Fwd: [beagleboard] Re: EMC2 on BBB?

35 views
Skip to first unread message

Drew Fustini

unread,
May 30, 2013, 6:19:17 PM5/30/13
to ps1...@googlegroups.com, Ed Bennett, Jay Hopkins


---------- Forwarded message ----------
From: David Anders <dande...@gmail.com>
Date: Thu, May 30, 2013 at 5:17 PM
Subject: Re: [beagleboard] Re: EMC2 on BBB?
To: beagl...@googlegroups.com


Drew,

one of the TI sponsored cape contest winners will have his project available through boardzoo.com soon:

http://hipstercircuits.com/replicape-revision-a2-ready-for-review/

this would provide an excellent control system for a pick-n-place....

Dave



On Thursday, May 30, 2013 5:11:12 PM UTC-5, Drew Fustini wrote:
Cool, we were brainstorming at my hackerpace in Chicago (http://pumpingstationone.org/) last night and we are going to try to get EMC2 working as a group project.  We have members that built a Pick and Place gantry and would like to control it with the BBB directly (they had previously been using TinyG connected to a Pi).  We were also looking at what HipsterCircuits has done for stepper control with the PRU.

Anyways, just wanted to ping the Beagle community to see if anyone else was similarly interested.


On Thu, May 30, 2013 at 4:57 PM, David Anders <dande...@gmail.com> wrote:
Drew,

no not yet, it is on my "todo list" with my copious amount of free time.....

i purchased one of the generic cnc units off ebay to test it with....

Dave



On Thursday, May 30, 2013 4:54:09 PM UTC-5, Drew Fustini wrote:
I'm about to start working on getting the CNC sofware EMC2 (http://www.linuxcnc.org/) to run on the BBB.   Is anyone else working on this, too?

Thanks!
Drew

--
For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to a topic in the Google Groups "BeagleBoard" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/beagleboard/ck4J3tl_GMs/unsubscribe?hl=en.
To unsubscribe from this group and all its topics, send an email to beagleboard...@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to a topic in the Google Groups "BeagleBoard" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/beagleboard/ck4J3tl_GMs/unsubscribe?hl=en.
To unsubscribe from this group and all its topics, send an email to beagleboard...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Drew Fustini

unread,
May 31, 2013, 10:14:47 AM5/31/13
to ps1...@googlegroups.com
---------- Forwarded message ----------
From: <mvca...@gmail.com>
Date: Thu, May 30, 2013 at 5:31 PM
Subject: [beagleboard] Re: EMC2 on BBB?
To: beagl...@googlegroups.com


FYI - The LinixCNC gang is already a long ways down that road.
You can search the developer mail list archives to see what's been going on:
http://dir.gmane.org/gmane.linux.distributions.emc.devel

Dave

Drew Fustini

unread,
May 31, 2013, 10:15:08 AM5/31/13
to ps1...@googlegroups.com
---------- Forwarded message ----------
From: Eric Keller <eeke...@psu.edu>
Date: Thu, May 30, 2013 at 11:12 PM
Subject: Re: [beagleboard] Re: EMC2 on BBB?
To: beagl...@googlegroups.com


On Thu, May 30, 2013 at 6:31 PM,  <mvca...@gmail.com> wrote:
> FYI - The LinixCNC gang is already a long ways down that road.
> You can search the developer mail list archives to see what's been going on:
> http://dir.gmane.org/gmane.linux.distributions.emc.devel
>
> Dave

quite a bit of discussion on the linuxcnc users list as well
https://lists.sourceforge.net/lists/listinfo/emc-users

There is work to be done on the real time kernel at this point.
Eric

Drew Fustini

unread,
Jun 5, 2013, 11:30:39 AM6/5/13
to Ed Bennett, Jay Hopkins, Shawn Blaszak, ps1...@googlegroups.com
[ed, jay, shawn: do you want me to email you directly with interesting things or just the ps1bbb google group?]

---------- Forwarded message ----------
From: Michael Haberler <habe...@gmail.com>
Date: Wed, Jun 5, 2013 at 12:15 AM
Subject: [beagleboard] Re: EMC2 on BBB?
To: beagl...@googlegroups.com


yes

I'm involved with the LinuxCNC project. My works in progress are:

1. getting LinuxCNC to run on non-x86 platforms, and with a wider range of realtime kernels (when I joined it was RTAI only, which doesnt run on non-PC platforms yet)
2. splitting the package into the GUI/Interpreter part, and a standalone open source motion toolkit which can be talked to over ZeroMQ and protobuf; LinuxCNC per se remaining a user of that toolkit. This effort is tentatively called 'MachineKit'.

(1) is very far down the road; LinuxCNC now runs on RTAI, Xenomai, RT-Preempt and on vanilla kernels too (with severely reduced tool path quality, but the full configurability of the HAL environment nonetheless)

Right now we have per-kernel-flavor builds; this will be replaced by a unified binary which runs unchanged on any kernel by sporting a runtime-loadable portable realtime kernel API.

The Xenomai build for the Beaglebone is just being wrapped up from the 3.2->3.8 kernel transition, and I think we will have a ready-to-go SD image with batteries included with a very low number of weeks. This build has a stepper generator which runs on the PRU, and I think it's the best performance you'll get short of adding an FPGA. Also, the PRU code is done very clever - it supports 'tasklets' if you will, meaning you can plug in other jobs into the PRU 'task' as needed.

For folks interested in this, I suggest to keep an eye on Charles' blog which will have the latest and greatest: http://bb-lcnc.blogspot.co.at/ and the emc-developers mailing list. Now that I know that there's interest here I will announce here too.

Our goal is to provide a LinuxCNC kit for the 3D printer community, and I have started cooperating with the local reprap crowd here to get immersed in this specific mode of CNC - I have so far mostly used LinuxCNC to drive 'heavy metal', and thats still the largest user base.

(2) MachineKit: the foundations are complete (protobuf infrastructure, zeroMQ transport, extensions to the LinuxCNC realtime environment to support a message-based interface in addition to the current shared memory model of interaction), and work has started on migrating parts of the code away from the old middleware (NML) to the new scheme. That effort I expect to yield results 'later this year'; I'd be happy to hear from folks interested in cooperating on this.

- Michael


Am Donnerstag, 30. Mai 2013 23:54:09 UTC+2 schrieb Drew Fustini:
I'm about to start working on getting the CNC sofware EMC2 (http://www.linuxcnc.org/) to run on the BBB.   Is anyone else working on this, too?

Thanks!
Drew

--

Drew Fustini

unread,
Jun 5, 2013, 2:40:23 PM6/5/13
to Ed Bennett, Jay Hopkins, Shawn Blaszak, ps1...@googlegroups.com
from: http://www.element14.com/community/community/knode/single-board_computers/next-gen_beaglebone/blog/2013/05/22/bbb--working-with-the-pru-icssprussv2#comment-22799

"yes, I did the inital work for non x86 installs and non-RTAI kernels, including exploring the PRU and the debugger; Charles did the heavy lifting on the PRU stepper generator.

 

The realtime requirement isnt limited to stepgens; path quality suffers badly if you dont have that.

 

EMC2 was renamed to linuxCNC because of trademark issues with EMC Corp."

Drew Fustini

unread,
Jun 5, 2013, 9:39:29 PM6/5/13
to Ed Bennett, Jay Hopkins, Shawn Blaszak, ps1...@googlegroups.com
Okay, so I understand, do you think we should still try to do PRU stepper control for the hackerspace challenge (July 5th deadline)?

thanks,
drew


On Wed, Jun 5, 2013 at 4:38 PM, Ed Bennett <e...@kineticsandelectronics.com> wrote:
Until we clarify a few goals and get some traction I think it would be best to lie low, relatively speaking. IMO, the additions that would be useful to the pick and place goal would be Norm (from Panasonic's PnP R&D office) and JP DeBauge, the ME from Navistar who was working with Jeff and the gang earlier on. Whadda you think?

Other (smart, talented, well meaning) people may have different agendas that might or might not mesh with the effort we've already put into this. If in the name of being inclusive we spend all our time re-negotiating our objective from first principles it will be much harder to make progress. Perhaps not a politically correct mindset but but you did ask ;)

HH
-e

Shawn Blaszak

unread,
Jun 6, 2013, 1:53:23 PM6/6/13
to Ed Bennett, Drew Fustini, Jay Hopkins, ps1...@googlegroups.com
I don't, presently, have the Google Group messages sent to my e-mail inbox.

As for the project, now that we know that the LinuxCNC community is already well down the road of porting to the BBB it does, unfortunately, take some of the wind out of our sails when it comes to differentiating ourselves.  I definitely, like the idea of being able to use the PRUs to eliminate the need for a kernel hacker to port the real-time patch on the 3.8 kernel.  I've seen that, consistently, be a stumbling block to LinuxCNC being implemented on a number of non-x86 systems in the past and think it would be cool if we could put control of that part of it into the hands of the LinuxCNC community. 

I like the idea of focusing on how this can relate to the Pick-and-place project and think it could be a cool way to help provide the P&P project with software to run the machine.  My thoughts, in that regard, were that we could:

* use EMC2 and the stepper PRU code to control the cartesian robot hardware
* use Wayland s the display server for the linux install (assuming it there is a working Wayland for BBB), since wayland, specifically, provides hardware acceleration using OpenGL ES).
* Use Qt to create the GUI since it is supposed to also hardware accelerate under a properly functioning Wayland install.
* Integrate the PnP machine closely with EagleCAD so that someone can do all their design work in one package and they, easily, move the existing project files into the PnP machine software environment.  The reason I think this is a good idea (besides the general ease, for the user, of moving from design to assembly) is that EagleCAD's file format/database and component libraries already store a lot of the info that the PnP machine will need to do it's job (the precise physical dimensions of the components; the location of  the soldering pads on the components; and obviously the precise physical layout of the PCB board being populated).  The reason I think EagleCAD might be preferable to any of the other packages is that it's wide-spread popularity means it already has many many more specific component specs included in it's libraries.  The biggest problem I've seen to adopting EagleCAD's file format is that, last I heard, there was a serious licensing issue where they had claimed it was a CC license but then had included a specific CC incompatible clause which broke it's compatibility with other open source licenses.  At the least, we would have to make sure they have fixed that issue, otherwise it may not be legally possible to integrate their stuff with any of the other SDKs/tools that would go into the project.
* Use OpenCV or, if at all possible, OpenCV through the much simplified OpenFrameworks interface to recognize the location and orientation of the ships on the tape, on the vaccuum needle, and once placed on the PCB.

That's most of my idea for now.  Sorry it's taken me a bit of time to get around to documenting it all.  If we did this, I'd be most interested in working on the GUI, OpenCV, and PCB CAD file format parsing end of it.

-Shawn


On Wed, Jun 5, 2013 at 4:38 PM, Ed Bennett <e...@kineticsandelectronics.com> wrote:
Until we clarify a few goals and get some traction I think it would be best to lie low, relatively speaking. IMO, the additions that would be useful to the pick and place goal would be Norm (from Panasonic's PnP R&D office) and JP DeBauge, the ME from Navistar who was working with Jeff and the gang earlier on. Whadda you think?

Other (smart, talented, well meaning) people may have different agendas that might or might not mesh with the effort we've already put into this. If in the name of being inclusive we spend all our time re-negotiating our objective from first principles it will be much harder to make progress. Perhaps not a politically correct mindset but but you did ask ;)

HH
-e


On 06/05/2013 10:30 AM, Drew Fustini wrote:
Reply all
Reply to author
Forward
0 new messages