#emc | Logs for 2008-02-14

[00:00:04] <SWPadnos> BigJohnT, that's what SolidCAM is for ;)
[00:00:13] <SWPadnos> another $3500 or so
[00:00:23] <gezar> see, no way am I paying that
[00:00:38] <gezar> and shoot, back way back in the day, I had an open dialogue with the bob cad guy
[00:00:44] <SWPadnos> gezar, well, that's if you already have the $6000 SolidWorks package
[00:00:47] <BigJohnT> damm that would be 7.5k
[00:00:57] <gezar> thats just retarded
[00:00:59] <BigJohnT> 4k for solidworks
[00:01:09] <BigJohnT> $3995
[00:01:10] <SWPadnos> no, its prety high quality software, and does a hell of a lot
[00:01:20] <SWPadnos> hmmm. is that a current price for new licenses?
[00:01:29] <BigJohnT> it's the best 3-d modeling software I've used
[00:01:38] <BigJohnT> yet
[00:01:40] <BigJohnT> yep
[00:01:44] <SWPadnos> oh - the software is $4000 but the more or less required service contract is $1300/year
[00:01:50] <SWPadnos> or some such
[00:01:59] <gezar> thats just too much money for me
[00:02:13] <BigJohnT> LOL well the service contract gives you the upgrades
[00:02:16] <SWPadnos> of course - it's away too much for anyone who doesn't make their living on it
[00:02:33] <SWPadnos> it's half of what I spent for a PCB design package
[00:03:28] <gezar> holy sheep shit batman
[00:03:28] <fenn> it was more than that actually
[00:03:37] <BigJohnT> you don't have to have it
[00:03:59] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[00:04:06] <gezar> just a big investment is all
[00:04:31] <BigJohnT> If you design things like I do it's just part of your cost to do business
[00:04:35] <SWPadnos> yep
[00:04:43] <fenn> <insert standard whining about FOSS CAD/CAM>
[00:04:47] <SWPadnos> heh
[00:05:15] <SWPadnos> if I really thought that my $12000 could have prompted a FOSS person/group to come up with something comparable, I might have donated it somewhere
[00:05:33] <BigJohnT> I can design a machine in 3-d and "know" that it will fit and work when I build it, that means a lot of time savings to me
[00:05:37] <gezar> foss?
[00:05:44] <fenn> free/open source sofware
[00:06:03] <fenn> also known as FLOSS (free libre open source software)
[00:06:10] <fenn> but FLOSS just looks weird to me ;)
[00:06:27] <gezar> I like using cad to spend time instead of spending time on metal
[00:06:40] <BigJohnT> I've seen some butt flossers on the beach...
[00:06:48] <SWPadnos> um. yay
[00:06:49] <fenn> depends how many you plan to make, and how many times you want to do it before getting it right
[00:06:50] <SWPadnos> ?
[00:07:25] <BigJohnT> I want it right when I deliver it to my customers
[00:07:31] <BigJohnT> that keeps them coming back
[00:07:43] <fenn> cad/cam/cnc is definitely slower than by hand for most things
[00:07:54] <SWPadnos> and coming back for more stuff, not service ;)
[00:08:05] <BigJohnT> Exactly!!!
[00:08:10] <SWPadnos> fenn, I wouldn't say that
[00:08:15] <SWPadnos> it's likely slower for most simple things
[00:08:19] <SWPadnos> but not for most things
[00:08:25] <fenn> well, most things are simple
[00:08:31] <BigJohnT> that is why I am writing the simple g code generators
[00:08:47] <SWPadnos> most things you do (or a hobbyist does) are simple
[00:09:07] <fenn> yep
[00:09:09] <SWPadnos> many things others do are simple, and something like a conversational (tm) program would suffice
[00:09:17] <BigJohnT> how come my torch height ain't simple?
[00:09:28] <SWPadnos> because we need to charge you more
[00:09:42] <BigJohnT> ok you can charge me all you want
[00:09:59] <fenn> eh wot?
[00:10:02] <SWPadnos> ok. next time, the price doubles
[00:10:03] <BigJohnT> I'd rather owe you than beat you out of it
[00:10:13] <fenn> BigJohnT: what's hard about torch height?
[00:10:47] <BigJohnT> well I need to take a 0-300vdc signal and convert it to binary
[00:10:51] <fenn> BigJohnT: if there were a GUI for hal do you think it would be satisfactory?
[00:11:16] <BigJohnT> I "think" that I have all of the hal figured out
[00:12:03] <BigJohnT> the biggest hurdle on hal was the man pages are kinda lean on info for non-electrical engineers
[00:12:20] <fenn> so, two resistors, an opto, and a DAC? or am i missing something
[00:12:46] <fenn> ADC i mean'
[00:13:02] <BigJohnT> ok I was typing wtf LOL
[00:13:07] <fenn> brain needs flushing
[00:13:14] <lerman> BigJohnT: Did I just hear you volunteer to enhance the man pages?
[00:13:16] <BigJohnT> the dividing circuit is simple
[00:13:34] <BigJohnT> I have before but I don't have the secret decoder ring...
[00:13:43] <BigJohnT> and yes I would
[00:13:51] <fenn> BigJohnT: know how to use diff?
[00:14:05] <BigJohnT> diff?
[00:14:17] <BigJohnT> no
[00:14:23] <SWPadnos> man diff
[00:14:25] <SWPadnos> :)
[00:14:30] <BigJohnT> nope
[00:14:30] <fenn> http://iihm.imag.fr/blanch/howtos/DiffPatch.html
[00:14:39] <BigJohnT> looking
[00:15:00] <BigJohnT> plan 9 from outer space...
[00:15:10] <fenn> no, that's an operating system
[00:15:23] <BigJohnT> ok LOL
[00:15:31] <fenn> er, a host in this case
[00:15:49] <BigJohnT> ok, reading it now
[00:16:18] <BigJohnT> the + - adds and subtracts text from the file?
[00:16:25] <fenn> so the idea is you make changes to your local copy, do a diff, and then you can post the patch file for anyone to apply to their stuff
[00:16:45] <fenn> yes + is new lines, - is deleted lines
[00:18:15] <fenn> my presence is requested at the dinner table
[00:18:17] <BigJohnT> ok, the man files are where? I've not used linux much so bear with me...
[00:18:23] <BigJohnT> ok
[00:18:31] <fenn> emc2/docs/src/man
[00:18:40] <BigJohnT> ok, enjoy dinner
[00:18:57] <fenn> \fB bold \fR roman \fI italic (i think)
[00:19:08] <BigJohnT> ok, cool
[00:19:19] <fenn> see 'man groff' for more
[00:19:41] <BigJohnT> I'd love to add the stuff that I posted on wiki on hal things
[00:33:05] <BigJohnT> hmmm, emc2/docs/src/man is that on my computer or on the web? I can't seem to find it anywhere
[00:33:29] <eric_U> really sorry about that, I have the only copy
[00:35:29] <SWPadnos> BigJohnT, do you have an emc2/docs/man directory?
[00:36:35] <BigJohnT> the only emc2 directory I have is under /etc/emc2
[00:36:57] <SWPadnos> ok, I assumed you were referring to acopy of the source code
[00:37:01] <SWPadnos> a copy
[00:37:41] <BigJohnT> hmmm I did a live CD install...
[00:38:23] <BigJohnT> I need to got to the CVS and get a copy?
[00:38:36] <SWPadnos> ah - that was from fenn's comment. I think the correct directory would be under emc2/docs/man
[00:38:56] <SWPadnos> yes, you need a cvs checkout or tarball (preferably a checkout so it's easy to update)
[00:39:20] <SWPadnos> you'll need to install cvs, build-essential, and some other stuff if you actually want to compile
[00:39:43] <SWPadnos> (like apt-get build-dep emc2, and apt-get install emc2-dev)
[00:39:50] <BigJohnT> just want to help with the man pages at the moment
[00:40:32] <SWPadnos> ok - those you should be able to check locally by just using the -M argument to man (I think it's -M "specify man dir" or similar)
[00:40:55] <BigJohnT> ok let me see
[00:41:03] <SWPadnos> man man ;)
[00:41:24] <cradek> man ./whatever.9
[00:41:32] <BigJohnT> yep man man
[00:41:55] <SWPadnos> oh, that's too easy ;)
[00:42:17] <SWPadnos> I guess man is also shortfor "make this groff-like markup human-readable" then :)
[00:43:09] <cradek> cvs is by the easiest way to make a diff, even if you only have an anon checkout and/or are not signed up as a committer
[00:43:28] <cradek> is by far
[00:43:51] <BigJohnT> ok, where/what do I need to do to checkout the man pages?
[00:44:11] <cradek> they are part of the normal checkout
[00:44:12] <BigJohnT> I'm looking at emc2/docs/man at cvs
[00:44:12] <SWPadnos> if you do a cvs checkout, you'll get the manpages
[00:44:24] <BigJohnT> ok let me look
[00:44:42] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_5_10_and_6_06_from_source
[00:44:56] <SWPadnos> that has a list of the things you need to install, and how to get a cvs checkout
[00:45:06] <BigJohnT> ok reading it now
[00:45:37] <BigJohnT> let me follow the bouncing ball on wiki
[00:45:42] <SWPadnos> heh
[00:45:57] <cradek> we'll sing along
[00:46:12] <BigJohnT> hmmm, password?
[00:46:47] <jepler> if a line starting "sudo" asks you for a password, it is your regular login password
[00:46:59] <BigJohnT> sudo password you dummy
[00:47:07] <BigJohnT> slap on forhead
[00:47:42] <SWPadnos> later versions of ubuntu are better in that regard - they say something like "sudo: [password for steve]:"
[00:47:45] <BigJohnT> been a long day... ice storm yesterday yada yada
[00:47:59] <SWPadnos> heh - we've still got a little bit coming down
[00:48:08] <SWPadnos> only 8-10 inches total though
[00:48:32] <BigJohnT> sometimes late at night I need extra prodding...
[00:48:36] <BigJohnT> no power for 24hours
[00:48:57] <BigJohnT> wife gets cranky when it gets cold/hot...
[00:49:29] <BigJohnT> Rugludallur: Hi
[00:50:01] <BigJohnT> studied your plasma a lot
[00:50:24] <Rugludallur> hi BigJohnT
[00:50:44] <Rugludallur> BigJohnT: Great, how is your project going ?
[00:51:01] <Rugludallur> BigJohnT: have you finished designing the THC circuit ?
[00:51:15] <BigJohnT> Rugludallur: I'm looking at using a digital input to a second parallel port
[00:51:24] <BigJohnT> to control the torch height
[00:51:48] <BigJohnT> adding to your hal file a few things to convert seems to work
[00:52:07] <BigJohnT> just need to complete the electrical end
[00:53:05] <BigJohnT> a question, the ddt and hypot part of your hal is to determine if the Xand Y are changing speeds?
[00:53:09] <Rugludallur> I see, well getting a digital number relative to the voltage diff would be better, that way you can make better decisions on when or if to move the Z axis
[00:53:32] <Rugludallur> BigJohnT: let me bring up the diagram, I don't remember :)
[00:53:38] <BigJohnT> the voltage diff from the set point
[00:53:56] <BigJohnT> I have several scattered on my work bench...
[00:53:59] <Rugludallur> BigJohnT: going to set the point via parallel to or via serial ?
[00:54:13] <BigJohnT> parallel is the plan so far
[00:54:18] <BigJohnT> no
[00:54:27] <BigJohnT> set via pyvcp
[00:54:41] <BigJohnT> read actual via parallel
[00:55:38] <Rugludallur> BigJohnT: ok, at the moment I'm setting mine via serial with pyvcp, it was actually very easy since it can be done from userspace
[00:56:16] <BigJohnT> you are using a THC to tell you up and down still?
[00:58:31] <Rugludallur> BigJohnT: yup, i'm using a commercial system and I would not want to mess with that stuff, to many things can go wrong with High Frequency and High Voltage around
[00:59:00] <BigJohnT> I'm a glutten for punishment then
[00:59:46] <Rugludallur> BigJohnT: The DDT changes the movement information from location to speed
[01:00:41] <Rugludallur> BigJohnT: and the Hypot figures out the total travel speed for X and Y axis
[01:00:51] <BigJohnT> ok
[01:01:36] <BigJohnT> from there you go to a minmac and a comp
[01:01:39] <Rugludallur> BigJohnT: those are then used to figure out the maximum speed (to determine if we are slowing down) and CHL trigger to Lock the height while turning around sharp corners
[01:01:41] <BigJohnT> minmax
[01:02:00] <Rugludallur> minmax: stores the max speed so we know what the feed rate should be
[01:02:04] <BigJohnT> I kinda thought that but wanted to verify
[01:03:24] <Rugludallur> BigJohnT: ask ahead, I'm happy to answer if I'm at the keyboard, and if I'm at work I will at least check every hour or two and respond
[01:03:39] <BigJohnT> thanks
[01:04:15] <Rugludallur> btw, Rugludallur == Dallur
[01:04:18] <Rugludallur> Dallur is me at work
[01:04:21] <Rugludallur> Rugludallur is me at home
[01:04:33] <BigJohnT> what does Dallur mean?
[01:05:01] <Rugludallur> BigJohnT: It's the name of my sailboat (that i'm building from steel, hence the plasma)
[01:05:31] <Rugludallur> BigJohnT: It's icelandic for old/crappy/leaky/unsafe boat
[01:05:35] <Rugludallur> :)
[01:05:40] <BigJohnT> Sweet!, I used to sail around in the Gulf of Mexico on my small sail boat
[01:06:03] <BigJohnT> singlehand mostly
[01:06:18] <BigJohnT> so your from Iceland I take it?
[01:06:29] <Rugludallur> BigJohnT: yup
[01:06:43] <Rugludallur> BigJohnT: did you cruise for extented periods or did you drop down every now and then ?
[01:06:52] <BigJohnT> cool, I have not had the chance to visit there yet
[01:07:06] <BigJohnT> I would stay out for a week or two at the most
[01:07:33] <Rugludallur> BigJohnT: That's probably plenty for a smaller boat, around 26' ?
[01:07:57] <BigJohnT> 25' boat
[01:08:05] <Rugludallur> BigJohnT: if you ever happen to be passing through Iceland or just in the area feel free to look me up and drop by the shop
[01:08:16] <BigJohnT> swing keel
[01:08:25] <BigJohnT> I will
[01:08:29] <Rugludallur> BigJohnT: Ahh, so no issues going up river or trailering
[01:08:42] <BigJohnT> or over sand bars LOL
[01:09:01] <BigJohnT> 1000 lb keel
[01:09:22] <BigJohnT> she would make some good speed on a close reach
[01:09:22] <Rugludallur> nice, lead ?
[01:09:31] <BigJohnT> cast iron
[01:10:02] <Rugludallur> * Rugludallur has been daydreaming about trying to source Depleted Uranium for his keel
[01:10:08] <BigJohnT> she drew 6' down and 18" up
[01:10:12] <Rugludallur> BigJohnT: if you want to see pictures of the plasma table and other such you can check out http://dallur.com
[01:10:37] <BigJohnT> ahh, that is one of my favorite sites...
[01:11:06] <BigJohnT> I prefer to make bullets from depleted uranium...
[01:11:10] <Rugludallur> BigJohnT: 18" hmm, not much that can stop you
[01:11:31] <BigJohnT> it was a nice boat
[01:11:38] <Rugludallur> BigJohnT: yeah, brilliant properties, self sharpening and such, have you made any ?
[01:12:12] <BigJohnT> not yet, but you can think about such things
[01:13:18] <Rugludallur> BigJohnT: yeah, but nobody wants to sell DU, even though it's about as radioactive as granite gravel
[01:13:31] <BigJohnT> I had the mast rigged so I could drop it in a minute, btw the power was a british sea gull outborad ...
[01:13:46] <BigJohnT> outboard
[01:13:49] <Rugludallur> BigJohnT: did you end up selling her or ?
[01:13:54] <BigJohnT> yes
[01:14:10] <BigJohnT> then I moved to Missouri
[01:14:19] <BigJohnT> in the middle of the USA
[01:14:37] <Rugludallur> BigJohnT: yup, not easy to sailing around there
[01:15:03] <BigJohnT> I've seen only a few small day sailers here on the lakes
[01:15:05] <Rugludallur> BigJohnT: although with a draft of 18" ...
[01:15:21] <BigJohnT> so how is your boat coming along
[01:16:15] <Rugludallur> BigJohnT: I have not had time to work on it in months, :( A couple of my friends got me to found a software company with them about a year ago and that keeps me to busy
[01:17:10] <Rugludallur> BigJohnT: but then again I get to work on embedded linux systems day in day out :)
[01:18:03] <SWPadnos> yay! :)
[01:18:10] <BigJohnT> how big is the hull
[01:18:27] <SWPadnos> how big are the embedded Linux systems? :)
[01:18:31] <Rugludallur> BigJohnT: the boat is a 33 feet
[01:18:34] <BigJohnT> LOL
[01:18:41] <BigJohnT> nice size
[01:18:49] <BigJohnT> single mast
[01:18:52] <BigJohnT> ?
[01:19:40] <Rugludallur> SWPadnos: 128MB with X, mplayer, firefox, flash, alsa, networking and everything else :)
[01:19:48] <Rugludallur> BigJohnT: yup, cutter rig
[01:20:27] <Rugludallur> SWPadnos: 48MB when compressed into a initramfs image
[01:20:31] <SWPadnos> 128M RAM? (I'm assuming all those apps won't fit on a 128M SSD)
[01:20:34] <SWPadnos> wow, not bad
[01:20:46] <SWPadnos> s/not bad/that's great/
[01:21:29] <BigJohnT> sweet
[01:21:52] <Rugludallur> SWPadnos: yeahh, i'm pretty happy with it, spent the better part of a year on the project
[01:21:57] <SWPadnos> heh
[01:22:00] <SWPadnos> what CPU?
[01:22:04] <Rugludallur> SWPadnos: AMD Geode
[01:22:15] <SWPadnos> ah, nearly almost x86-like ;)
[01:22:27] <Rugludallur> SWPadnos: actually it's x86 100%
[01:22:49] <BigJohnT> mine was a fractional rig sloop
[01:22:53] <SWPadnos> yeah - but there are some things added, and probably some others removed
[01:23:13] <SWPadnos> (at least, there are if you have one of the media processors - I think those are in the Geode line)
[01:23:20] <Rugludallur> SWPadnos it's X86, i586 but with the MMX extensions and 3dNow
[01:23:49] <Rugludallur> SWPadnos: yeah, the old geode processors were not exactly x86 but the newer ones (LX) are all x86
[01:23:50] <BigJohnT> * BigJohnT hears the dinner bell ringing
[01:24:07] <SWPadnos> how's the performance?
[01:24:31] <Rugludallur> SWPadnos: I can decode HD 720p
[01:24:48] <SWPadnos> ok, that's pretty good - is that due to hardware help, or CPU speed?
[01:25:06] <Rugludallur> SWPadnos: cpu and more optimisations than I care to think about
[01:25:11] <SWPadnos> heh
[01:25:55] <SWPadnos> is this a standard hardware platform or something custom?
[01:25:57] <Rugludallur> SWPadnos: at some point I really want to make a custom EMC build for one of these, would be sweet to buy a $200 machine and use it
[01:26:04] <SWPadnos> yes indeed
[01:26:07] <SWPadnos> any PCI slots?
[01:26:19] <Rugludallur> SWPadnos: Custom platform, but we share a lot of stuff with the OLPC
[01:26:41] <SWPadnos> oh, interesting
[01:26:46] <Rugludallur> SWPadnos: nahh, the one we are creating is custom MB, but you can still get Geode MB's with PCI, GenIO, parallel and all the other gizmos
[01:26:47] <SWPadnos> I haven't taken mine apart yet
[01:27:16] <SWPadnos> yeah, I've got a couple of those $60 Google OS dev kits, but the latencies are pretty bad
[01:27:38] <SWPadnos> most all-in-one PC platforms have crappy onboard video, and half of them have no slot in which to put a generic do-nothing video card
[01:28:45] <Rugludallur> SWPadnos: yeah, but then again if for example you can run the GUI remotely, and palce the all-in-one machine in the same box as your drivers ...
[01:29:16] <SWPadnos> yes. I did that with an off-the-shelf computer (a HAL application, so no GUI at all)
[01:29:33] <SWPadnos> edited runlevel 2 to be (the normal) text mode multiuser
[01:29:53] <SWPadnos> left out lots of stuff - I wonder if it will notice if I plug in a USB memory stick :)
[01:29:58] <Rugludallur> SWPadnos: hmm, not networking ?
[01:30:45] <SWPadnos> sure, there's networking
[01:31:02] <Rugludallur> SWPadnos: as far as I am concerned the best setup would be motors with built in drivers and step generators, just plug in ethernet and 230v
[01:31:02] <SWPadnos> it's just like the old days, when you had to type "startx" to get a GUI :)
[01:31:32] <SWPadnos> well, there are some nice aspects to that, but I don't think anything under $2000/axis is there yet :)
[01:31:40] <SWPadnos> plus the controller
[01:32:18] <Rugludallur> SWPadnos: It's quite doable to get x86 geode based chips on MB with the circuits needed for <$100 so ,, it could be done for quite a bit less
[01:33:17] <Rugludallur> SWPadnos: and i'm sure you can make do with far lesser chips
[01:35:32] <SWPadnos> oh - the google thing is a Via C7, not geode
[01:35:35] <SWPadnos> oh well :)
[01:36:14] <SWPadnos> http://clubit.com/product_detail.cfm?itemno=A4842001
[01:36:30] <SWPadnos> that's $60 including the CPU - just add RAM, PSU, and storage
[01:37:35] <Rugludallur> SWPadnos: $100 with onboard flash chip, ram and just needs 12v in :)
[01:37:42] <SWPadnos> heh
[01:37:47] <SWPadnos> but no PCI slots ;)
[01:37:51] <Rugludallur> SWPadnos: nope :(
[01:37:59] <SWPadnos> I can add a Mesa card to that one
[01:38:11] <Rugludallur> SWPadnos: hehe, you beat me hands down there
[01:38:15] <SWPadnos> and the latency is probably fine for that usage (I don't remember how horrendous it was)
[01:38:48] <SWPadnos> I wonder if something like DMX would be good for motion control on some smallish number of axes
[01:43:39] <SWPadnos> hmmm. probably not, it's only 250 kbaud
[01:50:24] <Rugludallur> Anyways, I need to go to sleep, almost 2AM :P
[01:50:30] <Rugludallur> l8r
[01:50:32] <SWPadnos> good night
[01:52:27] <lerman> SWPadnos: ethernet, ethernet, ethernet.
[01:54:52] <SWPadnos> yes yes, I know :)
[01:58:19] <fenn> etherGNAT
[01:58:30] <fenn> the license unencumbered variant of etherCAT :)
[01:58:42] <fenn> if only it were real
[01:59:51] <SWPadnos> heh - the license unencumbered *HARDWARE* :)
[02:00:26] <fenn> eh? the hardware has specific license restrictions, patents etc
[02:00:39] <fenn> seems to be mostly around protecting their trademark
[02:01:06] <fenn> but it's up in the air whether it's actually legal to use (self inconsistent)
[02:01:09] <SWPadnos> could be, but hardware is required for etherCAT
[02:01:19] <SWPadnos> regardless of whether they want anyone to use it
[02:01:37] <lerman> For our use a plain old 100baseT interface should be fine. No special hardware required.
[02:01:48] <fenn> dmx is some music protocol right?
[02:01:52] <SWPadnos> yes
[02:01:52] <fenn> like MIDI
[02:01:53] <SWPadnos> lighting
[02:02:37] <fenn> ethernet has that ubiquitous je ne se qua
[02:03:16] <lerman> Does anyone use token ring anymore?
[02:03:21] <SWPadnos> IBM maybe ;)
[02:03:39] <SWPadnos> on those RS6000 workstations they never threw out
[02:03:53] <fenn> SWPadnos: are you sure special hardware is required for ethercat?
[02:03:57] <SWPadnos> yes
[02:04:01] <fenn> why is that?
[02:04:27] <SWPadnos> the slave devices are supposed to stick their information into the packet as it gets passed around, and that's only doable within the timing slecs by hardware
[02:04:50] <SWPadnos> that's how they get such high cycle rates - there's one packet with all the command and feedback data, and it gets passed around
[02:05:10] <lerman> We need fast, but not that fast. :-)
[02:05:12] <SWPadnos> as it gets relayed by each device, the data from that device is added to the packet, leavign the rest of the data unchanged
[02:05:36] <SWPadnos> I don't know how they deal with checksums
[02:06:13] <fenn> could do a checksum for each segment
[02:06:40] <SWPadnos> hmmm - well, the checksum is at the end, so the hardware can compute it on the fly and insert it into the packet at the end
[02:06:45] <lerman> I would have each node just repeat the packet and add to it.
[02:06:57] <SWPadnos> that introduces large amounts of latency
[02:07:27] <fenn> i thought that was what it did
[02:07:30] <lerman> For our use, 5 or 10 nodes would be enough.
[02:08:09] <SWPadnos> each node does forward the data, but it's done before he full frame has been received
[02:08:25] <lerman> Heck, I'd be happy if we could build a pluto board that interfaced using pt to pt ethernet instead of parallel port.
[02:08:39] <SWPadnos> it's transmitting the beginning of the frame within some microseconds of receiving it, and sticking the new data in on the fly
[02:09:18] <fenn> lerman that's a great idea
[02:09:50] <SWPadnos> if you're going to do point to point, you might as well use SPI or something :)
[02:10:05] <fenn> SWPadnos where ya gonna get the SPI?
[02:10:10] <SWPadnos> pluto
[02:10:12] <lerman> Nope. You can't get a decent SPI board for PC.
[02:10:13] <SWPadnos> or mesa
[02:10:23] <fenn> with a mesa you dont need a pluto
[02:10:24] <SWPadnos> eaiser than adding the ethernet MAC and PHY
[02:11:09] <fenn> i refuse to believe ethernet is that difficult
[02:11:30] <lerman> I would use full duplex cat5 with special cabling. The output of machine 1 goes to the input of machine 2. The output of machine 2 goes to the input of machine 3. ... The output of machine N goes to back to the input of machine 1.
[02:11:31] <SWPadnos> ok, have it your way :)
[02:12:20] <SWPadnos> that's the topology of thercat, with the addition that the last device in a chain will automatically use the return hald of the full duplex link and return packets that way instead of forwarding them
[02:12:27] <SWPadnos> s/thercat/ethercat/
[02:12:42] <SWPadnos> at least according to wikipedia: http://en.wikipedia.org/wiki/EtherCAT
[02:12:56] <SWPadnos> it's been a while since I read any ethercat specs
[02:14:22] <fenn> so, er, why does that require special hardware? (assuming no insane speed requirements)
[02:14:25] <lerman> I'm waiting for the Arduino boards to have ethernet on them.
[02:14:38] <fenn> get an rtl8139..
[02:14:53] <SWPadnos> fenn, because most ethernet chips get a packet from the host processor and transmit it
[02:14:59] <fenn> or one of those microchip thingies.. enc28j60 or whateva
[02:15:15] <lerman> The special hardware is apparently necessary to allow them to tack stuff onto an existing packet instead of reading the packet and forwarding a new one.
[02:15:21] <SWPadnos> they don't get some data from ethernet, start retransmitting while still receiving, and stick some data into the outgoing packet on the fly
[02:15:29] <SWPadnos> right
[02:15:48] <fenn> what's wrong with reading the packet and forwarding the new one?
[02:15:49] <SWPadnos> and they don't tack extra data on - each unit has an address range in the packet which it's supposed to populate
[02:15:55] <SWPadnos> it takes too long
[02:16:01] <fenn> define 'too long'
[02:16:03] <lerman> They want to be faster than that.
[02:16:14] <SWPadnos> fenn, "longer than they want" :)
[02:16:15] <lerman> Their too long is not too long for us.
[02:16:38] <SWPadnos> could be true, but maybe not. consider that a minumum ethernet frame is 512 bits
[02:16:44] <SWPadnos> minimum too
[02:17:06] <lerman> That's pretty short at 100 mbits.
[02:17:17] <SWPadnos> and there's minimum turnaround time (which could probably be ignored on a special segment)
[02:17:25] <SWPadnos> sure, 200k packets/second
[02:18:44] <SWPadnos> you could do something that would work for us, but it wouldn't be ethercat I think
[02:18:49] <SWPadnos> which is fine
[02:19:11] <fenn> enc28j60 is 10MBit wah
[02:19:42] <fenn> better to use a processor with on-board ethernet anyway
[02:19:45] <lerman> Fine with me. It is not clear that ethercat will every be cheap enough for people who want a $60 pluto look alike.
[02:19:52] <lerman> ever
[02:19:53] <SWPadnos> heh
[02:20:22] <fenn> all we need is point to point
[02:20:25] <lerman> Although if they start using it in cars, it might be.
[02:20:31] <tomp> this was cool, a python script to convert a csv file into a gEDA sym file ( for hal config files built with gEDA CAD ). http://geda.seul.org/wiki/geda:tragesym_tutorial
[02:20:44] <fenn> that should be doable with an at91sam..
[02:20:49] <fenn> or similar
[02:23:09] <lerman> Found the web site. Supports 100 mbit. YES.
[02:24:23] <SWPadnos> which? the AT91SAM7X
[02:24:26] <SWPadnos> ?
[02:24:46] <SWPadnos> there are a bunch of ARM chips that could be very useful
[02:24:53] <lerman> 7XC256
[02:25:04] <SWPadnos> 256k to 512k (I think I even saw a 1M chip)
[02:25:17] <SWPadnos> ethernet, USB, SPI, serial, PWM, and on and on
[02:25:22] <SWPadnos> $9-15
[02:25:35] <SWPadnos> I don't think the C is needed - that has a crypto processor in it
[02:25:54] <fenn> or one could (theoretically) do the ethernet directly in the fpga
[02:26:03] <fenn> but then you need ROM
[02:26:07] <lerman> Whoops. That's expensive. It's hard to make money if you want to sell a complete board for $60 :-)
[02:26:24] <SWPadnos> you need a PHY even if the MAC is in the FPGA
[02:26:30] <SWPadnos> and magnetics
[02:26:52] <fenn> those come built into the jack nowadays :P
[02:27:05] <SWPadnos> the coils do, that's true, but not the PHY :)
[02:27:16] <fenn> ok, i never looked into that in detail
[02:33:46] <lerman> So, who makes a cheap board with one of these suckers on it? It would be worthwhile to get two of them and just connect them in a loop with a host to prove that we can send packets in a loop.
[02:37:02] <SWPadnos> a 7x256?
[02:37:52] <lerman> For proof of concept, any old processor would work. Heck maybe we should just connect 3 PCs in a loop.
[02:38:11] <SWPadnos> what are you proving again?
[02:39:15] <lerman> Proving that from within an emc system kernel we can send packets in a loop with low latency. (Using this ethercat like scheme.)
[02:39:35] <SWPadnos> oh, I think step 1 would be to get some sort of ethernet driver going
[02:39:41] <SWPadnos> probably based on the RTNet drivers
[02:39:58] <lerman> That should be fine.
[02:40:10] <SWPadnos> let me know if you need my help :)
[02:40:22] <lerman> Thanks. :-)
[02:40:27] <Jymmmm> Ug
[02:40:52] <SWPadnos> Jymmmm, you say that a lot - are you regressing to neanderthal?
[02:41:00] <Jymmmm> ug ug
[02:41:40] <Jymmmm> SWPadnos: It's IRC, Ug serves many purpose in the minimal about of time and effort =)
[02:41:57] <Jymmmm> amount
[02:41:58] <lerman> U U
[02:42:07] <SWPadnos> in the immortal words of Luke Skywalker: "Yo mama is so ugly, she put the 'ug' in 'ugnaut'"
[02:53:42] <dmwaters> {global notice} Good day all, I apologize for the netsplits, we had some maintenence due on a couple servers but it's over now. thank you for your patience, and thank you for using freenode!
[03:29:08] <eric_U> did I do that?
[03:31:19] <eric_U> so SWPadnos, have you actually used the google MB?
[03:31:27] <SWPadnos> no
[03:32:22] <eric_U> If I could get away with using it headless, it would be kinda nice
[03:32:54] <eric_U> needs a pci riser board
[11:57:38] <BigJohnT> hmmm
[12:00:44] <BigJohnT> Ok, I've installed CVS and have checked out EMC2. Just to start looking I'm trying to find the and2 man document and I don't find it.
[12:01:20] <BigJohnT> I've searched the CVS as well as my emc2-trunk directories
[12:07:22] <Vq^> it's listed in .cvsignore for some reason
[12:09:06] <BigJohnT> looking
[12:18:43] <BigJohnT> is it in emc2/docs/man/man9?
[12:21:16] <BigJohnT> there is a Mach3 ad in the EMC forum on cnc.zone...
[12:35:01] <alex_joni> BigJohnT: use . scripts/emc-environment
[12:35:20] <alex_joni> that will source the emc-environment file, and set some paths and folders
[12:35:27] <alex_joni> after that you can do : man and2
[12:35:58] <alex_joni> http://www.linuxcnc.org/docview/2.2/html//man/man9/and2.9.html
[12:36:06] <alex_joni> and yes, it's in emc2/docs/man/man9
[12:36:20] <alex_joni> but it might be generated by comp
[12:37:04] <BigJohnT> alex_joni: where do I use scripts/emc-environment?
[12:40:58] <BigJohnT> man and2 displays the and2 man page fine for me, I want to look at adding some info to it...
[12:53:53] <maddash> my steppers like to make massive vibrations @ .083 inches/sec, at 2000pulses/inch
[13:00:42] <maddash> the calculated step freq at that vel is .083*2000=166Hz, but halscope gives me 737.6182Hz
[13:00:44] <maddash> wtf?
[13:07:29] <alex_joni> 14:32 < alex_joni> BigJohnT: use . scripts/emc-environment
[13:07:40] <alex_joni> the "." was *not* a typo
[13:08:42] <maddash> argh stupid stepgen
[13:10:09] <maddash> http://img339.imageshack.us/my.php?image=200802140801201400x1050gr5.png
[13:10:18] <maddash> what kind of stepping pattern is that?
[13:12:15] <alex_joni> the normal one :)
[13:12:47] <alex_joni> when you request a step frequency which is not a multiple of base-period, it will dither between adiacent frequencies
[13:13:11] <maddash> how is that normal? don't you see the unevenness?
[13:14:10] <alex_joni> nope :)
[13:14:20] <alex_joni> first of all.. I'm not sure what you commanded..
[13:14:31] <alex_joni> adding a trace with stepgen vel or similar would be nice
[13:14:43] <maddash> getp stepgen.0.frequency 166
[13:14:52] <maddash> bah brb
[13:22:30] <alex_joni> bbl
[13:53:17] <jepler> also make sure you are capturing at the same frequency that stepgen is running
[13:53:29] <jepler> running it in a 2kHz thread is *cough* atypical
[14:01:16] <jepler> compare these traces: http://emergent.unpy.net/files/sandbox/stepgen-at-right-frequency.png http://emergent.unpy.net/files/sandbox/stepgen-at-wrong-frequency.png (hal details at http://pastebin.ca/903779)
[14:08:31] <jepler> when I set scale 2000, velocity 0.83, stepgen's internal report is a frequency of 166Hz -- http://pastebin.ca/903786 -- how did you find this figure of 737.6182Hz?
[14:12:30] <jepler> of course .. he's gone and I'm talking to myself
[14:12:49] <SWPadnos> don't worry, he often has the log in a browser window
[14:17:50] <jepler> aha, there *is*, however, a bug in the timestamp that halscope shows when you are mousing over the trace
[14:18:01] <SWPadnos> yay!
[14:18:05] <SWPadnos> or something
[14:18:27] <jepler> there was some unclear thinking there -- it was probably mine
[14:19:20] <SWPadnos> is the timestamp from the beginning of the buffer or something wlse?
[14:19:27] <SWPadnos> else
[14:20:11] <jepler> there's a sample number, the oldest in the buffer is sample 0; scope_disp.c's intent is to convert that into a decimal number of seconds before or after the trigger point
[14:20:43] <SWPadnos> ah, ok. seems simple enough ;)
[14:20:44] <jepler> however, it divided by the period rather than multiplying by it (and elsewhere there was a fudge of 1e6 which is what would make samples of a 1ms function act right)
[14:21:04] <SWPadnos> oops
[14:24:35] <jepler> yay, bug fixed!
[14:25:14] <SWPadnos> woohoo!
[14:25:51] <jepler> well -- changed at least
[14:25:53] <jepler> somebody else should test it
[14:25:54] <jepler> bbl
[14:25:56] <SWPadnos> heh
[14:45:41] <ALS> any idea what type of accuracy I could expect using photointerrupers as homing switchs?
[14:48:31] <SWPadnos> it will vary widely depending on the type of interrupter, the thing you use to interrupt, separation distance of the emitter and detector ...
[14:48:58] <SWPadnos> it would certainly be good enough to use along with an encoder index
[14:49:05] <SWPadnos> (I think :) )
[14:49:27] <ALS> no encoder stepper system
[14:49:40] <SWPadnos> ok
[14:49:56] <SWPadnos> what kind of opto are you thinking of using?
[14:50:16] <ALS> I want to home in the middle of my envelope
[14:50:44] <ALS> just looking for one that may work good
[14:53:28] <ALS> I could use a micro switch but I would running over it all the time
[14:53:58] <SWPadnos> the trick there is to use one with a lever arm and roller at theend, and you make a cam that will press the switch
[14:54:24] <SWPadnos> so you get "infinite overtravel" without damaging the switch
[14:54:41] <ALS> 100000 cycle switchs
[14:55:17] <SWPadnos> yeah, true
[14:55:23] <ALS> they are cheep though
[14:56:00] <SWPadnos> there are optical switches that are meant for this I think, I'm nust not sure where to get them :)
[14:58:02] <SWPadnos> you might think about a magnetic prox switch as well, but I'm not sure of their accuracy either
[14:58:56] <jepler> ALS: you can place the home switch anywhere and still set machine 0,0,0 at the center of the work volume -- not sure if that influences your decision
[14:59:33] <ALS> http://www.circuitspecialists.com/prod.itml/icOid/5164 this guy claims very accurite
[15:00:41] <ALS> jepler: no I stiil like home in the middle less motion always in the work zone
[15:01:44] <SWPadnos> ALS, one thing you do need to be sure of - the home switch must be "on" on one side of home and "off" on the other side. you'll need a cam (or interrupter) that's half the length of travel if you put the switch in the middle
[15:02:05] <SWPadnos> it can't be momentary or the homing code doesn't know which way to start off
[15:02:30] <SWPadnos> (though we made some changes in that area last year, I'm not sure what the end result was)
[15:04:05] <ALS> I was thinking slot type with a tab for interupt
[15:04:14] <SWPadnos> long tab
[15:04:26] <SWPadnos> http://web6.automationdirect.com/adc/Shopping/Catalog/Sensors_-z-_Encoders/Double-Insulated_(PBT)_Limit_Switches
[15:04:51] <SWPadnos> one of those that I looked at was rated for 30 million operations, with repeatability of 0.01mm at 1 million operations
[15:05:01] <SWPadnos> $29 or so
[15:06:02] <ALS> allen bradly want's you to take out a second morgage
[15:06:27] <SWPadnos> oh com on, you could buy 1000 switches for the price of a house ;)
[15:06:54] <SWPadnos> http://web6.automationdirect.com/adc/Shopping/Catalog/Sensors_-z-_Encoders/Double-Insulated_(PBT)_Limit_Switches/Side_Rotary_Lever_with_Polyamide_Roller_Actuator/AAP2T41Z11
[15:06:56] <SWPadnos> $13.75
[15:11:07] <ALS> you guys get davenull all straight'nd out with his glass scale and stepper's?
[15:11:25] <SWPadnos> dunno - he never responded to the latest round of emails
[15:11:36] <ALS> I seee that
[15:12:10] <jlmjvm> he emailed me yesterday,said it will be several months before he starts his machine
[15:12:26] <SWPadnos> ok, that's good
[15:12:33] <SWPadnos> plenty of time to plan :)
[15:12:51] <jlmjvm> or change plans
[15:16:07] <ALS> he got my hair up but I bit my lip
[15:16:37] <SWPadnos> heh, yeah. "this may be annoying, but your software is shit. I hope I don't offend" :)
[15:17:09] <ALS> what happend to pid
[15:17:40] <tomp> if you dont want to meet cranky old bastards, dont go near toolmakers
[15:19:01] <SWPadnos> corollary: if you do't want to meet cranky young bastards, don't go near programmers in the morning / before their first cup (pot) of coffee
[15:19:04] <SWPadnos> don't
[15:19:11] <tomp> :)
[15:21:12] <jlmjvm> i am looking forward to the fest
[15:22:13] <SWPadnos> yep - it's generally a good time
[15:22:51] <ALS> where yall comin from jimjvm
[15:23:11] <jlmjvm> huntsville,al
[15:24:36] <jlmjvm> my friend from nasa may attend also if he can get some time off
[15:24:53] <jlmjvm> but i suspect he will be busy
[15:25:39] <jlmjvm> will it just be emc people?
[15:26:24] <SWPadnos> http://www.cnc-workshop.com/
[15:27:00] <SWPadnos> no - the EMC stuff is in part of the shop, near the big mazak
[15:27:08] <jepler> emc people are a minority
[15:27:21] <SWPadnos> there are classes and demos in the entire other building, as well as the rest of the shop
[15:28:06] <jlmjvm> sounds good
[15:29:58] <jlmjvm> im looking forward to learning quite a bit while im there
[15:51:30] <tomp> a minority? can we get funded?
[15:57:41] <jlmjvm> will mach people be there also?
[15:59:18] <alex_joni> some
[15:59:28] <tomp> sometimes the guy from cardinal ( owns the show) lists registered people
[16:00:13] <tomp> Roland ...
[16:01:57] <tomp> http://www.cnc-workshop.com/ no list of attendees
[16:02:12] <jlmjvm> also says 2009
[16:02:44] <SWPadnos> 2008 and 2009 are on there
[16:02:51] <SWPadnos> he hasn't set up the site for this year yet
[16:02:58] <jlmjvm> never mind,should read all before blurting out
[16:03:23] <SWPadnos> too bad it's always in the middle of June - my wife can never make it because she has to work
[16:03:41] <tomp> there's a verb for ya, (v) to blurt
[16:03:46] <jlmjvm> havent had enough of the coffee referenced earlier
[16:04:07] <fenn> that's what cats do right?
[16:04:47] <JymmmEMC> * JymmmEMC lol @ SWPadnos
[16:05:07] <SWPadnos> why is that?
[16:05:20] <JymmmEMC> SWPadnos: Bring your wife to cnc workshop days?
[16:05:32] <SWPadnos> well, she's done more welding than I have ...
[16:05:35] <jlmjvm> to utter suddenly or inadvertently; divulge impulsively or unadvisedly (usually fol. by out): He blurted out the hiding place of the spy.
[16:06:07] <SWPadnos> (of course I haven't done any, but that's beside the point)
[16:06:22] <JymmmEMC> SWPadnos: Oh, my bad... After the "OMG The wifes is in the garage, call the police", I thought otherwise =)
[16:06:26] <SWPadnos> heh
[16:06:47] <SWPadnos> she does "clean" or "organize" from time to time, but otherwise she's cool ;)
[16:07:13] <JymmmEMC> Uh oh, that's even worse
[16:08:21] <JymmmEMC> Okey quick... what is the weirdess thing on your desk right now?
[16:08:33] <SWPadnos> open space
[16:08:45] <SWPadnos> luckily, I don't have much of it
[16:09:20] <JymmmEMC> The weirdess thing I got is a firestick. Anyone else?
[16:09:39] <SWPadnos> I suppose it could be the bottle of "Jade Windscreen" chinese medicine
[16:10:56] <JymmmEMC> which do you think is weirder, this http://www.thinkgeek.com/gadgets/tools/754d/ or your ancient chinese secret
[16:11:07] <JymmmEMC> to be on a desk that is
[16:13:30] <ALS> if it's a steel desk quick and easy to clean just fire it up and stand back
[16:14:50] <JymmmEMC> ALS: Heh, yeah. I had just orderd the firesteel and got it a couple days ago.
[16:15:50] <JymmmEMC> got one of these too http://www.thinkgeek.com/gadgets/tools/7f02/
[16:20:04] <tomp> get a big block of firesteel, mill it at the fest, tell everyone its edm
[16:27:37] <JymmmEMC> lol
[17:28:20] <jlmjvm> jepler:shouldnt the datum symbols go away after an estop or machine off?
[17:33:07] <SWPadnos> jlmjvm, not necessarily - some machines can keep track of position while in e-stop
[17:33:39] <jlmjvm> key word being"some"
[17:34:09] <maddash> jlmjvm: I agree with you, but cradek insists that "those who need to home should do so manually"
[17:34:25] <cradek> I hear my name
[17:34:49] <cradek> I like how I can participate in a fight without even typing
[17:35:09] <maddash> "homed flags NEED to be cleared after an estop" -- http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-01-03.txt
[17:35:14] <jlmjvm> didnt realize we were fighting
[17:35:22] <cradek> my bridgeport unhomes when I estop
[17:35:30] <maddash> ditto
[17:35:35] <maddash> hence the "bug"
[17:35:39] <cradek> in order to home, it has to move the quill down (off the home switch)
[17:36:11] <cradek> if I'm cutting near the top of the quill movement and I hit estop, I can be quite screwed and have to lower the knee in order to home it again
[17:36:32] <cradek> because the tool can't come out either, without clearance
[17:36:43] <jlmjvm> do you have emc on your boss?
[17:36:45] <SWPadnos> maddash, referencing yourself saying the same thing isn't that great a method of convincing people you're right
[17:36:49] <SWPadnos> at least not me
[17:36:57] <cradek> it's an annoyance and I avoid using estop in some cases because of it
[17:37:04] <cradek> jlmjvm: no, it's factory still
[17:37:07] <maddash> SWPadnos: no, i thought that link was the one where cradek argues against me
[17:37:10] <jlmjvm> k
[17:37:11] <SWPadnos> heh
[17:37:13] <maddash> SWPadnos: but now i can't find it
[17:37:37] <SWPadnos> in the immortal words of the penguind: "you didn't see anything" (waves wings)
[17:37:39] <cradek> probably because I didn't argue except to say that some machines can keep position throughout an estop, which is a matter of simple fact
[17:37:39] <maddash> cradek: i don't understand how you would be screwed
[17:37:40] <SWPadnos> penguins
[17:38:33] <maddash> crap, my bridgeport is making growling noises
[17:38:37] <cradek> maybe I didn't explain it clearly then
[17:39:31] <maddash> jlmjvm: its a very simple issue to set axis*.homed to 0 after an estop
[17:39:57] <SWPadnos> maddash, but that's not a universal solution
[17:40:14] <maddash> SWPadnos: why not?
[17:40:20] <cradek> maddash: sounds like you should do that in your tree. in general a thing is not a bug if the behavior is by design
[17:40:23] <SWPadnos> any machine with absolute feedback or with encoders that are powered while estopped will maintain position in estop
[17:40:32] <SWPadnos> making people with that type of machine re-home is annoying
[17:40:44] <SWPadnos> (my bridgeport will be in that category)
[17:40:51] <cradek> so will mine
[17:41:02] <maddash> psht rich kids
[17:41:12] <SWPadnos> no, it only costs an extra $50 or so
[17:41:14] <cradek> ooh! ad hominem!
[17:41:38] <SWPadnos> but most people don't bother to think about what should be powered in what stop/run state
[17:41:40] <cradek> jlmjvm: now we're arguing :-)
[17:42:10] <cradek> fwiw, it would be nice if this were an option, or if there was a way to unhome in general
[17:42:19] <maddash> ini?
[17:42:24] <SWPadnos> that could work
[17:42:25] <cradek> I bet a patch that makes it an option would be well-received
[17:42:30] <SWPadnos> yes
[17:42:36] <maddash> "[TRAJ]CRADEK_IS_WRONG = 1"
[17:42:39] <cradek> I'm sure a patch that just changes the behavior would not be accepted
[17:42:55] <maddash> cradek: seriously? I'm about to whip out my notepad
[17:42:56] <cradek> maddash: I'm trying to help, dude
[17:43:00] <SWPadnos> more like MADDASH_ACTS_LIKE_AN_ASS=1 ;)
[17:43:09] <maddash> cradek: that was a joke, btw
[17:43:24] <maddash> yay.
[17:43:44] <jlmjvm> dont want a behavior change,just thought it should lose the datum emblem after an estop
[17:43:53] <maddash> hm, head isn't head anymore
[17:44:03] <jlmjvm> i like the ini idea
[17:44:06] <cradek> jlmjvm: that icon represents that the axis is homed
[17:44:14] <maddash> SWPadnos: rofl
[17:44:16] <jlmjvm> agreed
[17:44:30] <maddash> ok guys, set the clock to 00:45:00
[17:45:37] <SWPadnos> [TRAJ]UNHOME_IN_ESTOP ??
[17:45:54] <SWPadnos> no wait, it's not TRAJ
[17:46:06] <maddash> SWPadnos: really, I have no idea. the names and labels are up to you guys.
[17:46:21] <SWPadnos> yeah - trying to think of something that makes sense
[17:46:24] <SWPadnos> to others ;)
[17:47:22] <maddash> HOMED_TAINTED
[17:47:24] <cradek> maybe machine off should unhome, not just estop
[17:47:32] <jlmjvm> agreed
[17:47:35] <fenn> feature request: ability to paste multiple lines in mdi
[17:47:36] <SWPadnos> I don't think so actually
[17:47:42] <cradek> on some machinery, disabling the amps might cause position loss
[17:47:47] <cradek> it's certainly that way on stepper machines
[17:47:49] <jlmjvm> exactly
[17:47:59] <SWPadnos> then it should be another option, or a mask-type option
[17:48:25] <SWPadnos> [EMC]CRAPPY_MACHINE
[17:48:26] <cradek> jepler: and also ESTOP_LOSES_POSITION?
[17:48:40] <maddash> jepler: TOO_FREAKIN_LONG_FOR_ME_TO_TYPE_/_NO_SHIFT_KEY_LOVE
[17:48:55] <SWPadnos> [EMC]ADLP
[17:49:01] <cradek> the more explicit, the better
[17:49:07] <SWPadnos> RHSWMNO
[17:49:17] <SWPadnos> (reset homed status when machine is not on)
[17:49:29] <cradek> now it's getting silly in here
[17:49:36] <SWPadnos> yep
[17:49:40] <SWPadnos> thank you thank you
[17:50:17] <maddash> actually, jepler's got the main point -- it's disabling the amplifier that causes unhoming, and ESTOP/OFF both disable the amp
[17:50:36] <jlmjvm> couldnt it be tied to machine on,then even if you estop it still works,beacause estop turns the machine off also
[17:50:41] <cradek> again, that's only true on certain machines
[17:51:08] <jepler> POSITION_LOSS = never,amplifier-off,estop
[17:51:10] <cradek> disabling an amplifier doesn't typically cause position loss on a servo machine
[17:51:30] <SWPadnos> jepler, exactly
[17:51:36] <cradek> on a stepper machine, it often does, especially with microstepping
[17:51:45] <SWPadnos> actually, never has two possible values :)
[17:51:47] <jlmjvm> mine does
[17:51:51] <jepler> SWPadnos: ??
[17:51:52] <maddash> cradek: I'm speaking about the rest of us who actually need this new feature
[17:52:06] <SWPadnos> it's either "never" - absolute encoders that are re-read at power-on
[17:52:07] <jepler> what are the two values of "never"?
[17:52:23] <SWPadnos> or "never" - as long as the computer is powered
[17:52:37] <SWPadnos> but re-homing is needed at startup
[17:53:03] <SWPadnos> so it's more like never, powerup, estop, amplifier_off
[17:53:06] <cradek> SWPadnos: I don't think any absolute encoder schemes are currently supported
[17:53:38] <SWPadnos> they aren't unsupported - many absolute encoders are read once at startup and then they act like incrementals
[17:53:40] <jepler> well .. there's POSITION_FILE but it doesn't allow emc to start with a joint marked as homed
[17:53:59] <SWPadnos> it's easy enough to add a module to read the initial value
[17:54:05] <cradek> interesting
[17:54:21] <cradek> so you could add hooks to the POSITION_FILE code
[17:54:35] <cradek> in that case, you might actually want it to start homed
[17:54:41] <SWPadnos> yes
[17:55:05] <SWPadnos> or you can add something to the HAL setup that does the initial read
[17:55:27] <cradek> or maybe that would better be done as a new 'style' of homing
[17:55:40] <SWPadnos> ok, so what this tells me is that there should probably be an input or two to the motion controller (?) that tell it that the machine is homed, or that the machine is not homed
[17:55:59] <SWPadnos> separate inputs, probably RW so they're one-shots
[17:56:09] <cradek> I'm afraid we're feature-creeping so much that maddash cannot make a suitable patch
[17:56:23] <SWPadnos> nah, we can reset the timer
[17:56:39] <cradek> I don't think we need to write startup-already-homed today
[17:56:45] <maddash> it's not quite clear anymore what you guys want
[17:56:50] <cradek> that seems like a separate problem to me
[17:57:16] <cradek> 'when to optionally transition from homed to unhomed state' is where we started
[17:57:19] <SWPadnos> HAL pins would solve all the cases - just connect machine-on (inverted maybe) to the "unhomed" input
[17:57:46] <cradek> I don't think that would allow unhome-at-estop-only
[17:58:05] <SWPadnos> connect inverted estop to the unhomed input
[17:58:06] <cradek> I may be wrong but I think hal doesn't know 'off' from 'estop'
[17:58:17] <cradek> oh hmm
[17:58:35] <cradek> you mean an input "unhome" [the verb] not "unhomed" [the adjective]?
[17:58:38] <jepler> so you'd drive axis.0.position-lost from estop, or from amplifier-enable, or never drive it away from FALSE depending on your machine?
[17:58:42] <maddash> there's an unhome input pin?
[17:58:53] <SWPadnos> ie, "any external event can tell the motion controller that it is homed, or that it isn't homed. it's up to the integrator to decide which events do what"
[17:58:53] <jepler> maddash: no, that's an alternate proposal SWPadnos is making..
[17:58:54] <cradek> maddash: no, it's a suggested implementation
[17:58:58] <maddash> ah, ok
[17:59:10] <maddash> * maddash twiddles his thumbs and waits for a consensus
[17:59:28] <jepler> would driving position-lost TRUE in machine-on state cause a transition to another machine state (like amplifier fault does)?
[17:59:30] <cradek> lunchtime
[17:59:34] <jepler> ooh lunchtime
[17:59:35] <maddash> just be careful of ... *ahem* HAL bloat
[17:59:58] <jepler> whatever, dude
[18:00:08] <SWPadnos> jepler, good question, my first response would be that homed/unhomed shouldn't cause a machine state change
[18:00:09] <cradek> (I think this belongs in task, not hal)
[18:00:28] <cradek> there should be an UNHOME_AXIS emcmot message
[18:00:35] <cradek> task and/or guis can send it
[18:01:01] <cradek> fwiw, I want unhome on the gui, but not hooked to estop etc
[18:01:42] <maddash> 'EMC_UNHOME_AXIS'?
[18:01:49] <SWPadnos> since position and homing are tied to hardware, I think there may need to be some HAL hooks
[18:02:25] <cradek> he's sure abrasive
[18:02:28] <jepler> yeah
[18:02:57] <jlmjvm> gotta go to the trade school and try to get the index homing working today,and start training the instructors how to run it
[18:03:00] <cradek> SWPadnos: it's true hardware is involved, but a message can trigger homing (which does complex things), it can also trigger unhoming (which is very simple)
[18:03:20] <SWPadnos> I guess I'm more concerned with the hardware side initiating one of those messages
[18:03:25] <cradek> jlmjvm: do you mean index-only (no home switch)?
[18:03:33] <jepler> SWPadnos: it's hard to imagine the situation that would cause you to decide position had been lost (that's what "unhome" means, right?) that wouldn't make you want to abort a program in progress just like you would for an amplifier fault
[18:03:47] <cradek> I was just typing what jepler typed
[18:03:50] <SWPadnos> heh
[18:03:50] <jlmjvm> no it has home switches also
[18:04:09] <SWPadnos> I can see that, but hardware may be able to tell you that it's already homed
[18:04:28] <jlmjvm> i could only make the z work before,gonna try for x and y today
[18:05:17] <jlmjvm> wish me luck
[18:06:23] <jepler> good luck
[18:06:28] <jlmjvm> btw,the teachers are blown away with what they have seen so far,even heard 1 say its better than the new haas they have
[18:06:30] <SWPadnos> rotsa ruck
[18:06:44] <SWPadnos> hmmm. too much caffeine I think ;)
[18:07:09] <cradek> lunch! lunch!
[18:07:12] <cradek> bbl.
[18:07:19] <SWPadnos> oh, good idea! bbl
[18:07:32] <jlmjvm> later guys,thanks
[18:13:52] <fretless85_> hey guys
[18:30:25] <ALS> hows the hand?
[18:48:17] <fretless85_> not that good as it used to be...
[18:49:14] <fretless85_> if want to see photos without bandages, let me know :P
[18:53:30] <ALS> thanks anyway
[18:54:17] <fretless85_> :)
[18:55:23] <ALS> off to coach a wrestling match see 'yall' latter
[18:55:35] <bill20r3> is that a metaphor for something?
[18:56:02] <ALS> ?
[18:56:12] <fretless85_> ALS, have fun
[19:31:09] <maddash> whatever happened to that irc bot that recorded all cvs changes?
[19:31:31] <jepler> maddash: it now is on the -devel channel
[19:32:50] <maddash> are VFD's less costly than steppers?
[19:35:18] <SWPadnos> it depends on the VFD and the stepper
[19:35:39] <SWPadnos> but the important question is, what is the relevance between steppers and VFDs (other than "motors" in general)?
[19:36:01] <maddash> they turn and spin
[19:36:33] <SWPadnos> well, have fun. time to run some errands
[19:40:30] <jepler> the motors driven by VFDs are typically in an entirely different (higher) power category than the stepper motors used for positioning in CNC machines.
[19:41:43] <maddash> so VFDs can't drive CNCs?
[19:42:08] <maddash> aren't VFDs similar to servos?
[19:42:53] <cradek> a vfd is used to drive a (generally high-power) 3 phase motor at various speeds in a limited range
[19:42:58] <seb_kuzminsky> http://en.wikipedia.org/wiki/Variable-frequency_drive
[19:43:00] <cradek> it's good for a spindle
[19:43:20] <alex_joni> or a mincing machine
[19:43:44] <fenn> or .. air conditioners, big ones
[19:43:49] <jepler> if I understand correctly, 'VFD' is a driver technology (more like saying "microstepping chopper drive" than "stepper motor")
[19:43:52] <maddash> I used 'VFD' to denote the driver AND the AC motor at once
[19:44:01] <maddash> whoopsies
[19:44:52] <maddash> but a VFD can make an AC motor spin, turn slowly, stop, or even go the other way, right? isn't that the essence of controlling, say, a milling table?
[19:45:10] <jepler> jon elson has a brushless servo driver board (http://pico-systems.com/acservo.html) which he demo'd at last year's CNC workshop for positioning on a CNC mill
[19:46:34] <maddash> so VFD-driven motors are the same kind as servomotors?
[19:47:05] <jepler> brushed vs brushless are different
[19:47:34] <seb_kuzminsky> vfd is to ac as pwm is to dc (if i understand correctly)
[19:48:41] <maddash> whoa...cool
[19:48:48] <jepler> I understand this little enough that I should probably keep silent..
[19:49:07] <maddash> so i can build a servo system using vfd component
[19:49:07] <maddash> s
[19:52:09] <fenn> vfd's drive sinusoidal AC motors
[19:52:14] <fenn> generally three phase motors
[19:52:39] <seb_kuzminsky> also bldc, right fenn? externally commutated dc motors?
[19:52:39] <fenn> brushless servos are usually trapezoidal three phase (not sinusoidal)
[19:52:45] <seb_kuzminsky> ah
[19:56:37] <maddash> does the univ stepper controller use an fpga? i can't quite see it in the pic
[19:58:03] <jepler> maddash: yes, I believe jon elson's boards use xilinx FPGAs. however, the board is not intended to be reprogrammed with new fpga configurations by the end user (unlike, say, the mesa 5i20)
[20:00:35] <maddash> ah, that explains the recent obsession with 5i20 instead of USC
[20:02:51] <seb_kuzminsky> mesa releases their FPGA firmware open-source style, which is pretty sweet
[20:03:37] <maddash> poor jon
[20:06:25] <jepler> I think Jon does OK.
[20:09:35] <cradek> he could relicense his firmware if he wanted
[20:09:46] <cradek> not sure how it's programmed though
[20:09:56] <cradek> the mesa has it easier
[20:11:35] <jepler> I believe there's an eeprom on the board but I'm not entirely sure
[20:15:59] <maddash> hm, so i suppose someone has written a stepgen firmware for the mesa?
[20:16:38] <maddash> holy crap, someone was rich enough to buy the 7i43 and write stuff for it
[20:18:16] <seb_kuzminsky> mesa gave me a couple of 7i43 boards and some other stuff in exchange for writing the emc2 driver for it
[20:18:43] <seb_kuzminsky> they wrote the firmware, it does step/dir generation, pwmgen, and encoder counting
[20:18:43] <fenn> uh, 7i43 is only $60
[20:18:44] <maddash> how could they know that you wouldn't flake?
[20:18:55] <seb_kuzminsky> they couldnt, really
[20:19:01] <fenn> with no driver it's useless
[20:19:31] <seb_kuzminsky> i'm negotiating with them now to write drivers for their new firmware on their other FPGA boards (5i2?)
[20:19:53] <fenn> seb_kuzminsky: have you seen the stuff jmkasunich was working on?
[20:20:52] <maddash> hm i'm trying to decide between the gecko g100 and 5i20
[20:21:02] <seb_kuzminsky> i looked at his 5i20 driver when i first started my 7i43 driver
[20:21:28] <fenn> g100 is probably not what you want
[20:22:25] <seb_kuzminsky> jmkasunich did his own firmware, right? or he tweaked the old mesa hostmot firmware?
[20:22:38] <maddash> fenn: why's that?
[20:22:48] <fenn> i'm not sure.. it was supposed to be a more flexible, easy to configure sort of thing
[20:23:11] <fenn> maddash: first of all, dont you hate rabbits?
[20:23:18] <fenn> and secondly, it doesnt work with emc
[20:23:19] <seb_kuzminsky> i know next to nothing about fpgas, but i'm pretty impressed with mesa's new hostmot2 firmware
[20:23:20] <maddash> fenn: wabbits, yes
[20:24:22] <seb_kuzminsky> with hm2 there's generic "functions" (stepgen, pwmgen, encoder, and gpio), and you can select (at fpga firmware compile-time) how many of each, and what I/O pins to hook them to
[20:24:29] <maddash> fenn: doesn't the g100 use a fpga?
[20:24:47] <seb_kuzminsky> then when you go to use it, you can dynamically turn some of those functions off, and their pins revert to gpio
[20:25:05] <seb_kuzminsky> pretty impressive to this nerd who doesnt know what to expect from fpgas :-)
[20:25:40] <fenn> maddash "the g100 is a 6 axis motion controller"
[20:26:51] <fenn> i dont think you have access to the firmware, or at least not without hacking it
[20:27:08] <fenn> and in any case, the software is why it costs so much
[20:27:47] <fenn> and all those terminal blocks :)
[20:28:02] <seb_kuzminsky> heh
[20:28:21] <maddash> ugh
[20:28:46] <maddash> it even uses a rabbit
[20:28:51] <maddash> despicable
[20:29:03] <fenn> i thought that was why you were playing around with rabbits
[20:29:41] <maddash> no i was playing with the ADC functionality
[20:29:43] <fenn> gecko does have a generic step multiplier board
[20:31:01] <fenn> g901
[20:31:21] <fenn> of course i have no idea what you are doing
[20:31:38] <maddash> re-re-trofitting the bridgeport?
[20:31:49] <fenn> is software step generation too low?
[20:31:54] <fenn> or are you going to servos?
[20:31:57] <fenn> or .. or..
[20:32:13] <fenn> too slow
[20:32:14] <maddash> not quite, no, .. ..
[20:33:09] <maddash> I thought that I could replace the 30-year-old drivers with newer geckos and get some hardware motmod, all in one shot
[20:33:19] <seb_kuzminsky> anyone know the status of jmkasunich's mesa drivers? there seem to be several
[20:33:20] <maddash> stepper drivers, I meant
[20:36:54] <fenn> maddash i think you just like to fiddle with things
[20:37:18] <maddash> fenn: as opposed to?
[20:37:53] <fenn> not breaking something that works
[20:39:50] <maddash> [gasp] you've exposed me for who I am! A programmer!
[20:40:29] <maddash> fenn: actually, i wasn't successful at all, during my last attempt at a retrofit
[20:41:03] <maddash> fenn: the machine GROWLED whenever i did g{0,1}x**y**[z**][f**]
[20:41:51] <maddash> this was when I hooked up emc to the ancient stepper drivers
[20:42:33] <fenn> could be noise in the lines
[20:43:12] <maddash> huh?
[20:49:34] <fenn> stepper A makes a step, current changes in the coil, which changes the magnetic field, which induces current in the "step" signal for stepper B, which steps.. etc
[20:55:47] <cradek> I'd throw out everything but the motors (or maybe them too) and hook up geckos
[20:55:58] <cradek> factory spec was 100ipm and you can get that with software step generation
[20:56:04] <cradek> it's the simplest possible retrofit
[20:57:01] <cradek> actually there might be a nice transformer to keep - don't know what voltage the DC bus was originally
[20:59:34] <maddash> cradek: no, the spec my dad has says 120 ipm
[21:00:03] <maddash> but the main problem to fix is the grinding noise
[21:01:32] <cradek> steppers make lots of odd noises when you drive them wrong
[21:01:55] <seb_kuzminsky> are steppers fairly quiet when drive right?
[21:02:02] <seb_kuzminsky> *driven
[21:02:04] <cradek> not really
[21:02:10] <maddash> right. you suggested last time that the power supply was too weak for the speeds that I commanded
[21:02:31] <cradek> microstepped drivers are fairly quiet but half or full step is pretty rough and loud
[21:02:32] <seb_kuzminsky> heh in your 5-axis videos on youtube they right sing :-)
[21:02:52] <cradek> yes that mill is loud - and they are small steppers
[21:08:49] <SWPadnos> seb_kuzminsky, the 5i2x is still a work in progress
[21:08:59] <SWPadnos> the driver and new FPGA anyway
[21:09:16] <seb_kuzminsky> ok
[21:09:25] <SWPadnos> actually, I'll move to the dev channel for this, I'm about to get technical :)
[21:09:43] <seb_kuzminsky> what's the relationship between the 5i2x driver and the 5i20?
[21:09:47] <seb_kuzminsky> ok
[21:42:49] <skinny77cncpuppy> http://www.reuters.com/article/oddlyEnoughNews/idUSL1283190120080212
[21:43:19] <skinny77cncpuppy> The thief, who also left a lit blow torch at the scene, is expected to be badly charred, spiky haired and not exactly the brightest bulb in the socket.
[22:42:25] <jymm> What is be like
[22:44:13] <cradek> have you been drinking?
[22:44:48] <jymm> No, why?
[22:44:58] <alex_joni> what is be like?
[22:45:01] <jymm> Well, diet Dr Pepper, but that's abou it
[22:45:09] <cradek> you dropped an "m" and you're talking gibberish
[22:45:18] <alex_joni> and an "t" now
[22:45:41] <jymm> I'm connected from multiple points
[22:46:01] <alex_joni> usually you're not lowercase either :)
[22:46:25] <cradek> bbl
[22:46:26] <jymm> case realloy doens't matter a whole lot, it's all the same
[22:49:10] <fenn> * fenn suspects jymm is a windaz user
[22:57:21] <jymm> fenn: Sure, but I'me moving away from it as best as I can =)
[23:43:00] <dmess> hi all