#emc | Logs for 2006-04-15

[00:04:04] <|Magician|> I am kinda curious....this may even be a dumb question, but why would the boards limit the performance...the only think I can think of is the encoder feedback
[01:07:37] <Jymmm> Jymmm is now known as Red70sShow
[01:07:37] <Red70sShow> Red70sShow is now known as Jymmm
[01:28:51] <Jymmm> les_w you lurking by chance?
[02:14:51] <|Magician|> it sure is quiet in here....everyone must be out having a better friday night than me
[02:15:12] <jmkasunich> friday before easter
[02:16:27] <|Magician|> any friday is good though
[02:16:42] <jmkasunich> that is true
[02:17:46] <|Magician|> what were you refering to earlier when you were talking about performance diffrences
[02:18:14] <jmkasunich> things like maximum step rates, maximum encoder count rates, etc
[02:18:47] <jmkasunich> for step and/or count rates under 10KHz you can get by _very_ cheaply using software counting and step pulse generation and a parallel port for IO
[02:19:13] <jmkasunich> on the other hand, there are boards ranging from $200 to $500 that will step and/or count up to 1MHz or more
[02:19:48] <jmkasunich> if you haven't done the math to figure out what rates you need, you can't possibly decide which is right for you
[02:22:17] <|Magician|> It just seemed to make more since to go with too much
[02:23:03] <|Magician|> ok, what boards were you refering to that had higher rates
[02:23:04] <|Magician|> ?
[02:23:33] <jmkasunich> unfortunately I don't know of any that so steps _and_ DACs for servo
[02:24:07] <|Magician|> Its the servo thats throwing me off so bad
[02:24:20] <jmkasunich> Jon Elson's USC board ($200) generates step pulses and counts encoders, up to about 300KHz
[02:24:42] <jmkasunich> lots of board, like vital, motenc, and mesa, do DACs and count encoders
[02:24:54] <jmkasunich> the mesa might be able to generate step pulses
[02:25:13] <SWPadnos> G-Rex
[02:25:17] <|Magician|> Yeah, I called Jon Elson.....He must have other things going on right now or something
[02:25:35] <jmkasunich> I think Elson's board has a connector for an external DAC, so if you are handy with electronics (or pay Jon to do it) you could run the servo with that
[02:25:36] <|Magician|> His boards was originally what I had in mind
[02:26:15] <|Magician|> but I cant seem to get one from him right now
[02:26:20] <fenn> why does usc count encoders? that doesnt make sense
[02:26:44] <jmkasunich> fenn: when you are using the steppers, the usc actually counts the outgoing step pulses
[02:27:03] <jmkasunich> but it uses the same hardware, and from the softwares perspective, it is counting encoders
[03:52:36] <SWPadnos> SWPadnos is now known as SWP_Away
[05:05:07] <CIA-4> 03jmkasunich 07HEAD * 10infrastructure/farm-scripts/check_commit: working on new farm scripts
[08:46:44] <Lerneaen_Hydra> slow day today?
[13:27:23] <les_w> morning all
[14:41:21] <alex_joni> morning les
[15:17:57] <skunkworks> hey alex.
[16:02:31] <samco> samco is now known as SkunkWorks
[16:21:10] <alex_joni> hi samco
[17:32:12] <Jymmm> This looks kinda interesting... both 6" and 8" http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=35098
[17:41:56] <Jymmm> http://www.craigslist.org/sby/tls/150964066.html
[17:45:15] <CIA-4> 03rayh 07HEAD * 10emc2/tcl/bin/halconfig.tcl: cleanup modify and start axis tune
[17:55:43] <SWP_Away> SWP_Away is now known as SWPadnos
[18:39:38] <CIA-4> 03cradek 07HEAD * 10infrastructure/cvs-server/chroot.html: document describing the basic chroot scheme
[19:21:59] <TorbaX2> TorbaX2 is now known as TorbaX
[19:31:02] <fogl> hello
[19:32:10] <fogl> can someone please tell me where (what directory) does apt-get install emc2 put all files?
[19:32:26] <cradek> it puts them in several places depending on their purpose
[19:32:30] <alex_joni> dpkg -L emc2
[20:30:06] <jepler> it's too bad the sherline table is so small
[20:30:56] <jmkasunich> it's too bad the sherline is so small, I think you mean
[20:31:02] <cradek> I wonder if they'd do well making slightly larger machines
[20:31:33] <jepler> jmkasunich: maybe so
[20:31:43] <cradek> their machines are sure useful for making small things, they have that market cornered well
[20:32:28] <jmkasunich> and there is something to be said for machines that can be picked up with one hand, and carried by one person without a hernia
[20:33:03] <jmkasunich> I'm not sure how much bigger you could get and still meet that criteria
[20:33:12] <jmkasunich> weight goes as size cubed
[20:34:39] <jmkasunich> whats needed to set up passwordless (non-interactive) ssh/scp?
[20:34:44] <jepler> my main "want" is to mill circuit boards, so a table at least as big as the blank PCB (4x6) would be nice.
[20:35:05] <jepler> jmkasunich: you have to have a private key on the side that will execute "ssh", and the public key in the authorized_keys on the other end.
[20:35:08] <Jymmm> jepler you couldn't add a plate?
[20:35:10] <cradek> jmkasunich: just copy a key
[20:35:19] <jmkasunich> you need more X and Y and less Z
[20:35:29] <cradek> more Y mostly I think
[20:35:44] <Jymmm> * Jymmm has that!
[20:35:44] <fenn> dont need much rigidity for circuit boards
[20:35:51] <jmkasunich> jepler: is it a question of travel, or table surface?
[20:35:58] <jepler> fenn: to mill away the copper? You need plenty of rigidity
[20:36:10] <jmkasunich> plenty is relative
[20:36:20] <cradek> you need nice repeatability, but no force is involved
[20:36:36] <jmkasunich> 0.005 deep in copper compared to 0.2 deep in steel are two different worlds
[20:38:24] <jepler> sherline's 5400 is 5" Y travel with a 2.75" deep table.
[20:38:44] <jmkasunich> you only need 4", so that works
[20:38:45] <cradek> is the throat depth >4"?
[20:38:48] <jepler> 2000 is 7" Y travel but the same 2.75" deep table.
[20:39:42] <jmkasunich> what do you guys use to hold down the board stock when milling? clamps along the edges? double-side tape? vacuum?
[20:39:48] <cradek> tape
[20:40:00] <cradek> clamps are useless - they don't flatten the board
[20:40:02] <jepler> jmkasunich: I just copy everything cradek does
[20:40:05] <jmkasunich> thought so
[20:40:25] <cradek> vacuum seems like it might work, except you lose it when you drill
[20:40:30] <jmkasunich> so all you need is a 4.5x6.5 chunk of 1/4" or 3/8" aluminum plate
[20:40:42] <jmkasunich> drill holes for four flathead screws that fit the tee-nuts
[20:40:49] <cradek> yep
[20:40:54] <jepler> hmms
[20:41:05] <jmkasunich> if you want to go first class, make a rectangular bar on the bottom that fits into a table slot so it goes on square
[20:41:08] <jepler> what does "throat depth" mean?
[20:41:14] <cradek> I bet you'd have to attack it with the flycutter to get it flat enough
[20:41:37] <jmkasunich> I have a 5-1/2" face mill ;-)
[20:41:47] <cradek> jepler: how far forward from the back edge of the work can you cut without the work hitting something
[20:41:50] <jmkasunich> anybody want "pc board tables" ?
[20:43:02] <jmkasunich> IOW, distance from centerline of tool to the column of the machine (or other obstacle)
[20:43:19] <cradek> jmkasunich: jepler won't actually buy a mill... he can mooch for free
[20:43:43] <jmkasunich> what do you put under the board to keep from drilling the table?
[20:43:55] <jepler> jmkasunich: perfboard
[20:44:01] <jepler> cradek: well yeah probably
[20:44:06] <cradek> another piece of circuit board would also work
[20:44:17] <jmkasunich> so the stack is table, tape, perf, tape, board?
[20:44:21] <cradek> yes
[20:44:28] <jmkasunich> seems like the tape would cause a lot of non-flatness
[20:44:33] <cradek> different tapes so the right thing comes apart when you pull
[20:44:40] <Jymmm> jepler: Lets say a drill press, "throat depth" is how large of a piece of material you can drill before it hits the back support (as example)
[20:44:43] <cradek> I agree, but it works surprisingly well
[20:45:04] <cradek> you have to apply the tape carefully
[20:45:06] <jmkasunich> I bet decent aluminum plate is flat enough without even flycutting it
[20:45:23] <cradek> yes, could be, if it's in good shape
[20:45:33] <jmkasunich> (not rectangular bar, that is extruded and tends to be wonky, but plate is pretty good)
[20:45:48] <jmkasunich> "good shape" = new
[20:46:05] <cradek> they make that new? I'm such a scrounger
[20:46:05] <jmkasunich> a 6x6 piece of 1/4 plate is something like $12 from metal express
[20:46:27] <jmkasunich> for some things I scrounge, for some I buy
[20:46:39] <jmkasunich> when a customer is paying for it, I usually buy
[20:47:01] <jmkasunich> I looked up the 6x6x1/4 a few days ago to do a quote
[20:48:18] <cradek> screwing it to your larger mill's table and cutting everything but the t-slot bars would give you a nice flat fit
[20:48:31] <jmkasunich> ?
[20:48:44] <cradek> leave two bars to go in the slots
[20:48:48] <jmkasunich> one bar
[20:49:25] <jmkasunich> (makes it easier to install, no real sacrifice in accuracy)
[20:49:39] <cradek> ok
[20:49:49] <jmkasunich> I would probably make the bar a separate piece
[20:49:51] <cradek> but same idea, the part you cut away becomes flat
[20:49:57] <jmkasunich> mill a slot for the bar
[20:50:13] <cradek> yeah, either way
[20:50:18] <jmkasunich> use 1/4" or 3/16" x slot width for the bar
[20:50:21] <cradek> depends on whether it starts "flat enough"
[20:50:33] <jmkasunich> that way I can just take a skim cut to make the thing flat
[20:50:43] <jmkasunich> otherwise it takes multiple passes
[20:50:57] <cradek> I forget you don't have cnc...
[20:51:26] <jmkasunich> and I have older machines, if I make a single pass without moving Y, no errors from slop in Y
[20:51:49] <jmkasunich> besides, why make a lot of passes when one will do?
[20:52:02] <jmkasunich> I can mill a 4x6 plate flat in about 2 minutes
[20:52:19] <fenn> btw a flycutter wont make a flat surface
[20:52:35] <fenn> it will be slightly curved because the mill isnt perfectly square
[20:52:59] <jmkasunich> btw, a flycutter won't make a flat surface _IF_ your machine isn't square
[20:53:22] <cradek> btw, nothing can make a _flat_ surface
[20:53:34] <cradek> :-P
[20:53:44] <jmkasunich> just a matter of how close to flat you want/need to be
[20:53:47] <fenn> your machine won't be perfectly square
[20:54:02] <jmkasunich> get out the blue and the scraper...... ;-)
[20:54:02] <cradek> I'm sure that's true
[20:55:00] <jmkasunich> anyway, unless this plate is an inch thick, holding tighter than a thou or so is silly, it will bend that much when you screw it down
[20:55:05] <fenn> i'm used to working with crappy equipment
[20:55:34] <jmkasunich> does the sherline come with a flycutter?
[20:55:40] <fenn> so i'd face it off while it was on the table to make sure it's flat and parallel anyway
[20:55:42] <jmkasunich> and how big can it cut in one pass
[20:55:44] <cradek> you can get a small one
[20:56:32] <jmkasunich> if I was doing it, I'd flatten one side of the plate on my big mill, and square the edges
[20:56:48] <jmkasunich> then mill the slot for the bar, and drill the screw holes
[20:56:51] <jmkasunich> screw it to the table
[20:56:58] <jmkasunich> and flycut the top on the sherline
[20:57:17] <jmkasunich> also, mill the back edge on the sherline, so its parallel to X
[20:57:40] <jmkasunich> then you can either mount a "fence" on the back edge, or just hold something there when mounting a board
[20:57:58] <Jymmm> pegs/pins?
[20:58:15] <jmkasunich> I was thinking more of something flat
[20:58:31] <jmkasunich> I have some parallels, 1/2" x 1" x 6", ground all over
[20:58:35] <jmkasunich> I'd use one of those
[20:59:13] <Jymmm> http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=789
[20:59:15] <jmkasunich> if I wanted a permanent fence it would be 1/8 alum, screwed into the edge of the plate
[20:59:59] <jmkasunich> mine are ancient, belonged to my dad
[21:00:06] <jmkasunich> probably 40+ years old
[21:00:16] <Jymmm> for $20, they could be mounted on his table
[21:00:48] <jmkasunich> the problem with a permanent fence is chips getting down in the 90 degree groove between fence and table
[21:01:22] <jmkasunich> with a removable one, you wipe the table clean, then hold the "fence" against the table with one hand, push the board against it with the other, and then lower it onto the tape
[21:01:51] <jmkasunich> if you do a permanent fence, put a 1/16" x 45 degree chamfer on the table edge first
[21:01:56] <jmkasunich> someplace for the chips to go
[21:02:08] <Jymmm> sorry... I menat get a big flate plate, pin it so the parallels sit solid, place the blank PCB againest it
[21:02:21] <jmkasunich> you can scrape out most of them, and the tiny ones left do no harm in the groove
[21:03:00] <jmkasunich> this is one of those cases where a pic is worth 1000 words
[21:03:04] <jmkasunich> I don't follow you
[21:04:11] <Jymmm> jmkasunich you have a base plate, maybe with a grid of holes. You put in pins/pegs that the parallels will fit on to. then you have a perfectly sqaure edge to place your stock againest.
[21:04:26] <Jymmm> like a lip of sorts.
[21:04:47] <jmkasunich> the holes in the table and the parallels need to be very good fit on the pins
[21:04:54] <Jymmm> right
[21:05:16] <Jymmm> he does have a cnc mill
[21:05:33] <jmkasunich> a cnc mill doesn't mean precision
[21:05:48] <jmkasunich> you want a 0.0005 fit
[21:06:05] <jmkasunich> so hole diameter and pin diameter and hole spacing all need to be better than that
[21:06:09] <Jymmm> jmkasunich he's (better) not making nuclear reactor parts
[21:06:28] <jmkasunich> ok, maybe he doesn't need 0.0005
[21:07:06] <jmkasunich> but since my method can give him 0.0005 squareness with no precision parts, might as well take it
[21:07:13] <Jymmm> true
[21:07:45] <jmkasunich> actually, there is another approach - just mill the fence out of the solid
[21:07:49] <Jymmm> I just figured reuse existing shit when you can =)
[21:08:02] <jmkasunich> IOW, flycut the entire plate down 0.060" except where you want the fence to be
[21:08:33] <Jymmm> ...and hope that there isn't one tiny sliver of swarf under the stock =)
[21:09:07] <fenn> i dont see how this fence stuff helps align the two sides of the pcb
[21:09:30] <jmkasunich> the fence takes care of alignment in one direction only
[21:09:44] <fenn> yeah but you're not using the same datum surface on the pcb stock
[21:09:46] <jmkasunich> (and it ensures parallelism
[21:09:50] <jmkasunich> yes you are
[21:10:00] <fenn> when you flip it over?
[21:10:01] <jmkasunich> one edge (back edge) is the datum
[21:10:12] <jmkasunich> flip from side to side, back edge is still back edge
[21:10:14] <fenn> by "two sides" i mean front and back
[21:10:26] <fenn> so you'd need 3 fences to align it
[21:10:31] <jmkasunich> nope
[21:10:41] <jmkasunich> one fence gets you Y and parallel
[21:10:46] <jmkasunich> all thats left is X
[21:10:55] <jmkasunich> which can be done by offsetting
[21:10:55] <fenn> or you could rotate it 90 degrees when you flip i guess
[21:11:12] <jmkasunich> then you lose everything, X, Y, and parallel
[21:11:32] <fenn> * fenn considers chicken-scratching
[21:11:54] <jmkasunich> if you have ONE straight edge on your board stock, and a fence on the table, you can do the flip and only have to align one axis
[21:12:20] <jmkasunich> gotta go away for a while
[22:04:26] <SkunkWorks> fenn: good idea using the same registered edges
[22:07:44] <fenn> i cant figure out what jmk is saying