Feedback Calibration vs Physical Alignment

[jdownie]

I have been puzzling over the issue of alignment/squaring of a gantry router (though originally I came up with the idea for a lathe), and had the following idea. Instead of precisely truing the physical frame, why not calibrate the XYZ system, using feedback, so it can simply compensate for any out-of-square conditions? This way one could concentrate on strength and rigidity/stability of the router, and allow for adjustment in software. Any thoughts?

John

[motomitch1]

I think some cnc lathes do work that way but the cnc software has to support it.

[ger21]

Originally Posted by jdownie Instead of precisely truing the physical frame, why not calibrate the XYZ system, using feedback, so it can simply compensate for any out-of-square conditions? This way one could concentrate on strength and rigidity/stability of the router, and allow for adjustment in software. Any thoughts? John

Not exactly what you mean by feedback, but here are two scenarios.

Mach3 (and Mach2) has the ability to home each side of a gantry to it's own switch, which will automatically correct for squareness. You need two motors to do this. You also need to make sure the gantry can twist or flex enough to allow this, if it's not square to start with.

I'm pretty sure Mach3 also has the ability to compensate for "out-of-squareness" through software using correction formulas. Screw mapping is also available and is somewhat related. These are some of the probably hundreds of new features, and are undocumented at the moment. But the ability is there.

[ViperTX]

No way.....just think about it.....what sort of profile would you need.....how far apart would each profile be.....just think about it.

[jdownie]

Seems to me that a fairly simple matrix transformation to the intitial toolpath coordinates would do the trick, assuming the errors were consistent through each plane. I will have to look in the Mach literature to see if there is any discussion of this.

[ger21]

Originally Posted by jdownie I will have to look in the Mach literature to see if there is any discussion of this.

I don't think it's in there yet. You can try searching the Yahoo group for info, which is where I saw the information, or contact Art directly.

[laserman]

How would you know where to comp the error in 3 dimensional space at any point? To correctly calibrate a machine tool the machine has to geometrically correct to start with. If you are talking about a CMM it is a different story, but even then it is better to have the machine tool flat, straight, and square first then calibrate each axis individually. This allows each axis to work with the other accurately.

[jdownie]

As has been written above, the machine does not have to be perfect to map it, so that corrections can be made. There is no question that having a machine that is perfectly flat, square, etc. is better than one that is not. The question revolves around the most cost-effective way to get a machine that will make the part.