#emc | Logs for 2008-11-08

[00:36:11] <markdashabout> The sherline connections in stepconfig agree with those in the best sherline docs I can find. But still no movement.
[00:36:34] <markdashabout> Have rebooted into RH6 and EMC1.
[00:37:00] <markdashabout> Doesnt even detect the network card, the lousy old kernel, but the real time still good.
[00:37:57] <markdashabout> All connections are therefore good.
[00:38:08] <markdashabout> Anyone any ideas or has everyone gone to bed?
[00:41:56] <BigJohnT> emc1? what version are you using
[00:42:06] <jmkasunich> he left
[00:42:29] <BigJohnT> crap missed that :)
[00:42:39] <BigJohnT> anyway dinner here
[01:27:01] <kirk_wallace> Hello, Will a brushed universal motor give back-emf when coasting?
[01:27:14] <archivist> yes
[01:27:46] <archivist> from the residual magnatism (not a lot but)
[01:28:30] <archivist> I assume you mean driven without power applied
[01:30:34] <kirk_wallace> [OT] I have a generator that uses a brushed armature as a starter, then uses the field to excite the AC rotor windings. I'm trying to figure out how to use the field winding after the engine starts.
[01:31:44] <kirk_wallace> It's similar to the bottom half of this page - http://rich.homeunix.com:443/onan-cck/048.jpg
[01:31:58] <archivist> ah, I have an old one that was designed for tanks(army) that is a starter come dynamo
[01:34:28] <kirk_wallace> After the engine starts, I get about a Volt on L1 and L2 (AC legs), then I re-apply 12 V DC to the "starter" to energize the field, but I get no increase in AC.
[01:37:41] <kirk_wallace> I was also trying to figure out how the battery gets recharged, there must be some residual magnetism to self excite the DC part also?
[01:38:18] <archivist> yup builds up same as a car dynamo
[01:39:01] <archivist> and in the mix is any regulation and contacts
[01:40:11] <archivist> and the dynamo action is the field for the alternator action
[01:41:58] <kirk_wallace> I think so. I think the DC "motor" is expected to drive the field windings. If there is a field then there has to be AC.(?)
[01:42:28] <archivist> field is dc
[01:42:31] <kirk_wallace> Did you see the diagram?
[01:42:48] <archivist> yes is it an Onan you have
[01:43:12] <archivist> * archivist also looking in own collection for something
[01:43:56] <kirk_wallace> Yes. All I have wired is the lower left, including the battery and starter relay. Nothing else.
[01:46:04] <kirk_wallace> The shut field shown is the outer stator housing. The rotor has a commutator and four slip rings L1 a,b ; L2 a,b.
[01:46:06] <archivist> no resistors or voltage regulator
[01:47:11] <kirk_wallace> Not right now, I'm trying to test by exciting the field with a lab power supply.
[01:48:03] <archivist> er what do you think is the field
[01:48:39] <archivist> as they are supposed to be self excited
[01:48:56] <kirk_wallace> I want to eventually use a solid state field controller - http://www.power-tronics.com/products/XR500B/XR500B.htm
[01:50:09] <fenn> kirk_wallace: the alternator "latches" like a latching relay.. the battery is recharged when the reverse current relay turns on, caused by the alternator's >12V output
[01:51:39] <kirk_wallace> I thought that it might self excite, but it doesn't seem to. I have been running at near idle, maybe that's not enough. I just don't want to burn something up. I've already smoked the AC winds once.
[01:51:55] <fenn> sometimes the magnetic field can go out in one of the coils while it's running, then you have to hit the start field coil again
[01:52:19] <fenn> if you smoked the shunt coil it might be shorted :\
[01:52:36] <fenn> = no field, no current
[01:53:28] <fenn> of course there could be other things wrong with the alternator
[01:53:58] <archivist> have you removed some of the circuit
[01:53:58] <kirk_wallace> I smoked the one of the AC windings because it was shorted.
[01:55:07] <kirk_wallace> If the field was shorted, I would think it wouldn't work as a starter.
[01:55:34] <kirk_wallace> I disconnected all wiring except the starter stuff.
[01:55:59] <kirk_wallace> I thought I would add more function as I go.
[01:56:05] <archivist> you have disconnected the self excitation as well?
[01:56:33] <fenn> um, it's a starter too? i dont see that on the schematic
[01:56:55] <archivist> start solenoid
[01:57:28] <fenn> wait a minute, does "starter" mean "DC brushed motor that spins up the engine"?
[01:57:36] <archivist> yes
[01:57:40] <fenn> or does it mean "coil to get magnetic field into the alternator"
[01:58:24] <kirk_wallace> It uses the generator as an engine starter.
[02:01:54] <kirk_wallace> My setup appears to be similar to the diagram except without a regulator and reverse current relay. Oh, and it is an RV genset.
[02:02:25] <archivist> RV?
[02:02:37] <fenn> camper trailer
[02:02:53] <kirk_wallace> Recreational Vehicle
[02:08:11] <archivist> I wonder if that gismo you want to fit will be ok as its rated at 50+volts for the field not the 12 yours may work at
[02:10:01] <kirk_wallace> That looks like problem, unless the the field needs to be in that range, but don't think it would work well as a motor if the field voltage need to be that high.
[02:11:53] <archivist> whether you can use the series coil at that voltage dunno, the supplier may know your genset and have a circuit suitable
[02:19:32] <kirk_wallace> Think I'll test it again. Start the engine, then put some DC on the output post of the starter relay and see if the AC legs come up. I'll try a higher RPM too. Thanks for the help.
[03:07:18] <tomp> trying to get vmserver 2 built on hardy.
[03:07:28] <tomp> err that gcc installed is 4.2.4 but needs 4.2.3.
[03:07:37] <tomp> gcc is a symlink to gcc4.2 which is 4.2.4, tried to chg symlink to 4.2.3 but it evades my effort,
[03:07:37] <tomp> is still -> 4.2.4. (need for cad under vmserver )
[03:08:07] <tomp> has anyone used vmware server on hardy?
[03:08:11] <tomp> how?
[03:09:20] <SWPadnos> you might try something like "GCC=gcc-4.2.3 make"
[03:10:01] <SWPadnos> or apt-get install --force-version gcc-4.2.3 (or however the option and package name are spelled)
[03:10:46] <tomp> thanks SWPadnos, needed some help ;)
[03:10:56] <SWPadnos> did one of htose work?
[03:10:59] <SWPadnos> those
[03:11:19] <tomp> aint there yet ;)
[03:11:28] <SWPadnos> ok. I was wondering :)
[03:11:45] <tomp> looking at the script beforehand
[03:20:05] <jepler> tomp: maybe this thread is relevant? http://ubuntuforums.org/showthread.php?p=6034959#td_post_6020268
[03:22:05] <jepler> (at least it doesn't seem to be an emc/rtai specific problem ..)
[03:23:12] <tomp> jepler: SWPadnos: yes the 2 symlink trick seems to be working.. you have to agree that it gcc is not correct and continue. ... time wil tell :)
[03:24:11] <tomp> and thanks i know this is all ot
[03:27:38] <jepler> I wasn't trying to criticise you for being OT -- vmware has to deal with kernel modules, and sometimes there are differences there between regular ubuntu and emc with the rtai kernel
[03:27:51] <jepler> just pleased (?) to notice the problem isn't unique to our system
[03:31:01] <tomp> i agree and not criticism taken :) just thanks
[03:32:43] <SkinnYPuPp> I've yet to install the newest vmware server and about the only stumbling block I recall was the header files were in a different path than the default install path
[03:35:08] <SkinnYPuPp> on all previous versions i had used that is
[03:39:29] <tomp> SkinnYPuPp: maybe the 2 symlinks jepler mentioned will help with that
[03:40:18] <tomp> i'm instaling the client now, so still untested
[04:00:34] <cradek> wow, nebraska split its electoral votes for the first time.
[04:06:06] <SWPadnos> huh. why would they (you) do that?
[04:06:56] <KimK> * KimK thinks SWPadnos asks an excellent question
[04:07:22] <KimK> * KimK is from Nebraska
[04:07:55] <toastydeath> some states were saying they'd modify the electoral college votes based on the popular vote
[04:08:01] <cradek> it's always been possible, but has never happened before
[04:08:04] <toastydeath> dunno any more about that though, it wasn't in the news much
[04:08:12] <SWPadnos> interesting
[04:08:19] <cradek> nebraska and maine can split, all others are winner-take-all
[04:08:33] <SWPadnos> it was ~57% McCain/41% Obama there, so I guess it'll go 3+2
[04:08:41] <cradek> obama won district 2 which is approximately omaha
[04:08:45] <SWPadnos> oh, I didn't realize it was a splitting state
[04:08:49] <SWPadnos> heh
[04:09:59] <KimK> Problem is, Nebraska only has 5 electoral votes to begin with, so what happens is, candidates (all sides) generally bypass/ignore Nebraska because they could only pick up one vote anyway, probably
[04:10:11] <cradek> (I don't think maine has ever split)
[04:10:53] <cradek> KimK: it's true that we get ignored, but that might also be because we always end up republican no matter what (except this time!)
[04:12:26] <cradek> I guess I don't care if they visit here - I know perfectly well enough about them - the telegraph has a daily report
[04:12:31] <cradek> or maybe the internet
[04:12:55] <SWPadnos> that's "the big interconnected tube thingy"
[04:13:00] <KimK> As one who thinks the electoral college is a good thing, I believe we should go back to winner take all, regardless of red/blue factor. I see no movement in that direction, however.
[04:13:41] <cradek> I don't have an (informed) opinion about that I guess
[04:13:45] <SWPadnos> I'm not so sure the electoral college is a good thing, nor is plurality voting necessarily a good thing
[04:14:10] <cradek> ... but I'm inclined to agree with SWPadnos
[04:14:24] <KimK> Ha, selection by lottery, perhaps?
[04:14:28] <cradek> these things encourage/enforce a two-party system which is detrimental to us all
[04:14:35] <fenn> SWPadnos: i never knew you were an anarchist!
[04:14:44] <KimK> Short straw is forced to be president, lol
[04:14:47] <SWPadnos> plurality voting because it lessens the need to reach consensus, since it's winner take all
[04:14:51] <SWPadnos> heh
[04:15:14] <SWPadnos> no, not voting at all, but making it so winning by one vote means you can ignore the ~50% of the people who didn't vote for you for the next 4 years
[04:16:15] <SWPadnos> it doesn't matter much anyway, we have two major parties that for the most part could be one party with some squabbling
[04:17:27] <KimK> Say, while you experts are in attendance, is there a good (easy) way to preview a bunch of text files in Ubuntu? I'm trying to do a crashed drive recovery for friend (no, no backup, of course) and I've ended up with a lot of text files of questionable validity. If I could just see the first few lines of each and then "catch or release", that would be great.
[04:17:53] <SWPadnos> zoom in a lot with the folder view
[04:18:01] <SWPadnos> the test in the icon is the actual text in the file
[04:18:06] <SWPadnos> s/test/text/
[04:18:11] <KimK> OK, I'll try it...
[04:18:26] <cradek> head -10 *
[04:18:57] <SWPadnos> for a in * ; do echo $a && head -10 $a ; done
[04:18:59] <SWPadnos> :)
[04:19:15] <cradek> SWPadnos: head gives you a nice header with the filename all by itself
[04:19:29] <SWPadnos> oh, hadn't noticed that before
[04:19:39] <SWPadnos> is that only with multiple files?
[04:19:48] <cradek> yes
[04:19:57] <SWPadnos> ok, no wonder I hadn't seen it :)
[04:20:10] <cradek> see also: tail -f *
[04:20:16] <cradek> a longtime favorite
[04:21:27] <SWPadnos> I'll have to try that one day
[04:21:35] <fenn> great for watching logs
[04:25:52] <KimK> thanks, icon viewer works, and easy to right-click the klunkers
[04:26:50] <SWPadnos> cool
[04:27:08] <SWPadnos> I was a little surprised to see that the icons had the actual text in them, but it is a cool feature
[04:32:22] <KimK> Has anyone used ddrescue? I can recover all but about 24 blocks of 512 bytes (3000 hex or 12,288 bytes) from an 80GB HDD, and it is still enough to cause problems. 2000 hex in one area and 1000 hex in another. If you have any suggestions (other than back up next time), I'd appreciate them.
[04:32:47] <cradek> dang, you excluded my only useful advice
[04:33:09] <KimK> Ha, yeah, I'm giving him that already, lol
[04:33:10] <cradek> it has just that tiny bad spot but still won't mount?
[04:33:59] <fenn> KimK: e2fsck should pull out the broken files and put them in lost+found, then at least you know what's broken
[04:36:03] <KimK> Oops, sorry, I forgot to mention its a crashed windows xp usb storage drive, so FAT32/vfat. Yes, it was the "backup" storage drive. (Wait, you mean the files weren't anyplace else?)
[04:36:20] <cradek> ohhh
[04:36:31] <cradek> boot a linux cd, mount the partition, copy everything off
[04:36:39] <cradek> and no I'm not kidding
[04:36:50] <cradek> linux will mount windows filesystems much more readily than windows will
[04:37:19] <fenn> reading is no problem, its writing to them that linux balks at
[04:37:28] <SWPadnos> it's USB - easier to just plug it into a Linux box :)
[04:37:30] <cradek> yeah, don't do that, just copy the data off
[04:37:38] <toastydeath> you cannot simply
[04:37:40] <toastydeath> copy data
[04:37:41] <toastydeath> into mordor
[04:37:47] <KimK> I know you're not, that's exactly what I've been doing. And using ddrescue to recover drives that are "difficult" for dd or dd_rescue.
[04:37:49] <SWPadnos> um yes, yes you do
[04:37:58] <cradek> mordor?
[04:38:10] <cradek> KimK: oh sorry, sometimes my reading comprehension is limited
[04:38:27] <SWPadnos> paraphrasing Boromir in a funny LOTR video
[04:39:51] <cradek> oooh, this velocity limit thing is cool
[04:40:59] <KimK> But I didn't entirely trust the USB connection, so I uncased and went direct to IDE. I've tried as a slave. Now about to try building another image on a solo master IDE connection. Not sure if it will help, but read of some difficulties with 2 drives where one is defective, so thought it might help.
[04:41:55] <fenn> you know electron microscopes go for quite cheap these days
[04:42:08] <KimK> cradek: Oh, you're working on your velocity limit? What news there? And any recent Mesa / hostmot2 news? (I've been out of touch a bit lately)
[04:43:01] <cradek> KimK: velocity limit works - currently AXIS gets a little confused if you change the vel limit in another gui, but it's just a display problem. The control part is done and working well.
[04:43:43] <cradek> jepler says he hopes to help fix up the gui issues later this weekend.
[04:44:22] <KimK> cradek: excellent and excellent! will we see it in the next release? And when will that be?
[04:44:23] <cradek> I don't know the latest about hostmot2 - all I have seen is the stuff on the lists.
[04:44:35] <cradek> it will be in EMC 2.3
[04:45:02] <cradek> along with the "spindle at speed" stuff
[04:45:12] <cradek> (and a zillion other things)
[04:47:24] <KimK> spindle at speed? also excellent, I didn't even know that was missing or that you guys were working on it, but it will certainly be needed. Is there a list of the "other things" posted anywhere?
[04:48:29] <cradek> http://cvs.linuxcnc.org/cvs/emc2/debian/changelog?rev=1.44
[04:48:47] <KimK> And spindle speed calibration for different gear ranges? Now that parameters are fading in favor of pins, will speed ranges be OK?
[04:49:08] <cradek> I use a mux2+scale for my lathe's two gears
[04:50:34] <cradek> that list is pretty much what will be new in EMC 2.3 (I guess vel limit isn't in there yet)
[04:50:43] <KimK> I haven't tried to do speed ranges in EMC2 yet, so I'm not sure what you're telling me there, but I'm looking at some spindles with gear changes that need retrofitting
[04:51:36] <cradek> the spindle speed request output is always in RPM. To get a voltage for a VFD for instance, you scale it according to the selected gear, and send that out a dac
[04:52:00] <cradek> you will have a different scale value for each gear, that's all there is to it
[04:52:26] <cradek> you can use mux2/mux4/mux8 to select the right one
[04:52:40] <cradek> if you have more than 8 gears you might have to make a mux16.
[04:54:40] <KimK> And this is done only in hal? (Assuming inputs from gear change switches to hal) Or does it also go through classicladder?
[04:55:20] <cradek> you can do it however you like - on my lathe, classicladder does the gear change, but the scaling for the dac output is in hal
[04:55:35] <cradek> not sure classicladder can deal with floats (the scale values)
[04:55:56] <cradek> setp mux2.0.in0 .004
[04:55:56] <cradek> setp mux2.0.in1 .0154
[04:55:56] <cradek> net ScaleValue mux2.0.out scale.1.gain
[04:55:56] <cradek> net MotorCommand scale.1.out => m5i20.0.dac-03-value
[04:56:00] <KimK> I don't think it should have to, so that's fine
[04:56:37] <cradek> and then select with the same thing that does the gear change like so:
[04:56:37] <cradek> net LowClutch classicladder.0.out-04 m5i20.0.out-03 mux2.0.sel
[04:56:39] <KimK> .004? what is that, volts/RPM?
[04:56:44] <cradek> yes
[04:57:01] <KimK> OK, sounds good
[04:57:49] <cradek> in this setup, ladder knows when it's safe to change gears
[04:58:30] <KimK> and that's a good thing, lol
[04:58:51] <cradek> yes
[04:59:04] <cradek> the original control would do WHAT YOU SAY when you say it
[04:59:15] <cradek> 3000 rpm and change gears? ok, here goes...
[04:59:27] <KimK> bzzzzzzzzz!
[04:59:36] <cradek> SQUUUUAAWWEEEWKKKKK
[04:59:54] <cradek> it uses ... clutches and belts and stuff
[05:00:10] <cradek> so it could get away with it, but it didn't sound nice
[05:00:45] <KimK> although there is at least one type of gearbox that can't crash (could do other things, I suppose)
[05:02:11] <KimK> so congratulations on the almost finished improvements. When is 2.3 due?
[05:02:31] <cradek> haha, you keep asking that, I keep not answering
[05:02:58] <cradek> a date is not set, but I hope pretty soon
[05:03:47] <KimK> I'm a persistent interviewer, lol. OK, better it's done when it's ready, I suppose. Don't want to be "fixed date" releases like the ubuntu folks, and then send out update after update.
[05:04:02] <cradek> we haven't quite roped anyone into committing to be the 2.3 release manager yet
[05:07:53] <tomp> the prev fix to vmware server wont get you past the broken links with the client, this (tho in chinese) will get you past that bugger http://metavige.blogspot.com/2008/05/ubuntu-vmware-server.html
[05:08:11] <KimK> BTW, do you guys have any pull with the ubuntu folks? I'm with the what, 35% who are still on dialup. And I see the suspiciously same-sized files coming in again and again from different repositories. archive.ubuntu.com and then us.archive.ubuntu.com and then archive.us.ubuntu.com and then, and then, and then... can't these guys fix this? How many times do I have to download the same files?
[05:08:41] <cradek> nope, sorry, but they have some great forums
[05:08:53] <cradek> I don't know how responsive the experts are in those forums, though
[05:10:25] <KimK> not to ask a stupid Q, but why is it hard for you to get a 2.3 release manager?
[05:11:30] <KimK> (sorry, I'm probably not aware of your development problems, which I'm sure are many, I just don't know the history)
[05:11:41] <cradek> it can be a lot of work, a time commitment
[05:12:04] <KimK> your = your group's
[05:12:11] <cradek> anyone can do the mechanical part of building releases, but doing the work that makes the releases GOOD takes time
[05:12:38] <cradek> you have to pretty much follow everything that's going on, and decide what's safe to put in the stable releases
[05:12:47] <cradek> nobody likes a regression (something broke that used to work)
[05:13:43] <cradek> then you have to do the building, uploading, announcing, write the changelog, etc etc. for each release you make
[05:14:28] <cradek> I did the 2.0 and 2.1 series, jepler did 2.2, we don't know yet who will do 2.3
[05:15:07] <ewilhelm> and none of that is nearly as fun as writing new code! :-D
[05:15:15] <cradek> so right
[05:15:35] <KimK> and the developers (all the usual suspects) want to stick with developing, and have someone else release? (Ha, I guess so!)
[05:15:36] <cradek> but it's so very important because that's what most users use
[05:15:50] <ewilhelm> * ewilhelm has some experience of that with maintaining CPAN modules
[05:16:54] <KimK> cradek: there you go
[05:17:14] <cradek> I should go to bed - nice talking to everyone
[05:17:51] <KimK> I would offer, but don't have enough experience w/ building, cvs, etc. OK, cradek, goodnight
[05:18:06] <cradek> oh yeah, you have to know cvs pretty well :-)
[05:19:28] <KimK> cradek: so ask ewilhelm, above
[05:20:02] <ewilhelm> * ewilhelm is not the droid you're looking for
[05:20:12] <KimK> ha
[05:21:47] <ewilhelm> the above was me sympathizing, not volunteering ;-)
[05:22:58] <ewilhelm> writing the changelog as you go might be a good habit, but it's difficult to stick to (and the releaser usually still has to check the logs)
[05:48:46] <tomp> is the rtc function compiled into the kernel used by the live cd? ( a pre-req of vmware server 2 )
[05:51:23] <tomp> this gawd awful ubuintu ... how can you 'su -' when theres no root account!?!?!
[06:43:44] <KimK> tomp: you're not sudo'ing? or gksudo'ing?
[07:31:32] <tomp> Kimk no idea of whatto try now... yes i sudo'd and i sudo -s -H but vmware rejects its own security certificate, saying it is self signed ( well yes Duh! ) who else woud be invoved in a singe server single client environment ) argh!
[07:49:06] <KimK> tomp: Sorry, I don't know what to tell you about the vmware certificate. Wait, you're trying to run the EMC2 live CD on vmware? Does that work? It seems like it wouldn't. You should ask someone besides me, though.
[07:51:07] <KimK> Well, maybe it would work if you just want to look at some of the GUIs and so on. But I'd guess the real-time part, not so much.
[07:52:34] <KimK> I believe EMC2 uses RTAI, if that helps
[09:29:12] <anonimasu> morning
[10:49:09] <kaien1> hello
[11:56:54] <steve_stallings> steve_stallings is now known as steves_logging
[11:59:09] <markdashabout> jepler: are you reading this ? wanted to continue last night discussion....
[12:08:37] <anonimasu> hm, think it's worth buying Machine Tool Reconditioning and Applications of Hand Scraping, Reprint
[12:10:41] <anonimasu> crap
[12:10:47] <anonimasu> they dont have precision machine design avaiable :/
[13:24:29] <[1]a-l-p-h-a> [1]a-l-p-h-a is now known as a-l-p-h-a
[13:27:47] <JymmmEMC> Hey a-l-p-h-a, ltns
[13:34:44] <anonimasu> toastydeath: formula 1 is awesome
[13:34:45] <anonimasu> :p
[15:01:49] <Vq^> i experience problems with exact path mode and exact stop mode (g61, g61.1) in 2.2.5
[15:02:39] <Vq^> it turns about 50mm from the commanded position when working in one axis
[15:03:20] <Vq^> i've also tested g64 with a number of different P values but the result is the same
[15:03:52] <skunkworks> that seems not right.
[15:05:12] <cradek> 50mm??
[15:05:40] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TrajectoryControl
[15:05:44] <cradek> put your test program on pastebin please
[15:05:59] <skunkworks> you can see from the first picture that the tolerance works...
[15:06:23] <skunkworks> good morning
[15:06:27] <cradek> hi
[15:07:36] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Released
[15:08:05] <cradek> I see there was a g61/g64 related bug fixed in 2.2.6. Be sure to test the latest version.
[15:09:11] <JymmmEMC> mornin all
[15:10:48] <skunkworks> I think winter finally came here.
[15:11:15] <cradek> whee
[15:18:39] <cradek> http://retrothing.typepad.com/photos/uncategorized/2008/11/05/edison.jpg
[15:19:19] <anonimasu> cradek: talk about coincidence I saw that in my physics book at tursday
[15:19:57] <anonimasu> cradek: think this is a good read http://www.powells.com/cgi-bin/biblio?show=HARDCOVER%3ANEW%3A9781114266612%3A92.95
[15:22:42] <cradek> yep, good but maybe boring
[15:22:57] <cradek> I want one but am unwilling to pay $80-$90 which is the going price.
[15:23:30] <anonimasu> I'm pondering ordering it today
[15:23:56] <anonimasu> that and precision machine design(but they done have it)
[15:24:00] <anonimasu> :/
[15:24:09] <archivist> * archivist reads anonimasu's copy
[15:26:15] <skunkworks> that is funny. There is a show I listen to on npr (this american life) and it was the halloween episode. There was a familly that moved in to a big old house that started hearing and seeing things (call them ghost). After a doctor hearing the story - he figured it was carbon monoxide poisoning. And it was - they found I think the furnace was leaking. I wonder how much of the ghost storys where because of crappy gas li
[15:27:09] <anonimasu> hehe
[15:29:08] <Vq^> cradek: http://arda.no-ip.org/rdump/emctest/
[15:30:12] <Vq^> as i said i tested with g64 as well, but there is no other difference than that the g61 code is replaced with g64 and a P value
[15:31:54] <skunkworks> you probably need to post your ini file also
[15:33:16] <Vq^> i've uploaded it to the same folder
[15:33:28] <skunkworks> This is the episode - it is quite funny - (I like david sedaris) (oct 31 show) http://www.thislife.org/Radio_Archive.aspx
[15:33:42] <Vq^> but why would the ini file make a difference?
[15:33:58] <Vq^> could it change the meaning of exact stop mode?
[15:34:27] <skunkworks> if it is a bug - there might be a clue in your accell settings and such
[15:35:08] <Vq^> ok
[15:35:20] <anonimasu> cradek: I think to get a clue about scraping :)
[15:35:27] <anonimasu> cradek: (the book I mean)
[16:35:26] <jadams_> can anyone suggest a good kit for building an EMC-compatible CNC? Or pre-built ones for purchase?
[16:35:40] <jadams_> is there such a marketplace where some portion of the proceeds go to the EMC project?
[16:35:41] <jadams_> etc.
[16:38:59] <archivist> so many ways....
[16:41:21] <archivist> what do you want to machine and make
[16:41:36] <jadams_> I'd like an xyz dremel basically
[16:41:50] <jadams_> z travel can be minimal
[16:42:52] <archivist> PCB making or general engraving
[16:42:56] <jadams_> engraving
[16:43:20] <jadams_> ultimately, I want to move that knowledge elsewhere. a giant xyz power drill would be fun.
[16:43:52] <archivist> well best way to gain knowlege is to do most yourself
[16:44:13] <jadams_> my dad actually runs a robotics engineering company, and I built tons of stuff for him in my youth. I'm not at a lack of ability to build
[16:44:20] <archivist> later on perhaps retrofit an old mill
[16:44:39] <archivist> ah source of parts!
[16:44:43] <jadams_> yeah
[16:44:45] <jadams_> I have parts
[16:44:52] <jadams_> but I lack time, in large part
[16:44:57] <jadams_> I run a software development company
[16:45:10] <jadams_> as such, I'm pretty much running full bore 90% of the time and sleeping the remainder
[16:45:38] <archivist> well some dremel grade stuff I see on ebay
[16:45:40] <jadams_> I might just build this thing with a friend of mine: http://lirtex.com/index.php/cnc
[16:46:04] <jadams_> but I kind of have a love affair with 80/20
[16:46:05] <archivist> add stepper drivers and a power supply
[16:46:42] <jadams_> I've also got an arduino handy if that helps
[16:46:56] <jadams_> figure it can maybe be a 3-axis plus home/limit stepper driver with minor software...anyone done that?
[16:47:47] <archivist> arduino is the hard way
[16:47:50] <jadams_> kk
[16:49:06] <archivist> download the live CD (see topic) and connect steppers via drivers to par port, make stuff (dont forget steps I missed)
[16:49:56] <jadams_> ha
[16:49:57] <jadams_> awesome
[16:52:29] <archivist> I put together one, added stepper drivers and a power supply connected up, ran stepconf, then it was a runner
[16:55:47] <archivist> jadams_, http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Case_Studies
[17:05:45] <fenn> yuck @ lirtex.com
[17:06:31] <fenn> the design he is following is called a "mcwire"
[17:07:08] <archivist> * archivist is looking at Kirks tool changer
[17:07:19] <DanielFalck> the 'pillow block' setup is pretty cheesy
[17:08:25] <archivist> * archivist saw plastic and.....
[17:09:58] <fenn> jadams_: here's a better example of a milled pcb, one of the guys in this channel: http://timeguy.com/cradek/cnc/pcb
[17:10:13] <jadams_> fenn: cool
[17:11:49] <fenn> basically, you need to use a single flute carbide V cutter, and get rid of any backlash in the mechanism
[17:13:49] <fenn> http://thinktink.com/stack/volumes/voli/store/mechmill.htm
[17:15:04] <fenn> i just use a laser printer and a hacked laminator
[17:52:23] <cradek> fenn: did you see this? http://lirtex.com/images/robotics/CNC/3_Axis_CNC_Controller_thumbnail.JPG
[17:52:32] <cradek> worst photoshop job ever
[17:55:02] <fenn> hahaha that doesnt even make sense
[17:55:44] <fenn> mostly the yuck was at http://lirtex.com/images/robotics/CNC/PCB_Isolated_Board_640x480.JPG
[17:55:57] <cradek> ooh, yikes
[17:56:21] <cradek> that's not a very satisfying result, is it
[17:56:48] <cradek> dang, that makes mine look good
[17:57:32] <archivist> by young fred aged 5.....
[17:59:16] <cradek> on the bridgeport_mill mailing list, some guy says was cleaning the grease out of his mill's head and quill housing, and kept finding little balls...
[17:59:20] <archivist> cradek found "subdividing simplices" in graphics gems, may be suitable for finding the middle of glyphs
[18:00:02] <cradek> archivist: what can I do to help?
[18:00:14] <archivist> dunno yet
[18:01:15] <archivist> was reading your source a bit to learn, and trying to find a spec for glyphs
[18:01:54] <fenn> looks similar to a quadtree
[18:02:22] <cradek> truetype glyphs are closed paths, possibly with holes (another path inside, going in the negative direction)
[18:03:43] <archivist> yes but whats the structure they are in, line/points/what
[18:04:31] <cradek> conic splines
[18:05:07] <fenn> conic == parabola and circle? (hyperbolas seem far fetched)
[18:05:28] <cradek> or maybe they are cubic
[18:05:38] <cradek> postscript is one, truetype is the other...
[18:06:25] <fenn> postscript = cubic, truetype = quadratic
[18:06:33] <cradek> conic = 1 control point, cubic = 2
[18:06:40] <fenn> right
[18:06:47] <cradek> ok
[18:06:54] <cradek> (ttt handles both kinds)
[18:07:02] <fenn> nice
[18:07:30] <fenn> so it wouldnt be much work to trace postscript outlines?
[18:07:48] <cradek> it already does that (.pfb files)
[18:09:16] <cradek> ah cool, that includes 'computer modern' (the basic tex font)
[18:09:21] <archivist> hmm I dont need to know as the library gets the outline :)
[18:09:28] <cradek> archivist: exactly :-)
[18:11:12] <archivist> but... I have not spotted where a two or more part glyph is dealt with eg I
[18:11:16] <archivist> i
[18:11:35] <cradek> http://timeguy.com/cradek-files/emc/computer-modern.png
[18:12:22] <alex_joni> cool :)
[18:12:45] <cradek> archivist: looks like you set up callbacks and then FT_Outline_Decompose does the dirty work (see the func_interface stuff)
[18:13:15] <archivist> ah ok
[18:18:12] <alex_joni> oh yuck: http://www.youtube.com/watch?v=Tzw5MAWSAyk
[18:19:07] <cradek> yuck?
[18:19:11] <alex_joni> looks like the scale is wrong
[18:19:18] <alex_joni> look at the letter at end of the video
[18:19:51] <cradek> oh, bizarre
[18:20:00] <cradek> I think that's machine flex
[18:20:10] <izua> that looks weird
[18:20:20] <cradek> yeah, it's pretty wrong
[18:20:30] <izua> maybe it should have done multiple passes at various Z
[18:21:15] <alex_joni> well.. he says he has 0.4 mm accuracy :D
[18:21:25] <fenn> maybe shouldnt have held his bearings in place with duct tape
[18:21:57] <izua> yes, i'm sure it does. on the rotating axis.
[18:22:07] <alex_joni> I'm surprised he didn't toast his motors
[18:22:31] <alex_joni> http://lirtex.com/index.php/stepper-motor-controller
[18:23:24] <izua> that is so wrong.
[18:23:54] <izua> the video definitely uses microsteping, while he explains fullstep in the tutorial
[18:26:06] <cradek> archivist: http://www.youtube.com/watch?v=PlkT8-AQ_MA&NR=1
[18:26:28] <cradek> looks like this does something in the corners - you can see it
[18:27:10] <cradek> maybe that's something to consider - just do the corners so they are sharp.
[18:30:24] <archivist> good find cradek
[18:35:58] <archivist> he seems to do corners as we/I are thinking but some odd ordering of the cuts
[18:36:31] <cradek> yes
[18:38:05] <archivist> hmm more machining this weekend or fonts!!!
[18:54:08] <jmkasunich> I think he's doing a roughing pass
[18:54:35] <jmkasunich> then the finish cut is the "skeleton" cut with corner ramps
[18:56:54] <dmess> right the corner ramps leave a sharp cornre rather than conical
[18:58:11] <archivist> just whats needed for a hand engraved look where the font is suitable
[19:01:09] <dmess> depending on size of tool and depth of letter they may require some hand finishing as the conical cutter doesnt dig out the whole cone.. it gets the corner tight... but as a function of tool diam. leaves mat'l at the edges of the cut
[19:36:58] <fenn> comic sans ftw!
[19:46:25] <anonimasu> http://www.youtube.com/watch?v=jAjRxIl2_bA&feature=related
[19:57:15] <dmess> anyone around know how to install a VM with win2000 so i can run catia on this box???
[20:01:47] <dmess> im still at 6.06 though... will that be a problem??
[20:02:58] <anonimasu> dmess: not really
[20:03:05] <anonimasu> dmess: vmware?
[20:04:55] <alex_joni> dmess: shouldn't be a problem
[20:05:04] <alex_joni> I ran vmware on 6.06 without problems
[20:06:29] <anonimasu> it shouldnt be a problem
[20:18:53] <archivist> hard disk in a tray is as close to VM as I need to get
[20:39:47] <Vq^> alex_joni: hello mr Joni
[20:42:57] <Vq^> alex_joni: do you recognize my problem with getting blending despite using G61 or G61.1 ?
[20:53:31] <Guest304> Guest304 is now known as skunkworks_
[21:00:10] <maddash> SWPadnos: are you busy?
[21:01:10] <SWPadnos> maddash, not especially
[21:02:17] <maddash> SWPadnos: I'm having some troubles with an HD44780-style LCD controller - can I ask you some questions about it?
[21:02:40] <SWPadnos> you can, but I haven't written any software for an LCD controller in years
[21:03:31] <archivist> about 3 years for me
[21:04:02] <maddash> ok. the chip reads parallel data 8-bits at a time from 8 pins of my uc...then, there are separate "strobe," read/write, and register select pins
[21:04:24] <maddash> hm, why don't i link the datasheet instead: http://www.crystalfontz.com/controllers/Sunplus_SPLC780D_Controller.pdf
[21:04:45] <maddash> er, this one's easier to read: http://class.ee.iastate.edu/cpre288/labs_fall2008/Lab1/nhd-0420h1z-fsw-gbw.pdf
[21:05:30] <maddash> ok, so is a strobe pin an external clock? e.g., is it supposed to oscillate between 1/0 periodically? or is it just a signal for valid data?
[21:05:41] <SWPadnos> oh, a module, not direct to the controller
[21:05:56] <SWPadnos> it says "load now, the data lines are valid"
[21:06:30] <maddash> so if I'm not sending data, it's not supposed to be raised (assuming that strobe=1 means valid data)?
[21:06:39] <SWPadnos> you set the R/W, SEL (sometimes called A0), and data lines, then you hit the E line to clock the data in
[21:06:57] <SWPadnos> I believe the rising edge loads data in, so you can leave the pin in any state
[21:07:21] <SWPadnos> but it's best to explicitly set it low, so you're guaranteed to get a rising edge the next time you set it high
[21:07:37] <SWPadnos> (or the other way around, I don't remember which edge is active)
[21:07:50] <maddash> okay, so I have the strobing part correct
[21:08:09] <maddash> I can't figure out why this damn thing isn't displaying anything
[21:08:20] <SWPadnos> 1) contrast
[21:08:28] <SWPadnos> 2) incorrect initialization sequence
[21:08:46] <SWPadnos> 3) init sequence or data are being sent too fast
[21:09:05] <SWPadnos> 4) you're continuously refreshing the display data, which can cause the display to flicker rapidly
[21:09:11] <archivist> there is a specified delay (slow beasts)
[21:09:12] <SWPadnos> (which changes the effective contrast)
[21:09:37] <SWPadnos> yes, the init sequence has a few things like "delay at least 10 ms but no more than 50ms, then send the next code"
[21:10:18] <SWPadnos> what's your MCU clock speed?
[21:11:12] <maddash> uc is clocked @ 72MHz
[21:11:40] <SWPadnos> ouch. you'll need a lot of micro delays then, just to keep pulse widths large enough
[21:11:55] <SWPadnos> you need at least 36 clocks high for E, and at least 24 clocks low
[21:12:05] <maddash> i have a 16-bit timer clocked @ 1:256 prescaler of 36MHz counting up to 0xFFFF to create a 466 ms delay
[21:12:26] <maddash> a 466 ms delay between every signal transition
[21:12:26] <SWPadnos> you also need to delay 8 clocks from the time you set the address and data lines before you hit the E line
[21:12:33] <SWPadnos> 466 ms?
[21:12:41] <SWPadnos> that's far too long for the init sequence
[21:13:02] <maddash> I've got the mcus hooked up to LEDs so that I can watch the init sequence and all the data I send out
[21:13:09] <maddash> MCU*
[21:13:30] <SWPadnos> I don't see how an LED would help there. you need a scope
[21:13:54] <SWPadnos> are you using the timer interrupt to do the init also, or is that inline somewhere in your startup code
[21:13:56] <SWPadnos> ?
[21:15:49] <maddash> I hardcoded the init sequence and put in timer interrupts between every step-up,step-down of E
[21:16:00] <maddash> technically, I'd need 6 scopes,, since I have an 8-bit data feed...but I watch the LEDs for digital highs and lows
[21:16:20] <archivist> nah 1 logic analyser
[21:16:22] <SWPadnos> hold on - for the initialization, is the timer running and controlling the E line?
[21:16:27] <SWPadnos> one MSO ;)
[21:18:03] <maddash> no, I use the timer by reseting the timer counter (TMR1) to zero and doing a tight loop: "while (TMR1<0xFFFF);"
[21:18:36] <SWPadnos> is that to prevent C from optimizing away busy loops?
[21:18:38] <maddash> does that make sense?
[21:19:01] <SWPadnos> I understand it, but it doesn't make much sense to do that for init ;)
[21:20:54] <maddash> NVM, CONTRAST SCREW-UP
[21:21:07] <SWPadnos> heh, step 1 ;)
[21:22:04] <maddash> omfg I'm such a moron
[21:23:09] <maddash> SWPadnos: thanks a zillion
[21:23:17] <SWPadnos> sure. happy LCDing
[21:24:12] <maddash> after init, "while (1) { sendchar('A');sendchar('P');sendchar('P');sendchar('L');sendchar('E');sendchar('S'); }
[21:24:13] <maddash> "
[21:25:38] <maddash> omg, wtf? the module is a 20x4 array of chars, but it goes from line 1 > line 3 > line 2 > line 4
[21:26:10] <SWPadnos> that's relatively common
[21:26:42] <maddash> :(
[21:26:43] <renesis> heh
[21:26:46] <SWPadnos> the controllers usually have 20 or 40 character wide internal buffers, so 2-line displays have line 1 at addresses 0-15 (or 19), and line 2 at 40-55 (or 59)
[21:27:02] <SWPadnos> for the 4-line display, they just use both halves
[21:27:23] <SWPadnos> it makes some of your life easier because a 20x4 display has a contiguous address space
[21:27:49] <SWPadnos> a 16x2 doesn't, so you have to know which line you want to be on, and whether your string needs to "wrap", since you have to wrap it manually
[21:28:21] <maddash> so much for "industry standard"
[21:28:21] <SWPadnos> the displays also support horizontal scrolling, so you could have a "virtual screen" that's wider than the display, if you have a 16x module
[21:28:28] <SWPadnos> that is the industry standard
[21:52:35] <JymmmEMC> I just like making alien creatures on them (like pacman)