Guest
Show Posts
91
VB and the development of wizards / Re: G68 Rotation point?
« on: April 09, 2011, 01:09:10 PM »
Hi Terry,
Here is some more info to add to your notes:
I've verified these as of mach 3.43.22 - there are more problems I've discovered than I've listed here. If you want the full G68 related list I can send it to you.
The model is that G31 is done in the current WC system. The WC is the entire universe from the G31 viewpoint, it just calls for movement in the WC and gets touch points reported in the same WC. At the level of the G31, it does not know if the WC is rotated or not wrt to the machine axes - the WC is just whatever it is - and the G31 should neither care or know how the WCs relate to the MCs.
G31 is allowed with G68 active (as it should be). The problems arise because mach does not implement handling of a rotated WC in a way that is mathematically consistent. Unfortunately, Mach's implementation only works correctly if a WC is not rotated (wrt to the machine axes).
Problem 1: Mach does not apply the rotation transformation when reporting the probe touch point - so when G68 is active, the touch points are not reported wrt to the active WC. This is what drove me to look for the rotation point, as with it I could correct for mach's reporting.
...but the problems are not all G31 specific - if you want to get surprised try this:
G53 X0 Y0
G54 X0.07071 Y0.7071
G68 Y0 Y0 R45
you should get a WC of X=1, Y=0 after the rotation. If you look at the multimode DROs, that's what you'll see for values (when they are in WC display mode). Now go look at the WC DROs - they will be wrong.
Problem 2: The WC DRos don't comprehend G68
Now for kicks go press the "X Zero" button.... You just told mach to zero the WC X axis - that should place the X0 point at the current position in the WC (which is rotated wrt to the MC). You expect to see the X DRO (and only the X DRO) go to 0 and the offsets change - That's not what happens.
Problem 3: Functions that cause offsets to change ignore G68.
Basically I've concluded that any use of G68 should cause a blinking "Warning, there be dragons here...." message ;-)
Oh, well, not much I can do except write it all up, report it to Brian and hope that fixes make it into a release sometime. In the mean time, I'm going to make my stuff check for G68 and either refuse to do some things or reset the system to G69 as a side effect - those seem to be the only available choices for now.
Dave
Here is some more info to add to your notes:
I've verified these as of mach 3.43.22 - there are more problems I've discovered than I've listed here. If you want the full G68 related list I can send it to you.
The model is that G31 is done in the current WC system. The WC is the entire universe from the G31 viewpoint, it just calls for movement in the WC and gets touch points reported in the same WC. At the level of the G31, it does not know if the WC is rotated or not wrt to the machine axes - the WC is just whatever it is - and the G31 should neither care or know how the WCs relate to the MCs.
G31 is allowed with G68 active (as it should be). The problems arise because mach does not implement handling of a rotated WC in a way that is mathematically consistent. Unfortunately, Mach's implementation only works correctly if a WC is not rotated (wrt to the machine axes).
Problem 1: Mach does not apply the rotation transformation when reporting the probe touch point - so when G68 is active, the touch points are not reported wrt to the active WC. This is what drove me to look for the rotation point, as with it I could correct for mach's reporting.
...but the problems are not all G31 specific - if you want to get surprised try this:
G53 X0 Y0
G54 X0.07071 Y0.7071
G68 Y0 Y0 R45
you should get a WC of X=1, Y=0 after the rotation. If you look at the multimode DROs, that's what you'll see for values (when they are in WC display mode). Now go look at the WC DROs - they will be wrong.
Problem 2: The WC DRos don't comprehend G68
Now for kicks go press the "X Zero" button.... You just told mach to zero the WC X axis - that should place the X0 point at the current position in the WC (which is rotated wrt to the MC). You expect to see the X DRO (and only the X DRO) go to 0 and the offsets change - That's not what happens.
Problem 3: Functions that cause offsets to change ignore G68.
Basically I've concluded that any use of G68 should cause a blinking "Warning, there be dragons here...." message ;-)
Oh, well, not much I can do except write it all up, report it to Brian and hope that fixes make it into a release sometime. In the mean time, I'm going to make my stuff check for G68 and either refuse to do some things or reset the system to G69 as a side effect - those seem to be the only available choices for now.
Dave
