#emc-devel | Logs for 2006-07-12

[13:00:32] <alex_joni> morning jeff
[13:16:56] <cradek> hi alex, jeff
[13:17:11] <cradek> got an easy 25mm/sec (60ipm) on my lathe with a new power supply yesterday
[13:17:17] <alex_joni> COOL
[13:17:33] <cradek> I think it's going to work really well.
[13:18:07] <alex_joni> that's really nice
[13:19:12] <cradek> what's the cheapest/simplest way to get more digital IO from emc? other than adding parallel ports, that is
[13:19:22] <alex_joni> multiplexing them
[13:19:35] <alex_joni> or using jeplers serial protocol
[13:19:45] <alex_joni> and a small Atmel
[13:19:52] <cradek> I mean at the base period (encoders etc)
[13:20:07] <alex_joni> oh.. only a card
[13:20:16] <alex_joni> but you can move the encoders off-PC
[13:20:34] <alex_joni> say you put an LS7166 on a micro, and get position feedback through serial
[13:20:54] <alex_joni> although that might be a bit more expensive/complicated than a second parport
[13:21:08] <jepler> cradek: don't you even have a PCI parallel port you've never used?
[13:21:18] <cradek> yeah another parport is probably the way to go. with one in input mode and the other in output, it makes quite a bit.
[13:21:29] <cradek> jepler: yes if I can find it
[13:21:29] <alex_joni> right
[13:21:39] <alex_joni> it probably costs around 20$ new
[13:22:30] <jepler> writing a hal driver for this should be easy: http://www.futurlec.com/PCI8255.shtml
[13:22:37] <alex_joni> which means you might get it for under 10$ on ebay
[13:22:49] <alex_joni> right, the 82C55 is part of the STG driver
[13:23:25] <cradek> that's a cool card
[13:23:54] <alex_joni> cradek: you could even build one of your own (on ISA)
[13:23:58] <jepler> or put one on the parport .. you're stuck writing a driver again though. http://jap.hu/electronic/c64pio.html
[13:25:00] <jepler> (is that CON44 the centronics connector?)
[13:25:34] <cradek> probably
[13:26:21] <alex_joni> http://www.boondog.com/
[13:26:49] <cradek> 404
[13:27:22] <alex_joni> http://www.boondog.com/artwork/8255artwork.pdf
[13:28:21] <alex_joni> * alex_joni runs home
[13:28:24] <alex_joni> later guys
[13:28:31] <jepler> bye alex_joni
[13:28:32] <Lerneaen_Hydra> bye
[13:29:02] <cradek> I'll just add a second parport if I need it, thanks for reminding me that's the easy thing to do
[13:33:45] <jepler> bah -- you should buy a $65 card instead of a $20 card, especially if it means you have to write software
[13:34:10] <jepler> especially if you actually have the $20 card already and, worse yet , you have a good idea where to look for it
[13:34:25] <cradek> haha
[13:36:22] <cradek> jepler: do you remember if we programmed that divider with your laptop?
[13:37:00] <cradek> also do you remember if you can have different division ratios for the different channels?
[13:37:22] <cradek> I don't think I even have the source - maybe you can put it on your blog or something
[13:38:00] <jepler> cradek: yes, not with the source as it is, yes it's on my blog
[13:38:38] <jepler> http://emergent.unpy.net/projects/01149271333 updated mkstate.py: http://emergent.unpy.net/files/sandbox/mkstate.py
[13:39:25] <jepler> if you want more than one table, you'll have to change the table-copying code or make several copies of it
[13:39:51] <jepler> each divider has a different index register for its table, so you'd load XHI for channel A, etc
[13:40:38] <jepler> 'table' is the location of the (currently only one) table in RAM, and must be 256-bytes aligned. 'states' is the location of the (currently only one table in ROM) which has no alignment requirement
[13:40:56] <cradek> cool, sounds doable
[13:41:19] <jepler> lower divider for X and Z axes, higher for spindle?
[13:41:29] <jepler> oh, and we never did finish/test the "stretch index pulse" code
[13:41:58] <cradek> oh I counted wrong - if I /8 the spindle i'll still get 512 edges/rev which should be fine
[13:42:12] <cradek> so I bet I can make everything /8
[13:42:57] <cradek> I think I can probably hook the index pulse straight through and not worry about it
[13:43:06] <jepler> 10240 edges/second when threading 20TPI at 60IPM?
[13:44:27] <cradek> that's 1200 rpm on the spindle, should be very doable
[13:46:02] <cradek> that will be *very* impressive
[13:48:55] <jepler> $ units '(3 rotation / mm) * 1 ips * (500 encoderline/rotation) / 8'
[13:48:55] <jepler> Definition: 19050 edge / s
[13:49:05] <jepler> it's so cool that you can define your own units
[13:49:16] <cradek> slick
[13:50:02] <cradek> so it looks like at /8 I can increase my base period a tad, which would be nice
[13:50:23] <jepler> did you run it on your 24V PS yet?
[13:50:30] <cradek> yes last night
[13:50:36] <cradek> it works really great
[13:50:37] <jepler> how fast does it go?
[13:50:56] <cradek> easily 25mm/sec as hoped
[13:51:06] <cradek> s/hoped/designed/
[13:51:08] <cradek> haha
[13:51:11] <jepler> whee
[13:52:09] <cradek> that's on Z, the crappier one, X will be better (but I probably don't want to go any faster anyway)
[13:52:24] <cradek> Z is where the speed is needed for threading
[13:52:30] <cradek> brb
[13:58:22] <jepler> aha, on the 8255 PCI board, the screw terminals are apparently for a relay and an LED indicator
[14:14:39] <cradek> I wondered
[14:17:29] <jepler> enough of the words are in english in this document for me to make a guess: http://www.etteam.com/download/10PC_INTERFACE/1008/Man_ET_PCI8255_V3.pdf
[14:18:08] <jepler> #
[14:18:09] <jepler> #
[14:18:10] <jepler> RELAY 1 CH ON BOARD ?????????????????? NO, NC, COM, LED POWER ,LED ON/OFF RELAY
[16:55:18] <Lerneaen_Hydra> jepler: 1200rpm and 20TPI sounds similar to that of commercial lathes, (depending on the diameter of the workpiece of course)
[18:21:34] <cradek> Lerneaen_Hydra: it remains to be seen whether my little sherline will be able to do that, but with a light enough cut it looks likely
[18:39:33] <Lerneaen_Hydra> cradek: you mean if the motor is powerful enough?
[18:40:08] <cradek> the servo mostly
[18:40:09] <Lerneaen_Hydra> cradek: I've been spoiled, the cnc lathe I worked with has a main motor with something like 10kW
[18:40:18] <Lerneaen_Hydra> oh, is there a lot of load on it?
[18:40:26] <cradek> sure it can jog at 60ipm, but can it cut at that speed?
[18:40:36] <Lerneaen_Hydra> servo or stepper?
[18:40:43] <cradek> servos
[18:40:49] <Lerneaen_Hydra> hmm, I see...
[18:41:01] <cradek> I'll see, no point in guessing
[18:41:09] <Lerneaen_Hydra> 60 isn't *that* fast though, is it=
[18:41:14] <Lerneaen_Hydra> s/=/?
[18:41:47] <jepler> it's faster than direct-drive stepper systems seem to go
[18:41:47] <cradek> well on a small machine it's full travel in a few seconds
[18:42:00] <Lerneaen_Hydra> hmm, you have a point there
[18:42:10] <cradek> it looks pretty fast to me
[18:42:31] <Lerneaen_Hydra> 30 ipm on my lathe at home looks just fine, but in the large one looks punitive
[18:42:50] <Lerneaen_Hydra> rapid moves are scary with that one though, as g0 is 1000 ipm
[18:43:35] <Lerneaen_Hydra> especially with safety distances of 1mm or so...
[20:02:33] <alex_joni> safety distances?
[20:02:55] <Lerneaen_Hydra> alex_joni: g0 to 1mm from the material, then g1/2/3
[20:03:33] <Lerneaen_Hydra> so in effect, 1000ipm to 1mm from the material
[20:03:37] <Lerneaen_Hydra> quite formidable
[20:05:16] <alex_joni> ahh, I see what you mean
[20:05:26] <alex_joni> approaching distance
[20:05:50] <Lerneaen_Hydra> right
[20:11:09] <alex_joni> nice fellow Marc le Douarain
[20:11:51] <cradek> I'm glad he's interested in emc
[20:14:26] <alex_joni> seems like he's really glad someone is using CL for real stuff ;)
[20:14:32] <cradek> yes
[20:16:57] <alex_joni> it's always a rewarding experience to learn about users
[20:17:18] <cradek> too bad dallur wasn't around
[20:17:41] <cradek> he speaks very highly of emc+cl
[20:18:07] <jepler> I think it's neat
[20:18:07] <alex_joni> I think dallur was the one who got marc's attention
[20:18:10] <jepler> now that I understand it a bit
[20:18:19] <cradek> me too
[20:18:49] <cradek> we've still got some integration problems but it's usable
[20:19:06] <cradek> maybe one of us should try to finish fenn's work
[20:19:28] <cradek> that's the biggest thing missing (being able to see the connected signal name)
[20:25:08] <cradek> that sure killed the conversation
[20:27:05] <alex_joni> not really..
[20:27:06] <alex_joni> :)
[20:27:14] <alex_joni> * alex_joni was away setting up email forwarders
[21:10:57] <alex_joni> good night all
[21:11:08] <cradek> night alex