#emc-devel | Logs for 2007-08-24

[00:03:32] <cradek> jepler: how's your driver box coming?
[00:11:10] <jepler> cradek: I found that I neglected a detail
[00:11:16] <cradek> oops
[00:11:28] <jepler> cradek: the distance from the xylotex to the back panel is a bit longer than the short ribbon cable to DB25 connector
[00:11:45] <jepler> (I put it nearly at the front because I anticipate eventually putting the pluto "in between")
[00:12:24] <jepler> the front-of-box bezel has screws so when I want to make a front panel for the motor wiring etc I will not have to replace the whole bezel
[00:13:14] <cradek> are both of the connectors on the ribbon idp so you can just replace the cable?
[00:13:26] <cradek> idc? idp? whatever that's called
[00:13:27] <jepler> yes I could
[00:13:39] <cradek> that's good
[00:13:41] <jepler> but once I put the pluto into the mix it won't matter, so I'll just put it off instead
[00:15:02] <cradek> off to hang drywall...
[00:15:05] <jepler> ooh fun
[01:49:02] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[01:49:18] <SWPadnos> hi jmkasunich
[01:49:25] <SWPadnos> sans _ :)
[01:49:41] <jmkasunich> hi SWPadnos
[01:50:10] <SWPadnos> I'm finding out that you really need a lot of bypass caps for 16-bit analog stuff
[01:50:15] <jmkasunich> heh
[01:50:41] <SWPadnos> that nice 6-channel A/D needs dual caps on 16 pins
[01:50:43] <jmkasunich> what are your analog signal swings? +/-10, or 0-5V?
[01:50:51] <SWPadnos> +/-10
[01:51:08] <jmkasunich> so 1 lsb = 0.3mV
[01:51:14] <SWPadnos> yep
[01:51:26] <jmkasunich> you can't even measure 10mV signals anywhere in our lab
[01:51:36] <SWPadnos> heh - I can, but only barely
[01:51:53] <SWPadnos> I can also borrow a network and/or spectrum analyzer, which certainly can
[01:52:46] <jmkasunich> thats only good for repetitive stuff
[01:52:52] <SWPadnos> true
[01:52:57] <jmkasunich> many noise sources are impuleses
[01:53:02] <jmkasunich> impulses
[01:53:24] <SWPadnos> yeah - I'm hoping the shield can is enough to keep it relatively quiet
[01:53:57] <jmkasunich> that an good board layout
[01:54:00] <jmkasunich> and
[01:54:00] <SWPadnos> I hadn't really planned on the cost of 40 tantalum caps per board either ;)
[01:54:09] <jmkasunich> tantalum? why?
[01:54:17] <SWPadnos> the 10uF bypass caps
[01:54:29] <SWPadnos> paralleled with .1uF ceramics
[01:54:33] <jmkasunich> usually when I think bypass I think 0.1uF ceramics
[01:54:45] <SWPadnos> got those too :)
[01:55:03] <jmkasunich> we probably use a 10:1 ratio of number of ceramics to number of larger caps
[01:55:28] <SWPadnos> all the datasheets (for the op-amps, instrumentation amps, DACs and ADCs) tell you to parallel one of each for each of the pins they want bypassed
[01:55:40] <jmkasunich> unless your power routing is absolutely horrible, that seems like overkill to me
[01:56:05] <SWPadnos> it does, but since the customer said "cost isn't an object right now", I want to make sure it works
[01:56:20] <SWPadnos> I can always leave some unpopulated, but it's hard to add more
[01:56:20] <jmkasunich> I suppose thats one approach
[01:56:52] <jmkasunich> true
[01:56:52] <SWPadnos> since I have no debug time, I want to be sure it "isn't my fault" if something doesn't work
[01:57:13] <SWPadnos> paranoia / CYA mentality, but hopefully it will work
[01:57:38] <jmkasunich> if you have that kind of time constraints, your approach makes sense
[01:58:01] <jmkasunich> you still need to make sure you have a good layout - all the caps in the world won't help if the ground plane isn't right, etc
[01:58:04] <SWPadnos> heh - I need boards completed by the end of next week, and I haven't finished the schematics yet, let alone the layout
[01:58:06] <SWPadnos> yeah
[01:59:44] <jmkasunich> ah, so this isn't a "I needed to add the caps to make it work" thing
[01:59:55] <jmkasunich> its a "wow, that datasheet sure calls out a lot of caps" thing
[01:59:58] <SWPadnos> nope, not yet ;)
[01:59:59] <SWPadnos> yep
[12:13:17] <Guest517> Guest517 is now known as skunkworks_
[12:32:26] <fenn> i think this should be a manpage instead, or maybe in addition http://linuxcnc.org/docs/html/config/emc2hal/index.html
[12:32:41] <fenn> in man 3hal emc
[12:32:47] <fenn> er, man 3hal motion i mean
[12:37:41] <skunkworks_> http://groups.yahoo.com/group/CAD_CAM_EDM_DRO/message/93580
[12:39:13] <fenn> or maybe section 9 would be better
[12:44:31] <fenn> skunkworks_: that's very sweet.. what do you say to something like that?
[12:46:40] <skunkworks_> Nothing. from what I read - he seems to always plug emc. And from meeting him at the workshop - he seems like a nice guy. :) (even though he doesn't do opensource..)
[12:47:08] <skunkworks_> nobody is perfect I guess
[12:47:48] <fenn> i bet emc cuts into his bottom line quite heavily
[12:47:59] <skunkworks_> I bet more now than before.
[12:48:23] <skunkworks_> I don't know - there is always going to be 'I don't know linux' people
[12:48:47] <fenn> mach is .. $150? how much do you think he could get away with if there werent a free+better option
[12:49:23] <skunkworks_> yah
[12:49:39] <fenn> i'm surprised that some enterprising individual hasnt made a prepackaged emc controller
[12:49:58] <fenn> with a pretty windows gui attached over ethernet
[12:50:51] <skunkworks_> heh - that would be cool. All you would have to charge for is hardware and your time+markup. I would think it would be pretty cost effective.
[12:58:48] <skunkworks_> why hasn't anyone done that?
[12:59:15] <fenn> they were too busy talking on irc and surfing the net
[12:59:52] <skunkworks_> heh
[13:08:45] <jepler> fenn: reportedly that's what this system does: http://atelierrobin.net/p41.htm -- I don't know whether the interface between the embedded linux and the win/mac GUI is nml or something else
[23:35:05] <cradek> jepler: is there anything we can do about http://sourceforge.net/tracker/index.php?func=detail&aid=1767781&group_id=6744&atid=106744 ?
[23:35:24] <cradek> other than disable that warning
[23:43:18] <jepler> I have been trying not to think about it
[23:43:37] <cradek> I think it's the last thing that routinely bites me
[23:45:19] <cradek> we certainly can't get it right if using G43.1