you guys are boring tonight
shall we crash a few machines for your amusement? ;)
the thrill! the excitement! the impending repairs!
you could just lie and describe an imaginary crash
and not have the divots on the table to show for it?!?!?!?!
toastydeath, do you know if the keys on a machine tool vice are hardened?
or all the ones i've seen are
the ones that came with my vice are too big
no grinder eh
I have a tool post grinder :)
I was offered a grinder free
then he asked me if I would pay to have it rigged out of the building
then he asked if I would pay $500 and pay to have it rigged out of the building
price on that deal was going the wrong way, so I don't have a grinder
free for only $500?
plus at least $500 rigging
at least $1k
I thought of all the ways I could tell my wife that I was getting a free grinder for $1k
plus at least $250 to rent the forklift to get it into my basement
rigging is always the killer
yeah the university sold a huge machining center in trade for the rigging to remove it
quit funny - the lady in charge of the accounting department is not very versed in the ways of computers and internet.. (scary I know) but I told here that I won an ebay auction an that we need to pay $ by paypal. She said - 'if we won it - why do we have to pay for it?'
ebay's new messages when you bid are mildly amusing
the latest commercials are mildly funny also.
I just saw today that marriss is working on $30 stepper drivers
if he can get over his god complex
You get that too? ;)
if you run a business, discussing politics with your customers is not a good idea
funny thing was, he tried to build a brushless motor drive and failed, now he's going to save everyone by replacing servos with steppers
but I'd be interested in $30 stepper drives, even with limited capabilities
archivist: gcam seems to import gerber.. gpl http://www.cnczone.com/forums/showthread.php?t=33159&page=2
[02:35:56] <skunkworks> http://gcam.js.cx/index.php/Main_Page
last time i tried it, it crashed
haven't updated recently though
just did for me.. :)
that was when the guy came on here and said emc was garbage because it didn't load gerber, so it's been a while
(crashed on window$)
so it didn't load?
I was just thinking about that - but emc supports RS274...
after an hour or so he came around
it loaded - I was just playing with the menus and it crashed
but still upset
well - I guess I don't have the latest vertion. Heh - now I just got a virtual memeoy too low.. ;)
how is the tuning going?
but I think it'
that's your problem
s stable for now
you need voodoo
vodoo will cause oscillation every time
it would seem =-O
for future reference, does anyone make a 'dumb' drive that works with brushless motors?
they still have feedback
ah, the 'deck of cards' amp? :)
little black boxes, I
'm staying away from their newer stuff
I have one, but my motors are 240V
they have ones that go that high
I think the amps on Ebay are usually 80V max
that's the most common
you need anti-voodoo
I suppose the top end might be nice
The tuning is complete, i can accept an occasional following error when jogging manually.
jogging has a higher accel profile than g0 moves
(i think its square)
no it doesn't
er, maybe i'm thinking the other way around
well I don't think so...?
g0 this way, g0 that way
the max vel might be different by a few percent
It faults if I tap the jog button, the loop never has a chance to stabilize from the accel before the decel is thrown at it
you'll occasionally get errors running gcode too then
GNieport1 try a short g0 move one way then the other way
I've speak a whole week on the tuning, I'll eat my lump
I understand - maybe later then, right?
why so much trouble tuning?
you can't put the drive in "dumb" mode?
it's not _that_ smart
I cannot get a stable oscope reading from the amplifier tuning pin
no dumb mode
it has it's own pid though
I need the stable scope signal to tune the amp internal loop exactly.
maybe you should zero out the integral gain in the drive and see what happens
i'm not sure i understand what that means
another great internet theory down the tubes
a series of tubes
The amplifier has an output terminal that, among other things, is used to monitor the velocity error during moves.
If I can't see the error, I have to just guess at values for the PIV loop
can't just use halscope for that?
P=1 would give you output equal to the input
yes, not really
i agree fantastically
eric_U I tried, really, I did
I haven't played with the drive I have like that yet
the ones I am playing with use the serial port for their scope
anyhow, it handles g0 and g1 at max velocity, both long and short moves
how much error do you get during those moves?
seems like short jogs are pretty useful
so set the ferror limits to 3 or 4 times that so you won't get errors
it's at 0.001 right now
drive is still going into oscillations on short moves correct?
it is stable
set your MIN_FERROR to 0.004" or so and call it done
knock on iron
its foolish to risk an ferror in the middle of a long program
the idea behind ferror is to trip if you run into something, or a motor fails, or an encoder fails, etc
so the difference between setting it at 0.001 and 0.004 ain't much
Hopefully my mediocre tuning job will not result in a bunch of scrap parts, or worse, variability between duplicate parts
Later on, I'll replace the amplifiers, no more budget now :)
I'm an amplifier collector, my wife is gonna kill me
I'm done for a while... have a newborn
that will do it
I need to sell some and raise the paypal balance
meh, the hospital bills, miraculously just at the end of 07 to use up my deductable, did it
time to get homing to work
will EMC allow a Machine On when a limit switch is tripped (outside of 'ignore limits')
I believe so
it just won't allow you to move that axis
axis is homing
sweet. I can leave my limits in the Estop loop.
if you have independent high and low limits, you can move back toward them, but not farther away
that's version 2.0
jmk is not right - you have to 'override limits' to turn Machine On
but he is right - it will smartly keep you from making the problem worse by jogging in the wrong direction AFTER you override limits
(I even just tried it)
hmm 50a peak 25a cont 50A20DD http://advancedmotioncontrols.com/download/datasheet/50a8dd.pdf
jmkasunich: nice van norman #6 with linear scales http://mtechdro.blogspot.com/
kimron is now known as kimron_
kimron_ is now known as kimron
kimron is now known as kimron_
...kimron is now known as kimgon
kimron_ is now known as yk
yk is now known as yk1954
please stop it, that's irritating
hello, i need some help. i am using the emc2.2 live cd and got an error
Can not find -sec DISPLAY -var INTRO_GRAPHIC -num 1
and more ...
you can paste longer error messages using http://pastebin.ca
- you put the text there, and then paste the URL it gives you here
any young(er) people expected at illini fest?
yk1954: if you're using a custom .ini file, make sure INTRO_GRAPHIC is defined
twice2: why do you ask?
young is only a state of mind
wondering about future
perhaps you could consult a futurologist
kurzweil is a conservative bet i think
twice2: define younger
I saw ages 8-80
last few years
dangit, reminds me of a johnny cash song
under 25 is definitely a minority though
i'd take it out one piece at a time.. it wouldn't cost me a dime
ten year old graduated from veterinarin school ?
a ten year old med student eh?
that some guinness world record?
anyone still awake?
my comment is not meant to detract from the most excellent work on going here
thanks cradek see at http://pastebin.ca/887129
you need those two variables in your ini file I think
yk1954: do you have a pluto interface?
duh, ignore me, cradek spotted the real error
we ain't got to the real error yet
if you don't have any hardware, pick one of the "sim" configurations
pluto is a special interface for servos/encoders
if you don't have it, you can't run that configuration
quite a clue in this stuff , where should i pick one of the "sim"
on that initial screen with the penguin
when you start EMC, you get a GUI window with a list of configurations
(and a penguin)
ok, i understand now
I just shot footage of machine moving, what video format is best to post
I have had good luck sticking stuff on youtube
cradek: iirc on your sherline conversion you chose 3:1 ratio. why, nnfs, more torque or resolution?
torque - the motors are small
(the resolution is very silly)
scale is 1/6000 mm!
I would not have picked these encoders - but they work fine
van gogh cnc
[03:59:12] <GNieport> http://home.cinci.rr.com/bog/stuff/100_2228.mpg
I think 3:1 is about the most you can get with one simple pulley reduction
that is a 300 IPM G1 square :)
this is scary http://cgi.ebay.com/15-000-30-000-Watt-Power-Inverter-Extremely-Powerful_W0QQitemZ110219563099QQihZ001QQcategoryZ32814QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
yeah, k, i have no hardware at the moment, but I've been playing with the sdp center distance calcs
oh man - that is outright fraud
A bit :)
looks ul approved also
GNieport: cool - is it a spiral or circle at the end?
he admits that it would draw 125A at full load
and he sells it with clip leads that would heat up at 5A and burn up at 20A
125A*120V = 15000W what's the problem?
cradek it jogs to a center-ish area, and does several square shapes, then jogs back to the origin; all g91 moves
'have it easily power an entire home'
echo "18/360"|bc -lq
GNeiport ... what accel are you using
vs. echo "1/360*18"|bc -lq
25 IPS iir
GNieport: sounds great! congrats, how sharp are the corners?
maddash: which one is right?
dunno, I need to put a pen and paper down
fenn: it should come out to .05
thank you all for the hand-holding :)
fenn: i think this is a floating-point thing
but your scale variable is not set to the correct number of decimal places
xfmedia puked on GNieports video :/
twice2 what format is better
fenn: so what should scale be, then?
glad you got it going
mplayer was happy with the vid
inverter ... lets see ... 2/0 Cu ?
yeah, don't worry about me, I'm hobbled at the moment
maddash: hmm nevermind scale doesn't get rid of the problem, just moves it further down in magnitude
dave_1: not quite - 4 AWG is good for 100A, so 2 AWG would be good enough for 125A
seb: I wouldn't have gotten it without your help
(he's right about 120V being far superior to 12V)
fenn: it's a rounding error
guess I like overkill
when trying to get the p gain to oscilate, why does this number vary so much?
oh yes i've burned up those clip leads at <400 mA
what are you changing, and how does a gain oscillate?
nec ampacity rules for conductors http://www.allaboutcircuits.com/vol_1/chpt_12/3.html
GNieport hey u get ur machine's pid done?
ZN wants the lowest figure that will sustain osc
Gamma-X: PID gains come from the physical properties of the machine - gear ratio, motor current->torque transfer function, inertia
then go 0.6 times that for a starting P
not oscillate, overshoot with damped oscillation, correct?
fenn well i read a howto to find the number were the pgain starts to oscilate then u back it down 1 and edit i and d
Gamma-X when you went out with your woman and got sushi, I stayed in the below 32F garage and kept working with the guys here ;-)
nope ... it should sustain ... non-dampened
so what's the big deal with GNieport's vid?
GNieport nice! lol but i got laiiiiidd
maddash: his new retrofit is working
what did he have to do?
or try at_pid
well oscilation refferes to when the frequency spikes,
besides slapping on the steppers and emc2?
er, are those servos?
dave_1: sustained oscillation is scary for most folks, so its usually an unknown. ZN wants people to wet thier pants:)
Gamma-X: oscillation means it moves back and forth
maddash: I had a bit of trouble tuning with the samrt-amps
fenn so i have to find where the p starts to oscilate... then edit i and d correct?
the osc may be noisy but amplitude is low
I'm trying to think if you are guaranteed oscillation
GNieport you're using the mesa?
sustained oscillation - like eating sushi for the first time
maddash: yep, and 7i33 analog output card.
just have to wait and see what happens
Gamma-X: yeah i guess.. seems sorta like commuting to work by cannon launch
twice2: did you see the vid
GNieport: why don't you upload to youtube?
GNieport got a diff link to ur vid?
GNieport: not yet, let me switch to it
because I'm going to delete it in a couple of hours
[04:14:54] <GNieport> http://home.cinci.rr.com/bog/stuff/100_2228.mpg
er, I was speaking for twice2's benefit
sorry, the upload apparently failed
I already wgetted the vid
what does halscope look like ... do you ever get close to max speed?
hehe, i've got dead xfmedia window all over the place
those are ll max velocity G1
GNieport is it plausable to actualy mill sumtin that fast? lol
Yes, you ever heard of high-speed
dave_1 i personally have not edited yet with halscope
I won't be, the machine is too old.
I may zip a bit on aluminum
halscope is good stuff
GNieport: and no shake of the machine a-tall, very nice
it's possible to mill any material almost as fast as the machine can go
it's a series I bridgeport, correct?
GNieport: mind if I upload it, then?
[04:18:05] <GNieport> http://home.cinci.rr.com/bog/stuff/100_2228.mpg
i fixed it
had to clear out the cruft from the ftp account
what is a good video player for linux?
maddash: I'd reather not post it
maddash but whatever
GNieport: is that still dovetails& adjustable gibs? did you change to ball screws?
maddash It's just not a very good copy
tomp: all original, the machine came from Japan in 1970 with ballscrews and steppers
do the knee gibbs extend all the way to the head?
what's going on the z axis?
maddash, would you please pull the vid again from the link I posted? it's actually longer
are you going to drive the knee for z, or is there some way to drive the quill?
I am going to use the 2-axis to machine a servo mount for Z
Quill driven by ballscrew
pulled himself up by his bootstraps
and a 3000lb counterweight
GNieport: sure, holdon
hah, 4 sec longer
[04:26:51] <GNieport> http://home.cinci.rr.com/bog/stuff/spindle_assembly.jpg
driving where the depth stop nut is usually located?
does it have a ballscrew concentric to the quill?
yes, I don't have a detail drawing, but I believe that was how the factory did it for powered Z quills
that's what is on my bridgeport
maddash: this will kill yah, i type the url in and got a porking video :o
eric_u how do you mean? the ballnut will have a bit of lateral float, if that's what you're asking
sorry, I misunderstood one of your answers
twice2: that's b/c I removed to old entry and still uploading the replacement
outrigger drive for quill http://www.elrodmachine.com/images/QU1C%20on%20a%20Millenium%20Edition%20Bridgeport.jpg
[04:32:58] <GNieport> http://cgi.ebay.com/Shizuoka-Turret-Vertical-Milling-Machine-Fontana-Ca_W0QQitemZ140202626424QQihZ004QQcategoryZ58224QQssPageNameZWDVWQQrdZ1QQcmdZViewItem#ebayphotohosting
I only have photos like this of the originally available setup
tomp: looks the same :)
with a bandit !
the guy I bought it from had cut the servo mounts off with a sawzall, and converted it to a manual machine
[04:35:21] <maddash> http://www.youtube.com/watch?v=WLctVdLJ4gE
apple II's and bandits, the 2 big users of 6502's
In the auction I just referrenced, do those appear to be step motors? I have never been quite sure what the OEM used
yep a detail photo is 6 wire
GNieport: i'm impressed, i'll watch for Shizuoka's now
oh, Yamazen i recognize
like kanematsu goshu, the importers, not the mfctr
your knee dovetail goes higher
nah, utube says it's not available. I'll look for it tomorrow. Glad you got it up and going on GNieport.
If I lived in Cali, I'd have many Shizuoka's to choose from. They go up about once a month on ebay
[04:42:28] <GNieport> http://www.youtube.com/watch?v=dQXTRKmwIis
Yamazen tells me that there are no docs availble. I real let-down.
shizuoka speed retrofit http://www.parsons-technology.com/aData.asp?sMode=41
apparently Shizuoka shipped all kinds of different options ;-)
geez why is the centroid box so big?
it's got a PC inside of it, maybe
hah they dont include a monitor
My enclosure is that large, I left room for 4 axis amps and a VFD
it includes peltier beer cooler
[04:53:13] <GNieport> http://i15.ebayimg.com/05/i/000/d5/e4/11fe_1.JPG
is this a stepper?
i think so, 6 wires
using that centroid mill must suck
fenn, what makes you say that
no buttons knobs wheels..
they have a pendant available
you mean, "it isn't like my Fanuc"?
i mean just a keyboard would be taxing
I run Fanuc 21i at work and it's all touchscreen except for the jog wheel pendant
Well, besides estop and machine on, a 'function activate' button
ah but touchscreen is different from a keyboard
i think i'd rather have the wheel boltd to something
unless its a really big machine
agree. A long way from a 1980's NC controller with 55 buttons, and a keyboard, and a jog wheel, all on one huge slab :)
lots of buttons isn't necessarily a bad thing
you can train your hands to do stuff faster with physical controls
no, not really bad at all
Touchscreen = 2 or 3 nested screens sometimes to get to a function
and you have to look at it
bbl, nite all
wonder if i should buy an air hose reel or make on..
yo-yo auto-wind? or like a garden hose wind up reel?
well, i'm too cheap to buy an auto-wind reel, but i might make one
maybe i'll put that off until i have a functioning cnc mill
a single bicycle hook works, just hang the coil on it, thats cheap
not when you have other people using the shop
k, thought this was your dungeon
i'm the only person doing anything constructive
the rest just make a lot of messes
bring back tool chips ( big coin like things you handed into the tool crib to get a tool ), keeps those who cant use tools away from 'em
i'd like to implement a computerized inventory system but i have a lot of prjects
and most tools that people borrow are too big to fit in my standard box
i suppose i could just switch to all pneumatic tools.. then they'd be small enough
but again, that's not what gets borrowed
bar code reader, bar codes on all tools,and one the back of the borrowers neck
please insert neck into CNC tatoo station
i wonder what insurance would be like on that
i'm such a sucker for chinese air tools
CNC tatoo would destroy the "art" of it
tomp: you asked earlier how square the corners were. The corners were heavily blended at F300, but I set the controlled blending to P0.1 and it followed spot-on
thats very interesting, at F300? ( meaning 30 or 300ipm, dunno the emc convention uses implied decimal or not )
the default P is 0.001 inch or 0.01mm (i think)
so increasing P should increase the rounding
fenn: I dont't believe so, it was shaving inches off
tomp 300 IPM
emc controled machine?
wicked! with a .1" radius at right angle turns
it slowed down :)
or is that true?
sweet, depth of cut and dia of tool?
depth of cut -1, tool dia infinite ;-)
[05:15:36] <GNieport1> http://www.youtube.com/watch?v=WLctVdLJ4gE
nice, very nice
in the video, it is doing ovals at the default blending
i take it you are a machinist right?
does P0.1 mean allow 0.1" deviation from programmed path? ( hurries to docs )
nah, not good enough for that title yet
[05:17:11] <GNieport1> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TrajectoryControl
well, if you have machining questions im normally around to catch them, im okay I guess
and you can hang out in #machine
there is a #machine?
as of this week
it's toasty's project
im just so so, ive got a lot of stuff, but I dont like to give specifics
a lot of stuff?
i know how to use a hacksaw
fenn making a hose reel will be an interesting challenge
I can never remember which way the teeth face on those
which ever way you want
gezar: maybe you were japanese in a previous life
or just put it on the wrong way
typical hacksaw has 4 possible orientations, all have some use
well, come on now
seems like you'd want to pull if the blade were facing up
otherwise your back would hurt
well, i guess it would hurt either way
think inverted saw, blade threaded over a pipe inbetween floor joists
pro's just use a saber saw
cold now, sleepy time
i hope to be pro someday
i hope to get a cushy research job
i'd settle for 'pro' at anything
pro R&D excellent. I'd love it
not that i have any idea what that would be like
i understand that the P0.1 allows 0.1" deviation from programmed, but i dont understand ever allowing any deviation from programmed path, the programmed path is what i asked for (dangit) i program 'tangential' paths always, and never look for high velocity thru small turns
tomp: you can never exactly follow the path unless you stop at every corner
(emc provides this too)
yes, you can, most machine tools can follow the path exactly, just not fast.
and not stop
i can hand crank exact paths come on
the machine just lies to you and you have no way to check it
you check the part
tomp: try driving your car 50 mph around a 0.1" corner
ever heard of wire edm, the path ios exact
one feature, I dont know if emc has this, is positioning always from the same direction?
wire edm is just so slow you can't measure the deviation
uhm the error you cnat measure is not an error
"exact" doesnt mean "as good as you can measure"
exact position stop check, is exactly that, move to point, stop, move to point stop
exact then doesnt exist
at that point you have mechinal errors
one way to define exact is by the resolution of the machine
and then issues with cutter dig, and so on and so fourth, but its not a methamatical proposition any more
but i'd agree that it doesnt exist
ok, this is moot, but swinging 0.1" radiii makes parts that are unacceptable ( peg wont fit into hole )
if you tell it 0.1" and get 0.1" what's the problem?
the die and punch are fskd
if the default is "emc deviates a bunch from commanded" ithat's no good
your cutting hairs here, and if your making punch/die sets with 0 clearance you better be using them to cut saran wrap
then set your tolerance lower
0.1" is not saran wrap, it's cowhide
so .1 is your clearance, .9 in a 1. hole, no problem
wait a minute, so one of you, either tomp or fenn, has a problem with emc's ability to generate movement pulses that make something move a fixed distance?
er.. i'm just saying no machine can move along an exact path
not pulse size, path deviation. and i say machien can move exactly along paths, the velocity suffers, and the mpotion can be continuous
you have to stop unless you can change directions in the time it takes one encoder tick
(which is impossible)
thats not emc's issue, that physics
emc has to work within the limits I would imagine right?
gezar: no emc is Open Source, it can do anything!!! :)
i hope folks arnt saying that
it may be possible to change directions open loop (in the time between encoder ticks) with very well calibrated feed forward
but what's the point of that?
fenn your talking about very small deviations on measured part right?
depending on your definition of small, yes
let's say you are machining individual atoms..
what is your definition of the deviation, well, hahaha
let's ignore the fact that atoms move around on their own
emc is exempted from the laws of physics
ds2: it's called "simulator mode"
fenn: i didnt know you were referring to what happens between axis position updates, the circle made up of a zillion lines idea... yes there is deviation in all digital controls ( and the analog controls are flaky )
i think that deviation is below the 0.1" threshold
tomp: i was setting up a hypothetical situation where your desired path was less than the encoder resolution
so i could then go on at length about the open loop behavior between encoder ticks
k, another limit, true, my encoder is really 0.001mm but say its coarser, and yes, between the command and the check, the axis is 'on the loose'
if you define 'exact' as within the atomic diameter then yes you can follow an exact path
but your velocity around an atom-sized radius would be very low with any sane accel
fortunately small things can accelerate quickly
i can't find out what you guys are talking about.
toastydeath: tomp sez you can follow an exact path without stopping
i didnt like that a path deviation approached 0.1"
and presumably "so why would anyone ever want to deviate from the path?"
you can follow a damn near exact path without stopping as long as you have no sharp corners.
but that's not practical on machines under $250k
fenn, true, thats what i meant
it's practical if you put rounded corners on everything
any sharp corner is going to give you a problem.
toastydeath: i think it happens on cheaper machines at low velocity ( low low )
i think that's how plasma cutters do it
tomp: it doesn't
you dont need a $250k machine to do it
it follows the minimum encoder resolution
which is competely different from "true path"
damn stop agreing with me ;)
toastydeath: how's a $250k machine going to beat encoder resolution?
velocity driven rather than point driven
and even then, there will be some error
the error is small
small but not zero
they're actually getting very near atomic accuracy
and they do have atomic accuracy over small scales
a couple mm
agie dem250, open loop, 1/2 micron steps, just built well, < 250K when new
gage blocks are already below atomic accuracy, and have been for decades
but this is motion, not a block.
i want to say hundreds of years but i'm not sure
egyptians wringing little metal slabs together
it hasn't been hundreds of years because we haven't had measures that close except since the 70's
after we abandoned krypton 86 as the length standard
heh egyptians had better accuracy than most machine tools
what does that have to do with the conversation?
* fenn blames tomp
they knew flat
anyway without an external reference frame, you can't know how accurate your machine is, even with perfect encoders.
theroetical encoders, that is.
so i'm still kind of confuesd as to what kind of accuracy you are talking about
a true, perfect profile? Perfect from the perspective of the encoders?
wel crystal lattice spacing is around 2-20 angstrom (0.00002-0.0002 micron)
there are machines that are accurate in positioning under an atom's diameter
oops .0002-.002 micron
so that's really not a limit
so in the strictest sense, there is no such thing as "perfect," because you can keep adding zeroes, and we are now able to measure and move under an atom's diameter
but, does it matter?
well, im going to bed, machine accuracy is probably years away from the accuracy of PI
i suppose until we get down to the planck length
fenn: that's what i'm asking you
and thats probably the best limit you can define
what are you guys talking about when you say perfect
toastydeath: i complained that the program didnt have the 0.1" deviation in it. That the >>desired form<< differed from the >>actual form<< by a whole bunch.
0.1" is a wacking huge bunch
and the machine didn't show that on the error?
if the program said '0.1" tolerance than that's the specified path
your crazy error following thingy that hal has
tomp buys toasty a singha
best beer around
toastydeath: it will not show up on hal's ferror because emc asks for the rounded corner
is the error dimensional or geometric?
translate for us mere mortals please
it is not what was asked for
sorry, is the error like "the thickness of the part is .1" big"
or is it geometric, like, an axis is drifting/cutting corners/angles are not right
the movement of the axis, not the dimension of the part
it doesnt matter if there is metal in the way, different issue
'orders must be obeyed at all times' journey into space BBC 1950
i'm just trying to figure out what, precicely, is .1" oversize
or whatever it is
toastydeath: emc will round the corner by 0.1" (if you tell it to)
oh, thus your accel convo
i get it.
if it cant meet the requested path with the given accels and feed
does it slow down early, or sacrifice the corner?
a little of both
it slows down until it can make the corner to within tolerance
computed ahead of time of course
see that makes perfect sense
so what's the issue?
different strokes for different folks
users think they have good control but the allow the deviation to be large. a good machine is one that is acceptable speed and tight to path
Fanuc just never makes it to full speed
tomp: ever machined work hardening stainless?
is that what EMC does, as well?
hard milling yes
i suppose that wasn't phrased well, nevermind
toastydeath: and how much path tolerance might you have on that fanuc? 0.1"?
haha a lot less than that
tomp: i mean if you stop in the corners it will work harden, so you have to keep moving
fenn, yes, keep moving, lighter cuts, never stop
and plasma, if you stop in the corner it will burn a divot
edm divot too
its a plasma
the path tolerances on all our machines are damn tiny
and the speeds are very slow ( not )
they'll leave sharp corners
well no, they stop in the corners
or come to a near stop
now i wouldn't set g64 p0.1 on a machine tool, but i might on some kind of welding robot
or with thin sheet material in a plasma cutter
well the other issue is with a machine that has a high enough accel you don't get crappy things happening
(better solution is beefier motors)
with momentary stops
VERY momentary, a given
with higher accel you can set the blend tolerance lower for the same velocity
my stupid laptop keeps spinning up the hard drive
i thought laptop_mode was supposed to prevent this
it requires a stiff machine and torque ( if you feel it turn the corner its not stiff enuf ) (eg rokuroku )
i walked around imts with a glass of water, and set it on the machine while the salesman yacked. i watch the ripples
where on the machine?
any solid ( not wobbly door )
table or saddle, i suppose
that can be fixed o some degree with jerk limited motion
non motion part
what are you trying to deduce from that
a very stiff machine will transmit very small ripples through it.
yes, very small ripples
static stiffness anyway
no, during cut
that's not really true either man
yes imts and at timtos
depends on the mass distribution too
and the wright who set it up
because you don't define what's a synchronous motion versus asynchronous motion
from a water on machine test
I've had a bridgeport shaking so badly, you'd be wearing that cup of water
shocked the heck out of me
i put big yaskawas on one, and it 'rocked'
opened my eyes that's for sure
i just cant believe there's not a better way to cut metal than to throw more mass/stiffness at it
fenn: there is!
throw LESS metal at it
is actually a very sucessful strategy
uh, what? toastydeath?
fast and hard need big and stiff, slow and easy dont
they sacrifice static stiffness in favor of dynamic stiffness
eric_U: that's what i'm ostensibly working on
part of dynamic stiffness is lowering the mass of the machine
the crap deflects more, but chatters less
because the frequency is higher?
and in a finish pass it doesn't matter because the forces are low
steam cylinder hexapod, crank up the steam, crank up the stiffness
but doing the whole high dynamic stiffness thing takes a lot of analysis
the guy who came up with the equations for chatter pioneered that kind of machine design style, i believe
he's the one who managed to use a 6" long x 1" diam endmill for roughing
on some prototype machine
freq higher because its sorta damped with all that tension?
freq is higher because the frequency is ~ 1/(distance*mass)
simple harmonic motion
and one of the more interesting things is that when you increase the frequency of vibration, you have fewer usable frequencies for machining
but when you do find a usable frequency, the depth you can cut at is huuuuge
so how does the control deal with this stuff?
it just does what the program tells it to do
and they handle all the forces/etc by doing FEA on the toolpath itself
and adjusting crap.
does it do FEA on the partially cut workpiece?
it actually just does a couple cases
what software you talking about btw?
uh hold on i think i have a flyer from one of the folks
here was one
company called Third Wave Systems
advantedge FEM 5.0
there's also another package hold on
i can't find the flyer
* fenn looks at their website
you can do a lot of the stuff with a couple cutting tests
and doing oldschool tool wear analysis
to figure out what is going on and make guesses based on that
you don't get 100% of the capacity out of stuff, but you can get 75% of the way there
if the distance is constant, and the mass decreases, the freq increases (f=1/(d*m)) so how does old iron fit in? all low freq? ( 200-400 Hz )
tomp: cast iron is a damper.
there's more damping capacity in cast iron per pound than there is reduced dynamic stiffness
so even though your mass goes up, the damping capacity of iron is always greater
that's why crap isn't made out of mild steel.
its also why they use granite
there are weird hybrid machines now using granite epoxy bases/members combined with ceramic trusses
there's other ways to get damping too: viscoelastic shear, piezo actuators
for high precision, they're super cool looking
yeah but piezoactuators have been done to death
what's the verdict?
what freq is the hexapod or these hybrid material machine tools
it's called fast tool servo, it works ridiculously well
but it's also more expensive than the machine tool itself
corrects both vibrations and any error motion in the machine due to geometry
i.e. changing loads and bending members
re: frequency, i have no clue, i just know it's very high
tomp: no idea, i'd need fanc cad software or lots of time with a pencil to figure that out
haha and a twelve pack of beer to motivate you
or i could just build one and hit it with a hammer
its amazing how often the hammer is an appropriate technology
the fast tool servo crap would be right up your alley fenn
hmm. fast software at ~50kHz
corroberating 7+ inputs synchronously
7 minimum for a lathe, more for a mill
7? is 6 not enough?
any leads on who might repair dynamic research linear encoders? ( 2 bad 1 good, ttl output is very wavy not square)
because there's more than one kind of error motion per axis
you have to know the direction and magnetude of some errors and it takes more than one sensor to get that
but there's only 6 degrees of freedom in 3d space
but that's degrees of freedom, not error motions in an axis
ugh.. this is why i'm doing hexapods
also those 7 sensors do not account for the spindle
which has it's own fleet of sensors
usually another 8 or so
cant you just stick some mirrors on the workpiece and bounce lasers off it?
that's what they do
each of those 7 sensors is an interferometer
measuring the path length of some member
with refrence to a flatness standard
i dont get it
who cares where the individual components are
i care about: the position of the tool relative to the work
because if one of those components changes length (bends, twists, shifts) the tool moves.
the sum is where the tool is to the work
you need all those path lengths to calculate the tool tip
so put your sensors on the tool and the work!
doesn't work that way
what sensors would you use?
oof 1am nite all
depends what i'm doing, but interferometers and non-contact probes sound like a good bet
that's what they use
but you have to put them in meaningful places
and you have to make them look at something equally meaningful
if you have it look at the spindle, god alone knows where that thing is
no, have it look at the metrology frame
but you can't do it from the tool tip.
or you get errors from the atmosphere
* fenn picturing a hexapod attached to the tool
as the metrology frame
you have to have knwon paths
a metrology frame is a seperate frame around the machine
that is free of stresses
and kinematically correct, so you can mount mirrors and standards on it
the machine has unpredictable stresses
but by measuring the machine frame with reference to the unstressed metrology frame, you get an absolute tool position
and you need a minimum of seven sensors to do that on a lathe.
just to get absolute tool position?
because you can't measure twist with a single beam
some of the members, the main X and Y supports most notably, require two beampaths to show twist/sag/etc
and i forget where the other three go
if i send you one with 6 beams will you give me a million dollars?
no, because it won't work
at least not well
to that i say: harrumph
every measurement you take from the metrology frame will increase your accuracy
so you will have a machine that is accurate, but will have at least one unchecked error
and then you have the fun subject of the spindle.
assuming you can predict deflections within the workpiece/spindle, that seems pretty straightforward
the spindle's error is larger than the rest of the machine and it's more unpredictable.
6 non-contact probes around a sensor surface, arranged in an octahedron
that's kinematic, not necessarily checking errors
that gives you the position of the sensor surface relative to the reference frame
i'm assuming the sensor surface is a cylinder that is machined into the spindle
there's a radial and an axial metrological surface.
oop,s right it should be a cone
it's better if you seperate it out
that way you don't have to figure out the components of what your sensor is seeing
meh, it's going to be lots of math either way
that's why it costs so much and uses two seperate controls
it will be cheap in 5 years
that's what they said 5 years ago
and 5 years before that
and uh, all the way back to the 80s when they came up with it
it's still expensive as balls
fsdafsd is now known as Unit41
with some of the new nanolithography stuff it may actually get less expensive
well i assume "precision" means something different now than in the 80's
same general ballpark
hurr.. why is it expensive?
because getting crap to move reliably in nanometers sucks
but piezo sounds so easy
and piezo controllers/drives are expensive
it does sound easy!
they so desperately wanted it to be
then they tried to do it
they tried, and it's just still really expensive.
the demand is low
anyway it is 2 and a i have school tommorow
we must discuss this again sometime!
hello all, stepper config questtion...
what's up gene
the last time I tried to run that configger, whose name I've forgotten, it locked me up trying to run it, need a tutorial while
i'm doing it
they have it working well now
it's stepconf and it's on the menu from 2.2.2 onward
what version are you running?
2.2.3 but its not showing in either of the three menu pulldowns
gene: applications->cnc->configuration wizard
alex did you change the menu on 2.2.3?
I might be wrong about the name :D
on 2.2.2 its stepconf
ok, stepconf it is then :)
and I'm not at my linux box to see... if I'm having a senior moment or not LOL
[19:33:37] <alex_joni> http://cvs.linuxcnc.org/cvs/emc2/debian/extras-Ubuntu-6.06/usr/share/applications/emc2-stepconf.desktop?rev=1.2
oh, not an emc menu but gnome then, no wonder i couldn't find it, brb
it's called: EMC2 Stepconf Wizard
gene: you run it before you run emc2 :)
gene: you can also run 'stepconf' from the terminal
* alex_joni calls it a day
By Alex, didn't intend to chase you off though. :)
gene it's like 7pm for Alex
I found it in lost+found, no applications menu here. Its running and working on a new config, where do I start with the timings as I was running base at 30,000 ns before
did you run the latencey test?
That is step one
not yet, i just installed the nvidia drivers and i get two notices on startup of emc, but it runs Alex's logo code just fine with my old config
nothing about a latency test ib any of the gnome menu's though
one moment I get you the link
[19:41:07] <BigJohnT> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TweakingSoftwareStepGeneration
you just open up a terminal window and type latency-test
let it run a while while you abuse the computer then it will tell you the magic number you need
what kind of stepper drives do you have?
xylotex 262's & drivers
I don't know anything about them but I think there is a selection for them so you don't have to know the timing specs
i have them running at about 25 ipm with my old stepper_inch.ini
looks like the max is about 37 u-secs
looks like the max is about 37 u-secs, with the test at 25u-secs
how do i divide that up?
there is only one number you need but I forget the name...
I think it is the one on the right... maybe someone can chime in here
stepconf I think tells you which number to use...
max interval? 38u-s here
any way let it run a while
what kind of machine do you have?
micromill very heavily modified, bigger tables, new belt driven z axis
The number you enter in stepconf is the "Max Jitter (ns)".
technically, the one from "base thread" but they should generally be about the same
ok lemme back up
good luck .. bbl
now i have the breakout box so i assume i use sherline pinouts?
toastyde1th is now known as toastydeath
depends on how you wired it up...
from 'std' to xylotex
now, i have joints1 and 2 limit switces error, but do not have them wired. fix?
unselect them from the stepconf
in each axis
ok, but its not obvious where
the pinout config page lists all the pinouts, just select none on the pins with an axis limit switch
BigJohnT: it was more like 9Pm
and I was still at work
alex_joni: I was close LOL
so before it was 21:30 when I left home
* BigJohnT grabbs his calculator
err.. left work for home
I'm GMT + 2
7 from here
you're around -6?
oh, -5 then
so you're east cost then?
thanks bigjohn, found it, then had to fix the limits with vim. looking good now, and with less display latency too. :0
buit its about 35f here in the shop & the only thing that would hold me he3re much longer is a couple of virgibs to sacrifice over my xorg,conf to get 1600x1200 to work
i'm on the east side, central WV
see yall later & thanks guys
did anyone get the monthly spam from enco with the free shipping code yet?
(I notice it's no longer January)
msc, penn tool so far for snail spam
what's up SWPLinux ?
not too much - just hanging out in the hotel
hoping I'll be able to fly home tomorrow :)
yeah you should try to get home before the next time you're scheduled to fly out
anything going on on the lists? I don't have access to those while I'm traveling
and that's only two weeks away or so
heh - which is Tuesday
nope, I haev two trips before Wichita
oh -- nevermind
my advice stands, though
I should return from Bombay 12 hours before I leave for Wichita
oops - Mumbai
Hi guys. Do you know if it possible to hack the Sherline 4 axis parallel port control box? I'd love to add spindle control to my mini mill. thanks
Hugomatic: you "just" need to add an optoisolator in the right place and hook it up
the sherline spindle motor driver is not isolated so you have to know what you're doing
you can find the schematic for it online ... somewhere (I did)
if you're slightly less adventurous, you can buy an add-on isolator board for the KBIC motor drive
this brings everything out to nice screw terminals
you would just have to rig up a different enclosure for it
[21:53:50] <cradek> http://www.galco.com/scripts/cgiip.exe/wa/wcat/itemdtl.r?pnum=KBICSI-5-KB&mfgr=KB%20ELECTRONICS
I *think* it's this one
this board is very nice - it takes switchable 0-5 or 0-10v analog input
if you use 0-5 you can hook it *directly* to a parallel port pin and easily use pwmgen. it gives nice linear control of the spindle speed
(so you still have to know what you're doing, but it's safer/easier)
cradek: but I only have a single parallel port on my computer
as far as getting the signal off the parallel cable, one of those db25 jumper boxes might be easier than opening up the sherline 4-axis driver box.
a single parport has 12 outputs. your four step/dir axes only use 8
So I don;t need a second port?
depends whether those remaining four are used for anything currently
I sincerely doubt they all are
something like this: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=140196822036
(also used to be a staple at Radio Shack in the US)
I could simply add a connector and use 4 lines from emc
yes it would be easy to tap into the parport lines with that kind of box
(in fact I see that this is a Radio Shack item)
in fact that's what I did on the nist lathe
ooh it's 4PM!
it's the weekend!
Thanks for all the info guys, and have a nice weekend
well, I ought to run. see you folks later this evening
weekend! mill building time!
apparently there really is such a thing as a centerless lapping machine
i'll be damned
presumably for the polish after a centerless grind
it's got sizing control too
allegedly, i'm going from books
Ive never seen one
never even seen a picture
uses a really long copper cylinder as the lap and a brass cylinder as the regulating wheel
alex_joni: let me see it's 23:45 at alex's house...
BigJohnT: one hour off again :P
I don't have a watch
00:43 < BigJohnT> I don't have a watch
hmmm back to the drawing board
hab mal ne frage.... also bin hier am einstellen von meinem emc2 für eine 3d-step endstufe dran.... hab soweit auch alles hinbekommen also boost not-aus endlagen kühlung ein spindel ein (auf extra relais karte gelegt) doch die sleep funktion macht nicht das was ich will
when you fire up EMC a pop up manual tool change window shows up... What is that for?
BigJohnT: manual tool changing :)
also beim ganz normalen starten ist die endstufe auf sleep... doch wenn ich irgendwas verfahre in eine endlage komme oder den not aus drücke stoppt die bewegung doch die endstufe bleibt im normal modus und wechselt nicht auf sleep
when you have a M6 Txx it'll tell you to change to tool xx and push a key
ravennb: hmm, koenntest du mal die hal file auf pastebin.ca stellen?
(hope the others will forgive our german conversation)
ohhhh sorry sollte ich englisch sprechen?
ravennb: if it's alright with you :P
(if there's something you can't translate.. you can use german :)
okay i'll try to write in english right now
[22:49:58] <ravennb> http://www.pastebin.ca/888189
here's my hal file
hmm. I see you have motion.motion-inpos for sleep
that's not quite sleep
and you're correct that it won't work while estopped or on a limit
because inpos is not true for those cases
maybe you want to and2 the motion inpos with machine enabled?
yes i've found that already out^^ but I found nothing on google which is better
is the sleep a pause button?
BigJohnT: no, it's a low power on the stepper drives
no there are no buttons for sleep
reduce current or something like that
ok, my gecko's do that automaticly
reduce current when idle
so what should I use instead of motion.motion-inpos??
ravennb: maybe you can set sleep when XYZvel is 0 ?
hmmm i'll give this a try
XYZvel is a float
so you probably need a wcomp
with something like -0.0001 and +0.0001
there is already a wcomp which comparates XYZvel with 0.1 and 3
right, that's for boost
i'll try this right now and see what happens
there is wcomp.0.in I've to add a new wcomp how do I do this? It displays me "pin wcomp01.in not found"
i've tested wcomp.1.in wcomp.0.1.in and wcomp.01.in
add count=2 (or the desired number) to the line which is 'loadrt wcomp'. Otherwise, only wcomp.0 will be created.
also, use 'halcmd show pin' to find what pins exist, rather than just trying things that you think might exist -- saves time
okay the first tip helped me but how do I use 'halcmd show pin`?
ravennb: open a terminal
then write: "halcmd show pin"
and you'll get all pins
or write halcmd show pin wcomp (and you'll get only pins starting with wcomp)
ähhhm you mean a terminal from ubunutu (Applications->Accesoires->terminal)??
okay this gives me some error messages
had emc2 to be started at this point?
ahhhhhhhhhh it works^^
ähhhm I had to invert the output of the second wcomp and all my tryings don't work
ravennb: setp parport.0.pin-xx-invert TRUE
"parameter or pin 'parport.0.pin-17-invert' not found
Perhaps it's easier to invert my new "sleep" bit
use 'halcmd show param' to see a list of parameters. you'll probably find that you made a simple mistake in the name.
(again, do it while emc is running)
hmmm okay it works but the sleep bit did nothing its only 0
i've set the wcomp.1.max to 50 and the value is 30 but its false.... the same thing when i set it to 0.1
is the function in a thread?
how do I make a thread?
emc makes threads already - base and servo
the core EMC functions are automatically added to threads, and the sample configs add the functions for the HAL blocks they use
but if you add more hal blocks, you need to add their functions yourself, using an "addf" command in your hal file
the function is in the standart_pinout.hal if this information helps
I'm coming into the conversation late and didn't read back yet
I didn't think standardpinout.hal had a wcomp in it?
I've modified it
ravennb_: you need to addf your new wcomp function
"addf wcomp.1 servo-thread" or something like that
if you already have wcomp.0, that only covers the first wcomp
ravennb_: you already have one (for wcomp.0, need one for wcomp.1)
ahhhhhhhhhhhhhhh yes I've seen this yet
this is for looping the wcomp or?
for calling the wcomp code
otherwise it's just some software component which doesn't get run
jeeeahaaaa thanks it works
I'll try some values now and then I could go sleeping^^
it's early :P
yes but tomorrow (or today if we are correct) I'd planned a stress test für my cnc soft/hardware
that's nice :)
hardware means the electronic in this case the other hardware I've to build on work
that's great everythings works fine now
good to hear that
thanks you a lot
ravennb_: keine ursache :P
ja hab mich auch angestrengt nicht so noob like rüberzukommen kenne das von nem anderen chat wie die dann fertig gemacht werden^^
ravennb_: das kommt hier sehr selten vor
das runtermachen oder die noobs?