Axis compensation

[Justin]

I came here to find a solution to the same issue and noticed the question wasn't really understood.

On some Milling machines the X and Y axis's can be out of square to each other rather than being 90 degrees.

This can be due to wear or errors during production and it makes it very difficult to for example produce a square shape, it'll wind up being trapezoidal.

Similar problems can also happen with a Z axis on a milling machine. You can sometimes shim the column into square but it's difficult and affects rigidity, other options can be even more complicated especially when your talking about a column and head weight of 200kgs or more.

To be able to input an offset value over a specified length of travel would be an excellent solution.

Z axis compensation would have to be for two directions to compensate for squareness between the Z and X, and Z and Y.

Example for every 300mm of travel along the X axis add or subtract a certain number of steps evenly distributed during that specified length of travel to the Y axis.

[cncdrive]

No, we have not planned to implement this.
And we now have a few different projects going on and even if those are finished we want to implement S curve motion profiling instead.
I think that most machines have precise enough driving mechanism to not need unlinear axis compensation.
Even a cheap C7 ballscrew can do better than 0.005mm per 300mm nonlinearity.

But I understand that it would be the cheapest to use old rusty cheap screws with low tolerances and still achive perfect precision using the software compensations,
however in reality this is usually only a dream, because if the screw is really that unlinear then it is 100% sure that it also has backlash which is never good.
However UCCNC has backlash compensation, but backlash always involves issues.
So, currently we concentrating on other things including an S curve motion planner which would give advantage to more users than nonlinearity compensation.
Maybe one time we will try to implement this too, but not in the near future for sure, sorry.

[Justin]

CNCDRIVE

You are not understanding the issue.

It has nothing to do with the inaccuracies of some screws.

It relates to the WAYS, for example, on a MILLING machine X and Y not being 90 degree's to each other.

[Battwell]

fix the mechanical issue. then its right forever!

[Justin]

Your comment isn't particularly helpful Battwell.

[Karnage]

I came here to find a solution to the same issue and noticed the question wasn't really understood.

On some Milling machines the X and Y axis's can be out of square to each other rather than being 90 degrees.

This can be due to wear or errors during production and it makes it very difficult to for example produce a square shape, it'll wind up being trapezoidal.

Similar problems can also happen with a Z axis on a milling machine. You can sometimes shim the column into square but it's difficult and affects rigidity, other options can be even more complicated especially when your talking about a column and head weight of 200kgs or more.

To be able to input an offset value over a specified length of travel would be an excellent solution.

Z axis compensation would have to be for two directions to compensate for squareness between the Z and X, and Z and Y.

Example for every 300mm of travel along the X axis add or subtract a certain number of steps evenly distributed during that specified length of travel to the Y axis.

Have you considered fixing the fundamental issue with your machine? I'd love to see how this crazy implementation would work on a manual machine, does the machinist do tha calcs on the fly while spinning the handles?

[naijin]

Hi there, just wondering... if the machine is still works in progress, would it be easier to try and square it up before actually trying to feed formulae etc.? and if the when the machine is all up and running it may be possible to actually use the Z axis
and engrave or draw a line the full length of the X axis and use that line to square your job to.. who knows it may even work.
Cheers
Nick

[thegreatwaldo]

Hi there, just wondering... if the machine is still works in progress, would it be easier to try and square it up before actually trying to feed formulae etc.? and if the when the machine is all up and running it may be possible to actually use the Z axis
and engrave or draw a line the full length of the X axis and use that line to square your job to.. who knows it may even work.
Cheers
Nick

Hi Nick
I think I can see what justin is aiming at. A sort of "backlash" kind of compensation in the software that would compensate for a wonky (badly made machine ) in the software. So if the the rails were out of parallel say by 5mm (exageration) at one end the software would adjust accordingly. I don't how practical this would be as if there was a bow in the rails I can't see it working. Obviously fixing the machine is the correct way to go as cnc is all about accuracy parallel rails should be parralel. Anyone who uses a metal working lathe knows what sloppy loose slides are like.
Cheers
Andrew

[cncdrive]

It is also a question how to measure out non-parallelness to feed in a formula properly?
And if you have a method to measure the non-parallelness then why can't you adjust it on the mechanics to get rid of the problem?

[Justin]

It is also a question how to measure out non-parallelness to feed in a formula properly?
And if you have a method to measure the non-parallelness then why can't you adjust it on the mechanics to get rid of the problem?

The issue has nothing to do with Parallelness or backlash, as stated it relates to the squareness of, for example X and Y axis to each other and is aimed mores so for users of Milling Machines not routers.

I've included an exaggerated image showing the issue viewed from above down the Z Axis

Milling machines (most, apart from some cnc's) don't use linear rails that can be simply unbolted and realigned, they usually use ground cast iron ways that are integrated into the machine.

Out of squareness can happen for many reasons, it's a lot more difficult to deal with on a milling machine because fixing the physical problem means having the ways reground.

The amount of out of squareness is easily measured and can be expressed as an amount of apparent travel on a stationary axis during a measured amount of actual travel on the other axis that should be at 90 degree's but is out of square.

If that amount were 1mm over 100mm then for every 100mm of travel on X, 1mm's worth of steps needs to be added or subtracted distributed evenly along the Y axis of travel.