#emc-devel | Logs for 2010-05-30

Back
[00:10:17] <pengui1> i need to confirm something: in velocity mode, instead of G1x100, the command is G1X100 F200 ? (F for speed)
[00:13:41] <andypugh> pengui1: I am not sure that question makes sense.
[00:14:06] <pengui1> how is the sintaxis in velocity mode?
[00:15:00] <andypugh> You are talking about a velocity-mode stepgen?
[00:15:05] <pengui1> yes
[00:15:41] <andypugh> Then it will create pulses at a rate determined by the signal sent to its velocity pin.
[00:16:30] <pengui1> so, i can specify the speed once
[00:16:38] <andypugh> So, if the velocity pin is wired in HAL to X-position then G1 X10 will cause it to ramp to 10 units per second velocity, and stay there. But that would be rather unusual.
[00:17:15] <pengui1> so, X10 means velocity=10 ?
[00:17:46] <andypugh> No, it means that axis.1.position-cmd moves to 10 at the set speed.
[00:18:13] <andypugh> (actually, even that is not necessarily true, because of coordinate offsets).
[00:18:45] <morfic> and because of acceleration/deceleration and distance not being enough it never reaches the set speed?
[00:18:46] <andypugh> What pin do you have wired to the stepgen velocity input?
[00:19:34] <andypugh> (I do rather think this is a #emc rather than #emc-devel duscussion, by the way)
[00:19:38] <pengui1> i like do something like: net x-velocity => stepgen.0.velocity-cmd
[00:20:37] <pengui1> assuming this:
[00:20:37] <pengui1> loadrt stepgen step_type=0,0 ctrl_type=v,v
[00:20:39] <andypugh> In that command, x-velocity is just a random signal name, it doesn't do anything.
[00:21:24] <andypugh> lets back up a bit. What are you actually trying to do?
[00:21:27] <pengui1> yes, i mean i in velocity mode
[00:21:35] <pengui1> rastering
[00:21:39] <pengui1> i need constant speed
[00:22:26] <pengui1> so, i need to confirm the syntax in velocity mode
[00:22:59] <andypugh> Ah, that again. Did you ever try M67?
[00:23:05] <pengui1> from X100 to X200: G01X100 F500, G1X200F500
[00:23:41] <pengui1> m67 is for digital output
[00:23:58] <andypugh> Analog output
[00:24:30] <pengui1> yes... i will need that later... first i need to solve the constant speed
[00:25:05] <andypugh> Your problem is (IIRC) that you were using spindle speed changes, and that was breaking the motion queue?
[00:25:47] <pengui1> i think my "delay" problem is the acceleration/deacceleration in each position
[00:26:01] <pengui1> this can be solved in velocity mode
[00:26:13] <andypugh> Did you try making very small Z moves to drive the intensity? (They have to be small, or you get cosine changes in X)
[00:26:21] <pengui1> yes, i try
[00:26:27] <pengui1> didn't works
[00:26:38] <andypugh> Very, very small Z moves?
[00:26:38] <pengui1> (works, but the delays remain)
[00:26:47] <pengui1> Z=0.0001 and Z=0
[00:27:23] <andypugh> That seems very odd.
[00:27:27] <pengui1> i only want to confirm the sintax....
[00:27:34] <pengui1> in velocity mode
[00:27:35] <andypugh> There isn't a syntax
[00:27:49] <andypugh> There is no "velocity" mode.
[00:28:01] <pengui1> yes, exist
[00:28:20] <andypugh> You can use velocity-mode stepgens, but that isn't the same as there being a "velocity mode" in G-Code
[00:28:40] <pengui1> i know velocity mode is in stepgen
[00:29:14] <pengui1> but the speed is set up with F command'
[00:29:18] <pengui1> ?
[00:29:44] <andypugh> Yes. But that is totally separate from the stepgen velocity.
[00:30:21] <pengui1> what do you mean'
[00:30:26] <andypugh> (How to describe something I barely understand myself)
[00:30:45] <pengui1> doesn't mind
[00:31:49] <andypugh> The G-code interpreter produces a list of motion commands. These eventually appear on a set of HAL pins listed here: http://www.linuxcnc.org/docview/html//man/man9/axis.9.html
[00:32:42] <andypugh> Then in the HAL file those pins are linked to a freely-chosen set of HAL components such as stepgens to do whatever you want to do.
[00:34:06] <andypugh> Typically you wire axis.0.joint-pos-cmd to stepgen.0.pos-cmd. But you don't have to.
[00:34:51] <pengui1> in this case it will be something like: axis.0.joint-vel-cmd to stepgen.0.vel-cmd
[00:35:43] <andypugh> With a servo system you wire the position-cmd output to a PID controller input, and that outputs a signal propertional to the position error. You can wire that to a stepgen velocity pin, and that makes sense.
[00:37:42] <andypugh> There is an axis.N.joint-vel-cmd pin, but I don't know if it will do what you want.
[00:38:24] <andypugh> (It's mainly for looking at in Halscope, to see what the axis is doing)
[00:38:52] <andypugh> But there is nothing to stop you trying it.
[00:39:22] <pengui1> yes... i'm searching sample hal programs
[00:39:37] <pengui1> there is a few examples in velocity mode
[00:39:43] <andypugh> net x-vel axis.0.joint-vel-cmd => stepgen.0.vel-cmd
[00:40:39] <andypugh> You will find that they almost certainly get the velocity out of a PID controller block that compares commanded position to the encoder position.
[00:46:01] <pengui1> in the emc-users people know this things?
[00:46:50] <pengui1> or is people that say "i bought a machine i don't know how install emc" ?
[00:47:37] <andypugh> no, this is the sort of discussion we get on emc. emc-devel is more for discussing exactly how to code new features or to fix bugs.
[00:48:05] <pengui1> ok... i will go to the other channel
[00:48:33] <andypugh> If you end up writing a component to look at current position and modulate laser power to suit, then that is new code, and here is the place.
[00:49:19] <andypugh> (well, that's may asessment of the split, I am open to be corrected)
[00:49:21] <pengui1> i'm doing something new... rastering is unknow here
[00:50:29] <andypugh> Yes, but you are trying to do something new with existing software components. If and when you write new components, here is the place.
[02:52:56] <CIA-1> EMC: 03jepler 07v2.4_branch * r89354bb9a045 10/ (VERSION debian/changelog): release 2.4.1
[02:53:03] <CIA-1> EMC: 03jepler 07v2.4_branch * rba0ccfb684cc 10/docs/src/drivers/pluto_p.lyx: fix lyx markup broken in 416b833
[02:58:26] <jepler> jepler has changed the topic to: EMC2 development -- http://linuxcnc.org/ | Latest release: EMC 2.4.1
[03:35:09] <cradek> Yayy!
[04:12:42] <morfic> cradek: you used nurbs on a cnc before, no? how well do they work right now?
[06:04:36] <Dave911> Cool .. 2.4.1 is out! Nice job guys!
[08:10:55] <alex_joni> jepler: cool