Newbie- What kind of accuracy and precision can be expected?

[refuser]

Hi,

I have been lurking these forums for a while now as I am planning to build my own CNC mill.

My question is, what kind of accuracy and precision can be expected from a DIY CNC?

When I started to think about building, I was going for MDF, now I'm thinking more in the lines of aluminum t-slots, would the material choice affect accuracy a lot or would it just make a difference in load handling capabilities.

I am considering proper linear bearings and belt drive.

Einar

[Harpye]

MDF is working - so the tolerances could be changing 0,xmm just by dependence on the humidity of the atmosphere! In addition positionning tolerances of spindels e.g. there could be a tolerances range up to mm fail positioning . I also would invest in ballthreads ( pre loaded and selected ballset) Tolerance class as you could effort or as you need - you 'd have to make a tolerances calc!


Hansjeorg

[Richster]

From my own experience (which is a fairly low-budget affair in the big scheme of things) the bulk of my inaccuracies originate from backlash in the threaded rod/leadnut assemblies. If you're going the belt drive route, that would certainly be a step in the right direction. I probably have an error about 0.1mm to 0.3mm (that's a guess - not measured!) on each axis because of leadnut backlash. I guess it comes down to what you want to manufacture with the machine, and how important the accuracy is to you. That doesn't sound much, but the theoretical accuracy (resolution) based on a single motor step with 8mm threaded rod is 0.0006mm! Other than that, machine flex is another cause, so the more rigid you can make the structure, as short axes as meet your needs etc., the better.

[Harpye]

It is difficult to give an estimation without knowledge of the system ...


very coarse...


rolled Trapezoid (7e) spindle/ nut fails up to 0,x mm on 300 mm length dead angle on change of direction, there is no preload applyable!

grinded Trapezoid spindle with matched nut -tol down to 0.05mm on 300mm length reduced dead angle ( by semicountered 2nd nut) but no preload possible

rolled balltread spindle tol 0.0x mm on 300mm - depending on quality class matched ballset possible - no preload, but dead angle compensation in noload situation.

grinded balltread tol 0.0xmm on 300mm - depending on quality matched ball sets possible and preload possible on higher quality classes. recommended: preload countered double nut, to higher the bearable load ... depends on application if necessary.

THere is also an option to make a length measurement on your spindle and then teach the system to compensate the spindle failure by the much higher theoretical step angle resolution - but this works just fine with a on machine measurement and this is much effort .... and over a certain level of accuracy you should think about a climate chamber for your machine ( just a joke but thats necessary for high precision manufacturing )

Linear guides and rails:

dependent on the quality of the spindle the lin tec should keep at least the same requirements -> preload possible on some ball and rail systems

the structural design should also be stiff enough to make the spindles and rails worth to buy - accuracy means solid and precise in built!

all quality related columns of the machine should be equal to get a good and costefficient result!!!


Hj

Short story about MDF

I have a solid Aluminum base on my machine, but once used a MDF lost plate in which I milled a pocket to receive an Aluminum part, I wanted to work on. I applied very little coolant on the aluminum with a spray gun during work. --- the MDF sucked it like a sponge and lifted the alu part about 0.18mm and the pocket was for the bucket!! Since then I use polycarbonate for a lost plate!

and now apply this experience on an MDF machine in a humid cellar!!! Ok - that is my experience but - do you want to have a "Sponge Pop"

[eSilviu]

With the one from linked thread, I get 0.0xmm backlash. I don't even bother to compensate-it when milling pcb's.
I used cheap M8 acme threads, and nut-spring-nut to reduce backlash (without this, backlash was 0.12-0.2mm).
Cutting speed is 250mm/min, and traveling speed is 400mm/min (can go up to 800mm/min, but may loose steps).

And it was very (seriously) cheap: a few old printers + 1 sheet of MDF for base + Al frame + other cheep components.

http://www.cnczone.com/forums/showth...176#post335176


PS: it was my second attempt. from the first one I've learned many many things (it was far away from perfect)

[refuser]

Thanks for your answers!

I understand that it can be hard to estimate, but with the figures you guys gave at least I have a clue.

Thanks!

Einar

[Harpye]

Silviu: that is what I call low load application - low speed low feed and not much material to move! I have a similar contruct on my hot wire cutter works find but also low to no load!

refuser - could you tell us some more spc of you planned built???

-Material to be machinned
-Operating space/ dimensions
-Accuracy target
-necessary speed of macining
-has the machine to mill 3D, 2D or just to shape

[refuser]

"refuser - could you tell us some more spc of you planned built???"

I'm afraid I can't, the more I think about it the more elaborate of a machine I seem to need even if I don't know what I will machine. But I will try anyway.

"-Material to be machinned"
Mdf, plastic, foam, PCBs and occasional aluminum.

"-Operating space/ dimensions"
600x600 mm XY and maybe 400 mm Z, the big z is because i want to be able to push a 500x6000x350 mm foam block through. (No, it's not a typo, it really says 6000 mm. I want to be able to mill kayak plugs out of foam, but since they don't need extreme precision i could just feed them through at 600 mm lengts at a time, up to some sort of stop.)

"-Accuracy target"
I might be naive but i would really like to be able to keep it within 0,05 mm over the entire work piece.

"-necessary speed of macining"
Might be naive about this one to but if i could do foam at 600 IPM I would be happy.

"-has the machine to mill 3D, 2D or just to shape"
3D

Einar

[Harpye]

In Germany we call such a phenomenon "eierlegendeWollmilchsau"-
I guess you would say - "all in wonder "!

Material as to be machinned - MDF, foam, and "semi solid" things shouldn't be a problem in 2D or 2,5D for 3D you should have a death angle compensated system!

PCB is in need of high acurracy in Z - also in x and y but mainly in z - x and y depend on the pitch you want to mill death angle compensation recommended ( at least the nut spring nut principle for it is just engarving / low load).

Aluminum is the benchmark - as the mill "likes to hook" and the surface goes worse, if there is any dynamic movement between the material and the mill without the path to mill. This would need a very stiff structure and a preloaded (ball thread) system to get good results ( my opinion)!

Dimensions x and y are good - 400mm z should eat much resources on construction weight and money... the forces caused as reaction on load - e.g. when milling Aluminum create a torque by a at least 400 to 500mm long lever and to compensate this and still have precision left, is really hard ! For foam there is rather no reactio and as precisionrequirements are much lower it could work!

But for 400 height to mill you need a mill wit about a > 400mm length - very expensive tool and you would finally need an axis length of 800mm!!

For your accuracy target I would say - depends on the costs but is really hard to keep!

I guess what you want is a really monster - great axis moves end up in high masses - rigid structures that you want to move rather fast ....

... and at least to get lucky you would need real mills ( not just 6mm shaft mills) on high speeds to get a 6m 3 d job done in time ( such a machine is surely no hobbyist work - and should cost some x00000 $ when built professional for commercial use!)

This is finally out of my horizon to estimate about the posibilities of such a machine. But I would recomment - if you have a buddy with a CAD CAM Engine including machining time calculation plugin give it a trial try to find a possible speed that can be machined in foam and afterwards parametrize a kayak model with this data - and get the estimated mill time - .... for a 3D job in this size should take lets say " just some time!"

If precision is 2nd prio on the kayak, why not a 6m span hot wire cutter! Much faster than a mill and much cheaper! You could place several blocks in line - and that you want to cut stays and the others are removed until they are cut afterwards you would have to sand a bit but I guess this could be the more efficient solution ( however - you should find this out for yourself ( I don't sell that stuff - I'm just a hobbyist !!)

here is some information ( I don't know about this machines for I built my own)

http://www.step-four.at/hp2/index.php?action=268

Here are some pics to watch : they actually do a giant scale sailingplane but a kayak should be similar..... afterwards it is sanded, finished and could be used as a solveable positive plug or for a glassfibre/ resine form!!

and here is my last hot wire cut from epp ( not solveable)



Hj