Re: [Emc-users] Synchronised cam turning
Hi Jack, I have done this. A couple of years back I built a 2' x 4' router for a local wood worker. It also had an A axis that he used for turning barley twist legs and such like. It took a bit of settings up but it worked reasonably well. I created a SheetCam post to do the appropriate axis swapping. I don't currently have an A axis for my mill and I can't really justify the time and money to build one just for this job. The idea of using a lathe with a few code tweaks is much more appealing :-) I did try playing around with turning the spindle into an A axis but I had lots of trouble with tuning. The combination of inverter drive and a fair amount of backlash in the gearbox made things rather difficult. I might have another go and see if I can get some semblance of stability at speed. The spindle encoder gives 4000 counts/rev so if I can get the PID loop stable it should have enough resolution for this job. Les Jack wrote: I went to a ShopBot event last year and saw where folks did 'turning' using their indexer. The process they used was to determine the 'final diameter' and layed out a rectangle of 'unwraped' surface (length as the length of the turning, width of the 'final diameter'). Then put whatever they wanted on the 'triangle' using various methods. Once done, the pathed it, and used the just routed the piece as a plane, using the indexer as the 'x' axis, and the 'y' axis as the indexer. You might be able to take that thought and make your 'unwraped surface' to be the profiles of the 'unwrapped' cams. Let us know how you attack this (if you do). -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
Hi Jeff, Sorry, I meant do you know of any components that use shared memory to communicate between a user space part and a real time part. Les Jeff Epler wrote: No, I don't think there are any. I proposed something similar (but it doesn't have the adding/removing offsets problem) to some folks doing a scanline laser with emc, but to the best of my knowledge they didn't go that direction with their implementation. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
On Tue, Jul 21, 2009 at 08:42:24AM +0100, Leslie Newell wrote: Hi Jeff, Sorry, I meant do you know of any components that use shared memory to communicate between a user space part and a real time part. Oh, yes. There are several: Userspace Realtime halscopescope_rt halsampler sampler halstreamer streamer and of course emc's task (userspace) and motion (rt) communicate through such a shared memory area. halstreamer is probably the most similar to what I'm suggesting, because a userspace process loads data and a realtime process consumes it. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
Thanks Jeff, you have been very helpful. One last question for the moment ;-). What is the best way to compile C components? With .comp components you use comp with it's various switches. What is the correct command line for a C component? Les Jeff Epler wrote: On Tue, Jul 21, 2009 at 08:42:24AM +0100, Leslie Newell wrote: Hi Jeff, Sorry, I meant do you know of any components that use shared memory to communicate between a user space part and a real time part. Oh, yes. There are several: Userspace Realtime halscopescope_rt halsampler sampler halstreamer streamer and of course emc's task (userspace) and motion (rt) communicate through such a shared memory area. halstreamer is probably the most similar to what I'm suggesting, because a userspace process loads data and a realtime process consumes it. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
On Tue, Jul 21, 2009 at 01:20:48PM +0100, Leslie Newell wrote: Thanks Jeff, you have been very helpful. One last question for the moment ;-). What is the best way to compile C components? With .comp components you use comp with it's various switches. What is the correct command line for a C component? it should be comp --compile foo.c comp --compile --userspace foo.c though at the moment some bugs break 'comp --compile --userspace' for C files. In another message I posted a link to a Makefile that works for userspace C components: http://mid.gmane.org/20090720161919.ga3...@unpythonic.net Jeff -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Synchronised cam turning
Has anyone tried turning a cam on a lathe using EMC? I tried using lots of G33 moves but they seem to get out of sync. My machine has pretty high acceleration and top speed but even at very low spindle speeds (14rpm) it doesn't seem to keep up. Does EMC have any other way of doing synchronised moves apart from G33/G76? Les -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
On Mon, Jul 20, 2009 at 11:23:55AM +0100, Leslie Newell wrote:
Has anyone tried turning a cam on a lathe using EMC? I tried using lots
of G33 moves but they seem to get out of sync. My machine has pretty
high acceleration and top speed but even at very low spindle speeds
(14rpm) it doesn't seem to keep up.
Without seeing your gcode, I am going to take a guess that the problem
is one of two things: you have lots of tiny moves, or you did not
properly calculate your K values.
emc's spindle synchronized motion is more or less the traditional motion
planner, but using spindle angle instead of time to control the forward
progress. Specifically, it has the same behavior when there are a lot
of small segments: it will plan each individual move so that it can stop
by the end of that move and stay within machine constraints.
Does EMC have any other way of doing synchronised moves apart from G33/G76?
emc and gcode don't, but of course hal does. You could create a
component which offsets the commanded X according to the spindle angle
and the shape of the cam. The component would include the mathematical
description of the cam shape, r=f(theta)+r0. It would take as inputs
the spindle angle and the commanded motor position (r0), and output the
modified motor position (r)
net Xpos-cmd = camcutter.pos-cmd-in
net Xpos-cam-cmd camcutter.pos-cmd-out = stepgen.0.motor-pos-cmd # or
whatever
net Xpos-cam-fb stepgen.motor-pos-fb = camcutter.pos-fb-in
net Xpos-cb = camcutter.pos-fb-out
net spindle-revs = camcutter.spindle-revs
net cam-enable motion.digital-out-nn = camcutter.enable
In emc 2.3 and above double precision numbers are used for hal floats,
so you can probably skip messing with index-enable and just use theta =
2*pi*spindle_revs for a periodic function f().
In your component, you have something like this:
if(enable) {
double theta = spindle_revs * 2 * M_PI;
double displacement = f(theta);
motor_pos_cmd_out = motor_pos_cmd_in + displacement;
motor_pos_fb_out = motor_pos_fb_in - olddisplacement;
olddisplacement = displacement; // (instance variable)
} else {
motor_pos_cmd_out = motor_pos_cmd_in;
motor_pos_fb_out = motor_pos_fb_in;
}
but this gives a step discontinuity when enable changes. You need to
embellish it with code to slow apply and unapply the displacement. Such
code is in emc/motion/simple_tp.{c,h} in libraryish form. (I'd be
tempted to just copy both into a single header, change simple_tp_update
to be static, and include the header)
Jeff
--
Enter the BlackBerry Developer Challenge
This is your chance to win up to $100,000 in prizes! For a limited time,
vendors submitting new applications to BlackBerry App World(TM) will have
the opportunity to enter the BlackBerry Developer Challenge. See full prize
details at: http://p.sf.net/sfu/Challenge
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
Hi Jeff, I would have replied earlier but I blew up my X11 settings and it took me this long to get it working again... is one of two things: you have lots of tiny moves, Yes. There isn't much choice in this. To even get a simple offset circle you need to generate lots of short lines. or you did not properly calculate your K values. I don't /think/ I did. For this job, K gets rather confusing. However changing the spindle speed affects the results so I would guess that K isn't the problem. emc's spindle synchronized motion is more or less the traditional motion planner, but using spindle angle instead of time to control the forward progress. Specifically, it has the same behavior when there are a lot of small segments: it will plan each individual move so that it can stop by the end of that move and stay within machine constraints. Is there any way to trick the motion planner into thinking the machine has infinite acceleration during synchronized cutting? As long as you are careful with the code, following error shouldn't be too bad. emc and gcode don't, but of course hal does. You could create a component which offsets the commanded X according to the spindle angle and the shape of the cam. The component would include the mathematical description of the cam shape, r=f(theta)+r0. Hmm, interesting idea. Not too bad for a simple circular cam but unfortunately I have been supplied with CAD drawings and they are wacky profiles. Several different profiles, all one-offs. Is there any way for a RT component to read a file when it initializes? I could store the profile in a text file which I could generate with SheetCam. Thanks, Les -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
An excuse to put a milling spindle on the crossslide maybe. will the clearance angle be ok if turning Dave Caroline -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
On Monday 20 July 2009, Dave Caroline wrote: An excuse to put a milling spindle on the crossslide maybe. will the clearance angle be ok if turning Dave Caroline I think that is something I would do with the A axis. I have turned stuff with it, but haven't tried to make a cam yet. I got stuck needing the laathe for something, but had a large piece of alu I didn't want to lose the center of, so I did the other small piece on the mill. Slow, but worked well. -- Cheers, Gene There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) The NRA is offering FREE Associate memberships to anyone who wants them. https://www.nrahq.org/nrabonus/accept-membership.asp A gambler's biggest thrill is winning a bet. His next biggest thrill is losing a bet. -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
emc and gcode don't, but of course hal does. You could create a component which offsets the commanded X according to the spindle angle and the shape of the cam. The component would include the mathematical description of the cam shape, r=f(theta)+r0. On Mon, Jul 20, 2009 at 05:25:19PM +0100, Leslie Newell wrote: Hmm, interesting idea. Not too bad for a simple circular cam but unfortunately I have been supplied with CAD drawings and they are wacky profiles. Several different profiles, all one-offs. Is there any way for a RT component to read a file when it initializes? I could store the profile in a text file which I could generate with SheetCam. rtai has shared memory between realtime components and userspace. This is how the communication between task and motion works, and how it works between the userspace and realtime parts of scope, sampler, and streamer. In this case, you'd have to arrange for userspace to load a 1D array of f(theta) values before beginning the cam cycle, then the realtime f(theta) becomes: find nearest table values to theta, and use an interpolation algorithm on those values (such as linear interpolation on two values) to give the offset. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
Heh. An A axis would be too easy :-) TBH I think I may pass on this job. Too many unknowns. I will check if the cams can be made in pieces. If so then I could simply mill the lobes. Les Gene Heskett wrote: On Monday 20 July 2009, Dave Caroline wrote: An excuse to put a milling spindle on the crossslide maybe. will the clearance angle be ok if turning Dave Caroline I think that is something I would do with the A axis. I have turned stuff with it, but haven't tried to make a cam yet. I got stuck needing the laathe for something, but had a large piece of alu I didn't want to lose the center of, so I did the other small piece on the mill. Slow, but worked well. -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
Thanks Jeff I'll look into it. Do you know off-hand if there are any current components that do this? Les rtai has shared memory between realtime components and userspace. This is how the communication between task and motion works, and how it works between the userspace and realtime parts of scope, sampler, and streamer. In this case, you'd have to arrange for userspace to load a 1D array of f(theta) values before beginning the cam cycle, then the realtime f(theta) becomes: find nearest table values to theta, and use an interpolation algorithm on those values (such as linear interpolation on two values) to give the offset. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
I went to a ShopBot event last year and saw where folks did 'turning' using their indexer. The process they used was to determine the 'final diameter' and layed out a rectangle of 'unwraped' surface (length as the length of the turning, width of the 'final diameter'). Then put whatever they wanted on the 'triangle' using various methods. Once done, the pathed it, and used the just routed the piece as a plane, using the indexer as the 'x' axis, and the 'y' axis as the indexer. You might be able to take that thought and make your 'unwraped surface' to be the profiles of the 'unwrapped' cams. Let us know how you attack this (if you do). -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Synchronised cam turning
On Mon, Jul 20, 2009 at 10:25:58PM +0100, Leslie Newell wrote: Thanks Jeff I'll look into it. Do you know off-hand if there are any current components that do this? No, I don't think there are any. I proposed something similar (but it doesn't have the adding/removing offsets problem) to some folks doing a scanline laser with emc, but to the best of my knowledge they didn't go that direction with their implementation. Jeff -- Enter the BlackBerry Developer Challenge This is your chance to win up to $100,000 in prizes! For a limited time, vendors submitting new applications to BlackBerry App World(TM) will have the opportunity to enter the BlackBerry Developer Challenge. See full prize details at: http://p.sf.net/sfu/Challenge ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
