0.01mm (which would be weird given that your table flatness is 0.01mm) you end up with 0.09mm which is still a realistic number for a 5 axis machine.

Of course, the machine is unlikely to be the absolute worst case, but these numbers also ignore flex, perpendicularity and other alignments.


Accuracy is a very complicated topic and trying to express it with a single number is very difficult. A lot of factors influence how much error you actually see in a given part. For example, if you were just trying to bore two holes an inch apart I'm sure the machine could do 0.01mm or better. If you were cutting some crazy impeller that filled the entire cutting area things would be different.

[hive8]

I understand where you coming from... Maybe i stated the requirements a little to close what i would like to have... Of course these are all worst case scenarios... which will not always come into play.

On other example of something like this is these machine, this machine must be crazy accurate
WL400

Let me say this then, lets design build one that has a position accuracy of lets say 0.05mm (about 0.0015") which is possible... If someone tells me that not possible at home then i will prove them otherwise... I even did it at my practicum at this company Zumtobel. I had to make parts that where within +0.02mm / 0.00mm on a manual lathe. or a cube of 25mm with 0.01mm surface flatness and parallelism with a metal file (no machine used).

I do make telescope mirrors and some of the most accurate mirrors are made at home in garages from very talented people... not in labs, climate controlled, these would be optimal conditions which we not always have but we still get it done.

At the end of the day it comes down how well the operator knows the machine, his craft to get the maximum out of it.

For me it is like this... i just want to generate a starting point to have something like this developed for future use.

[wizard]

I understand where you coming from... Maybe i stated the requirements a little to close what i would like to have... Of course these are all worst case scenarios... which will not always come into play.

For your basic X,Y,Z movements you should be able to get better than 0.05 mm over a mitied distance ignoring thermal grow and other issues. That means a good lead screw and most likely lead screw mapping.

Almost allof the really accurate machines I've worked on over the years had hand scrapped ways. So doing it by hand isn't impossible. The problem comes in the metrology how and with what do you measure squareness and other geometric elements of the machine.

On other example of something like this is these machine, this machine must be crazy accurate
WL400

Let me say this then, lets design build one that has a position accuracy of lets say 0.05mm (about 0.0015") which is possible... If someone tells me that not possible at home then i will prove them otherwise...

You aren't specifying over what distance but I have to agree it isn't impossible. It does take dedication and frankly not many have that dedication. There is a separate issue of positioning accuracy which is another ball of wax as there are so many choices these days.

I even did it at my practicum at this company Zumtobel. I had to make parts that where within +0.02mm / 0.00mm on a manual lathe. or a cube of 25mm with 0.01mm surface flatness and parallelism with a metal file (no machine used).

There was time when machine shops use to hand scrape all of their flats and squares. I knew a guy that use to do this in the 40's. The problem these days is that this is not the norm, instead it is expected that a machine will do the work.

I do make telescope mirrors and some of the most accurate mirrors are made at home in garages from very talented people... not in labs, climate controlled, these would be optimal conditions which we not always have but we still get it done.

Yes but do remember the work that goes into those mirrors! Further you are working on one parameter for the most part, a radius.

At the end of the day it comes down how well the operator knows the machine, his craft to get the maximum out of it.

Absolutely true. The problem is if you go the hand built route you have a lot of work ahead of you, some can't contemplate why you would do such.

For me it is like this... i just want to generate a starting point to have something like this developed for future use.

Nothing wrong with that. The net is filled with plans from the early part of the last century. The big issue in my mind is getting together a fool prof set of instructions to go along with the drawings. Also I suspect your spindle speeds are going to be hard to hit at the overall price range you set. Thus I would reccomend a bolt on spindle capability.

[hive8]

Thank you wizard for confirming the points i had made, as you see in my first post i would like to build it with a work cube of 200mm to 250mm so the travels are not crazy far hence the screws don't have to be super long, i am talking a desktop machine here not something huge...

Regarding spindle i am talking to a spindle maker that can help out on this part and would share his designs...

Also i am not planning on making a machine that can be build in a couple of days. For me and for others its also the joy of building a machine, to see it all come together and at the end use it to make parts.

[LongRat]

Bear in mind that you will have to be clever about how you design your rotary axes to get a decent sized part on the machine. With linear axis travels as small as you are planning, the largest part that you will be able to machine using all 5 axes will be very much smaller than the 200 cube.