Brother TC-225 / TC-229 adventure!

More
23 Aug 2017 09:08 - 23 Aug 2017 09:10 #97910 by andypugh
The belong to "milltask" and have been there since version 2.6:
linuxcnc.org/docs/2.6/html/man/man1/milltask.1.html

Though there are more in 2.7:
linuxcnc.org/docs/2.7/html/man/man1/milltask.1.html

"milltask" is (indirectly) loaded by a reference in the INI:

[TASK]
TASK = milltask

I think that there was originally an intention to have different versions of "task" for different applications.
Note that milltask is user-space. This means that it might well not update reliably if you change values from G-code.
(And it isn't clear when the system updates its internal opinions from the values).
You might well still have to use the "W extends Z" approach.
Last edit: 23 Aug 2017 09:10 by andypugh.

Please Log in or Create an account to join the conversation.

More
23 Aug 2017 14:57 #97932 by ihavenofish

Please Log in or Create an account to join the conversation.

More
23 Aug 2017 15:20 #97936 by ihavenofish
this is all just giving me a head ache :P
on the w axis, I'm also not sure how that works. I cant make another axis that uses the same pins, it wont let me.

so what do you do instead, just have it reference the outputs and inputs of z?

wont z trip its f error if w starts moving?

Please Log in or Create an account to join the conversation.

More
23 Aug 2017 20:02 #97946 by dgarrett

This means that it might well not update reliably if you change values from G-code.


The values of ini.* hal pins are only read only at the start of a gcode program

I've tried to add some clarification in the man page:

linuxcnc.org/docs/2.7/html/man/man1/milltask.1.html

Ref:
github.com/LinuxCNC/linuxcnc/commit/51a8...cbd0d16fcaafb9fc37c5

Please Log in or Create an account to join the conversation.

More
23 Aug 2017 20:11 #97947 by ihavenofish
does the toolchange ngc remapped from m6 not count as the start of a gcode program then?

thanks

Please Log in or Create an account to join the conversation.

More
24 Aug 2017 05:33 #97957 by ihavenofish
finally beating the z servo into submission. its still got a little resonance, but it dissipates fast. I can run at all speeds with tracking within a thou and only a little "whine" during downward deceleration.

I made a video of adaptive clearing ill post later.

I noticed some oddness in it.
during the "helix" outward moves, we get super smooth 400ipm motion. on the last perimeter pass, the motion is super choppy. g64 is on, P0.001" Q0, but behaves the same with p and q undefined.

its also clunky around the reversals which have a 1-2mm radius.

is there any settings to deal with behavior here? look ahead, or the like?

other than having slightly shorter segments (about 1/2-2/3 the size of the smooth running segments) I cant see why it may decide to behave like this. and even then, id expect the thing to just slow down like my router used to.

Please Log in or Create an account to join the conversation.

More
24 Aug 2017 06:42 #97958 by ihavenofish

Please Log in or Create an account to join the conversation.

More
25 Aug 2017 16:55 #98005 by ihavenofish
so, working a little more, I discovered 2 things while trying to get the resonance out of my servo tuning.

1st, I have backlash. not huge, but, enough to be annoying and ill have to address that by repacking or replacing screws.

theres in the .0003-.0006" range in the X. up to 0.001" in the Y, and about .0004" in the Z.

so, question: would this impact tuning ability? I suspect other than a brief moment of zero load, it should not matter, as the encoders are on the motor.

second thing I noticed, is my analogue voltages are high. for example, I get 6600rpm on the spindle at 10v, instead of 6000. I could imagine this causing the axes to overshoot then get into a loop trying to compensate. is there a way to calibrate this? I see it referenced in the manual, but whats in the manual is NOT in the pncconf wizard.

thanks.

Please Log in or Create an account to join the conversation.

More
25 Aug 2017 17:06 #98007 by Todd Zuercher
A little backlash shouldn't matter much on a velocity mode system, when the encoders are on the motors.

Normal tuning should easily compensate for the analog scale being off. (just uses a little less P and FF1)

Please Log in or Create an account to join the conversation.

More
25 Aug 2017 17:17 #98008 by ihavenofish
ok, thanks.

so why cant I get this smooth? what am I doing wrong here? why is there no autotuning! :P

all that jerking in that last video is all tuning. its doing aggressive reversals at each stop or slow down in the code. they may only be 0.0005", but it amounts to pretty bad vibration when its doing it ever 1-2mm. in sections of the code where g64 is able to keep the speed smooth, everythings happy.

I'm at a threshold now where no setting seems to improve it beyond where it is now. just make it slightly different.

Please Log in or Create an account to join the conversation.

Moderators: cncbasher
Time to create page: 0.171 seconds
Powered by Kunena Forum