Jogging at a certain angle (lathe mode)?

I know some people will likely be rolling their eyes but a custom kinematic might be a solution for this.

cmorley wrote:

The two jobs are not that far apart. setting an MDI G1 X1 f1 and then aborting is very similar to a keyboard jog.

If you are okay with stopping completely between each new jog target position, then yes, they are very similar. But think about jogging with an MPG. You really need to be able to "pick up" the current motion and continue to a new target.

Aciera wrote:

I know some people will likely be rolling their eyes but a custom kinematic might be a solution for this.

Is this what is being done with the W axis in the 5 axis tilted drilling thread? While reading that thread I though it might be applicable to this topic, but I'm not exactly clear on what exactly is being done there, how the W axis works.

It's what many custom kinematics do. Including "5axiskins" it uses an angle value (usually A B and/or C axis) angles to transform a movement in X,Y and/or Z to the joints (ie motors).

branch available for test: dgarr/anglejog
sim config: configs/sim/axis/anglejog/anglejog.ini

Ref:
github.com/LinuxCNC/linuxcnc/commits/dgarr/anglejog
github.com/LinuxCNC/linuxcnc/commit/f186826988

see man page for usage info: $ man anglejog

dgarrett wrote:

branch available for test: dgarr/anglejog

Very cool!

I'm not able to test this at the moment, but looking at the code I'm wondering if it would be a good idea to decouple the input jog scale (anglejog.scale-in) from the axis & joint jog scales, and instead use a very fine-grained scale for the latter. E.g. fix those at 1 µm/count and do the appropriate scale conversion in anglejog. I'm imagining a rather horrible stair-step "line" to be the result of not doing this if using a coarser jog scale (say 1 mm/count). (Maybe I'm missing something though, from only reading the code.)

> I'm wondering if it would be a good idea to decouple the input jog scale

yes --
The original sim config addressed stairstepping by incrementing
MPG counts by 100 units so that a scale of .00001 would
correspond to .001 units. This type of s32 count-multiplication
is typically required when using an integer lowpass filter ($ man
ilowpass) on axis.N.counts and reduces stair-steps due to integer
counts handling at useful scales.

The updated branch includes a pin iscale-factor (default=10000),
and multiplies counts-in internally by the factor, divides the
current scale in-use by the factor, and outputs a corresponding
output pin current-scale-out for connection to axis.*.jog-scale.
The updated sim config increments counts-in by 1.


The branch update also provides for a jog-accel-fraction input pin
and scaled output pins cos-accel-fraction, sin-accel-fraction for
connection to axis.[NM].jog-accel-fraction pins. This facility is
useful in reducing shake for fine positioning (single count unit)
moves and often eliminates the need for filtering counts from an
MPG.

ref: github.com/LinuxCNC/linuxcnc/commit/d69b15cfe1ce40

arvidb wrote:

Aciera wrote:

I know some people will likely be rolling their eyes but a custom kinematic might be a solution for this.
Is this what is being done with the W axis in the 5 axis tilted drilling thread? While reading that thread I though it might be applicable to this topic, but I'm not exactly clear on what exactly is being done there, how the W axis works.
 

Just saw this:

It is the 5 axis bridgemill demo which appear to do exactly that. Meaning that the W axis is actually moving the tool along its rotational axis. Also mentioned on this thread: forum.linuxcnc.org/38-general-linuxcnc-q...ted-drilling?start=0

I'm far from being an expert but maybe this worth investigating the kinematics allowing this.

@dgarret thank you for your work, I will test it within the next days!

I tried the approach of executing an MDI command then aborting, the problem is that it kills the spindle too. Since I have physical joystick, when I return to the neutral position, I do the following: feed-inhibit = true, switch from MDI to manual, feed-inhibit=false.
This way I am not turning off the spindle with abort.