#linuxcnc Logs

Sep 25 2022

#linuxcnc Calendar

01:59 AM Deejay: moin
04:48 AM JT-Cave: morning
05:26 AM Tom_L: morning
07:40 AM thunder-gunner[m: Morning
07:50 AM boozeman[m]: hi
09:35 AM andypugh_ is now known as andypugh
10:27 AM * boozeman[m] uploaded an image: (70KiB) < https://libera.ems.host/_matrix/media/r0/download/jauriarts.org/qRYLGSAvVxwaFwznfXYLdIkF/unknown.png >
10:27 AM boozeman[m]: Some sketching with Kikad. I will replace series connected endstops with PNP ones. on my POC-foam cutter This board sits back of the Y and V-axis.
10:29 AM * boozeman[m] uploaded an image: (819KiB) < https://libera.ems.host/_matrix/media/r0/download/jauriarts.org/lPuCqLQwMsLtlwPPcllIwzbX/20220925_182649.jpg >
10:29 AM boozeman[m]: XY-axis. The end stops are reused computer case power buttons 😎
10:31 AM boozeman[m]: * XY-axis. The original end stops are reused computer case power buttons 😎
11:04 AM Tom_L: JT-Cave, any luck?
11:10 AM JT-Shop2: nothing I've tried saved the comments
11:11 AM JT-Shop2: short of writing my own config parser I don't know what else to try
11:12 AM Tom_L: spend the effort on something else then
11:13 AM JT-Shop2: current project is to stop my arm from bleeding :)
11:15 AM roycroft: dip it in ca glue
11:22 AM roguish[m]: cauterize it
12:04 PM -!- #linuxcnc mode set to +v by ChanServ
12:04 PM Tom_itx: JT-Cave, https://thepipelineweb.wordpress.com/2017/05/16/python-save-comments-from-configparser/
12:06 PM JT-Cave: now that looks interesting
12:11 PM JT-Cave: it almost worked... but put the comments in different places
12:12 PM Tom_itx: that's an indexing thing you can figure out i bet
12:13 PM Tom_itx: it didn't look like you needed all those search criteria either
12:14 PM JT-Cave: it is not saving blank lines I think... royally screwed up the cat house wall with the door in it so time for a nap
12:14 PM JT-Cave: didn't take into account the connecting wall location... oh well shit happens
12:14 PM Tom_itx: woops
12:16 PM Tom_itx: i don't understand python speak for that search string
12:16 PM JT-Cave: yeah that's re speak and it's a bit cornfusing
12:18 PM * JT-Cave wanders up for a nap
01:50 PM * JT-Shop2 moves the door opening over a few inches...
01:50 PM JT-Shop2: Tom_L, that code almost worked but has some flaws that need fixing
02:28 PM Tom_L: but it's fixable?
02:31 PM JT-Shop2: yep
02:32 PM Tom_L: yay
02:33 PM JT-Shop2: you really don't need re to find ; or # at the start of a line
02:34 PM JT-Shop2: this issue now is the configparser puts only one blank line after each section so if there are more than one blank line that code get cornfused
02:35 PM JT-Shop2: I weighed all my options and the best one is to move the door over 3"
03:20 PM CaptHindsight[m]: https://forum.linuxcnc.org/21-axis/40659-axis-2-touchscreen-version-of-old-friend
04:19 PM XXCoder: fun frank howarth video today
04:49 PM andypugh: XXCoder: I started tim-stamping my scripts before feeding them to YouTube.
04:50 PM XXCoder: nice
05:54 PM JT-Shop2: Tom_L, https://pastebin.com/k3NeDTT8
05:57 PM roycroft: i had to sand a bunch of small parts today, and decided it was finally time to finish up my vacuum clamping system, which only took a few minutes to complete
05:58 PM roycroft: of course, the instant i started using i wondered why i hadn't done it years and years ago
05:58 PM roycroft: it reduced my sanding time by at least 30%
05:58 PM XXCoder: lol
05:58 PM XXCoder: nice
05:59 PM Tom_L: JT-Shop2, figure it out?
05:59 PM Tom_L: so you can test for multiple blank lines now
06:00 PM JT-Shop2: figured out step 1 get the index of comments and more than one blank line
06:00 PM JT-Shop2: yep the configparser will strip them out and that would confuse the comment inserter thingy
06:00 PM Tom_L: right
06:00 PM Tom_L: it shouldn't strip them but it does aparently
06:01 PM JT-Shop2: so they must be replaced to keep the index correct... easier than writing a config parser I think
06:02 PM Tom_L: heh i'd be lost either way
06:03 PM JT-Shop2: so find all the things that the configparser removes and replace after updating the ini file
06:03 PM JT-Shop2: makes me feel a bit like tron
06:04 PM Tom_L: heh
06:07 PM JT-Shop2: try running it on a copy of an ini file with comments
06:07 PM Tom_L: ini2?
06:08 PM JT-Shop2: that snippet of python code
06:08 PM JT-Shop2: <JT-Shop2> Tom_L, https://pastebin.com/k3NeDTT8
06:09 PM * JT-Shop2 listens to Iron Man then heads inside
06:09 PM Tom_L: how do i pass a file to it?
06:09 PM JT-Shop2: name the file test.ini and put in the same directory as the python file
06:10 PM Tom_L: yeah i just noticed that
06:10 PM Tom_L: booting
06:10 PM JT-Shop2: when testing code I try to KISS
06:10 PM -!- #linuxcnc mode set to +v by ChanServ
06:12 PM JT-Shop2: looks like python 3.7 retains the dictionary order but mesact only requires 3.6 so I'll use the OrderDict
06:13 PM JT-Shop2: OrderedDict
06:14 PM Tom_dev: 45 # comment
06:14 PM Tom_dev: 47
06:14 PM Tom_dev: hah i'd already put on in it
06:14 PM Tom_dev: that was the 7i96 config i've been using all along
06:14 PM JT-Shop2: cool
06:14 PM Tom_dev: and the actual comment is on line 46
06:15 PM Tom_dev: being called to be domestic...
06:15 PM JT-Shop2: lines start at 0
06:15 PM Tom_dev: ahh ok
06:15 PM JT-Shop2: me too
06:15 PM Tom_dev: later..
06:15 PM JT-Shop2: k
06:15 PM Tom_L: that much works at least
06:35 PM mrec: Tom_L: do you have any experience with the trajectory planners in linuxcnc?
06:35 PM acer is now known as _unreal_
06:36 PM mrec: my studies are advancing but it will still take some time until I'm firm with it
06:37 PM mrec: jogging just uses the simple trajectory planner for a single axis, however I'd like to have jogging implemented using 2 axis simultaneously
06:38 PM mrec: andypugh: in regards of our discussion last time, it is not that easy.
06:38 PM mrec: or maybe I did not exactly elaborate my target accurately enough
06:39 PM mrec: I want to simulate the compound movement (on my lathe) with linuxcnc, just by using X/Z
06:40 PM andypugh: If you simply send jog counts into the jog-counts pins, _and_ do it slowly enough that the axie can keep up, then the axes should follow the target positions.
06:42 PM andypugh: Just try what I suggested and see what happens, ie connect the same jog-counts source to the jog-counts of two separate axes, and see if you get straight motion. If that works then all that is needed is a component to send jog-counts in non-equal proportions.
06:46 PM mrec: well if I do it slow enough of course it will work to a certain degree
06:47 PM mrec: the path as shown in my screenshots yesterday happened when I used to jog at irregular speed (regular speed and slow speed should be okay)
06:58 PM andypugh: Have you tried the HAL-onlu method, at all?
07:03 PM mrec: hal-only? I'm basically doing the same already I'd say?
07:03 PM mrec: right now it works as you said - slow moves are okay
07:03 PM mrec: the point is if it's not directly mathematially synced then it's not synced
07:03 PM mrec: resulting in jerky moves of course
07:04 PM mrec: it runs through simple_tp and that also is what it is
07:04 PM mrec: simple_tp - This component is a single axis simple trajectory planner, same as used for jogging in linuxcnc.
07:52 PM andypugh: The HAL-only thing I mean is just connecting the same jog-counts source to two jog-counts pins.
07:58 PM mrec: I think you are still missing the point. I am more or less doing that inside the sourcecode using jog-counts with X/Z at the same time
07:59 PM mrec: the different acceleration settings of X/Z are the point, I guess you have more aggressive settings on your system mine is a tradeoff between space and motor power
08:00 PM mrec: I will revert all the modifications once I'm done and implement it in a better way, currently that's all just experimental here.
08:00 PM mrec: I'm not a mechanical engineer, so I'm still learning about the basics
08:01 PM mrec: motion planning... interesting topic
11:15 PM ZincBoy[CAON][m]: I was looking at doing this on my lathe. The idea was to create a kinematics module that has two virtual axis, one for the angle of the compound and one for the compound motion. The compound motion would be on the Y axis to keep the better motion planner but you wouldn't really use this in gcode and more for manual machining so it doesn't really matter.
11:32 PM mrec: exactly
11:32 PM mrec: I think this should be some basic feature of LinuxCNC for lathes
11:33 PM mrec: but I guess less people use linuxcnc for lathes (compared with 3axis standard CNC mills)
11:34 PM mrec: I have 2 mini lathes the compound of the Emco 5 is worse than the compound of the chinese lathe
11:35 PM mrec: yet I don't know about the keyboard input path for jogging (I did not look into that)
11:36 PM mrec: but for the pendant it seems to me there should be a pin exposed by motion/control.c for switching between simple_tp and another synchronized motion
11:39 PM ZincBoy[CAON][m]: With the kinematic component you don't need to switch motion planners. You just don't use the extra axis in NC mode.
11:51 PM mrec: I think it should not be so much work to use a different motion planner basically it's all there already (again I'm still studying, so I could be wrong too)
11:55 PM ZincBoy[CAON][m]: I think you are vastly underestimating the complexity of swapping out the motion planner. Pretty much everything ties into it. I have looked at it in some detail and the kinematic module approach is the way to go.
11:56 PM mrec: G1 vs jogging is doing exactly what you write - changing the motion planner
11:56 PM mrec: G1/G0 vs jogging
11:57 PM mrec: jogging is using simple_tp as you might know
11:57 PM mrec: and there are 3 branches in the code Free move, teleop and combined
11:58 PM mrec: Free and teleop are using simple_tp