#emc-devel | Logs for 2008-10-16

[00:32:08] <SWPadnos> Ian Wright's problem may be an extraneous unit conversion
[00:33:04] <SWPadnos> the difference between the two angles (68.023 and -1659.772) is 1727.8, which (possibly coincidentally) is 68.03*25.4
[00:33:32] <SWPadnos> (depending on rounding)
[00:58:27] <jepler> OK, it is a metric problem
[00:58:36] <jepler> changing "inch" to "mm" in sim_9axis.ini, you can see it
[00:59:32] <jepler> you can see it most clearly by going to A1, then touch off A, then enter 0 -- you get 24.4
[01:17:44] <jmkasunich> units suck
[01:24:19] <fenn> I disagree sir
[01:24:38] <fenn> it's redundant legacy units that suck
[01:40:45] <jmkasunich> you gonna buy me $1000 worth of measuring instruments?
[01:41:13] <jmkasunich> when you do, you can call my favorite units "legacy", till then, inches dammit!
[01:50:27] <cradek> jepler: how did that break?? do you see it?
[02:21:45] <cradek> fenn: first thing I had to do was convert cxf2cnc to inches :-)
[02:23:01] <cradek> and all uppercase, and %.4f everywhere...
[02:37:58] <LawrenceG> chris.... maybe a new mode for engrave.py would be useful.... a dumb mode where no vars or subs or math is used.... just g0 and g1 lines
[02:38:22] <cradek> LawrenceG: I don't know what engrave.py is...
[02:39:07] <LawrenceG> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators
[02:39:18] <LawrenceG> bottom of page
[02:40:17] <cradek> oh it's a gui for fenn's program? or something else?
[02:40:37] <fenn> gui
[02:40:42] <cradek> neat
[02:40:44] <LawrenceG> basically a gui with rotation, offsetts mirroring etc
[02:41:15] <LawrenceG> emc specific due to runtime rotation and subroutines
[02:42:24] <cradek> LawrenceG: it maybe still wouldn't work exactly... dialects can be weird
[02:42:33] <LawrenceG> it could use a checkbox to give dumb code to be super compatible
[02:42:49] <LawrenceG> g0 and g1.... maybe your dialect would be a good example
[02:42:52] <cradek> for example X1.23456 (one thing cxf2cnc was generating) causes my mill to freak out
[02:43:22] <cradek> also, G1 X1.2 Y2.3 F3 will not move
[02:43:42] <cradek> and G1 F3.000 will cause it to go at rapid speed :-/
[02:45:07] <fenn> your mill is stupid :)
[02:45:12] <cradek> yep
[02:45:34] <cradek> it's like the dilbert character tina the brittle tech writer. it wants to make a fuss about everything.
[02:46:32] <LawrenceG> I guess its time for an emc upgrade!
[02:46:41] <cradek> but I'd be hard pressed to make a better cnc control fit into a couple Kbytes burnt on a few eproms...
[02:46:55] <cradek> it's completely amazing that these old machines worked
[02:47:39] <fenn> i suppose it's better than control cams
[02:47:50] <LawrenceG> yea... the old scr servo controls amaze me.... pretty low bandwidth, but much better than cranking the wheels by hand
[02:47:57] <cradek> heck it has features that I think would be hard for me to put in emc2 and get right
[02:48:13] <cradek> s/I think//
[02:48:39] <cradek> (never mind that they might not be what I would consider "right", in the BOSS)
[02:48:43] <LawrenceG> can you guys get to imagebin.ca at the moment?
[02:48:57] <cradek> no
[02:49:19] <LawrenceG> pings work, but no web connections
[02:49:35] <LawrenceG> server must be down...
[02:51:02] <LawrenceG> where else can I stash a picture?
[02:51:41] <cradek> dcc it to me
[02:51:45] <cradek> I'll put it up
[02:52:17] <LawrenceG> dmess is trying to hook up a VFD without a manual..... was going to send him a typical connection diagram
[02:52:24] <cradek> ah
[02:52:29] <cradek> email then, maybe
[02:53:56] <LawrenceG> looks like he had to go... tomorrow I guess
[03:01:56] <jepler> cradek: no, I got distracted by the fact that my "stable" 2.2 tree won't build with some odd error
[03:02:00] <jepler> then the debate came on
[03:02:02] <jepler> so I haven't found or fixed it yet
[03:03:26] <jepler> if 210 in s.gcodes:
[03:03:26] <jepler> scale *= 25.4
[03:03:26] <jepler> p0 *= 25.4
[03:03:32] <jepler> this isn't conditional on the axis being a linear one
[03:09:47] <jepler> ah s*** -- I have unrelated stuff in my working file
[03:10:08] <jepler> here's the patch: http://pastebin.com/m5761e646
[03:10:22] <jepler> if someone else can test and commit it before I get a chance to commit it, be my guest
[03:10:25] <jepler> should go in 2.2 and TRUNK
[03:11:26] <jepler> test sim_9axis.ini, and the same modified for metric (which is the case where the original bug existed)
[03:11:40] <jepler> test all axes; test with inch or mm display; test with g20 or g21 as active modal code
[03:31:04] <SWPadnos> argh.
[03:31:26] <SWPadnos> it would be very nice if stepconf would show the status of parallel port pins on the setup page
[03:31:36] <SWPadnos> but I'm not sure how to do that well
[03:32:08] <SWPadnos> I don't think it would be good to load the RT system when stepconf is run
[03:32:30] <SWPadnos> I guess a simple parallel port read could be used, since stepconf only supports the parallel port
[03:32:30] <jmkasunich> doesn't stepconf do that during the "test" step?
[03:32:40] <SWPadnos> no, it only moves the selected axis
[03:32:52] <SWPadnos> AFAIK
[03:33:01] <jmkasunich> but it uses stepgen to do that, right?
[03:33:10] <SWPadnos> siggen and stepgen I think
[03:33:23] <jmkasunich> so there must be threads and all the other RT stuff going
[03:33:56] <SWPadnos> loaded when the test window gets loaded
[03:34:21] <jmkasunich> so display parport status when the test window is open
[03:34:24] <SWPadnos> you can run stepconf (except for the test window) on a pure sim install
[03:35:02] <SWPadnos> that could be done, but the real value would be to see the status of inputs when you're configuring switches
[03:35:13] <SWPadnos> err - you know what I mean
[03:35:17] <jmkasunich> yeah
[03:36:15] <jmkasunich> simple reads (a few per second) would be enough to display the port status, and let you manually poke at switches and see the bits change
[03:36:25] <jmkasunich> but I don't think stepconf as a user process can do inb()
[03:36:33] <SWPadnos> dunno
[03:37:00] <SWPadnos> but running on sim is relatively important IMO, so even the read might not be desirable sometimes
[03:37:17] <SWPadnos> (if the user has a printer for example, it may not be possible for a user app to read those ports)
[03:37:28] <jmkasunich> agreed
[03:37:44] <jmkasunich> reading the port is alot like test mode - needs to be optional
[03:37:55] <SWPadnos> like I said, I'd like to see it, but I'm not sure how to do it ;)
[03:42:55] <LawrenceG> jepler, I think the fellow on the emc list said he is running version 4.51 --- that is a BDI version supplied by Sherline is it not?
[03:43:03] <jmkasunich> yes
[03:43:36] <jmkasunich> pre-installed, pre-configured
[03:43:45] <LawrenceG> does it use hal?... I thought it was still emc1
[03:43:59] <jmkasunich> such that the people who have it are totally clueless if they try to use EMC2
[03:44:01] <jmkasunich> it is EMC1
[03:44:07] <jmkasunich> no HAL
[03:44:38] <LawrenceG> sounds like he will be a little confused getting emc2 advice :}
[03:45:16] <LawrenceG> if its preconfigured, I am not sure why he is messing with home/limit switches
[03:45:27] <jmkasunich> if you mean John Domville, he is replacing the EMC1 with EMC2
[03:45:59] <LawrenceG> ah ok... makes sense now
[03:46:26] <LawrenceG> how is jmk tonight?
[03:46:45] <jmkasunich> getting ready to walk the dog and get some sleep
[03:47:12] <LawrenceG> hey... my dog just brought me home from a walk in the rain
[03:47:50] <LawrenceG> good cnc dreams
[03:48:10] <jmkasunich> goodnight
[07:19:35] <micges> what is difference between emc with servo-thread at 1kHz and 1.4kHz ?
[07:20:41] <micges> what is expected in emc behaviour ?
[07:54:08] <fenn> I term in PID will be too low, D term too high
[07:57:17] <micges> what is term ?
[07:58:14] <fenn> the number
[07:58:38] <micges> ok
[08:00:01] <fenn> if it were me, i would just re-tune it from scratch
[08:01:46] <micges> thanks, I'll do this, just didn't know that servo period has impact to
[08:01:48] <micges> PIDs
[13:10:05] <SWPadnos> PID coefficients are not dependent on the PID function execution rate
[13:11:12] <SWPadnos> you shouldn't need to re-tune if you change the servo thread rate, unless you have extremely fast servos that actually need the faster rate (and you have to "de-tune" them for the slower thread rate)
[13:13:06] <micges> hmm
[13:13:31] <micges> I don't know speed of servos but Im drive them by analog signal
[13:13:41] <micges> velocity +- 10V
[13:15:20] <SWPadnos> I was just pointing out that the thread rate has no effect on PID coefficients - it's aware of how fast it's running
[13:15:31] <SWPadnos> so derivatives and integrals are correctly calculated
[13:17:26] <micges> I see , thanks
[13:18:12] <cradek> my experience is that changing the period changes the tuning
[13:18:37] <SWPadnos> unless you have very fast-response servos, I'd call that a bug
[13:19:17] <cradek> I don't know the cause, but it's pretty evident
[13:21:37] <SWPadnos> I wonder how to test that
[13:23:05] <micges> I can on real machine tomorrow but I don't know what to look for
[13:23:24] <SWPadnos> yeah, that's what I was wondering about :)
[13:23:31] <cradek> SWPadnos: if it's a purely software effect, you could test it in servo_sim.
[13:23:32] <SWPadnos> what to look for and how to measure it
[13:24:21] <SWPadnos> I guess you can look at PID output - "jagged" traces might indicate a problem
[13:24:39] <micges> jagged
[13:24:41] <micges> ?
[13:24:43] <SWPadnos> simulating a load isn't trivial though (since accel is synced to the PiD)
[13:24:45] <SWPadnos> not smooth
[13:27:34] <micges> ok noted
[13:28:44] <micges> other question: it is possible to jog with keyboard and joystick togheter
[13:28:45] <micges> ?
[13:29:04] <alex_joni> what do you mean together?
[13:29:11] <alex_joni> at the same time?
[13:29:15] <micges> yes
[13:29:17] <alex_joni> no
[13:29:31] <alex_joni> the jog commands come from different locations
[13:29:36] <alex_joni> and only one is accepted
[13:29:56] <alex_joni> so if you start a joystick jogging, a keyboard jogging will stop the joystick one
[13:30:07] <micges> ok and jog x with kb and jog y with joystick ?
[13:30:19] <alex_joni> it's the same message for any jog
[13:30:34] <alex_joni> doesn't matter if X or Y or XYZABCUVW
[13:30:43] <micges> ok thanks
[13:31:00] <alex_joni> any reason why you need that?
[13:31:23] <micges> I think I dont need
[13:31:33] <micges> new laser have joystick to jog
[13:31:39] <micges> and have kb too
[13:32:02] <micges> I wonder if they can cooperate safely
[13:32:06] <alex_joni> they can
[13:32:08] <SWPadnos> you can use both, just not at the same time
[13:32:20] <alex_joni> you can try but only one of them will work
[13:32:27] <alex_joni> as in .. bad things shouldn't happen
[13:32:50] <micges> if its only stopping actual jog its ok
[13:33:14] <alex_joni> I guess the most precise answer is "try it out" :)
[13:33:22] <alex_joni> bbl.. running home
[13:33:33] <micges> will be tested
[13:34:10] <micges> I writing interface to it now and I wondered if I must lock keyb jog when using joy
[15:59:31] <skunkworks> doesn't anyone do research anymore?
[15:59:43] <SWPadnos> no
[15:59:54] <SWPadnos> it's not profitable]
[15:59:57] <SWPadnos> -]
[16:00:11] <skunkworks> I meant - joe emc.
[16:00:17] <SWPadnos> yeah, I know :)
[16:00:20] <skunkworks> ;)
[16:00:24] <SWPadnos> (it's still not profitable ;) )
[16:00:40] <skunkworks> heh
[16:03:30] <SWPadnos> I've been trying to figure out how to be diplomatic in responding, but haven't managed it yer
[16:03:32] <SWPadnos> yet
[16:04:05] <SWPadnos> on another note - man, the Ubuntu 8.10 server image is almost as big as the desktop image
[16:04:17] <SWPadnos> ~620M vs. ~680M
[16:04:41] <SWPadnos> pr maybe ~690-700M for the desktop ISO
[16:04:43] <SWPadnos> gah
[16:04:45] <SWPadnos> or
[16:40:30] <issy> hi all
[17:33:39] <skunkworks> I am assuming the sense voltage will not really relate to exact current limit?
[17:34:35] <skunkworks> I actually - I probably need a better current meter
[17:34:48] <SWPadnos> you need a scope
[17:36:16] <skunkworks> I have one.
[17:36:39] <skunkworks> that is how I found out the current limit was tripping when it shouldn't be.
[17:42:35] <skunkworks> cool - the voltage across the sense resistor seems to clip exactly at the set voltage. Go figure ;)
[17:43:04] <skunkworks> I love the drive enable..
[17:43:22] <skunkworks> (75mv)
[18:03:09] <skunkworks> http://www.electronicsam.com/images/KandT/ampmessagain.jpg
[18:06:32] <alex_joni> skunkworks: 2-3 more iterations and it'll look like this: http://www.strippittech.com/PWMC8-2.jpg
[18:06:56] <skunkworks> na - that is too many parts..
[18:07:25] <skunkworks> I think I draw the limit at 40 or so. ;)
[18:07:36] <alex_joni> :P
[18:16:42] <alex_joni> I wonder where Kirk Wallace got OUTPUT_SCALE from
[18:17:58] <SWPadnos> non-stepconf configs :)
[18:21:23] <alex_joni> even that.. for steppers it's useless
[18:21:31] <alex_joni> and on servo's it doesn't have much benefit either
[18:21:33] <SWPadnos> yep
[18:21:45] <alex_joni> it has been dropped since before 2.1.0 I think
[19:41:04] <CIA-40> EMC: 03cradek 07TRUNK * 10emc2/src/emc/motion/command.c: give errors of the style "Linear move on line 9 would exceed joint 2's positive limit" instead of the existing error which often leaves you befuddled about what's actually wrong.
[19:41:30] <SWPadnos> cool
[19:43:13] <skunkworks> bfuddled. I lik it.
[19:43:15] <skunkworks> like
[19:43:33] <cradek> less befuddlement is always good
[19:44:26] <skunkworks> more use of the word befuddlement is always good.
[19:44:37] <cradek> ha
[19:44:38] <SWPadnos> elmer befuddle
[19:44:44] <skunkworks> :)
[19:45:05] <skunkworks> cradek: so far so good. No smoke. Hopefully this weekend I can put the screws to it.
[19:45:13] <cradek> nifty
[19:45:16] <SWPadnos> just make sure they don't touch each other
[19:45:21] <cradek> when can I get my order in?
[19:47:31] <skunkworks> I can sell you the eagle files for work already served on emc2..
[19:47:55] <skunkworks> cradek: what are the volts/amps for your lathe servos?
[19:48:21] <cradek> skunkworks: I don't remember...
[19:48:38] <cradek> actually I don't need any amps - both my machines have nice working amps now
[19:48:57] <skunkworks> nice - did you get the tach working then?
[19:49:05] <cradek> yes
[19:49:05] <skunkworks> for the one lathe axis?
[19:49:10] <skunkworks> what was the issue?
[19:49:30] <cradek> I think the slightly loose screw (backlash, not me) messed up the amp tuning
[19:49:45] <cradek> I told the amp to relax a bit (added a resistor)
[19:49:53] <skunkworks> nice
[19:50:16] <skunkworks> I got the cute 10k trimmer resistors today. 12 turn.
[19:50:18] <cradek> if I ever manage to get exactly zero backlash on it, I might try taking that resistor back off
[19:50:28] <skunkworks> ah - I see
[19:50:37] <cradek> not because it needs it, but because I'm curious whether that was the original cause
[19:50:49] <cradek> the performance is really good the way it is
[19:50:56] <skunkworks> neat
[19:51:04] <skunkworks> still looking for some videos
[19:51:11] <skunkworks> ;)
[19:51:26] <cradek> I need to start blogging again
[19:51:34] <skunkworks> :)
[19:52:27] <skunkworks> nobody seems to be blogging/
[19:53:04] <cradek> it's thankless work
[19:53:44] <skunkworks> Yes. But it gives me something to look at. so Thank you. ;)
[19:53:54] <cradek> :-)
[19:54:47] <skunkworks> So - bets on if 150v 20a will smoke the thing?
[19:55:27] <cradek> got heat sinks?
[19:55:36] <skunkworks> not mounted yet.
[19:55:39] <skunkworks> but yes.
[19:55:55] <skunkworks> I think I had bought some at last years emcfest also
[19:56:27] <skunkworks> for sure it looks like the drives will work great for the puma.
[19:57:28] <cradek> that would be slick
[19:57:35] <cradek> did you decide you need to replace the encoders?
[19:57:54] <skunkworks> if they are sin - yes.
[19:58:27] <skunkworks> which is what I have read that they are.