#emc-devel | Logs for 2009-04-17

[03:28:57] <garage_seb> the stepper motors stutter a bit with maxaccel=0... :-(
[03:29:35] <garage_seb> i think they can't handle the per-servo-loop accelerations, especially at high speed (cruis)
[03:29:39] <garage_seb> +e
[03:30:38] <SWPadnos> is the AXIS_N setting correct for the motors/drives?
[03:31:19] <garage_seb> step timing and [AXIS]MAX_VELOCITY are correct for the stepper driver
[03:31:25] <SWPadnos> hmmm
[03:31:57] <SWPadnos> what happens if you lower the [AXIS]MAX_ACCEL a little?
[03:35:36] <garage_seb> it's maybe a little better
[03:35:50] <garage_seb> but the problem is not during accel or decel, it's during cruise
[03:36:16] <SWPadnos> oh. interesting
[03:36:38] <SWPadnos> does a ddt on the position output show anything?
[03:36:45] <SWPadnos> (should be a nice flat line)
[03:37:45] <SWPadnos> when there's no accel limit, the stepgen shouldn't bother doing any accel math at all - it should just calculate the number of steps needed in one period, and set the rate generator to that number
[03:37:50] <SWPadnos> give or take 1 :)
[03:38:16] <garage_seb> subject to maxvel
[03:38:17] <garage_seb> yes
[03:39:15] <garage_seb> ddt of pos-fb isn't perfectly flat, i think it's having a hard time matching both vel & pos, so it weaves around a bit
[03:39:42] <garage_seb> i think this "weaving" is causing more accel than the motor can deliver
[03:40:33] <SWPadnos> I was thinking you should look at cmd-pos, just to be absolutely sure you know it's not the TP
[03:40:40] <SWPadnos> (I'm sure it's not, but it helps to be sure)
[03:40:51] <garage_seb> it's not TP, it's me ;-)
[03:40:57] <SWPadnos> heh
[03:41:38] <garage_seb> i was looking at axis.N.joint-vel-cmd, which i think should be the same as (or better than) ddt(axis.N. position-cmd)
[03:41:58] <garage_seb> joint-vel-cmd is rock-steady at the configured [AXIS] maxvel
[03:42:00] <SWPadnos> well, it should be, but that's not the input to stepgen
[03:42:13] <SWPadnos> I'd look at a ddt on joint-pos-cmd, which I'm sure is also rock steady
[03:42:31] <SWPadnos> but it's like metering your lab power supply voltage before troubleshooting your circuit - just peace of mind
[03:42:45] <garage_seb> yeah, good point
[03:42:58] <cradek> * cradek hides
[03:43:13] <SWPadnos> noworries, I'm sure it's not your fault
[03:43:21] <SWPadnos> you can probably go (back) to sleep
[03:43:31] <cradek> yay
[03:43:41] <garage_seb> conveniently, my [AXIS] maxvel is 10, so i can tell by looking at datapoints in the halscope trace that it's incrementing by .001 for each 1 ms servo loop
[03:43:47] <garage_seb> it's not cradek, it's me
[03:43:48] <SWPadnos> heh
[03:46:33] <CIA-2> EMC: 03cmorley 07TRUNK * 10emc2/configs/common/configurable_options/pyvcp/parportpanel.xml: add parport test panel for pncconf
[03:47:09] <cradek> $2.99 for wifi!?
[03:48:06] <SWPadnos> cheap
[03:49:23] <cradek> just like the motel!
[03:49:40] <SWPadnos> heh
[03:49:46] <SWPadnos> let me know ho breakfast is :)
[03:49:54] <SWPadnos> hmmm. or not. you are in Iowa
[03:50:55] <cradek> I think breakfast is coffee, and is included in the $31 cost.
[03:51:08] <SWPadnos> ooooh. nice
[03:51:22] <SWPadnos> powdered or liquid cream, what do you think?
[03:51:42] <cradek> powdered vegetable oil, of course
[03:51:49] <SWPadnos> heh
[03:51:59] <SWPadnos> as long as there's sugar in it too, it should be all rigth
[03:52:13] <SWPadnos> (I kind of like the taste of Cremora, but not in coffee)
[03:52:13] <cradek> sorry - you guys were devel-ing emc...
[03:52:29] <cradek> necer heard of it
[03:52:32] <cradek> v
[03:52:40] <cradek> pardon, thumbs.
[03:53:36] <SWPadnos> oh. cremora is to cream as Coffee Mate is to milk
[03:57:16] <garage_seb> the possibility exists that i have shitty stepper motors
[03:57:26] <SWPadnos> heh
[03:57:36] <garage_seb> it's sitting loose on my workbench
[03:57:51] <garage_seb> if i grab the motor body when i command it to move, it moves
[03:57:59] <garage_seb> if i release the motor body, it just whines
[03:58:00] <SWPadnos> if pos-fb is jittery and pos-cmd isn't, then it doesn't matter if you have motors attached at all
[03:58:20] <garage_seb> true
[03:58:21] <SWPadnos> though they do make good audible indicators of performance
[03:58:24] <cradek> does the mesa encoder have an up/down mode?
[03:58:36] <cradek> if so you could loopback
[03:59:16] <cradek> (sorry if that's obvious...)
[03:59:22] <garage_seb> the encoder has up/down, and the stepgen has quadrature ;-)
[03:59:22] <SWPadnos> isn't there a wuadrature output mode for the stepgens?
[03:59:26] <SWPadnos> quadrature
[03:59:29] <SWPadnos> heh
[03:59:58] <cradek> heh, cool.
[04:00:39] <CIA-2> EMC: 03cmorley 07TRUNK * 10emc2/src/emc/usr_intf/pncconf/ (pncconf.py pncconf.glade):
[04:00:39] <CIA-2> EMC: Fix method: make_piname to add proper HAL pin name endings (such as .in_not)
[04:00:39] <CIA-2> EMC: Add the start of a test panel for the parport page
[04:03:37] <cradek> goodnight, folks
[04:04:24] <garage_seb> goodnight cradek
[04:04:49] <SWPadnos> see yiouy
[04:04:52] <SWPadnos> or you
[04:07:08] <garage_seb> have i mentioned yet tonight that i like halscope?
[04:07:20] <SWPadnos> hmmm. lemme check my scrollback :)
[04:37:32] <garage_seb> ugh - save it for another night
[12:20:07] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/images/ (signal-direction.odg signal-direction.png): add image
[12:23:59] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/basic_hal.lyx: expand a bit on signal direction
[12:25:27] <CIA-2> EMC: 03bigjohnt 07v2_3_branch * 10emc2/docs/src/hal/images/ (signal-direction.odg signal-direction.png): add image
[12:26:43] <CIA-2> EMC: 03bigjohnt 07v2_3_branch * 10emc2/docs/src/hal/basic_hal.lyx: expand on pin direction