i'm thinking about buying a used electronic indicator setup
ebay always freaks me out a little
i should just do it and stop whining
Guest272 is now known as skunkworks_
easy enuf to make a new one with ttt
crap - thanks. I think I failed gnome file serching.
find ./ -name axis.ngc
actually locate works best for most things
you can even do regexp's and stuff
I was trying to use the file browser.. as I am a windows person :(
skunkworks_ is now known as skunkworks
[00:34:44] <skunkworks> http://www.youtube.com/watch?v=dKz81fY1OjY
even though I've seen it 1000 times, it's fun to see the machine and the cone doing the same thing - your video shows that nicely
doesn't the comment right at the top say "not for milling"?
sheesh ,some people can't follow directions
yeah but it seems to be the first thing a lot of people cut
* skunkworks never notice the comment
the begining of the video chunks - oh well. people will probably think it is the machine doing it ;)
only for about one second here
my minidv worked pretty well - I think this is the first time I have used the firewire port to download the video
Realtime : http://linux.slashdot.org/article.pl?sid=07/08/08/1810243
hows everyone doing tonight
full of rage and bile
haha uh oh
he's always like that
his name gives it away
it's so true.
anyway, I have a question about linuxcnc
on the site, it talks about "path deviation limited to a specified tolerance"
anyone read this: http://linux.slashdot.org/article.pl?sid=07/08/08/1810243
does that have anything to do with backlash?
no, it doesnt have much to do with emc at all really
backlash is a separate setting.
ok, but there is a setting for backlash right?
it's called "backlash" i believe
because I'm planning to go with ACME leadscrews instead of ballscrews
to save some money
and I'm concerned about backlash
you should make an anti-backlash nut
like a split nut design eh
backlash comp is a bad solution. Like fenn said - try to remove it. split nut - spring loaded nut..
what's the least backlash I can get with leadscrew?
doesn't work for large forces, though.
but it involves a tremendous amount of work
what's a realistic number?
that's pretty good
if you keep a split nut adjusted, it will stay pretty low
here is an interesting design (the white ones) http://www.aanrip.nl/abacklash2.jpg
[01:55:54] <fenn> http://www.aanrip.nl/backlashmoer2.jpg
how does backlash comp work exactly?
it knows it needs to move the screw X amount before the axis will move
when you change direction it advances the screw that amount
however consider what happens if you're doing climb milling
the cutting forces will take up the backlash, and then you advance the screw suddenly, making a very high chip load for a brief period
oh I see
but the cutting forces don't take up the backlash consistently?
in climb milling the forces reverse
instead of pushing the axis, the machine pulls it along
and the screw holds it back
well - maybe
also you can do the same sort of thing in the z axis with spiral cutters
an 1/8" endmill isn't going to pull the table out of backlash
having more than 50% cutter engagement cancels the climb pull too
thats not really climb milling though
except at breakout
sure it is
what if you're 80% cutter engagement? 99%? 100%? what's the cuttoff point?
direction of cut is by is the clip thicker or thinner at the end of the cut
this is all over my head
isnt irc great?
does the sieg x3 actually weight 700 lbs?
Skullworks-PGAB: we're talking about the cases where backlash is a problem
climb milling always has the thinnist clip as cutting edge departs the work
climb milling is just an example of the problem being detailed
fish i can imagine it weight 700 lbs
but x2 is only 160 lbs
from the pictures, it doesn't seem like the x3 is that much bigger
X3 wighes about 390 lbs
shipping wieght it under 450
* skunkworks would love a x1/x2 for my office..
well looks like I will be getting the x2
skunkworks: why not go for an x3? seems a more useful size
x3 is too big to fit in my apartment
I have stuff that big
or access to it anyways
ah but what if you don't!
Approximate shipping weight: 372 lbs. ( from the Grizzly web sight
it is big
then I would have a bridgeport sized mill in the garage :)
mine is all over the house
bridgeport sized stuff takes.. infrastructure
infrastructure == garage
cant just put it in a car
rent a wrecker
get some tubes
and a toe jack
I pulled the table and saddle
removed the head & motor
then the column
so I could move it myself and fit everything in my mustang
do spider couplers give backlash?
I've had an X3 since january
Skullworks-PGAB: how'd you fit the knee in?
damn, everything gives backlash
X3 is a mini -bed mill - there is no knee
I don't understand why people like to use lovejoy couplers.
no, not everything
membrane couplers don't, preloaded ballscrews dont
well the steppers aren't rotating that fast, anyone have experience with just a solid piece of coupler?
[02:10:36] <Skullworks-PGAB> http://images.grizzly.com/grizzlycom/pics/jpeg500/g/g0463.jpg
Skullworks-PGAB: oh i thought you were talking about a bridgeport
hmm - I thought the x3 was bigger than that for some reason.
the HF clone is 12-7/8'' (X), 5-3/4'' (Y), 8-5/8'' (Z) 18" swing
no - If anything I'll pick up a Hurco KM3 with ballscrews ready to go - they show up cheap from time to time - and they usually have a 3hp motor
Almost every part of a BP weighs more than I want to lift alone...
we have two horizontals at work
i wouldn't want to try moving one of those alone
i'd probably be okay moving a bridgeport with two dudes
me being one of those two dudes
they're the biggest machines in the shop, 50 hp
* skunkworks loves horizontal machining centers - love the chip removal.
well, except for the planers, which are 100
I used to run an old Cinci
but the planers aren't mills
nor do they turn on
they are neat machines!
we have some big cutters for them
nobody really uses them much
one of the cutters is 12" diameter, 7" wide
can someone explain to me how this thing works?
stagger tooth cutter thing
[02:17:50] <fish> http://www.aanrip.nl/backlashmoer2.jpg
I see some O-rings...
the o-rings provide clamping force
on the threads in the nut
which takes the backlash out
and that reduces backlash?
doesnt seem like a lot of clamping force
it's not, but you're not really going to be hogging metal off
well what's hogging metal off
supposedly the o-ring can stretch after a while with that design
fish, 5 horsepower cuts
i would say "hogging" starts
so for the x2, that would work?
here is a design that doesnt have that problem, but its more complicated: http://www.dumpstercnc.com/leadnuts.html
seems like a simple solution
I like the dumpstercnc address
that appeals to me
good for engraving, PCB and light routing
I use those nuts and they work well
would doing PCBs and engraving work with a CNC converted X2, assuming your piece is small enough?
sure, they're both mills
you dont need very much rigidity for pcb's
the x3 is just bigger and beefier
ok just checking
any backlash will kill a pcb though
a mill is actually probably a bad choice because the spindle doesnt go fast enough
you can use weights for pcb drilling
and get very good positioning
will a few thou backlash really affect pcbs?
to hold it in place you mean?
that's a trick they used on jig borers
they attached a line and a pulley to the table
and hung weights on it
kept the table up against the leadscrew
cradek: you maxnc - you made some sort of spring loaded double nuts - right?
consider that it's common to want to cut traces .006-008 wide
skunkworks: yes I use those dumpstercnc nuts
so if I wanted to do pcb work, definitely go with ballscrews?
that's not necessary, you just need a solution for backlash
definately use something that takes the backlash out of the leadscrew, or go to ballscrews
before I got those acme screws and dumpstercnc nuts, I used a spring between two nuts on triangular threads
are the solutions you guys provided me enough?
any of those spring/o-ring thingies will work
* maddash just cracked wep
i KNOW there are people using those who are doing really light stuff in aluminum
why not just use bigger springs for the dumpstercnc screws
and then use 400 oz in steppers
higher static friction
wear I can deal with
and you still get follow error
just replacing the parts every once in a while
I've cut Al with those
they're not as weak as you think
i wonder how much bandwidth #emc takes up
i bet there's a point where you get more "backlash" because of stiction/elasticity in the screws
I was planning on doing light cuts in steel
fenn: yes, follow error
that the machine can't see
it will show up in cuts
points on a circle
or rounding in sharp corners
if the feed changes direction at a good clip
you can see it if you use a scale on the table instead of the leadscrew
well thanks for the help guys
i guess they really hate cleaning out t-slots: http://machineshop.olin.edu/equipment/media/original/bridgeport_mill.jpg
doubt it ever gets used tho
what's the point of the t-slot?
to collect swarf so it doesnt fall on the floor
mounting crap to the table.
thats silly, if they were serious they'd just put an array of tapped holes
judging from the corners of that room
that machine indeed does not get used
there's a plate covering the entire tray of that bridgeport
i vacuum after i use the crap at work, and there's still chips stuck in the corner
yeah, it keeps the chips out of the slots
fenn: I don't think an array of tapped holes could never be as good as t-slots
aluminum isn't magnetic?
sucks to be you
flush it out with oil
maddash: why not?
or just keep it out of the slots in the first place.
fenn: tapped hole array don't allow arbitrary positioning
besides the fact that everything is designed for t-slot mounting
there is stuff with holes
usually measurement stuff
t-slots allow the most rigid setup possible
while tapped holes are fast to use
master rotary tables and angle masters use tapped holes
some granite surface plates have a section of holes
why to which part
why do "master" rotary tables use tapped holes?
two reasons - one, it's faster and you don't need t-nuts
because you can use locating pins in a counterbored hole?
fenn - given a choice between slots and holes I'd take slots every time
they don't want anyone using any sort of indicator
to try and use the slots to indicate the table in
no center hole, no slots, just a bunch of tapped holes in a grid
slots: can position clamps anywhere, easy to clean
holes: limit where you can put clamps, get crap and chips in them and are a bitch to clean out
but they're fast!
not when they're dirty
isn't that what I said before?
yes, it is
do they make "rotate in" nuts like for 8020, but to use on machine tables?
you tilt them about 60 degrees and it slides into the slot.. err
tee nuts that have the corners of the flange cut off, so you can drop em into the middle of a slot
* fenn finds a picture
as you tighten them then turn into position to grip
that's kind of cool
once 3d-printing becomes practical, machining should be mostly obsolete
perhaps, but i'm not going to hold my breath
i for one welcome our laser overlords
* fenn gives up on finding a picture
it would be interesting to see a laser printer beat a progressive die
or see it beat a multispindle screw machine
well you just gotta parallelize
or make the dies for either
don't be silly
but i like being silly
have you ever seen progressive stamping?
i meant for lasers
pew pew pew
but laser is the wrong way to do it anyway
they dont make silicon chips with lasers
OR DO THEY
damn keyboard on this portable finally died
jymmm: did you see the video?
there's more ways to etch silicon than you can shake a stick at
I used to work at NSC
it will be interesting to see if 3d printing in metals ever makes stuff faster than traditional machining
skunkworks: yeah, cool
it certainly won't make stuff fast enough in plastics to beat mold injection
well it depends on the shape
holy crap: http://www.youtube.com/watch?v=W_YnaHNcISw
hahah i saw that
they are sitting IN the fscking press!
lets see what happens when the management decides to raise production rates
the back guy leans out from under it to get the next piece just as it comes down
the very fact that they can affort to have about 8 guys loading and unloading a single press boggles the mind
relatives of the owner
compared to the cost of that die
they still load dies like that by hand here, too
oh come on, it looks like a food pan or something
Gaqwd lets hope you never have to sneeze
that's probably a half million to three quarter million dollar die
doesn't matter that it's dog pans
the size of it is what determines the cost
toastydeath: they only load by hand for fairly short runs
if it was volume they'd have some kind of automation
not necessarily, man
the cost of manpower to run the die is small compared to maitnence, and is completely puny compared to the profit generated by it
even loaded by hand, every part
high speed presses are autofed, sure
but slow ones like that, it still makes sense to load them by hand
hmm - looks like $17 + $20 shipping will get me a keyboard on ebay ;)
like I said, it depends on the run size
even for a million parts, dude
I find it hard to believe an american company would do a run of a million of anything with manual loading
some places just don't use automation.
i've got aquaintences who just retired from the die industry, working on large production dies
places that make a million of something do, or they see the job go to china, or a competitor that does
and again, that's a generalization
automation is the darling of trade magazines
not necessarily actual shops
plenty of places do it, plenty of places don't
I don't neccessarily means robots and such
what do you mean, then
ppl sitting in a press machine
no robots required
people don't sit in the die here in america
mechanical feeders, pneumatic widgets, transfer lines
they wait for the die to open, go over, pull the thing out, and move it
someone else drops the new sheet in
I saw a press making rocker arms at the Ford engine plant here - about one stroke per second, flat blanks in one end, rocker arms out the other
yeah, that's also common
about 10 stages of transfer in the press, all pretty much mechanical
that is not, strictly speaking, an automated press
nfw humans could do that, chinese or otherwise
it sure ain't manual
autmated presses are ones with robots to load and unload sheets
the only operation action was to dumb huge bins of blanks into one hopper and remove hoppers of finished parts (forklift driver did that)
and automatically lubricate the die, make sure it wasn't double fed, et cetera
ok, no wonder we're arguing - we have different definitions of automated
i'd say automated is when you dont have a guy pressing a button
those presses still have human attendants, is what i am saying
even the high speed ones
to load the rolls into the die, et cetera
and hit "go"
and wait for the roll to run out
right, but they don't handle every part
screw machines aren't automated
you should see the 12 station rotary dies that makes machine gun belt links...
automated vs automatic
automatic has a guy pressing a button
ribbon sheet in - links FLY out - 1100 a minute
that press was not automatic
* skunkworks still likes watching how chain is made.
i like rolling mills
big stuff there
i know some of you guys were interested in error compensation
does emc has compensation maps or something?
for work I had to go the LTV steel rolling mill in Cleveland - there was over 100,000HP of motors in one room
yes, emc2 can do screw error comp
like, how so?
is it a lookup table
or like a per-inch error
read the find manual for details, but yes, its a table
up to 256 points
that was all i wanted to know
you enter the nominal, and either the actual or correction, for forward and reverse
(so it can also correct for backlash, even lash that is not uniform over the length of the screw)
it interpolates, right? (doesn't step)
linear or some kind of curve fit?
i missed a big chunk of something i think
what's the command to ask the bot for the log
[03:07:00] <Jymmm> http://www.youtube.com/watch?v=AqKvYUO7C7w&mode=related&search=
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-08-09.txt
linear is a good way to do it
best fit curves have shown to have poorer performance in experementation
the adjustments are usually small, the variation in adjustment from one point to the next is even smaller
so the change from one very small slope to another very small slope is nothing to worry about
and even then, best fit tends to generate a larger true position error than linear methods do
jmkasunich: where you running your geckos directly from the printer port initially - or did you only hook them up to the mesa card?
geckos almost always require a breakout board
I think I ran them from the parport
I definitely ran them using software stepping with the 5i20 simply action as an I/O board
I think I'm gonna do something smart for a change and go to bed early
I think I am going to call it a night also.
pew pew pew
does anyone who is left know how shipping works on ebay
i am looking at this indicator and the seller says shipping is 12 bucks or so, but it's not included in the buy it now price
you add the shipping to the buy it now price.. that's the total
if the shipping price is specified, it'll get added in at checkout time
it will probably show up when you check out.
if it's not specified and there's not a way to calculate it on the page (like entering your zipcode) the seller probably screwed up
it's a flat rate so i hope it will be in the total
yeah it'll work then
electronic levels are more expensive than autocollimators
Skullworks-PGAB: Why do they require a breakout board?
Skullworks-PGAB: I read your earlier message why do you require a breakout board for geckos?
a-l-p-h-a2 is now known as a-l-p-h-a
whom05:19 < Skullworks-PGAB> geckos almost always require a breakout board
or someone :)
I am running the parker drives directly from the printer port. No issues.
yes - it was him that said that
I just wanted to know why he said that..
they work great from the parport..
did you see my video?
Jymmm winding him up about breakouts?
he seems to like em
someone should beat sense into him with a stick.
a big one
[12:20:50] <skunkworks> http://www.youtube.com/watch?v=dKz81fY1OjY
I dont get what the big problem is, I mean making parts is what im interested in with a cnc...
I have not blown out a printer port yet.. :)
geckos are optoisolated..
so are the parkers
then they dont need a thing
he does seem to worry his little head about the wrong things
he is a 'over doer':)
but we let robin take care of that ;)
[12:27:13] <anonimasu> http://www.youtube.com/watch?v=1AMZlDHwebw&mode=related&search=
I dont have any sound right now so I cant hear how it sounds
uh oh - his evil twin
* archivist copies to self to look http://www.youtube.com/watch?v=1AMZlDHwebw&mode=related&search=
thats not messing about!
[12:36:18] <archivist> http://www.youtube.com/watch?v=dKz81fY1OjY
jepler pointed out that the comment at the top of the 'EMC2 AXIS' program says 'not for milling' ;)
I never noticed that
steppers are nearly as noisy as the cutting head
or - the cutting head is pretty quiet..
or the microphone was near the steppers
The 2 S6 drives are definatly a bit noisier than the OEM650 drives. The S6 drives are a bit older.
the microphone was on the cam corder
I am going to use all OEM650's but I need to make a power supply. I grabbed 2 of the S6 drives as they have power supplies internally.
seem to work just fine though :)
you can see at the end of the video - I pan behind the engraver - the 2 big black boxes are the S6 drives and the small grey thing is the OEM650.
I want to cut steel like that.
yeah that was good
* anonimasu is still struggling to get good cuts
im too scared of chiploads ;)
a good solid machine is needed for that sort of cut
hm I need to cad this roller throttle body idea out..
skunkworks, design issue with the bridge http://news.bbc.co.uk/1/hi/world/americas/6938245.stm
I thought it was standard to load test when built almost like they were when repairing
It's too quiet in here
* anonimasu is happy
I got the mill at work to interpolate corners..
short line segments, with < 10 degrees of direction change
so im going to try doing some 3d contouring at it someday
I got kind of pissed off last time I tried because it'd exact stop :)
exact stop instead of contoure?
it's not the most amazing control out there but it works well
Sorry to say it, but I wish that was the worse of my troubles.
stupid old machine has less then 0.001 of backslash.
one of the axes has 0.004
I think the screw's more worn there
that's still damn good :)
Replace the balls
* bill2or3 nods off.
that's be overkill
Gawd I love PS
PS == PostScript
PS != EPS
sorry, didnt mean to brag or anything.. it's just so cool with a machine that's damn old :)
anonimasu: Oh no, it's all good.... That's awesome for brand new too!
that is cool. has it just not been used much, or is it really that durable?
it's durable I guess
though the control has been replaced a few years back
I wish I had any time to do machining.
anonimasu: that's the easy part=)
so they might have replaced screws and stuff :)
im trying to learn
ok, maybe not stupid, but still.
LOL @ bill2or3.... you play, you pay!
does anyone know polish?
[15:18:14] <anonimasu> http://cadcam.fme.vutbr.cz/?page=heidenhain&sub1=labyrint
of course, poland is a huge country, they all speak it over there.
bill2or3: Say it with me.... CON-DOM
nah Jymm, it was planned, but that doesn't mean I'm any less exhausted.
I want to know what that part is
bill2or3: So, you PLANNED to have a stupid baby huh? I think you need to fire your planners =)
whats the G code for baby
* bill2or3 fires himself.
that part is pretty cool :)
anonimasu, interesting part, if you figure out what it is, tell me.
i trying to
I think it's a labyrinth seal
or a mold for something
[15:27:28] <JymmmEMC> http://pastebin.ca/651187
some of the other pages there show what looks like parts of a rotary engine
perhaps it's related to that.
anonimasu: That be PS, so easy to use/edit even in text form
* anonimasu nods
I love the fact that's really a programming language too.
this will translate it, kind of. http://www.translation-guide.com/free_online_translators.php?from=English&to=Czech
but it still doesn't tell me what the part is.
LOL, convert it to PDF and it's only 821 Bytes!!!!
it looks very similar to a labyrinth seal
I dont know what that is.
* bill2or3 googles
and the title of the page is near enough for me
i recieved the most crushing news at work today
we apparently were all set up to own the LODTM, from lawerence livermore
at our shop
but couldn't afford the move
at least we know the folks who are getting it
and can go visit =(
pew pew pew
does anyone know about the bridgeport boss-8 (says R2E3 on the front)?
jone had posted some info on them.. let me look. I think it is steppers.
this one is definitely servo
X Axis 30" Y Axis 15"
Quill Travel (Z) 5" Knee Travel 13"
Horse Power 2 3 Phase
^^ the intarwebs gives this information
for one thing I wonder if it has encoders
[19:03:05] <jepler> http://www.cnczone.com/forums/archive/index.php/t-9643.html
says it homes to an index pulse
BOSS 3 through 6 are stepper. Only the BOSS 7 amd 8 are
servo. Then there's the R2Ex series, etc
jepler: ah, nice find
here's a bridgeport that's actually *in* bridgeport .. http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=290038162008
there's a couple of these machines very nearby but the guy is very "optimistic" about their value
says it's boss 8 with "nc400 servos"
I hate 'optimistic'
cradek: did you buy the servo interface for the mesa card?
[19:07:07] <cradek> http://www.cnczone.com/forums/archive/index.php/t-26311.html
maybe it's tach + encoder
sweet. are you hoping to use the boss controller? ;)
skunkworks: part of the reason he's so "optimistic" is that the controllers are still working
not that I care about that...
I wish someone could tell me if I can use the motors, encoders, amps, etc as-is - that really would add a lot to what it would be worth to spend on it
[19:37:43] <skunkworks> http://groups.yahoo.com/group/CAD_CAM_EDM_DRO/message/6222
looks like +/- 10v
that's good news
Scroll down until you find a line that says:
static int PERIOD=20;/* fundamental period for time interrupt */
... holy cow
looks like Jon E knows a lot about these machines, maybe I should talk to him
(maybe that's what his is)
cradek: is that how you used to have to change BASE_PERIOD?
the period was hard coded at one point?
apparently in 2000
New in 15-Mar-2000:
So, PERIOD = 0.0000167 sets the frequency task period to 16.7
microseconds, equivalent to the original "static int PERIOD = 20" that
was hard-coded in C.
that wasn't that long ago.. you guys have come a long way.
before my time...
sounds like they are noisy - low pwm freqency
better than steppers...
right - I would be able to put up with it :)
I am good at ignoring things.
the original control rapids at 250
I wonder what the max contouring is
like our laser is like 100ipm cutting - 400ipm rapids
I think he drip feeds from a PC, so it might stop after each move
he wasn't real clear about it
depends on how many blocks it can read in at a time.
good read http://finance.groups.yahoo.com/group/geckodrive/message/12405
I hate cnczone so much
[20:34:13] <anonimasu> http://www.cnczone.com/forums/showthread.php?t=41553&page=3
DUDE I AM MAKING HELIKOPT3R PARTS!
if someone sets out to build a cam measuring machine and fails
they are idiots
it's a damn simple thing
making a CNC version would add complexity, though
but that guy NC_CAMS turned it(like all other fscking threads he's involved into into a "IM THE L33T CAMGRINDER"
toast: measure/rotate the cam measure jerk/acceleration and lift..
I mean, it's a rotart to linear translation..
and a angle encoder..
no - it is way more complicated then that ;)
the stuff that measures accurately
is kind of nonlinear
so you could get a general overview of the cam that way, yes
but you'd need a function generator and a differential probe
and, a spindle, natch
toast: I'm not following..
ha ha cam joke
"i'm not following!"
toast: the commerical cam measurement machines has a probe(roller) that goes on top of the cam.
and a angle encoder for the rotation of the cam..
like i alluded to
it depends on how accurately you want to measure the cam
couple tenths, that's an awesome way to do it
i'm telling you how the followers work when they need to be more accurate than that
it doesn't just follow the cam, because the equipment that measures more accurate than tenths is nonlinear
so it requires a little more fenagling
is that how you spell that?
yeah, apparently it is
toast: encoders are linear afaik..
hmm i think art fenerty is full of crap
they're not using encoders
to measure the linear distance past tenths
LVDT's, capacetence sensors, etc
there are a couple ways to get that kind of measurement, but it all kind of sucks
fenn: He didn't seem to be slamming emc - that atleast counts for something
well its hard to slam someone else when your software is obviously inferior
I wonder why people are so scared to just try it - it's free after all
toast: yeah, there are lots of ways to measure, though good/cheap..
the whole pick two thing.
I have some 0-20mm inductive sensors ;)
i still love analog
if you have a 16 bit adc..
we have analog amps
for that kind of gear
they're very versatile
that gives you 20/65535 mm
ofcourse not optimal for doing that kind of thing ;)
what kind of thing
a analog inductive sensor..
for measuring cams?
or for digitizing
* anonimasu is mostly pissed off because of the attitude..
pew pew pew!
I mean, if someone wants to build something and you keep talking about your buisness
and how much better you do it..
ok, measuring a cam, how is that hard?
measuring it accurately is hard
unless you spend a lot of money using tools that are linear over the entire cam's throw
how bout a dial indicator
how about a encoder?
and a lever..
a dial indicator _with_ an encoder :)
for anyone in home shop machining that would need to make their own cam, both of those methods would work!
make an eccentric EMC measuring machine
i guess if feeler gauges are "yardsticks" then so are gauge blocks
well, or just buying a lvdt..
Our linear encoders break a lift down to millionths of an inche. Do we need to do it that fine? Probably not. Then again, some guys measure piston clearance with yard sticks (feeler gages), some use micrometers to 0.001" and others to 0.0001". How accurate do you want/need to be?
my reseller selld ones down to 1.5um..
and this brings me back to my poing about nonlinearity
LVDTs are ridiculously nonlinear
.2% over the entire envelope?
is that total error along the entire measure
or is that at the two limits
0,3% is standard
0,2% is optional
because that's .0005" over the travel
[20:55:40] <anonimasu> http://www.hemomatik.com/bild_pdf/Oversikt/59/SM_LVe1_h.pdf
and that's not very linear for something outputting in micrometers
toast: that's 0.0127mm..
a .25" lvdt
with .2% linearity
has five tenths of error
if that lvdt is outputting millionths, which is almost assuredly is, that's five hundred units of measurement
that's a specific case with one of our LVDTs
* anonimasu nods
that's what i am saying, nonlinear
because you can resolve at millionths, but you can't use it as a length measure
unless of course you hook it up to something like EMC
and do an error map and correct it digitally.
well, you end up doing that for any measurement...
not in the same way
LVDT error maps look like a drunken college student trying to find their way home
there's no real coherence for it, is what i'm saying
not that it's an inherently DIFFERENT kind of error map
just that a lot more points are necessary
than say, leadscrew errors.
* anonimasu nods
pew pew pew!
anonimasu: your mentioned http://www.cnczone.com/forums/showpost.php?p=329418&postcount=30
i think if you looked in the books, you'd know how to interpret the data
that's the cool thing about reading, there's usually related information in books
where does the nonlinearity come from btw?
physical inaccuracy in the coils?
they're not uniform
is it consistant nonlinearity - or does it change each time?
and you have three of them
oh, it's ultra-consistant
ah - then just map it once.
they make awesome dial test indicators
so you can use it to measure position to within a millionth?
for that reason
fenn: if you correct them, yes
but you really do need a very dense error map
skunkworks: the same guy runs in/out at all threads with IC discussion talking about how _great_ he's skills are..
im sure he really are.. but it's not like you have to tell everyone about it
and also temperature control
and all the other crap that goes along with millionth measurement
you could probably get .00002-3 in a home shop environment with a little work
and an error map
anonimasu: did you read the tread in emc? where he had no clue what emc could do - but was sure it wasn't good enough
without spending too much money
kill him with kindness I say
ah well, I'll just ignore it..
I think I should kill my cnczone user someday soon
kill him with an EMC voltron
don't you like the mach biased message board? ;)
i'm very very tempted to write back
"I hope you realize that you already told us about yourself at several other places at this board", and that hobbyists have vastly different requirements then professional teams do.
but, it's not worth the pain.
you know nc-cams has tons of supporters.. So it would be a loosing battle.
I guess he's yet another engine builder.
they are all kind of odd ;)
skunkworks: btw http://www.audietech.com/
he had mentioned that company I think
i'm back with more questions :)
How is the backlash work coming?
I think I'm gonna go with ballscrews actually
because they're cheaper than I thought
hmmm - cheap and zero backlash ball screws normally don't go togather..
well not zero backlash
the ones on homeshopcnc
he claims the standard nut/screw has a backlash of 1-4 thou
If I where to make a machine with cheap ball screws - I would use 2 nuts and preload them with belvel springs. But that is just my crazy talk.
so the preloaded ones are even better than that
preloaded ballscrews if done right - have no backlash
* anonimasu nods
or you could get oversize balls and load them
i've read some good things about homeshopcnc
until you end up at a reasonable level or precision
though that's lab work almost
I have no experience with home shopcnc stuff
they seem to be the cheapest source for ballscrew
anyone know how long the Z-axis screw on the x2 would have to be?
[21:29:38] <anonimasu> http://www.cnczone.com/forums/archive/index.php/t-38452.html
skunkworks: nice project :)
heh - I read that tread a bit.
he's there too
fish: I don't know of anyone that has an x2 on here. maybe jonE but he isn't on here much at all.
what kinda mills do y'all have?
[21:32:44] <skunkworks> http://www.electronicsam.com/images/KandT/DSCCurrent.JPG
[21:32:53] <skunkworks> http://www.electronicsam.com/images/KandT/DSCgantry.JPG
[21:33:06] <skunkworks> http://www.electronicsam.com/images/engraver/engraver.JPG
you dont' mess around
the first one isn't running at the moment. It is going to be converted to emc22
or emc2 even
* fenn gets the whip
yah yah - I am slow
i want that thing rigid tapping by friday
I can't wait.
that thing is going to be a beast
compared to its functionallity before
skunkworks: 20cm cuts in steel ;)
we have ripped a bunch with it. 10" facing mill and such
well - I have to drive home. BBL
what are you going to make with it?
got any plans?
I don't understand the question.. ;)
or do you just want it?
There is stuff I want to make - large stirling engine and such
_large_ is relative :)
what type of sterling?
I want to start with a 4 cylinder - where the hot from one gets fed into the cold of the next.
oh, four stage
I have to take pictures of the solar collector yet..
I take it you have a regenerator on each stage?
Yes - probably experiementing initially with stainless wool
and going from there - I have piston shape ideas also. - to transfer the heat better
hmm I see
but - remember I like to wing it.
which temperatures (or delta temp actually) are you going to run?
skunkworks: maybe you should ask nc_cams about it :)
skunkworks: im sure he has some good input on pistons
oh - good idea
I will be back later.
lerneaen_hydra: ambient and 120 square foot of sunlight to begin with
what medium to transfer the heat?
or something less exotic?
nothing that exotic - to begin with - I am putting the hot side into the focal point of the solar collector
oh, it's a reflective collector
the cold side will be cooled with water.
how are you going to focus on all four heads?
got a lake nearby?
heh - no
nice and fast: http://video.google.com/videoplay?docid=-2285657714531903438
all 4 heads will be in the same focal point (same assembly)
let me find a picture
hmm - not finding one.. BBL
pew pew pew
lerneaen_hydra: like this in effect http://www.whispergen.com/main/wobbleyoke/
and this http://www.whispergen.com/main/stirlingcycle/
the second flash down
skunkworks: oh, I see
so no middle piston
that's quite a neat setup
right - sort of 'double acting'
what size are we talking about
over or under 10kw?
I don't really know yet..
I have a mosfet with a top row 'M T' Next row 'P2955E' bottom row R404 M < motorola symbol
The 9 Maybe a g?
I can't find anything on it.
*reading the thread about emc with nccams involved..
I think I got it.
anonimasu: link to thread
anonimasu: funny huh
[22:44:24] <anonimasu> http://www.cnczone.com/forums/showthread.php?t=40309&highlight=nc_cams
I kind of think the discussion misses the whole point..
also, who said that integration is simple..
why does that one guy ramble about cars for like twenty minutes every post
toast: because he knows he's shit.
dude... that thread's... insane
lerneaen_hydra: I managed to piss him off in another thread :D
that's why we are at it..
what are you called there?
<- is atleast
[22:57:21] <anonimasu> http://www.cnczone.com/forums/showpost.php?p=329418&postcount=30
the other thread
* skunkworks samco
this nccams dude is like
like, i am an idiot
and i know this guy is a bigger idiot than i am
* lerneaen_hydra seconds toast
err. the bit about nccams
* anonimasu is kind of disturbed at him
ty for the clarification =)
We have and see them in our masters du to the logic we use to cut them (point to point milling as oppsed to curve fitting - we've tried both, long story). Via hand lapping and polishing, we get around the issues. Why bother? CNC ground masters take 7-10 days plus 1-2 days shipping and cost 2-3 time what it costs us to do them ourself - we get ours in 24-48 hours and they are made on site.
uys have tried to use step/direction to operate cam and OD grinders over the years. You can literally "see" the difference in the finish between a servo controlled grinder and a S/D controlled grinder.
well, woohoo what a fucking relevation
and that has nothing to do with the input_scale, noo
I should add him to my buddy list
have a 1:1000 timing belt and I promise you it'll be good enough
the movement by touching the table will be greater than a couple steps
well, obviously you shouldnt use step/dir on a ultra precise machine
of course, still it's not a flaw of step/dir
HAHA WTF "HOw you get counts per rev to mean something takes programming and math/calculus skills that not a lot of people posess. Heck, a lot of people don't even know what differentiation means let alone how to do it."
why do servos jitter
the jitter is dependant on the accuracy of the encoder
toast: because you have two phases, and no place for them to reach equillibrum..
as it needs to move a bit before it knows it's in the wrong place
and a encoder with counts it bounced between
then it pushes it back again
err, s/accuracy/number of counts per rev
lerneaen_hydra: hope that made any sense too
now this is kind of tangental but still related
whats all this about jitter? why not just set your deadband correctly
anonimasu: yeah, got it
having said the encoder issue, someone mentioned being able to use encoderless servos and use a different input
fenn: that dosent keep your motors from bouncing between counts or does it?
fenn: I don't see how you can't have some minimal movement of the servo with a changing load
i.e. table scales
or some other nonsense
is this true, and if so, would that eliminate the jitter
unless deadband is a type of hysteres and the load it low
toast: the mill at work uses resolvers on the servos(might be tachos) and feedback with linear scales
ah, so they're different items
anonimasu: resolvers, was that something that gives an analog sin/cos output?
I think it's tachos..
crap, it's late
could always make a HAL thing to control a fast tool servo
to correct the jitter
and Hi all
fast tool servo?
toast: the jitter is between two encoder counts..
anonimasu: there are a couple different types but the type i'm talking about is a piezoelectric actuator
or well a few.. depending on your pid..
so that vibration or other position error
any thoughts on hoe a linear motor will react to EMC2???
is cancelled out at the tool or some other spot
it's not position error..
it's vibration into the axis, right?
it's so extremely little..
ofcourse the coarser encoders the bigger it gets..
it's still positioning error
skunkworks: i think mach steprate is actually faster than emc, or maybe that was turbocnc that was faster. anyway there is a stepper driver in tuxcnc that is faster than either of them, i dunno if it ever got integrated into emc
put a fast tool servo between the main servo and the leadscrew
have it watch the jitter and counteract it
toast: the jitter is _extremely_ fast..
the acuators are faster
toast: how will you compute how much to actuate?
toast: is this actually a real problem?
yes, it's really in real machines
toast: tell me about what commercial machines that does this..
dont bring diamond turning lathes into this..
ultraprecice grinders, for one
roll lathes use BIG ones
toast: if you have a 5000cpr encoder..
to turn around keydrives
in quadrature that's 4x the amount of pulses..
that's 20 000 counts per rev.. if I remember it right
let's say the jitter is 5 counts..
that's 360/20 000
how is jitter going to be 5 counts?
fenn: it's probably more like 1 count..
piezoelectric actuators can position in tenths of nanometers
fenn: bad pid loop..
i thought it was <2 by definition
at very high frequencies
fenn: you are probably right
0.018 degrees in that case..
that's how much the servo is ~
if one rev of the leadscrew is 5mm
it's obviously more than that if it's affecting surface finish
fenn: please correct my math is it's wrong..
okay, so a millionth of an inch or so
a) that's a positioning error
and b) a fast tool servo will cancel it out
practically, they're used for surface finish
so you can use sub-par stuff in other parts of your machine, where you don't need real darn good true positioning
but want a nice surface finish
i'm not trying to make any sort of like, arguement here that people should be going out and buying all sorts of equipment
anonimasu: i get .0005mm
i'm just saying that there is that option out there
fenn: that's probably more accurate
that would def. affect visible surface finish
toast: well, I guess cnc machines arent good at all..
cnc makes fast tool servos even better
yep cnc is crap
what is wrong with you people
because then you can develop comprehensive error maps
toast: I dont know of any cnc mill that does that kind of stuff.. or has a need to..
okay, i said this in response to a grinder statement by nccams
toast: Okay, but what he really were talking about were stepper motors..
not servo motors..
all i am saying, and have been saying
is that fast tool systems are used to correct surface finish and small positioning errors
i don't know how this grew into "toasty hates cnc"
toast: what about active vibration damping?
and "no mills use it"
that damps vibrations, not necessarily positioning or asynchronous errors that would hurt surface finish
vibration isolation is certainly an excellent thing
yeah I see them in use in dtl's..
they're used in all sorts of things!
even on big roll lathes!
wow nice tape reader skunkworks
semiconductor fabrication equipment
i'm curious about active damping for hexapods
since they dont have the mass or cast iron, you need to use a hi-tech solution
hm, im curious about the actuation of them
when you say active damping, what parts of it do you want to damp
like viscoelastic shear or piezo actuators
toast: well, i mean put piezo actuators where the struts meet the platform
* anonimasu nods
i thought they were pretty stiff?
* fenn shrugs
what if $200 of vibration damping saves you $3000 in monstrous ballscrews
what you are describing IS a fast tool system
anonimasu: that just uses servos doesnt it?
I dont know
Actuator Resolution to 0.005 µm
ah - yes - we only used it for a bit before desciding it was for the birds. The computer (386sx) turned into the tape emulator.
the M-850 uses custom high-load precision screws and servo-motors
anonimasu: resolution doesnt mean active damping, you have to have the actuator bandwidth too
it were pretty good :)
200kg load capacity
piezos are several orders of magnitude faster than servos
they have one with screws or direct drive actuators
10000g acceleration i read in some article
which I think means piezoelectric/linearmotor
no direct drive just means the motor is attached to the screw directly
ah, that way :/
not sure that's an advantage either.. you get heat from the motor warping the screw
but it looks pretty :)
[23:33:04] <anonimasu> http://sfb528.tu-dresden.de/c3/4-Y%20Verzweigung%20mit%20HEXAPOD1_.jpg
the motor might be low temp
or they might be heated ballscrews
they might brag about it if they had such fancy stuff
but they dont
toast: I think that it's just you :p no offence but you are precision insane..
anonimasu: wood :(
anonimasu: it's kind of what i do?
i hate carpenters too
fenn: nist has one with plastic
[23:35:26] <anonimasu> http://www.hskworld.com/hexapod_1.jpg
nist uses theirs for making aluminum calibration fixtures
that thing is so ugly
looks like robbie the robot
01:38 < toast> anonimasu: it's kind of what i do?
he carps about precision
it's true, i do!
at work, that's the kind of stuff i do
[23:37:08] <anonimasu> http://tww.fh-duesseldorf.de/DOCS/FB/MUV/Hexapod.JPG
carp is the opposite of carpent
and that i'm getting into as a career
im happy if my parts are > 0.01mm
from where I want
my current chafe is the lathe i've been working on will turn surfaces flat to 20 millionths
but the workholding warps them
can emc do hexapod?
send it in the machine...LOL
i could DO em
Ziegler: yes (at least they say it does) ;)
i Love tome tolerances...makes me remember why i am...
is this hydraulic: http://sfb528.tu-dresden.de/c3/4-Y%20Verzweigung%20mit%20HEXAPOD1_.jpg
you see the motors on the side
I DO NOT KNOW
see the orange and green wires at the bottoms?
fingers are disfunctional this evening
big servos :)
oh, i thought those were servohydraulics
dmessier: we could do them too, if we redid the workholding
so what does a config look like for that??!
well, it could be hydraulic servos, but i dont see why you would do that in a research project - too fiddly
the problem is known, nobody wants to spend the time to make the chuck to hold the part
hydro-static cnc machines DO exist... MOOG-hydra-point...is 1
so you need tool holding help??
dmessier: no, we know what the problem is and what the fix is
nobody will spend the man-hours to do it
it's in tolerance at present, so we're going to leave it
toast: nothing's perfect
it's just irritating to know the machine will do much better
slap a plate on the lathe you are gonna use.. face it...c/bore to hold part.. drill hold down clamp holes and bolt FLAT..
and when you clamp it
the part distorts
and you don't get it any flatter than when you started
it sits at 2 tenths
some fixture require retruing every set-up
X Y Z R P W ?
shim 1 side.. skim and flip and skim..
i'm seriously telling you
that doesn't work in the range the machine will do
we worked some part from both sides /001"-.002" at a time...
dmessier: 0.05mm right?
and got it parallel to 20 millionths?
.001-.002" tool nose radius...
dmessier: what was your final flatness
shimming to millionths of an inch :(
you can't shim to millionths of an inch, that's the point
well, werm it with your hands..
we also used mag chucks if we could.. once it was getting really flat..
well with enough power it'll work ;)
alum.. Titanium.. hardened 4340... made no differance to us
dmessier: you still haven't told me what your final, verified parallelism and flatness numbers were
hardinge superlant lathes
obviously it'll pull tools from the machines in the shop too
dmessier: i'm getting +.0002/-.0000
4 micron finish
i want to get +.000020/-.000000
how do you use magnetic chucks on aluminum?
dmessier: there's another zero
some machines in that shop would do it..
aluminum was a pain..but the best air chuck in japan is made in the usa
the proper fixture to turn to that accuracy
is a special kind of 2-part chuck
hardinge SUPER-precision... chnc II.. conquest
you can load it vertically, stress free
get all your torque values even and your part stress even
and measure it to make sure it didn't bend at all
then reconnect the chuck, and it will hold
map it... ive seen it done on 18' pcs.. not 18"...
20 millionths of an inch over 18'?
are you drunk?
no.. but to holg .0005" over 18'... you map it
okay but i'm not talking about .0005" over 18"
dmessier: you are missing a zero to get where toast is talking about
we can talk about baseball
if you want to talk about things that are not what i am talking about
i aint drunk.. im just drinkin'... ; )==~~~~
this is why we should use the metric system :P
i'll convert it
can you grind it..??
grinding isn't as accurate as these lathes
anonimasu: .0005 mm
heh anon that's wrong
that's what google said =(
.00002 == .0005
i am using google do not blame me!
1/2 micron... all you had to say
i don't know metric yet =(
at least know metric quickly
i am trying!
[23:56:56] <anonimasu> http://www.google.se/search?hl=sv&q=0.00002+inch+to+mm&btnG=S%C3%B6k&meta=
0.002 inch = 0.0508 millimeters
0.00002 inch = 0.000508 millimeters
that is what i said!
and what I said.
though I calced dmess's value first :)
it is so frustrating
it goes from "damn this is a nice lathe"
dmessier: well, you are off by almost a planet :/
to "damn this is a crappy chuck"
toast: better get used to it :)
ok... so what material are you turning???
stringy shite....tool nose rad??
doesn't matter, it's diamond
surface speed?? and tool holder type??
dude, surface speed doesn't matter on diamonds