ELS - Using linear position to drive rotary movement

[IanVicAus]

Is it rotary position driving rotary axis.
Or is it rotary position driving desired linear position that is then "followed" by the rotary axis?
The difference between them is one is dependent on the machines accuracy (rotary to rotary)
The other is not dependent on the machine it is dependent on the accuracy of the "electronics"
That is the rotary and linear position and then how accurately the linear position can be followed by the stepper motor.
I am sure it is more work but if it is reasonable I would suggest using the second method.
The reason is that it then negates the issue of worn lead screws, backlash etc as it is using the "actual" position of the cutting tool derived by the linear scale rather than an assumed one by the rotation of the lead screw. Worn everything being an issue for hobbyists.

For absolute accuracy you would want to linear scales on the front and back so that any crabbing of the saddle could be accounted for, and in fact there should be two lead screws..... I have heard that some CNC machines do it this way now and so don't actually rely on the accuracy of the machine - rather they use the accuracy of the measured position and then adjust the commands to gain the desired output. Though perhaps this is sounding rougher than it is, perhaps they do it with an "accurate" machine and then do this to get a SUPER precision outcome. Basically increase machine precision not through build structure but through position measurement and stepper control.

Trust that all makes sense - I tend to make english hard and that is even without it being read by someone where it is their 2nd (3rd/4th...) language.

[thatch]

Hi Ian,

I think the gist of what you're asking for is supported today, e.g. on a lathe to make X depend on Z to cut a taper. That said, you're bringing in some physics topics, so:

• If someday you want to make X depend on Z and Z depend on spindle, and some combination of X and Z depend on a virtual compound axis, I totally see that as possible, and based on pairs of axes (measure->drive) not just forwarding mathematical values through.
• Backlash compensation is a possibility. The trick here is making it easy enough to configure/measure/validate that people can actually use it.
• If you have position chatter in X, you're going to have a bad time. Especially if there's also backlash on Z.
• If you have even sensing chatter in X, you're going to have a bad time. Especially if Z movements end up being harmonic and that feeding back to X.

I'd suspect the limiting factor here is identifying the use case and coding it up. If you can give a more worked example or maybe a video explaining what you're asking for, I could know whether this is really a new feature or if we just need to document better what already works.