#linuxcnc-devel Logs

Jun 05 2018

#linuxcnc-devel Calendar

01:26 AM rmu: skunkworks: i need to reorganize and cleanup first, it is becoming confusing.
07:39 AM skunkworks: rmu: check your post.. feedback from rob
08:03 AM rmu: thanks, interesting
08:07 AM rmu: it should be possible to continue the bezier (or bernstein) interpolation at any point with continuous velocity and acceleration, i just didn't want to clutter things too much
08:08 AM rmu: code will be updated in the evening, that is in about 6-8 hours, i fixed some corner cases, acceleration in 3d chips looks much nicer
08:45 AM skunkworks: rmu: cool!
09:40 AM skunkworks: rmu: what made you look at this? What is your backgroud?
10:02 AM rmu: skunkworks: longterm linuxcnc user, just recently got a "real" machine
10:06 AM skunkworks: cool - I am mainly a linuxcnc user/tester
10:06 AM rmu: tinkering around with the trajectory planner is just an excuse to avoid working on the toolchanger ;)
10:07 AM skunkworks: what 'real machine'?
10:07 AM rmu: a biesse rover woodworking thing
10:07 AM skunkworks: ah - I think I remember that. A router would certainly benifit from jerk limiting
10:07 AM rmu: with really scary speeds, x-axis makes 80m/min
10:08 AM skunkworks: I do have programing background - mostly vb but when I look at some of the linuxcnc guts my head spins
10:09 AM skunkworks: We have a matsurra that could go faster if it didn't 'thump
10:09 AM skunkworks: ' the ground when moving :)
10:11 AM rmu: in a former life i did all kinds of software stuff, mostly C++, in industrial control systems and other strange stuff
10:12 AM skunkworks: so this is right down your ally :)
10:12 AM skunkworks: I got into linuxcnc as they were just tranistioning from emc1 - emc2. good time to get in
10:13 AM rmu: it's fun ;) to work on something "real" and not just implement some stupid interface to some borked whatever from the 70ies
10:14 AM rmu: i'm using it since around 2005, it was already emc2 back then IIRC. didn't like the dos-based control system that came with a "käsefräse", aluminium extrusion router thingi with stepper
10:15 AM rmu: in fact, the control software didn't really work, it had some problems like always moving all 3 axes when interpolating a circle/arc
10:15 AM skunkworks: heh - I wrote my own in quick basic (I think) and then found turbocnc - then linuxcnc
10:15 AM rmu: found emc and never looked back
10:15 AM skunkworks: yep
10:17 AM skunkworks: I don't understand why people use anything else :)
10:30 AM skunkworks: Dad actually came across emc somehow in the mid to late 90's I think. But at the time it was way over my head. First install was the bdi emc1 - then quickly started to help test the emc2 version. (so so much better)
12:56 PM rmu: i have to figure out what's happening in tpHandleArcBlend and when and why it generates trajectory components with duplicate ids...
12:56 PM rmu: some improvements are on github
12:56 PM rmu: not yet cleaned
01:46 PM skunkworks: I am still seeing violations
02:11 PM rmu: yes, but they should be "nicer" and fewer
02:11 PM rmu: oscillation should be gone
02:13 PM rmu: something i don't quite understand yet is goind on with path blending in some cases, it seems target velocity is changed after a trajectory component is "activated", and i don't handle that case yet. so at the end of the segment there can be a huge discrepancy that leads to acceleration violation
02:14 PM rmu: and additionally my intended acceleration limit is off by a constant factor
02:14 PM rmu: probably
02:15 PM rmu: enough for today ;)
02:15 PM Tom_L: you must be on the other side of the pond
02:15 PM skunkworks: rmu: cool!
02:16 PM rmu: Tom_L: I'm on the early side ;)
02:16 PM skunkworks: rmu: still getting osolations...
02:17 PM rmu: skunkworks: still with the chips 3d?
02:18 PM skunkworks: yes - running it at feedrate factor of 1 also (not 1000 ;) )
02:18 PM rmu: i think i have to try with the imperial simulation perhaps there is something different
02:19 PM skunkworks: http://electronicsam.com/images/KandT/testing/rmu/better.png
02:20 PM skunkworks: if I run chips faster - it violates harder
02:32 PM rmu: just FYI this is what the first acceleration violation here looks like https://unfoo.net/~robert/Screenshot_20180605_205120.png
02:49 PM andypugh: rmu: Fancy coming to the LinuxCNC-fest in Stuttgart in Juli?
02:51 PM andypugh: https://doodle.com/poll/rrzknkwiaukwgrar
02:57 PM seb_kuzminsky: andypugh: that reminds me, what do you think about announcing/advertising the linuxcnc-fest on linuxcnc.org? we could add a "post" about it
02:57 PM rmu: andypugh: I'm thinking about it
02:57 PM andypugh: seb_kuzminsky: Good idea.
02:58 PM andypugh: (Though it is probably worth mentioning that it is pretty casual, no badges or speakers or clue)
02:59 PM seb_kuzminsky: sure
02:59 PM seb_kuzminsky: do you want to do it, or do you wnat to send me the info and i can add the post?
03:00 PM andypugh: All the info is at that URL above. (And when I last tried to do anything on wlo it took me 6 iterations to add a new Showcase, so I would rather you did it)
03:00 PM seb_kuzminsky: heh, ok
03:00 PM seb_kuzminsky: i just copy/paste the previous post and edit that ;-)
03:04 PM rmu: i will not bring my cnc machine ;)
03:07 PM andypugh: There will be several machines and robots there. You can test finite-jerk on non-trivial kinematics.
03:08 PM andypugh: And have you considered converting your machine for road use?
03:08 PM rmu: hehe... a bit large and heavy
03:09 PM andypugh: You are allowed up to 32 tonnes…
03:09 PM rmu: let's see if i manage to produce something stable till then
03:09 PM andypugh: Sorry, 42 tonnes.
03:09 PM rmu: i am only allowed 3,5 metric tons ;)
03:09 PM andypugh: Tie an anchor to the gantry, wind it back, re-deply the anchor...
03:10 PM KGB-wlo: push to master branch: http://linuxcnc.org/
03:10 PM seb_kuzminsky: hmm, that title is too long
03:11 PM KGB-wlo: push to master branch: http://linuxcnc.org/
03:11 PM andypugh: Thanks Seb
03:11 PM seb_kuzminsky: grr, and the link isnt clickable!
03:11 PM seb_kuzminsky: dont thank me yet
03:13 PM KGB-wlo: push to master branch: http://linuxcnc.org/
03:14 PM seb_kuzminsky: ok, i think it's good now
03:14 PM seb_kuzminsky: let me know if i got anything wrong (or just push the fix)
03:18 PM andypugh: Looks OK to me.
03:18 PM seb_kuzminsky: cool
10:23 PM seb_kuzminsky: does anyone know anything about computing speeds & feeds? I've been using https://fswizard.com/ and their android app
10:23 PM seb_kuzminsky: it's easy to use and seems to get reasonable numbers
10:23 PM seb_kuzminsky: but it's free only as in beer
10:23 PM Tom_L: i like that app
10:24 PM Tom_L: i generally shoot for a chip load per tooth and work out from the desired rpm etc
10:24 PM seb_kuzminsky: without that app i've usually resorted to looking for the tool manufacturer's table of chip loads
10:25 PM seb_kuzminsky: once you have the surface speed and the per-tooth chipload everything comes easily
10:25 PM Tom_L: i generally shoot for .001-.002" chip load on my little mill
10:25 PM seb_kuzminsky: but how do you get those two numbers?
10:25 PM Tom_L: HP and machine rigidity
10:26 PM Tom_L: i just do it by trial and error
10:26 PM Tom_L: i get to know the machine a bit
10:27 PM Tom_L: http://www.onsrud.com/plusdocs/Doc/index.html?model.code=FeedSpeeds
10:27 PM Tom_L: has quite a few different materials
10:27 PM Tom_L: what material are you cutting?
10:28 PM Tom_L: i have to run slower than most apps call out because of the machine size and spindle motor
10:29 PM Tom_L: rule of thumb, i use 2 flute on aluminum and 4 on steel
10:29 PM Tom_L: for chip evacuation
10:30 PM Tom_L: and you can buy a high helix cutter for aluminum to help that over a standard helix
10:33 PM seb_kuzminsky: i'm not looking for the speeds and feeds for a particular machining problem, i'm looking for the general method of figuring them out, so that i can capture that in an open source program
10:33 PM Tom_L: i se
10:33 PM Tom_L: e
10:34 PM Tom_L: https://www.kennametal.com/content/dam/kennametal/kennametal/common/Resources/Technical%20Tips/Machining%20Knowledge/TechTip_030_CuttingToolFormulas-1.pdf
10:34 PM Tom_L: bottom of the page might help
10:35 PM seb_kuzminsky: thanks
10:36 PM Tom_L: the machinist's handbook has them as well i'm sure
10:37 PM seb_kuzminsky: yeah, those formulas are straight forward and well known, what i'm really confused about is how to select the radial and axial depth of cut, and the surface speed and chip load
10:37 PM Tom_L: fwiw, my cad cam will calculate based on what you input and adjust the others as you change parameters
10:37 PM Tom_L: so you can start with a chipload or a rpm etc and it will figure as you input more parameters
10:38 PM Tom_L: so they interact with each other
10:38 PM Tom_L: lemme take a screen shot of that page just as a reference
10:39 PM seb_kuzminsky: yeah, those four fundamental number all affect each other
10:39 PM seb_kuzminsky: through model for tool life, tool deflection, surface finish, available spindle power, etc
10:39 PM seb_kuzminsky: rigidity of the machine, the tool, the work piece, and the setup
10:42 PM seb_kuzminsky: i read the relevant chapters of 'fundamentals of modern manufacturing', it had some good low level info, but not much in the way of implementable algorithms (except for those equations that you linked on kennametal)
10:42 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Seb/
10:42 PM Tom_L: to give you an idea how they layed it out
10:43 PM Tom_L: that's kinda a bad example but it was the tool that was loaded when i opened it
10:45 PM seb_kuzminsky: that's a skinny center drill!
10:45 PM Tom_L: the 'operation page' and 'tool page' have the bulk of it and the 'add/edit step' is the summary of the other 2
10:45 PM Tom_L: probably the point of it
10:46 PM Tom_L: i may have set that for an engraving tool
10:46 PM seb_kuzminsky: and if you select a milling cutter, then the 'step' or 'operation' page would have radial and axial depth of cut?
10:46 PM seb_kuzminsky: like fswizard does
10:46 PM Tom_L: i believe so, let me pull up a mill and repost those
10:48 PM seb_kuzminsky: i think you can get the surface speed by cross referencing the tool material and the work piece material
10:48 PM seb_kuzminsky: each cell in that table has a 'low sfm' and a 'high sfm', where low sfm is optimized for long tool life and high sfm is optimized for short cycle time (this is from Fundamentals)
10:50 PM seb_kuzminsky: each cell in that table also has a list that maps tool diameter to desired chip load
10:50 PM Tom_L: i renamed the first ones xxxxx1.jpg
10:50 PM Tom_L: refresh and there are some with a mill cutter
10:50 PM seb_kuzminsky: thanks, i'll take a look
10:52 PM Tom_L: i generally just set the 'important' parameters for f/s
10:52 PM seb_kuzminsky: hmm, i don't see the depth of cut numbers anywhere
10:52 PM Tom_L: i don't think it has those
10:52 PM Tom_L: the program is probably older than you
10:52 PM seb_kuzminsky: heh
10:52 PM Tom_L: alot of the new ones do all sorts of fancy stuff like that
10:53 PM Tom_L: down to calculating HP etc
10:54 PM Tom_L: download fusion 360 and have a look at their tooling setup
10:54 PM seb_kuzminsky: i've got to run, thanks for your input!
10:54 PM seb_kuzminsky: bye for now!
10:54 PM Tom_L: np, later
11:00 PM Tom_L: seb_kuzminsky, for reference the center EM is high helix for aluminum: http://tom-itx.no-ip.biz:81/~webpage/cnc/Seb/Endmills.jpg
11:12 PM hazzy: Seb, if you are interesting in optimizing material removal rate and calculating power requirements etc, this is an excellent resource:
11:12 PM hazzy: https://www.dropbox.com/s/tdn2x01414o3lzt/Tooling%20%20Feeds%2C%20Speed%20Power%20Removal%20Rate%20calc%20eq%20and%20tables.pdf?dl=0
11:12 PM hazzy: I apologize for the sloppy scan. The first page is actually the last page of the PDF, but I think you can figure it out :)
11:19 PM * Tom_L screws his head on sideways...
11:20 PM Tom_L: that's not the machinist's handbook is it?
11:22 PM hazzy-dev: Kinda looks like it doesn't it!
11:24 PM Tom_L: :)
11:24 PM hazzy-dev: It was a handout in a tool design course I took, to it probably is, LOL
11:24 PM hazzy-dev: I have kept it all these years for no reason!
11:25 PM Tom_L: i forget what the standard is but i think all tool MFG are going to a standard and are generally available in most cad packages
11:25 PM Tom_L: i've got a sneaking suspicion that's what seb is up to for lcnc... only a speculation.
11:25 PM hazzy-dev: Or PyCAM ...
11:25 PM hazzy-dev: Boy, that would be nice!
11:26 PM Tom_L: there was some talk about he here some time back iirc
11:28 PM Tom_L: pretty soon you'll have a page similar to the tool page for your machine.
11:28 PM hazzy-dev: Isn't Seb also the PyCAM maintainer, or something like that?
11:28 PM Tom_L: enter the HP here:, enter max rpm here: etc
11:29 PM Tom_L: i'm not sure
11:29 PM Tom_L: enter your work envelope here:
11:29 PM Tom_L: sorry that part is too big for you son.
11:29 PM hazzy-dev: That would be great, but it seems like it would have to be done in CAM, not the controller?
11:30 PM hazzy-dev: LOL, all my parts
11:30 PM Tom_L: yeah
11:30 PM Tom_L: some of them may be doing it already
11:30 PM Tom_L: i haven't looked into it enough to know
11:31 PM Tom_L: i keep a cad file for my work table so i can lay a part on it if i need to
11:33 PM hazzy-dev: I have a entire model of my machine in HSMWorks so I can simulate the fixtruing and operation, kinda silly for a 4 axis machine, but its fun!
11:34 PM Tom_L: i've also stored several common work offsets for it in the unused spaces G59 and down so i can pull them up quickly
11:35 PM Tom_L: mine is in catia
11:35 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Column_Mill_VMC.jpg
11:36 PM Tom_L: it's helpful sometimes
11:36 PM hazzy-dev: Very nice!
11:38 PM Tom_L: it was very helpful building it
11:39 PM hazzy-dev: Here is mill without the head (I stole the head from somebody else): https://github.com/KurtJacobson/RF45-CNC/tree/master/CAD%20Models
11:40 PM Tom_L: oh, i think i've seen that.. pretty nice
11:40 PM Tom_L: i was gonna put my Y motor out like that but it fit on the other end so that's where it went
11:41 PM Tom_L: i think alot of big machines hide that one in the back like that
11:44 PM hazzy-dev: Yes, I tried to put mine at the back under the column, there is room, but is was going to be a pain to mount it so I went the lazy rout
11:45 PM hazzy-dev: Tormach has the Y motor stick out the back, which I guess I could have also done
11:47 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Mill_Steel/Assembly/Y_Motormount2.jpg
11:47 PM hazzy-dev: That looks nice and clean!
11:48 PM hazzy-dev: The mazak also has the motor out the back, I had to remove it to fit it thru the door: https://i.imgur.com/Wd2yzhn.jpg
11:48 PM Tom_L: is that a small machine?
11:49 PM Tom_L: those slides don't look that big
11:49 PM Tom_L: nor the screw
11:49 PM hazzy-dev: Relatively small, but it weights 10000bls! the screw is 2"
11:49 PM Tom_L: the pic is deceiving then
11:50 PM Tom_L: that's pretty hefty
11:50 PM Tom_L: my bud's tree has 3" screws
11:52 PM Tom_L: headin to sleep here. ttyl
11:53 PM hazzy-dev: Its very hefty, but not that large: https://i.imgur.com/I1DKS6B.jpg
11:53 PM hazzy-dev: nite!