There's a post on Bob W's CNCCookbook site about him running into an issue he felt was related to uneven accelerations on the axes.  He noted poor contour following with uneven accelerations and a marked improvement at the visual level by setting the accelerations the same across the three axes.

Is this a confirmed  phenomena, do the axes really need to have very similar accelerations?

Seems like it would be more apparent in the X-Y axes, which the accelerations are easy to get the same.  For most of us it's the Z axis that is difficult to maintain the same accelerations.  Right now my servo based RF45 conversion can support acceleration of 50 on X and Y, and 20 on the Z even with gas springs to assist the head.  I'm planning on running the X and Y and around 35 for acceleration and was going to leave Z at 20.  It's a little harder to find the issue when looking to couple it with Z moves.

Anyway - insights are welcomed, I searched and found nothing on the subject.

I would say that the "SYSTEM" is only as good as the weakest link. If one axis is restrictive then one should work within that restriction and not push the weakest link into unreliable operation. If not satisfied with the outcome then change the system to satisfy your needs.

RICH

Sure, that is what you deal with if needed.

I guess my question is this: does the Mach 3 toolpath generation algorithme account for each axis having a unique velocity and acceleration profile?  Or does it assume that all accelerations are set equal?

The slowest axis will prevail. For example, if it was a small equal move how can you expect the weaker axis to accelerate to match the requested feedrate. So back to my original reply. Unless, you want to have the axis do work that it may or may not be able to do realiably.

RICH

OK. What you are saying is that Mach does not calculate realtime velocity during the acceleration ramp and look to keep each axis on path based on that acceleration/velocity.  Do you know this for a fact?  I'm not arguing, just looking to understand the math behind this  performance characteristic.  This isn't really about the weak system link, it's about the system architecture.  The architecture could support each axis having unique velocity and acceleration, but it sounds like it doesn't have that in the algorithym. 

Yes it does. I think what Rich is saying is that a multiple axis move can only accelerate as fast as the slowest axis is capable of.  If every axis accelerated at different rates, then it wouldn't be following the correct path.

No, there was a bug in Mach3 that would cause it to violate the acceleration limit on an axis under certain circumstances, which could cause position loss if one axis had lower acceleration than the others.  I was the one who found and reported it about two years and a half ago.  Brian has said it has since been fixed, but I'm not sure if anyone has verified that.  I certainly have not, since I no longer use Mach3.  But, certainly there is little or nothing to be gained by having either X or Y having higher acceleration than the other, since there will be few opportunities to take advantage of the higher acceleration on that one axis, even without the bug.

Regards,
Ray L.

Ok, I got it, thanks.

I found the section in the manual that states mach will create the motion based on the acceleration and velocity of each axis, supporting linear rates in all axes, and arc based in your choice of two. So it was a bug, that in theory is fixed.

I agree that X and Y should always be the same, and mine are. Its Z that wants lower acceleration. I plan on keeping with my  35 for X and Y, and 20 for Z.

Thanks for the infoz.

Ok, I got it, thanks.

I found the section in the manual that states mach will create the motion based on the acceleration and velocity of each axis, supporting linear rates in all axes, and arc based in your choice of two. So it was a bug, that in theory is fixed.

I agree that X and Y should always be the same, and mine are. Its Z that wants lower acceleration. I plan on keeping with my  35 for X and Y, and 20 for Z.

Thanks for the infoz.

That was exactly the setup I had, and when it transitioned from a 2-axis move in X/Y to a 3-axis move in X/Y/Z, it ignored the lower Z acceleration limit, which caused the Z axis servo to fault every time.

Regards,
Ray L.