#emc-devel | Logs for 2007-05-14

Back
[00:00:18] <cradek> or write an arc in mdi
[00:00:41] <jmkasunich> that just tells you what EMC does
[00:00:49] <cradek> g0x0y0; g3x0y1i0j.5f99
[00:00:54] <jmkasunich> and I'd be rather astonished if EMC doesn't do what the docs say
[00:01:15] <cradek> well that's what petev said - that emc's docs are wrong
[00:01:53] <jmkasunich> apparently the CAM software allows you to specify "center to start" or "start to center", and that might be ambiguous, or just plain wrong.
[00:02:19] <jmkasunich> either way is better than looking at the cam input and emcs result and ignoring the g-code, and then saying EMC is wrong
[00:02:39] <cradek> I understand
[00:03:33] <cradek> going from "the cam makes me set a surprising setting before it works right" to "emc's docs are wrong" is quite a jump, that's all I was saying
[00:04:05] <jmkasunich> in that case, we're saying the same thing, just proposing different tests to see what the truth is
[00:06:08] <cradek> yep
[00:08:14] <cradek> I just permanently screwed the DB25 breakout thingy you gave me to the lathe. I figured it would be nice if I could unplug it from the driver box.
[00:08:58] <jmkasunich> the board that had green plastic DIN rail mounts?
[00:09:55] <petev> well, the way I would interpret things, it sure seems like the cam labels the vectors backwards
[00:10:08] <petev> cradek, did u get to testing the auto tune?
[00:10:55] <cradek> not quite yet - maybe in a few more days
[00:11:04] <cradek> I'm getting close to having this together
[00:11:10] <jmkasunich> we want photos!
[00:11:21] <cradek> you'll see it in a few weeks...
[00:11:28] <petev> what happened that it got all torn apart?
[00:11:33] <jmkasunich> we want photos NOW!
[00:11:36] <petev> did you decide to change motors or something?
[00:11:39] <cradek> petev: "upgrade"
[00:11:52] <cradek> it was too simple and straightforward (and working) before
[00:12:11] <petev> good reason for an upgrade ;-)
[00:13:10] <cradek> now I'm going to need another of those rare beasts that is a fully wired straight-through parallel port cable
[00:13:38] <jmkasunich> you're not actually using it to go from parport to anything are you?
[00:13:46] <jmkasunich> its going from drive box to machine?
[00:13:52] <cradek> one from parport to box, one from box to lathe
[00:14:05] <jmkasunich> the pluto is now inside the box?
[00:14:08] <cradek> yes
[00:14:15] <petev> what connector on the box?
[00:14:28] <cradek> db25s everywhere
[00:14:50] <jmkasunich> don't cross them up, that would be bad
[00:15:02] <jmkasunich> (is motor power going thru the box-to-lathe one?)
[00:15:14] <cradek> yes
[00:15:41] <cradek> I may use a couple pins
[00:17:32] <jmkasunich> its kinda late to mention this now, but would using centronic style SCSI connectors have given you a better chance of getting properly wired cables?
[00:18:05] <cradek> it's only a tad too late. That was my plan until I saw how easy it would be to do the lathe end using this breakout thingy
[00:18:11] <petev> if finding the right cables is hard, why not get some round ribbon and IDC connectors from digikey?
[00:18:19] <jmkasunich> db25 for PC to pluto, and centronics for box to lathe
[00:18:46] <jmkasunich> petev: molded and shielded cables are "nicer" than homebrew IDE ones
[00:18:55] <cradek> yeah that was my plan like I said
[00:18:57] <petev> the round ribbon is shielded
[00:19:07] <petev> and the bolt on covers are not that bad
[00:19:08] <cradek> wish I had a breakout thingy
[00:19:22] <jmkasunich> although personally I don't like ribbon for power stuff, the wires are usually 26GA or even smaller
[00:19:51] <petev> yeah, I wouldn't use DB25 pins for power either
[00:19:55] <jmkasunich> cradek: I wish I had sent you the cable that was inside that SCSI box
[00:20:31] <jmkasunich> it was two 50 pin (IIRC) centronics, with about 4" of ribbon cable between them, with a 25x2 0.100" IDC connector in the middle
[00:21:38] <cradek> there's some chance I could find one of those here - but still no nice screw terminals
[00:21:47] <jmkasunich> thats not a breakout though - its really nice to be able to screw wires down
[00:21:52] <cradek> yeah
[00:22:56] <cradek> I even have a centronics-to-db25 cable here, but those are wired very strangely
[00:23:47] <jmkasunich> very strangely as in less than 25 distinct wires?
[00:23:56] <cradek> yes
[00:24:08] <cradek> and various things tied together I think
[00:24:11] <jmkasunich> seems the actual pinout doesn't matter since you have a breakout, but if it doesn't have enough conductors...
[00:24:37] <jmkasunich> didn't you pick up a mated pair of MS connectors at the fest?
[00:24:51] <jmkasunich> but that still has the "no screws" problem
[00:25:19] <cradek> I think I need 22 wires altogether
[00:25:38] <cradek> 8,4,4 for encoders, 4 for motors, 2 for home switch
[00:25:40] <jmkasunich> that includes both motors and encoders doesn't it?
[00:25:57] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[00:26:00] <jmkasunich> motors and encoders in the same cable would make me very nervous
[00:26:28] <cradek> yeah I know it's a calculated risk
[00:26:38] <cradek> (they're pretty low power)
[00:27:00] <jmkasunich> they still have brushes, and they'll still be PWMed
[00:27:52] <jmkasunich> there is a DB25 on the box, right? where do they go from there? both motor and encoder go to your interface board?
[00:28:10] <jmkasunich> or is the DB25 part of the interface board?
[00:28:26] <cradek> motors and encoders both go to the interface/driver board
[00:28:48] <jmkasunich> by discrete wires, ribbon cable, or is the DB25 soldered to the board?
[00:28:58] <cradek> wires to screws
[00:29:23] <jmkasunich> in that case, I'd run the encoder wires thru ferrite beads, between the interface board and the connector
[00:29:47] <petev> why no put a cord grip on the box and attach to the screw? then you only need the db25 on one end and a cable should be easy
[00:29:47] <jmkasunich> so any crosstalk from motor leads to encoder leads in the cable will have to fight its way thru the beads to get to the interface board
[00:29:50] <cradek> ok
[00:30:50] <jmkasunich> your encoders are differential aren't they? that will help a lot - keep the pairs together as you go thru beads, and common mode noise will be attenuated quite a bit
[00:31:06] <cradek> just the spindle is differential
[00:31:39] <jmkasunich> hmm
[00:32:05] <jmkasunich> its a calculated risk, but at this point I'd be nervous
[00:32:17] <jmkasunich> like you say, they're small motors, and maybe I overestimate the risk
[00:33:04] <jmkasunich> but I'd be looking for separately shielded cables for either the encoders or the motors
[00:33:17] <jmkasunich> sharing a connector is one thing, sharing 6 feet of cable jacket is another
[00:33:17] <cradek> centronics is definitely better. I'll look around and see what I can come up with. Maybe I could make a new board that fits in the green mounting thing
[00:33:41] <jmkasunich> you can easily have 10's to maybe 100pF of stray C from conductor to conductor in a cable
[00:33:48] <cradek> if I use centronics, I think there will be predictable twisted pairs in the cable
[00:36:35] <cradek> I hate this kind of "last part" of projects like this
[00:36:56] <jmkasunich> we have a saying at work: "labels and cables will get you every time"
[00:37:51] <jmkasunich> product is ready to ship, except you don't have the label that says its UL approved, or the catalog number label, or some other silly piece of sticky paper
[00:38:17] <SWPadnos> the "423" on one end ofa wire ... :)
[01:11:18] <Guest308> Guest308 is now known as skunkworks_
[01:27:41] <skunkworks_> skunkworks_ is now known as skunkworks
[04:04:58] <a-l-p-h-a> SWPadnos, how many times do you feel like thread crapping in geckodrive?
[04:05:13] <a-l-p-h-a> wrong chan again.
[16:38:26] <jepler> command (EMC_AXIS_HOME) cannot be executed until the machine is out of E-stop and turned on
[16:38:32] <jepler> I wish I knew why this happened intermittently
[16:40:37] <jepler> aha -- it seems to be correlated with starting homing just before exiting emc
[19:40:25] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[23:44:52] <Guest164> Guest164 is now known as skunkworks