#emc-devel | Logs for 2006-04-09

[00:00:16] <alex_joni> unless someone has very huge units/counts, but then the jump is still small
[00:01:07] <jmkasunich> alex: you are confusing me
[00:01:13] <jmkasunich> counts are irrelevant
[00:01:50] <jmkasunich> suppose the home position is at the left end of the table
[00:02:17] <jmkasunich> and someone shut the machine down when it was at the right, 500mm away
[00:02:23] <alex_joni> ok..
[00:02:27] <jmkasunich> when you turn on, it will treat that as 0
[00:02:46] <jmkasunich> when you do the home, the position command and feedback will both be at -500 when you hit the index
[00:02:53] <jmkasunich> suddenly both are reset to zero
[00:03:24] <alex_joni> right
[00:03:34] <cradek> the craptacular maxnc software would save the position when you exit, and it would be there next time
[00:03:40] <cradek> why don't we do that?
[00:03:45] <alex_joni> cradek: different topic
[00:04:03] <cradek> can't argue with that
[00:04:06] <alex_joni> but I thought about that too ;)
[00:04:07] <jmkasunich> that could be nice, but if the machine has cranks, somebody could still move it while off
[00:04:08] <cradek> some other time
[00:04:17] <jmkasunich> so you still need to home
[00:04:30] <cradek> sure, it doesn't make home unnecessary.
[00:04:32] <alex_joni> jmkasunich: if you don't have brakes
[00:04:49] <alex_joni> on robots we check the index pulse against the stored position
[00:04:56] <cradek> but for steppers without home switches, it would be pretty nice
[00:05:00] <alex_joni> if it matches we assume machine is ok
[00:05:13] <alex_joni> if it doesn't .. then you need to home
[00:05:38] <alex_joni> of course this is with brakes applied while machine is off..
[00:05:42] <cradek> sorry, I'll bring this up some other time (or I'll just write it)
[00:05:48] <alex_joni> back to topic ;)
[00:06:41] <jmkasunich> cradek: if "just writing it" is gonna require the ability for emc to stuff a value into the encoder counter, then it will affect the "canonical encoder interface" that we are discussing, and is appropriate
[00:06:54] <jmkasunich> if it can be done without stuffing a value, thats much better
[00:07:26] <alex_joni> jmkasunich: it will probably write the value to a file, and reload on startup
[00:07:38] <jmkasunich> reload where though...
[00:07:55] <jmkasunich> into the encoder counter, or into something inside the motion module
[00:08:07] <alex_joni> motion I'd say.. but rather in task or higher
[00:08:14] <jmkasunich> ok
[00:08:52] <jmkasunich> back to homing
[00:09:33] <jmkasunich> today the axis encoders never change (except when the shaft moves) and we home and offset and everything else by adding offsets inside the motion controller
[00:09:48] <jmkasunich> now we're talking about having the encoder counters change
[00:10:18] <tomp> if rotary are used, several absolute positions with several 'near home' dogs can be used, eleminating very long homing traverses
[00:10:43] <tomp> always less than 1 rev
[00:11:00] <jmkasunich> tom, please don;t make things any more compilcated than they already are
[00:11:04] <alex_joni> tomp: that may or may not be what you want, depending on the app
[00:11:21] <tomp> ok
[00:12:33] <alex_joni> jmkasunich: sorry, can't stay up longer & be usefull at the same time :(
[00:12:40] <jmkasunich> ok, goodnight
[00:12:45] <jmkasunich> I'm replying to Jon again
[00:12:53] <alex_joni> let me know if you have any more on this
[00:12:59] <jmkasunich> maybe I can convince cradek to brainstorm with me on this
[00:13:49] <alex_joni> good luck ;)
[00:13:51] <alex_joni> night all
[00:13:59] <tomp> bye alex
[06:58:20] <etla> anyone here ?
[13:00:17] <alex_joni> morning ray
[13:00:54] <rayh> Hi Alex.
[13:41:01] <alex_joni> hi matt
[13:41:21] <rayh> Well hi Matt.
[13:41:25] <mshaver> hi alex!
[13:41:39] <mshaver> hey ray, I got on here to see if you were awake yet
[13:43:37] <mshaver> rayh: your phone's busy...
[13:44:24] <rayh> Yes it is. Ma's talking to the kinds in Germany.
[13:44:47] <rayh> What's happening?
[13:45:17] <rayh> She's off the phone now.
[13:45:30] <rayh> kids in...
[13:45:34] <mshaver> just thought i'd catch up with you, been busy but i figured we should get our heads together re fest & hnc
[13:45:34] <rayh> can't type at all.
[13:45:39] <mshaver> ok
[13:46:34] <alex_joni> mshaver: everything ok lately?
[13:47:28] <mshaver> alex_joni: yep!
[13:48:16] <alex_joni> mshaver: nice to hear that.. same here, just busy
[13:53:32] <alex_joni> ok, going away for a ahile
[13:56:43] <steves_logging> steves_logging is now known as steve_stallings
[13:57:14] <steve_stallings> morning all, quick look in before going off to re-roof my shed
[13:57:43] <steve_stallings> Matt - we have room reservations Wed to Sat nite, same place
[13:59:24] <mshaver> good!
[13:59:58] <mshaver> talking to ray on the phone
[14:10:42] <steve_stallings> Gotta run in 5 min, see you later.....
[14:19:27] <steve_stallings> steve_stallings is now known as steves_logging
[15:25:13] <jmkasunich> morning guys
[15:25:35] <jmkasunich> won't be here long today, we're leaving for a baseball game when my wife gets home from church
[15:27:05] <rayh> Hi john
[15:27:13] <jmkasunich> hi
[15:27:37] <rayh> I was talking with matt this morning. We plan to bring a couple of modbus devices to fest.
[15:27:49] <jmkasunich> modbus?
[15:27:51] <rayh> One is simple PLC, the other a spindle drive.
[15:28:01] <rayh> Serial communication.
[15:28:05] <jmkasunich> no such thing as a simple PLC
[15:28:18] <rayh> How about cheap.
[15:28:22] <jmkasunich> heh
[15:28:33] <jmkasunich> spindle drive, now that can be simple
[15:28:51] <jmkasunich> a few digital pins for start/stop, a few analogs for spd cmd, feedback
[15:28:58] <rayh> I'll put together an ac motor and encoder.
[15:29:13] <rayh> Matt already has the AC drive.
[15:30:30] <jmkasunich> my project for the next few days (driven by jon's attempts to get his stuff to do threading) is to define a canonical encoder, make homing work with it, and redo all the HAL encoder drivers to match
[15:30:33] <rayh> I'll try to assemble complete info on them.
[15:30:48] <rayh> That sounds like a great plan.
[15:30:50] <jmkasunich> so hopefully when we get to fest, we can make the mazak home in index, not just switch like it is now
[15:31:00] <jmkasunich> home _on_ index
[15:31:10] <rayh> That would be a good step forward.
[15:31:28] <jmkasunich> yeah, index is nasty enough that we've been avoiding it
[15:32:49] <rayh> Matt says that he is working on FredP to attend.
[15:33:35] <rayh> brb brunch.
[16:03:28] <cradek> jmkasunich: in your canon encoder, will the motion controller still be able to tell it whether or not it should be resetting on index pulses?
[16:04:02] <jmkasunich> yes, the canon encoder will probably be identical to (or very similar to) the existing software encoder counter
[16:04:16] <cradek> as you already know, that's how threading works right now
[16:04:31] <jmkasunich> the work at hand is to fix the homing code to work with the "reset on index" behavior
[16:04:39] <jmkasunich> right, threading _needs_ reset on index
[16:04:43] <cradek> ok great, having a simulated encoder that does everything right is vital for testing
[16:04:52] <jmkasunich> homing (so far) doesn't like that
[16:05:13] <cradek> and giving us a reference implementation to point to when someone wants to add an encoder driver is nice too
[16:05:25] <cradek> ok, I understand now, thanks
[16:05:44] <jmkasunich> just added limit and home switches to my more complete simulation (which will be used to test homing to index)
[16:05:57] <cradek> cool
[16:05:58] <jmkasunich> gonna commit that if you want to play with it, its time for me to leave for baseball
[16:06:03] <jmkasunich> back in about 4-5 hours
[16:06:07] <cradek> I've got a date too
[16:06:13] <cradek> ok later then
[16:30:40] <alex_joni> hello
[16:42:28] <rayh> hi alex.
[16:42:43] <rayh> sounds like jmk is making progress
[16:42:52] <alex_joni> yeah..
[16:43:09] <alex_joni> this will be lots of thinking before it's right, though
[16:43:39] <rayh> Yes it will. Years ago I argued for a third axis type in emc itself.
[16:43:58] <alex_joni> LINEAR/ROTARY/?
[16:44:02] <rayh> we have linear and angular which is wrapped linear.
[16:44:07] <rayh> Yep.
[16:44:45] <alex_joni> what was the 3rd one?
[16:45:16] <rayh> Oh. the rotary now is named angular
[16:45:30] <alex_joni> yeah, it's the same..
[16:45:32] <rayh> so you'd have linear/angular/rotary
[16:45:49] <alex_joni> hmm..
[16:46:03] <alex_joni> I can see very limited usage of angular vs. rotary
[16:46:15] <alex_joni> at least at the current state of emc
[16:46:18] <rayh> I don't at all
[16:46:41] <alex_joni> IF it were a bit smarter you could define some run-time kinematics, based on that
[16:46:48] <alex_joni> but that's far from trivial ;)
[16:46:53] <rayh> wrapped linear is the normal way to operate stuff like z,b,c
[16:47:08] <rayh> A B C
[16:47:11] <rayh> sorry
[16:47:15] <alex_joni> yeah
[16:47:24] <rayh> Rotary is stuff like spindles.
[16:47:33] <alex_joni> RPW might be handled different than ABC though
[16:47:42] <alex_joni> ABC is workpiece turntables
[16:47:54] <alex_joni> RPW is tilting/rotating the head/spindle
[16:48:04] <rayh> Right. That is why angular is different from rotary.
[16:48:25] <alex_joni> but then again you'd still need kins to do RPW properly..
[16:48:59] <rayh> You would for stacked A/B also
[16:49:42] <rayh> I think that Graham-Sperry guy in England was working on the stacked thing.
[16:50:24] <alex_joni> never heard of him
[16:51:19] <rayh> Paul met with him a time or two. He's a college prof. Planned on his students building the 4'th and 5'th setup and then writing the kins and tool tip feedrate.
[16:59:06] <alex_joni> yeah, that's a nice thing to do ;)
[16:59:11] <alex_joni> not many users though:D
[16:59:31] <alex_joni> * alex_joni is thinking of building a Z axis for that sherline slides
[17:00:45] <rayh> I've often thought that a little gantry with z would be the ticket
[17:02:44] <alex_joni> gantry might be too much ;)
[17:02:48] <alex_joni> or you mean a fixed gantry?
[17:03:00] <rayh> fixed.
[17:03:13] <alex_joni> oh, so only a support for the Z axis?
[17:03:20] <rayh> yes.
[17:04:30] <alex_joni> is that needed for those small sizes?
[17:10:56] <rayh> Not to much. The stock Sherline Z mount is really pretty solid.
[17:35:14] <rayh> Does your Y base have an extension and 4 mounting holes opposite the motor side?
[17:37:53] <alex_joni> hmm.. not sure
[17:37:56] <alex_joni> don't think so
[17:42:22] <rayh> That is where the standard z base bolts down.
[17:42:40] <rayh> but you can build a plate below the xy table and bolt z to it.
[17:43:03] <alex_joni> yeah, that's my plan
[19:16:53] <SWPadnos> hiya folks. how're things?
[19:17:23] <alex_joni> sweet
[19:17:33] <SWPadnos> that's good :)
[19:22:16] <alex_joni> * alex_joni installed cvs3
[19:23:27] <alex_joni> bbl
[19:26:10] <SWPadnos_> arggh - this wireless network is a bit unreliable
[19:26:28] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[19:52:52] <jepler> what's "cvs3"?
[19:53:44] <jepler> "The CVS3 is a contour vision sensor that can be taught in via a keypad and a ..." <-- this cvs3?
[19:57:30] <cradek> for us, cvs3 is a backup cvs server
[19:57:36] <cradek> just like cvs2
[20:05:42] <jepler> oh oh
[20:05:47] <alex_joni> jepler: what?
[20:05:57] <cradek> did someone remove simio?
[20:06:19] <alex_joni> cradek: I did a while ago.. a few weeks/months?
[20:06:26] <cradek> ok
[20:06:38] <alex_joni> cradek: any reason to still have it?
[20:06:42] <cradek> no
[20:06:50] <cradek> I'm just still sorting out all the changes
[20:06:52] <alex_joni> ok..you scared me ;)
[20:07:21] <alex_joni> I think all the needed packages are installed ont his freebsd
[20:07:42] <cradek> great
[20:09:01] <alex_joni> you know what I don't see in that list? /pkgs ?
[20:09:03] <alex_joni> cvsd
[20:09:09] <alex_joni> or is that part of cvs ?
[20:09:23] <cradek> I don't use cvsd
[20:09:27] <alex_joni> oh.. ok
[20:10:39] <alex_joni> wth is going on in #emc?
[20:10:56] <cradek> nfc
[20:11:15] <alex_joni> asciiart 4 dummies 101
[20:13:40] <jepler> maybe trying to get bo's questions to scroll off, I dunno
[20:14:29] <skunkworks> :)
[20:56:00] <alex_joni> nite all
[20:56:18] <cradek> night alex
[20:56:25] <cradek> maybe I'll have new packages done tonight
[20:56:41] <alex_joni> cradek: cool.. you can add a newsflash to www.linuxcnc.org if you like
[20:56:46] <alex_joni> or I'll do it tomorrow ;)
[20:57:04] <cradek> ok maybe I'll try to figure it out