#emc | Logs for 2007-04-18

[00:28:35] <skunkworks> woo hoo - wpa setup on ubuntu. almost painless
[00:34:43] <skunkworks> (figures I should add encryption now that we are in town. - if the hack wpa - they can have whatevers on my computer)
[00:34:50] <skunkworks> figured
[00:34:57] <skunkworks> they
[00:54:41] <Jymmm> skunkworks: iptables baby, iptables =)
[00:54:57] <Jymmm> or VPN
[00:55:46] <Jymmm> skunkworks you at 10 or 54 mbps ?
[00:56:20] <skunkworks> 54
[00:57:08] <Jymmm> skunkworks: Yeah, might consider a VPN, you have enough bandwidth to dela with the encryption overhead.
[01:00:01] <skunkworks> cool
[01:01:11] <Jymmm> skunkworks: I've just seen a lot, and am overly paranoid =)
[01:01:46] <Jymmm> skunkworks: Even more so than the NetAdmins at work =)
[01:03:06] <Jymmm> oh hell, their idea of security is nightly disk copy =)
[01:07:01] <skunkworks> I like xcopy ;)
[01:15:39] <skunkworks> and eagle works - why do I need xp anymore? ;)
[01:34:01] <ejholmgren> skunkworks: build a real tunnel, with concrete and stuff
[01:34:28] <skunkworks> :)
[01:34:52] <ejholmgren> and put a single run of 10baseT through it
[01:35:14] <ejholmgren> not even the squirrels would be bored enough to tap that
[01:36:33] <jmkasunich> conduit!
[01:37:01] <jmkasunich> thats a tunnel, just roach sized instead of people sized
[01:37:07] <jtr> conduit with welded joints!
[01:37:19] <ejholmgren> and a trained sentry racoon
[01:37:21] <skunkworks> encased in concrete
[01:37:27] <ejholmgren> rabid preferably
[01:38:01] <jmkasunich> rabid racoon ravishes running roaches!
[01:38:26] <ejholmgren> say that 5 times fast whilst wearing a tinfoil hay
[01:38:30] <ejholmgren> er ... hat
[01:39:08] <jmkasunich> if you do (say it 5 times with hat), put it on youtube!
[03:22:14] <CIA-19> 03cradek 07TRUNK * 10emc2/src/emc/kinematics/tp.c: revert an optimization that causes jerks after first and before last moves that are blended together
[03:24:07] <CIA-19> 03cradek 07v2_1_branch * 10emc2/src/emc/kinematics/tp.c: revert an optimization that causes jerks after first and before last moves that are blended together
[03:32:11] <steves_logging> steves_logging is now known as steve_stallings
[03:32:46] <steve_stallings> can someone refresh my memory about the favorite place to place screen shot images for use with our discussions
[03:33:00] <cradek> I think some people use imagebin.org
[03:35:16] <jmkasunich> or their own servers, if convenient
[03:41:00] <steve_stallings> I think I must use the latter, imagebin.org page does not want to work, submit just clears fields, no image posted...
[03:41:32] <cradek> I admit to having never used it
[03:42:01] <cradek> obviously works for some: http://imagebin.org/8213
[03:42:23] <steve_stallings> it sends and reads a ton of cookies, but they were all accepted by my system, but somewhere my security is probably too tight
[03:49:25] <steve_stallings> OK, posted image on my server. Users list message about 4 axis EDM on the way....
[03:53:32] <cradek> wow that is strange gcode
[03:53:52] <cradek> I think the two "extra" axes are relative to XY
[03:54:19] <cradek> so arc interpolation is not special at all (the "extra" axes don't move during the arc)
[03:54:29] <cradek> you can see that in the image
[03:57:09] <cradek> it appears the extras are PQ, I think usually they are UV, in emc they would (right now) have to be AB
[03:59:18] <cradek> switching between g41 and g42 on alternate lines is very bizarre and certainly won't work on emc
[03:59:32] <steve_stallings> cradek, I think you have it, at least for wire EMD,... the extra axes are offsets from the main axis, so interpolating an arc on the main axes will also move the extra ones which are actually sitting still at that time
[04:00:05] <cradek> yeah that's my understanding
[04:00:26] <cradek> a clever person might be able to get that from emc using hal trickery (and care)
[04:00:45] <steve_stallings> unfortunately I don't think that fulfills the needs of the foam cutters that are cutting airfoils
[04:01:32] <cradek> well you could interpolate AB during an arc - that works fine
[04:02:16] <cradek> ngc says it's allowed but "has no known use", and I was careful to get it right in the TP rewrite
[04:02:37] <steve_stallings> but foam cutters do not use offset axes carried on the main axes carriage, they actually need to interpolate two arcs at the same time
[04:03:11] <cradek> so the extra axes are or aren't offsets from XY in that case?
[04:04:16] <steve_stallings> no, on a foam cutter you have essentially X and Y at one end of a hot wire and X' and Y' at the other end
[04:05:14] <cradek> ok but how do you specify X'? my understanding is it's at X+U
[04:05:58] <steve_stallings> sorry, but I don't know how to specify it, and that is the point... the software that these guys use never generates G-code, it just runs the steppers directly
[04:06:13] <cradek> ok I see
[04:06:48] <jmkasunich> can X' and Y' be specified in as X' = X + A, etc?
[04:06:50] <cradek> my point was only that if X' = X+U, an arc in XY also makes an arc in X'Y', and if you change UV while making an arc in XY, you get a different arc-like(?) think in X'Y'
[04:07:12] <jmkasunich> arc-like is the word
[04:07:36] <cradek> if that's how it's supposed to work (?) then our current setup will do it except for the X+U (X+A) issue
[04:07:42] <jmkasunich> if you want XY to be a 2" circle, and X'Y' to be a 3" circle, UV needs to be a 1" circle
[04:07:45] <jmkasunich> hence two arcs at once
[04:08:03] <jmkasunich> the X+A part is the easy part - hal can do that
[04:08:22] <jmkasunich> simultaneous arcs in XY and AB is the hard part
[04:09:28] <cradek> yes your cone case just isn't going to work
[04:10:05] <cradek> wonder if dave could do a cone in synergy
[04:10:40] <jmkasunich> true cone, or truncated? true we can do (only one arc, the other end is a point)
[04:10:53] <steve_stallings> I am sure Synergy can design it. The question is can the post generate valid code for it.
[04:11:02] <cradek> not true if X'Y' = XY + AB
[04:11:15] <jmkasunich> whats not true?
[04:11:18] <cradek> you couldn't make X'Y' not move with that setup
[04:11:29] <jmkasunich> ah, right
[04:11:49] <cradek> we clearly don't have a spec to work from yet :-)
[04:11:51] <jmkasunich> you'd need an equal and opposite move on AB to cancel XY
[04:12:12] <cradek> yes an arc in UV with -R
[04:12:31] <cradek> * cradek shivers
[04:12:45] <steve_stallings> wish tomp was here, I am sure he could help explain if wire EDM machines ever do two independent arcs at the same time, I would bet modern ones can
[04:12:48] <jmkasunich> what if we said no arcs, break the curves into small straight lines
[04:13:06] <jmkasunich> the airfoil folks will be doing that anyway, foils aren't constant radius
[04:13:40] <steve_stallings> sure, breaking things down into straight lines is exactly what the stepper driver inside the foam software has to do
[04:14:28] <steve_stallings> for reference: http://www.foamwork.net/
[04:14:54] <cradek> an input filter in AXIS could easily convert XYUV (relative) to XYAB (absolute), even breaking up arcs if necessary
[04:15:32] <steve_stallings> foamwork is not the most popular software, but the one that is most popular is in French
[04:15:44] <cradek> (a clever emc user could tackle this numerous ways)
[04:16:11] <steve_stallings> once you decide on a way to specify to movement in the first place
[04:16:24] <cradek> yes you'd have to know how the heck your cam does it
[04:17:00] <cradek> that g41/g42 stuff in dave's output is troubling - I can't fathom what it's supposed to mean
[04:17:21] <cradek> maybe it's something not at all like radius comp
[04:17:47] <cradek> and emc doesn't have g44 at all
[04:17:53] <steve_stallings> probably similar, just some sort of offset for wire diameter
[04:18:32] <cradek> but it switches between sides every line
[04:19:12] <cradek> ok, bedtime for real this time, goodnight
[04:20:13] <steve_stallings> interesting that it switches, but notice the lines alternate between G01 and arcs also
[04:20:36] <jmkasunich> I wonder if they are applying tool offset to one pair of coords, to get the taper?
[04:22:27] <steve_stallings> keep in mind that most wire EDMs literally carry the secondary axes on the stage of the primary axes and this probably influenced the developmen of the command sturcture
[04:23:27] <jmkasunich> steve_stallings: that might be true
[04:23:58] <jmkasunich> its not relevant for us though, its very easy to go from the X' world to the X+A world or back
[04:24:17] <jmkasunich> just simple additions in HAL (and corresponding subtractions for the feedback)
[04:25:00] <jmkasunich> in fact I'm pretty sure we have a comp that processes both the outbound and inbound signals in the same "box"
[04:25:35] <jmkasunich> so you could do the "stacked axis" thing and emulate the edm structure easily
[04:25:52] <jmkasunich> (make normal axes act stacked)
[04:26:49] <steve_stallings> more links about foam for anyone with the interest and energy - http://www.foamwork.net/links.htm
[04:28:03] <steve_stallings> me, I am comfortable that EMC can handle the special case of stacked axes as used by the wire EDM machines that I have seen
[04:28:24] <steve_stallings> time for some shut eye....
[04:28:44] <steve_stallings> steve_stallings is now known as steves_logging
[04:46:45] <jmkasunich> goodnight all
[04:56:48] <chr0n1c> howdy folks...
[04:57:51] <chr0n1c> i'm trying to install vmware on my emc box... can i ask the location of the kernel headers? (i'm new to the whole compiling thing..)
[04:58:05] <chr0n1c> i tried google but got mixed answers for the rt kernel headers
[04:58:41] <chr0n1c> oh.. i found g-cam.. looks interesting, i got that compiled earlier
[04:58:54] <chr0n1c> anyone seen it?
[05:01:25] <chr0n1c> ok, i think i figured out the headers thing
[05:04:06] <chr0n1c> nope i was wrong
[05:10:03] <chr0n1c> ... i didn't have the kernel source installed even.. i'll be m'kay ;)
[05:10:10] <chr0n1c> i think
[12:18:53] <cradek_> cradek_ is now known as cradek
[12:19:01] <skunkworks> logger_emc: bookmark
[12:19:01] <skunkworks> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-04-18.txt
[12:50:05] <Dibblah> Hi. This isn't (yet :) ) EMC related, but has anyone experience of http://www.austriamicrosystems.com/03products/products_detail/AS5040/description_AS5040.htm
[12:50:48] <Dibblah> It's a magnetic encoder, can do quadrature, etc. Very, very nice for the price / ease of implementation.
[12:51:53] <Dibblah> 10 euro for the chip, 2 for a magnet.
[12:55:44] <Dibblah> However, my real question is if someone's done a better-than-raw PID software controller for this type of chip, preferably using it's SSI mode, since that means the 12 bit (rather than 10 bit) version is usable...
[13:30:10] <skunkworks> I know a few people here looked at them - I thought the .5deg accuracy at room temp and .9deg over its full temp range wasn't that great imho
[14:06:28] <anonimasu> hi
[15:53:14] <Dibblah> skunkworks: Thanks for the response. I'm trying them out anyway. The datasheets specify that as worst-case, so I'll find out, I guess :)
[15:57:57] <Dibblah> Just starting out on my first venture into CNC. A little bridgeport conversion.
[16:23:36] <skunkworks> Dibblah: cool. what kinda price are those modules?
[16:24:46] <skunkworks> (you can get discs and sensors for encoders from usdigital for cheap.)
[16:57:29] <Jymmm> Hmmmm, almost Yin-Yang, but not quite... http://melusine.eu.org/syracuse/metapost/animations/sarlat/watt/?animation=B
[17:07:26] <Dibblah> skunkworks: Not buying modules. Bought the raw chips and made PCBs. So about 14 Euro.
[17:07:53] <Dibblah> ... Which I couldn't beat with USDigital :)
[18:04:26] <anonimasu> fasdfasdfdsaf5hm
[18:08:48] <alex_joni> if you think so
[18:10:40] <anonimasu> heh
[18:10:49] <anonimasu> I have a nasty part I need to cut
[18:10:58] <anonimasu> it's kind of a 3d valley
[18:11:01] <anonimasu> and I need 20 of them
[18:11:32] <Jymmm> GET ER DONE!
[18:12:46] <Jymmm> mnbvmbvmbv
[18:13:21] <anonimasu> ?
[18:15:22] <anonimasu> I just need a new visualmill post.
[18:21:50] <skunkworks> * skunkworks needs to do some 3d cad work. I have been a 2d person too long. ;)
[18:22:34] <anonimasu> lol
[18:25:07] <skunkworks> parametric cad software scrares me ;)
[18:26:25] <anonimasu> lol
[18:26:26] <anonimasu> nah
[18:26:29] <anonimasu> it's nice
[18:28:33] <feoc> http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&rd=1&item=140105555750&ssPageName=STRK:MEWN:IT&ih=004
[18:28:40] <feoc> i spend far too much money on ebay
[18:43:31] <skunkworks> anonimasu: what cam do you use for alibre?
[18:43:52] <anonimasu> nothing as alibre wont export curves..
[18:43:55] <anonimasu> for some stupid reason
[18:44:03] <skunkworks> ah
[18:44:12] <skunkworks> so what do you use?
[18:44:18] <skunkworks> for cad/cam
[18:44:33] <anonimasu> pm
[18:45:14] <renesis> hehe @ backroom juarez discussions
[18:45:34] <anonimasu> ?help
[22:19:26] <robin_sz> meep?
[22:22:39] <alex_joni> good night all
[23:14:25] <ds3> the Linux RT waters are deep, black, and rough
[23:17:20] <Dibblah> But you can see far deeper than in Windows ;)
[23:17:46] <ds3> does not seem that way
[23:18:44] <ds3> so far, I can find at least 4 implementations
[23:19:19] <Dibblah> You say that as if it's a bad thing.
[23:20:07] <ds3> =)