#emc-devel | Logs for 2009-08-27

[08:06:49] <micges_work> hello
[08:09:05] <alex_joni> hi
[12:46:22] <skunkworks_> do I need to ask about the source? http://www.cnczone.com/forums/showthread.php?t=84346&page=4
[12:46:27] <skunkworks_> last post...
[12:50:34] <SWPadnos> the "simulation with the scrollbar" thing is cool
[12:50:43] <SWPadnos> someone should ask, but it doesn't have to be you
[12:51:21] <skunkworks_> heh - that would be nice. (I am not very elegent)
[12:53:19] <SWPadnos> he added some very cool features to the UI
[12:53:35] <SWPadnos> multi-select and the scroller/simulation thing are nice
[12:54:31] <alex_joni> hmm.. he's using o-words
[12:54:53] <SWPadnos> which more or less guarantees GPL source
[12:54:58] <skunkworks_> that sphire is something I wrote - it is on the emc o-word wiki
[12:55:12] <SWPadnos> and those aren't in the NIST manual, AFAIR
[12:55:29] <alex_joni> he does link to emc2 g-codes from the program
[12:55:52] <alex_joni> http://www.linuxcnc.org/docs/html/gcode_main.html (under Help-> emc2 g-code)
[12:56:19] <skunkworks_> alex_joni: you downlaoded it?
[12:56:23] <skunkworks_> downloaded
[12:56:28] <alex_joni> yeah, played with it yesterdau
[12:56:32] <skunkworks_> heh
[12:56:34] <SWPadnos> it wouldn't run on my machine - I'm not sure what's wrong
[12:56:44] <SWPadnos> and I don't care to fix it at the moment
[12:56:47] <alex_joni> you need to install Dx9c
[12:57:12] <SWPadnos> I have that
[12:57:19] <SWPadnos> and .net something or other
[12:57:23] <SWPadnos> but maybe not 2.0
[12:59:55] <skunkworks_> I remember when the guy from tormach was playing with mach - panning and zooming for a bit. ;)
[13:00:07] <skunkworks_> mach - wow that was a slip. axis.
[13:00:15] <SWPadnos> heh
[13:00:30] <SWPadnos> Mach is part ofTormach :)
[13:00:44] <SWPadnos> (the words, not the companies)
[13:00:47] <skunkworks_> right
[13:01:02] <skunkworks_> I am even on the second cup..
[13:03:48] <SWPadnos> post #25 is interesting
[13:04:06] <SWPadnos> ... "I am not saying it MUST be as good as the SS- it's slightly cheaper, and the SS is buggy and every day appears more and more to be a dead product since they appear to have abandoned the driver development in "Beta" state." ...
[13:05:28] <SWPadnos> also post 26, if the guy is right about USB frame make-up
[13:07:08] <SWPadnos> ah, and post #36 tells it all
[13:09:52] <SWPadnos> well, I may not be diplomatic before I've had coffee, but I think I made the point. :)
[13:09:54] <SWPadnos> shower time
[13:11:38] <micges_work> hehe
[13:54:46] <skunkworks_> totally written from scratch.
[14:01:59] <jepler_> jepler_ is now known as jepler
[19:28:46] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[20:52:22] <robh_> cradek, did you get your spindle and vfd working?
[21:05:13] <cradek> yes
[21:05:27] <robh_> does it work well?
[21:05:42] <cradek> yes
[21:05:45] <robh_> whats it like holding a position like in tool change
[21:07:20] <robh_> was it a bolder drive u had?
[21:07:51] <cradek> this machine has a mechanical orient, so the vfd doesn't do anything except turn slowly until the mechanical stop engages
[21:08:18] <cradek> the old drive was yaskawa early 80s technology
[21:08:54] <cradek> I wouldn't want to try to orient the spindle with a regular vfd - they don't generally have position mode
[21:09:04] <cradek> you'd need an encoder and pid, and even then it probably doesn't work very well
[21:09:43] <cradek> mechanical orient = yay
[21:10:10] <robh_> aaah i see, i through that yaskawa used the old mag sensor to rotate to postion
[21:10:27] <cradek> I see they had that option, but this mill didn't have it
[21:10:55] <robh_> yes, this is one problems we are fiding talking to techs but hard to get out of them what they are selling
[21:11:12] <robh_> did you go for a open loop vector drive ? or is it closed loop
[21:11:24] <robh_> taking it was AC motor or was it dc
[21:11:28] <cradek> I'm running it in v/f mode
[21:11:38] <cradek> the drive will do sensorless vector but I'm not using it
[21:12:34] <robh_> you find in that mode u get good speed match (command to atchal)
[21:12:48] <cradek> I do not currently have a way to measure it
[21:13:03] <cradek> soon I will have encoder (resolver actually) feedback from the motor to emc
[21:13:17] <jepler> the drive doesn't have a mode that reads out motor rpm?
[21:13:22] <robh_> sorry with all questions just looking into vfd for mill and its allways nice to talk to someone thats been down road so to speak
[21:13:46] <robh_> jepler, it was just be a estimated one with no encoder back to drive
[21:13:58] <cradek> you can only read frequency =~ rpm
[21:14:06] <cradek> robh_: don't mistake me for a motor expert :-)
[21:14:21] <robh_> in vecter mode they work out where motor is in terms of magnets etc and get speed from tht
[21:14:40] <robh_> thats fine cradek i woudlt call most of the sales mans we spoke to experts ether hehe
[21:14:43] <cradek> not really - in real vector mode, you have encoder/resolver feedback that the drive uses
[21:14:49] <skunkworks_> cradek: is the resover is a good place?
[21:14:54] <skunkworks_> resolver
[21:14:55] <cradek> in sensorless vector, magic happens
[21:15:06] <cradek> skunkworks_: no, it's on the motor, which is belted to the spindle (non toothed)
[21:15:17] <cradek> I hope it will be good enough for tapping.
[21:15:18] <robh_> thats it about 20% speed error they say in sensorless, and %1 better in closed
[21:15:32] <robh_> Vbelt?
[21:15:39] <skunkworks_> cradek: your going to give it a try? cool
[21:16:02] <cradek> no, it's the flat-with-lots-of-little-V-things (like a modern car serpentine)
[21:16:02] <skunkworks_> is there a gearbox?
[21:16:15] <cradek> no, only one belt ratio
[21:16:29] <cradek> yeah I bet it is good enough for tapping. it's not like it will slip.
[21:16:31] <robh_> aaah i see
[21:16:57] <skunkworks_> I bet it will work great
[21:17:12] <cradek> skunkworks_: hope so. if not I'll get a t/c tapper from maritool :-)
[21:17:14] <robh_> cradek, will be rigid tapin at 6000rpm no problem soon ;)
[21:17:23] <cradek> wellll 1000 rpm maybe
[21:17:37] <cradek> I'd really need a braking resistor to go faster
[21:17:41] <robh_> aaah
[21:17:50] <robh_> yes all that start stop dump DC buss
[21:18:20] <cradek> currently I tap at 150 rpm with the tapmatic on the bridgeport - ugh
[21:18:28] <robh_> whats ramp like at moment then?
[21:18:44] <cradek> it's still the default 5s to full speed, which sounds very slow
[21:18:54] <cradek> the old drive (when it worked) would do it in about 1s
[21:18:58] <robh_> 5s thats not bad with no feedback id say
[21:19:11] <skunkworks_> cradek: did you connect with jmk?
[21:19:12] <cradek> I haven't tuned it much except for the basic v/f settings
[21:19:18] <robh_> yes i woudl said around 2s or just under to 6k (if goes up there) like our old fanucs
[21:19:20] <cradek> skunkworks_: yep he was a big help as usual
[21:19:27] <skunkworks_> :)
[21:20:03] <robh_> does ur changer drop pocket in a sep step on tool change or is it cam linked in arm action
[21:20:34] <cradek> pocket up/down is air and totally separate from the arm
[21:21:18] <robh_> that should make for a nice rapid change time, as on our random machine the pocket is linked with tool crab/arm so speed it up and it starts droping the tools out from the momentum of pocket coming down
[21:21:46] <cradek> the arm and carousel are only one speed, run by 3ph ac motors
[21:22:08] <cradek> arm is only 1-2 seconds
[21:22:18] <robh_> i see what they do these days is drop pocket down ready for arm while Z is going home, then arm is speed limit in the change of tool time
[21:22:20] <cradek> orient 1 second maybe
[21:22:59] <cradek> that's a good idea but this machine also has to move the table splash guard out of the way before dropping the pocket
[21:23:15] <robh_> get the angel grinder out ;)
[21:23:54] <robh_> be very nice to seed a vid of it in action soon
[21:24:06] <cradek> I bet it will do < 10sec chip-to-chip
[21:24:27] <skunkworks_> what shuttle rate are you up to?
[21:24:52] <robh_> thats nice speed realy, ok its no 1.2sec by todays specs but hay
[21:24:53] <cradek> rapid is 450 ipm. I had it at 750 but had trouble with velocity loop stability.
[21:25:20] <cradek> hardly makes any difference with 22" or so max travel
[21:25:21] <robh_> 11m/min
[21:25:22] <skunkworks_> yikes
[21:25:36] <skunkworks_> do you remember the accelleration?
[21:25:40] <cradek> 30
[21:25:44] <robh_> so ment to be a 20m machine
[21:25:46] <skunkworks_> heh fun
[21:25:56] <cradek> 450 is still WAY too fast to hit estop before something happens
[21:26:11] <robh_> lol
[21:26:14] <robh_> wait till u use a 40m machine
[21:26:16] <skunkworks_> still think there should be a jaw stop
[21:26:25] <robh_> u turn rapids down and u still doing 20m or 10m
[21:26:26] <cradek> clench sensor
[21:26:37] <skunkworks_> heh
[21:27:09] <cradek> robh_: with emc2 you can set max velocity wherever you want for proofing - even down to zero
[21:27:24] <cradek> I often run it with the wheel and check each approach after messing with offsets
[21:27:35] <robh_> yes i like the way u have unlimited rapid speed in EMC
[21:27:41] <cradek> the 25% 50% rapid overrides are almost useless IMO
[21:27:56] <cradek> (on a fast machine)
[21:27:57] <robh_> i turn rapids down to 1/2m and leavs me free to inch in on feed
[21:28:04] <cradek> it just makes the crash a little less bad
[21:28:46] <robh_> yes, on fanucs im always on FO which is normal around 1 or 2m max then use feed override as that gos to zero which stops machine and inch it in etc on proove out, hardly use 25 and 50
[21:30:05] <robh_> these days cam proves most of concept out for us, but cam sims never run generated code so can never trust it but EMC shows up any big gcode probs just if it showed operater errors lol liek wrong tool length
[21:30:55] <cradek> ah so the use FO for rapid too
[21:31:05] <cradek> they
[21:31:27] <cradek> heh, yeah sometimes in EMC you see a rapid line going through the rest of the work - bad sign - good to know
[21:31:40] <robh_> no zero speed on rapid they call it FO meaning force off, whic is user settable but default is 2m/min generaly
[21:32:07] <robh_> feed over ride goes to zero which stops everything, but has no effect on rapid cap, visa versa just as EMC does it
[21:32:36] <robh_> i think the new fanucs and others are same still 18i etc
[21:33:14] <robh_> i know mazatrols are thats why its pain on 40m machine + as 25% is still scary fast
[21:33:23] <robh_> as u say
[21:35:09] <robh_> aah yes while u around (hope dont mind) other night i asked about lathe with twin turret if u recall
[21:35:18] <cradek> sure
[21:35:36] <robh_> kind missed each other, anyways
[21:35:49] <robh_> if i call buttom W emc can apply tool lengths to that?
[21:36:40] <robh_> as its only a Z axis if u like. as apsoed to top which is Z & X but i dont wish to use both at the same time. but both are 8 stations
[21:37:06] <alex_joni> if you don't use both at the same time, then it's easiest to call both Z
[21:37:12] <alex_joni> and have the separation done in HAL
[21:37:28] <robh_> didt know if i could do that
[21:38:14] <cradek> yeah I think alex is right
[21:38:33] <robh_> right now they are bothed called Z on old control, but then it has two brains so they could do it
[21:38:35] <cradek> you can have tool length offsets on X and [Z or W]
[21:38:43] <cradek> not all three
[21:38:59] <robh_> right
[21:39:00] <alex_joni> so you could have X&Z and a separate config X&W
[21:39:10] <cradek> if you move to tool change position and make it machine zero before switching, I think it'd be possible to do in hal
[21:39:36] <cradek> just set commanded to zero for the one you don't want to move. hook emc's Z command to the one you DO want to move now
[21:39:45] <cradek> you could base it off "tool in spindle" number which is available in hal
[21:39:46] <robh_> alex_joni, mean by sep config (just confirm we talking same)
[21:40:07] <cradek> make sure emc's command is zero before you switch!
[21:40:15] <cradek> otherwise you will get a jump and a following error
[21:40:47] <cradek> this is just a matter of a few hal muxes
[21:41:05] <cradek> * cradek waves his hands
[21:41:11] <cradek> details left to the student, haha
[21:41:15] <robh_> hmm
[21:41:16] <robh_> hha
[21:41:39] <robh_> i do know some time later i will need to use both at once
[21:41:42] <cradek> if you wanted to move both at once you have a MUCH harder problem, and you'd probably end up with two controls like before
[21:42:03] <cradek> some kind of synchronization ("ready! go!") between them, but otherwise separate
[21:42:25] <robh_> yea in 80s they needed the two becasue what they where doing etc
[21:42:35] <robh_> these days its all in 1, the just call bottom ones a diff letter
[21:42:50] <cradek> sure, but I don't see how you'd program it though
[21:43:05] <cradek> how would you do a drill cycle with one while turning a profile with the other?
[21:43:12] <cradek> seems like that's the kind of thing you'd want to do
[21:43:13] <robh_> Z and X as normal, then cal in W say
[21:43:24] <cradek> or, boring with one, outside turning or threadign with the other
[21:44:40] <cradek> shat do you mean cal?
[21:44:46] <cradek> w
[21:45:15] <robh_> just sec
[21:49:45] <jepler> there would be a number of things made easier by a gcode that says "expect a position change and wait for input"
[21:49:55] <jepler> details of that left to the student as well
[22:11:54] <robh_> trying find exsample but failing
[22:12:47] <cradek> for what?
[22:26:17] <robh_> gcode and axes exsamples
[22:26:38] <robh_> that is to work with one turretn and 2nd as u say
[22:27:50] <robh_> but to drill with 1 , then wait for it to finish, turn with 2nd, etc is good enough, to work on multi axes and spindle thats quite a bit of kit, £500+lathe not sure who has one near to us might know of one person i can ask
[22:29:24] <robh_> is a good question tho what does Gcode do, id exspect they might split it up on some controls then u just have Mcodes that wait for one to finish & start etc to talk to each other. abit like how our old control worked, not ment to turnand drill at same time on both but u drill with one, give a mcode to say done with that program. then top one comes in does it job, cycle goes on
[22:30:09] <cradek> for one at a time, except for the tool length offset problem, you would just use different axes with EMC
[22:30:12] <robh_> sliding head is a good exsample of multi axes lathe, two slides, power tooling, multi spindle etc
[22:30:18] <cradek> an X,Z,W machine would make fine sense
[22:30:28] <robh_> yes that i am fine with i can hack that,
[22:30:43] <cradek> you could do a drill cycle in W, then when it's done, do some other profile turning in XZ, etc
[22:31:04] <cradek> but doing both at once is not currently the least bit how our gcode works...
[22:31:08] <robh_> thats all id be looking for
[22:31:26] <robh_> no, that is why i am realy intrested to see how they do doit, u got me thinking now
[22:31:51] <robh_> if i could drill, tap in W like Z thats perfect
[22:32:08] <robh_> wait for W to finish before using X Z no problem, and visa versa
[22:32:12] <cradek> I think you can tap in W - not sure
[22:33:26] <cradek> nope
[22:33:33] <cradek> Z only
[22:33:41] <cradek> "simple matter of programming"
[22:33:57] <cradek> that would be nice to have for 5 axis rigid tapping too
[22:33:58] <robh_> is there a W offset in tbl?
[22:34:13] <cradek> you can trade Z,W but not have both
[22:34:20] <robh_> 5 axes rigid now that would give bigger exspenive controls run for money
[22:34:31] <cradek> W,X or Z,X tool lengths only (currently)
[22:34:38] <robh_> peck drilling in W?
[22:34:44] <cradek> we have that already
[22:34:54] <robh_> rigth so adding W coloum would make EMC cry as is
[22:34:57] <cradek> G17.1 G83 Z-1 ...
[22:35:05] <cradek> err
[22:35:11] <cradek> G17.1 G83 W-1 R... P... F...
[22:35:46] <robh_> so its pritty much all there minour few things
[22:35:48] <cradek> if I messed with tool length offsets again, I'd add them ALL
[22:36:03] <cradek> but how to specify the tool table is a big problem then
[22:36:09] <cradek> that's the real reason it's not there
[22:36:18] <robh_> oh just need two cones for two axis ;) but thats nit pickin
[22:36:36] <cradek> now you're just getting silly :-)
[22:37:02] <robh_> maybe need to drump humand editable for machine readable, so only GUI can read write
[22:37:13] <cradek> yes I think that's the answer
[22:37:27] <cradek> but lots of guis to fix
[22:37:49] <cradek> it's hard to give up the easy hand-editable we have now
[22:38:12] <robh_> yes i do understand
[22:38:41] <robh_> cant just take something away from ppl and exspect them all to love u for it or brake ppls setups etc
[22:38:49] <cradek> right
[22:38:54] <alex_joni> robh_: I think micges had a version of AXIS with 10 cones
[22:39:01] <cradek> that's the problem I have with my random toolchanger stuff
[22:39:10] <alex_joni> he has one machine with lots of Z's
[22:39:11] <cradek> it's different enough to make it unpopular
[22:39:15] <robh_> i throught i saw a screen shot of it on here, but i didt catch waht it was about
[22:39:24] <alex_joni> 10 or so
[22:40:01] <robh_> i keep meaning to grab latest git and try ur random code out cradek see how it functions etc
[22:40:18] <robh_> did you add a tool error input to stop emc tool change loop?
[22:40:30] <cradek> jeff has done that, but it's not working quite right yet
[22:40:42] <cradek> "Failed to load tool 6" [program aborts]
[22:40:52] <cradek> we'll get it one of these days
[22:41:19] <cradek> off to find dinner
[22:41:21] <robh_> thats grate, i made my PLC ready when i did to reject tools that dont exsist or any change error problem to output error pin, but right now it has to finish the change ok
[22:41:26] <cradek> nice chatting
[22:41:35] <robh_> sure thx for ur time cradek
[22:41:42] <robh_> i better get some sleep