#emc | Logs for 2007-11-10

Back
[00:42:33] <Guest847> Guest847 is now known as skunkworks
[01:44:49] <lerman_> lerman_ is now known as lerman
[02:30:44] <Alf> Alf is now known as mbaulfinger
[03:37:32] <cradek> wow, as I suspected, the body of my vise slopes .004 from one side to the other.
[03:37:49] <SWPadnos> bummer
[03:37:57] <SWPadnos> time to get out the carbide end mill I guess
[03:38:09] <cradek> you'd think that would be fairly easy to get right, considering it must be done on a surface grinder
[03:38:44] <cradek> I have a face mill with carbides... nahhh.
[03:38:44] <SWPadnos> have you looked for dings or corrosion on the bottom of the vise or the table under it?
[03:39:02] <cradek> yeah I stoned and checked it very carefully
[03:39:10] <SWPadnos> hm
[03:39:18] <cradek> I didn't necessarily trust the table - I measured it on the surface plate
[03:39:32] <SWPadnos> ok, so you stoned the vise and the table?
[03:39:40] <cradek> yes
[03:40:03] <SWPadnos> damn. it's really har dto stop eating Almond Joy and Reeses peanut butter cups
[03:40:06] <cradek> the table is as flat as I can measure by circling an indicator
[03:40:12] <cradek> .004 is very easy to detect
[03:40:13] <SWPadnos> ok, that's a good thing
[03:40:33] <SWPadnos> have you tried measuring the thickness of the vise base?
[03:40:38] <cradek> I've been using it with .003 feeler stock under the right side
[03:40:54] <cradek> the base casting is very rough except the top and bottom surfaces
[03:41:14] <SWPadnos> hmmm. so it's flat on the bottom, the table is flat on the top, and there's a gap on one side?
[03:41:25] <SWPadnos> that doesn't sound possible
[03:41:44] <cradek> the vise's base surface and top (working) surface are not parallel
[03:41:54] <SWPadnos> sure, that's always possible :)
[03:42:00] <cradek> I've been compensating by lifting one side of its base
[03:42:16] <cradek> yes, that's exactly what I measured
[03:42:29] <cradek> it's very evident
[03:43:09] <SWPadnos> ok. then I think a grinder/flycutter is your only solution
[03:43:12] <SWPadnos> or a new vise
[03:43:31] <cradek> yeah it definitely needs to go on a grinder.
[03:43:58] <cradek> but, easier said than done.
[03:45:09] <DanielFalck> cradek:what brand of vise?
[03:45:30] <toastydeath> you could learn to scrape
[03:45:32] <toastydeath> and scrape it flat
[03:45:47] <cradek> DanielFalck: chinese special
[03:46:10] <cradek> it clamps well, the jaw doesn't ride up, but it sure isn't flat
[03:46:10] <toastydeath> local community colleges, if they have a machine shop, usually have a little grinder
[03:46:25] <toastydeath> you might sign up for a class or ask them to use the grinder
[03:46:31] <DanielFalck> get a Kurt and use the chinese one for non-critical clamping away from the mill
[03:46:49] <cradek> DanielFalck: sure, all I need is $hundreds
[03:47:11] <DanielFalck> I know
[03:47:26] <DanielFalck> I have 2 identical kurts on my mill
[03:47:35] <DanielFalck> lots of $$
[03:47:42] <cradek> good, you can spare one
[03:47:49] <SWPadnos> actually, I have a friend who bought a Grizzly Sine vise, and it wasn't too expensive but is very very good
[03:47:57] <SWPadnos> he was quite surprised at it
[03:48:05] <toastydeath> it tends to be really hit or miss
[03:48:16] <toastydeath> at least from what i see
[03:48:28] <jlmjvm> a kurt vise and power drawbar are on my wish list
[03:48:39] <SWPadnos> I also have a spare Bridgeport vise
[03:48:46] <cradek> this probably had crud under it when it was ground. the next one in line may have been perfect
[03:49:04] <toastydeath> exactly
[03:49:59] <toastydeath> how out is it
[03:50:03] <toastydeath> a little, a lot
[03:50:08] <toastydeath> is it cast iron or hardened something
[03:50:23] <tomp> if real cheap, maybe shims/jackscrews till you get to a grinder
[03:50:31] <cradek> .004 from side to side
[03:50:36] <toastydeath> flycut it down
[03:50:42] <toastydeath> i support whoever said flycut it
[03:50:48] <jlmjvm> i reloaded ubuntu completely on my machine with the latency problem and it seems to be ok now
[03:50:48] <toastydeath> it will come out at least to .001
[03:50:50] <cradek> tomp: yeah that's what I've been using. I can get it pretty level.
[03:51:17] <jlmjvm> have both computers working with 2.2 now with no rt errors
[03:51:23] <cradek> jlmjvm: yay
[03:51:37] <cradek> jlmjvm: wonder if you ended up with a different video driver or something.
[03:51:39] <jlmjvm> that step conf is sweet
[03:51:44] <Skullworks-PGA1> .004" just find the right paper stock to use
[03:51:45] <tomp> hold a couple flat stock bars and cut yourself unparallels ;
[03:52:05] <jlmjvm> im actually using a vid card i couldnt use before
[03:52:06] <cradek> Skullworks-PGA1: been using a piece of 12" feeler stock
[03:52:13] <Skullworks-PGA1> or brand of masking tape
[03:52:17] <cradek> tomp: haha unparallels
[03:52:48] <cradek> pseudoparallels
[03:53:20] <Skullworks-PGA1> problem with un-parallels is not getting the sides mixed up and doubling the error
[03:53:35] <tomp> big crayon
[03:54:13] <Skullworks-PGA1> Skullworks-PGA1 is now known as SkullWorks-PGAB
[03:54:15] <jlmjvm> can a widget be made to show the lube level,or is there one already made in a sample config?
[03:54:25] <cradek> heck maybe I will try to cut it. It needs grinding now, it might still need it after I'm done with it, no loss right?
[03:54:42] <cradek> jlmjvm: you could use pyvcp to make a virtual LED to show that
[03:54:47] <tomp> clamp onto a stout angle plate and flycut bottom?
[03:55:48] <cradek> I think if I take it all apart, the top (base of the business area) would be flat, and I could clamp it to the table.
[03:55:52] <jlmjvm> thats what i was wondering,told it to make a vcp when i did the stepconf,but nothing showed up
[03:56:14] <cradek> jlmjvm: I'm not sure how stepconf deals with that. I bet you will have to create your .vcp file yourself
[03:56:21] <cradek> err .xml or whatever it is
[03:56:29] <SWPadnos> non-pareils?
[03:56:48] <jlmjvm> k
[03:56:52] <tomp> aka jimmies
[03:58:50] <SkullWorks-PGAB> I woulld lap the base on a surface plate then grind the jaw ways on a 6x18 grinder - but thats my quick guess coming late into this topic.
[03:59:43] <tomp> the web is wonderful http://its.foxvalleytech.com/MachShop3/BasicMill/squaring1.htm
[04:00:36] <SWPadnos> oh hey - thanks. now maybe I can make that shim plate for my X axis motor mount
[04:02:22] <SWPadnos> of course, that method relies on having a square vise in the first place ...
[04:03:14] <SkullWorks-PGAB> tomp - too bad its so hard to find horizontal mills now days... ( as recommended in the link.)
[04:03:32] <tomp> or a shaper
[04:04:50] <SkullWorks-PGAB> I used to run a shaper - such a unique sound...
[04:05:37] <tomp> cute, a not-a-vise a rike-vise http://www.littlemachineshop.com/products/product_view.php?ProductID=2304
[04:05:48] <SkullWorks-PGAB> when the sound changed you knew your tool was going/gone
[04:08:57] <tomp> i forgot a shaper had a 'clapper' http://www.geocities.com/euxineseaweed/shaper.html ( not the hand/light thingy )
[04:09:52] <toastydeath> do some shapers have hydraulic clappers like planers do
[04:21:11] <SkullWorks-PGAB> The one I ran was WW2 surplus - had a 50" full stroke, to handle 48" work. - top of the ram was 7'6" from the floor
[04:30:21] <jmkasunich__> cradek: just saw the vise woes
[04:30:36] <jmkasunich__> do you have a decent pair of parallels that are at least as tall as the jaws?
[04:30:54] <jmkasunich__> turn it upside down and flycut the base (assuming it isn't hardened)
[04:57:06] <toastydeath> or use carbide or cermet
[04:57:10] <toastydeath> if it is hardened
[05:25:22] <user__> That rike vice is interesting I have a few parts I "corner" at two 123 blocks
[13:50:15] <lerman_> lerman_ is now known as lerman
[16:44:20] <lerneaen_hydra> is stepconf in 2.2.x?
[16:46:49] <fenn> yes
[16:47:04] <fenn> in 2.2.1 but not 2.2.0
[16:53:01] <lerneaen_hydra> ah, ok
[16:55:43] <danielbr> hello guys, great work in thin last release, I'm testing 2.2.1, all is fine, except Axis "cone tool" don't spin anymore when using an angular axis
[19:13:03] <[JMS-OPC]> uhhhh anyone seen where i put my soldering iron? (wtf?)
[19:48:06] <notranc> notranc is now known as kc6lbj
[19:48:35] <kc6lbj> kc6lbj is now known as notranc
[20:46:58] <skunkworks_> hmmm http://www.starrotor.com/Engine.htm
[20:47:20] <skunkworks_> 60-80% efficency.. Since when did the laws of thermal dynamics change?
[20:47:44] <anonimasu> yesterday
[20:48:08] <skunkworks_> hmmm Quote 'a conventional luxury car getting about 25 mpg on the highway would get about 75 mpg'
[20:48:24] <skunkworks_> I didn't get that memo.
[20:48:36] <skunkworks_> I am sure it was all covered up by big oil.
[21:15:48] <fenn> that's "projected efficiency" which means nothing
[21:30:02] <skunkworks_> fenn: am I correct with the 50% thing?
[21:30:33] <skunkworks_> because someone then posted this http://people.bath.ac.uk/ccsshb/12cyl/
[21:32:25] <skunkworks_> Fuel consumption at maximum economy is 0.260 lbs/hp/hour. At maximum economy the engine exceeds 50% thermal efficiency
[21:32:33] <skunkworks_> maybe I just am not understanding the terms
[21:32:33] <fenn> well, brayton cycle engines are different
[21:32:35] <fenn> also i think they are fiddling with the definition of "efficiency" a bit in some of the charts
[21:32:44] <fenn> "The basis of comparison for the energy efficiency was an isentropic compressor."
[22:01:22] <christel> [Global Notice] Hi all, we're experiencing some routing problems between our US hub and the machine services reside on, I'm about to do some re-routing and will also move services. There will be NO services while we swap things around. Thank you for using freenode and have a great day!
[22:01:46] <cradek> wow, freenode is sure a bit off lately
[22:01:46] <SWPadnos_> you think?
[22:02:44] <fenn> hah popsci drew the star-rotor motor running backwards
[22:04:00] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[22:16:03] <fenn> this is neat http://en.wikipedia.org/wiki/Fire_piston
[22:38:17] <SkullWorks-PGAB> ISP's are packet spoofing torrents - doubling the traffic load to bog down torrent users - and everything else is suffering as result
[22:38:24] <rayh> I'm thinnking through a axis that moves a fixed distance forward and back.
[22:38:56] <rayh> I believe I can set position in HALCMD and not run EMC2
[22:39:10] <rayh> But how do I handle accel/decel
[22:39:24] <jepler> use a limit3 block?
[22:40:18] <jepler> if you are using stepgen, it has acceleration and velocity limits that you can use as well
[22:40:32] <jepler> and properly-tuned pid will have a good response to a "step function" as well
[22:40:57] <rayh> I was planning on stepgen.
[22:41:09] <rayh> I guess I didn't look at the params for it.
[22:41:11] <jepler> in stepconf I use stepgen's accel and vel limits when an axis is being tested .. the command to the axis is always a step function, but stepgen gives a nice trapezoidal velocity profile
[22:41:23] <danielbr> limit3 works like a charm with stepgen
[22:42:19] <rayh> ah okay I can see them. this should work fine.
[22:42:35] <danielbr> i used it to control a conveyor in a silkscreen printer
[22:43:09] <rayh> Great. That's at least a cousin of what I'm thinking.
[22:44:32] <danielbr> in 10 days i 'll upload a video for this printer working with emc2 +hal in control
[22:45:10] <SWPadnos> hmmm. I should see if I'm allowed to take photos of the power supply that's going to be controlled by HAL
[22:47:16] <rayh> Hey thanks guys.
[22:47:24] <jepler> anytime ray
[22:47:30] <jepler> SWPadnos: that'd be neat
[22:47:31] <danielbr> jepler: Axis "cone tool" don't spin anymore when using an angular axis in 2.2.1
[22:47:53] <SWPadnos> yeah, it's pretty cool - all stainless cabinet, about 12x6x8 feet
[22:48:04] <SWPadnos> pipes and stuff coming out for cooling
[22:48:18] <SWPadnos> all in all, a pretty small-looking supply for 720 kW
[22:53:13] <jepler> danielbr: untested, but I believe you can restore that behavior by applying this patch to the source code and recompiling: http://emergent.unpy.net/index.cgi-files/sandbox/restore-rotating-cone.patch
[22:54:11] <alex_joni> hi all
[22:54:29] <danielbr> jepler: so i'll need RIP after, correct?
[22:54:37] <SWPadnos> hi Alex
[22:54:40] <danielbr> hi alex
[22:55:58] <jepler> danielbr: yes
[22:57:08] <danielbr> ok
[22:59:24] <alex_joni> danielbr: unless you are brave enough to make install :D
[23:00:14] <danielbr> i'm not brave enough :D
[23:00:30] <alex_joni> yeah, RIP sounds smarter anyways
[23:02:07] <danielbr> jepler; that change is in axis.py? why is necessary compile it, if this is a python program?
[23:02:40] <jepler> danielbr: you could modify the copy of the program in /usr/bin/axis, but your changes would be lost if you installed a newer package version
[23:03:18] <danielbr> that is easy for me now, thanks!
[23:03:22] <jepler> ok
[23:04:15] <danielbr> i'll find some time this week and translate some news strings in Axis
[23:10:48] <jepler> danielbr: looks like my change is not correct ..
[23:11:02] <danielbr> ohh
[23:11:56] <danielbr> i'll test later i don't have the last emc in this box
[23:12:26] <jepler> it just makes axis print errors instead of showing anything
[23:12:53] <danielbr> why this was changed?
[23:13:51] <jepler> danielbr: it conflicted with the change to allow up to 9 axes (UVW)
[23:14:36] <alex_joni> good night all
[23:14:37] <jepler> so it was removed but never restored
[23:14:42] <jepler> good night alex
[23:14:55] <danielbr> so no way for this come back
[23:15:11] <danielbr> gn alex
[23:16:37] <SWPadnos> hmmm. cone rotation isn't so simple when the ABC "axes" could be at any index
[23:16:48] <jepler> I revised the patch, and tested it this time. It seems to work in the same limited way it worked in 2.1.x: http://emergent.unpy.net/index.cgi-files/sandbox/restore-rotating-cone.patch
[23:17:04] <danielbr> great
[23:17:18] <SWPadnos> ah, so the axis mask is positional for XYZABCUVW?
[23:17:40] <jepler> SWPadnos: yes
[23:17:40] <danielbr> i only need it for an A axis
[23:17:43] <SWPadnos> (ie, XYC will be 000100011)
[23:17:45] <SWPadnos> great
[23:18:11] <SWPadnos> I was going to suggest using a find on the axis string, but axis mask already has that for you
[23:19:11] <SWPadnos> what would that be in python? something like axis_string.pos("A") ?
[23:19:56] <jepler> SWPadnos: that's the 'in' operator
[23:19:56] <jepler> >>> "X" in "XYUV"
[23:19:56] <jepler> True
[23:19:56] <jepler> >>> "Z" in "XYUV"
[23:19:56] <jepler> False
[23:20:23] <SWPadnos> ah. is there not a function that will return the index of the substring?
[23:20:24] <jepler> (string) in (string) is a substring search. (anything) in (other sequence types) is a search for a specific item.
[23:21:22] <jepler> sure. .find() returns the zero-based index, or -1 if not present. .index() returns the zero-based index, or raises an exception if not present
[23:21:50] <SWPadnos> ah, ok. so that would work then, but not as simply as I'd like (though the exception version would probably be fine)
[23:21:51] <jepler> strings have both find and index, while lists have only index and tuples, dicts, and sets have neither
[23:22:08] <SWPadnos> ok (maybe I'll retain some of that information :) )
[23:22:49] <jepler> oh, and lastly, 'in' for dicts and sets is of course a constant-time operation and not a linear or even logarithmic complexity
[23:23:02] <SWPadnos> so something like try glRotatef(s.position[axisnames.index("A")], 1, 0, 0) might almost work
[23:23:21] <SWPadnos> cool
[23:23:25] <SWPadnos> I wonder how they managed that
[23:23:46] <SWPadnos> ah - non-constant-time insertion could do that
[23:24:23] <jepler> (well, unless you seek out keys that are have pathological behavior)
[23:24:27] <SWPadnos> heh
[23:24:38] <jepler> SWPadnos: sets and dicts are hash-based in Python. insertion and access are amortized-constant-time
[23:24:43] <SWPadnos> I sense a new book: "Pathological Python Programming"
[23:25:19] <jepler> (removal is too, I think)
[23:25:41] <jepler> it's probably the most obsessively-optimized part of python, because dicts are used for absolutely everything
[23:25:46] <SWPadnos> yep
[23:27:33] <jepler> the comments at the top of dictobject.c, and the notes in dictnotes.txt are very fascinating, if you're an algorithm nerd. http://svn.python.org/view/python/trunk/Objects/dictobject.c?rev=58925&view=markup http://svn.python.org/view/python/trunk/Objects/dictnotes.txt?rev=53782&view=auto
[23:30:01] <SWPadnos> damn, that's a lot of comments
[23:30:18] <SWPadnos> and pretty densely packed with information
[23:31:47] <jepler> I think the architect of that code is Tim Peters, who reminds me in some ways of our John K -- he thinks very carefully about everything
[23:31:54] <SWPadnos> heh
[23:32:20] <SWPadnos> I'm just looking at the comment on the pseudorandom number sequence - j=((5*j)+1) mode 2**i
[23:32:25] <SWPadnos> and the comment reminded me of the stepgen comments :)
[23:33:04] <SWPadnos> err - mod, not mode
[23:34:53] <jepler> remind me why we're talking about this?
[23:35:02] <SWPadnos> rotation of the cone in axis ;)
[23:35:22] <SWPadnos> looking up the A axis index in axisnames to calculate the rotation matrix
[23:35:23] <jepler> oh of course
[23:35:26] <SWPadnos> heh
[23:50:37] <danielbr> good nigth, and thanks again to you jepler