I also don't completely understand the rules of linking. Many times I could link in the hal box manually, but when I put it in a script, It would say that signal (or pin) was already linked.
Goslowjimbo: you need to read the hal manuals, the rules are simple ( tho i dont rememeber ) and exdperiment on a vrey small basis ( not inside emc, but inside hal as the manual tutor works)
in general, if hal sez its already conneted, it is ;)
and it will get clearer with simpler testing
i been stymied by similar, and it was relaxing and examining the lists presented by halmeter that showed me... the real name, the real connections, the things i wrote but didnt attach to a thread...
Why do you not get the error when you enter it manually? The pins or signals are still connected.
Halmeter shows connections? I thought it just showed values of signals or pins.
you choose he signal from one of 3 (lengthy) lists signal functions params (??)
as to why when you dont... wrapping my decaffinated head around that...\
Oh, OK. you're talking about what I would call the hal box which shows the relationships and the values if you work down to the simplest level. I've been using that to get this far.
sorry but 8:30am here, gotta get to work. i'll be looking at hal and pyvcp next few days maybe i'll run into same & get some experience
Poinca814 is now known as Poincare
i put on my robe and wizard hat
no, google says it's a bloodninja shtick
[01:08:16] <eric_unterhausen> http://bash.org/?search=robe+and+wizard+hat&sort=0&show=25
it was a resubmit from another, much older webpage
bash.org is a great way to waste time and lose brain cells
...yet the chuck went fly... ;)
is a special cutter the only practical way to put a radius on the outside edge of something?
X/Z or Y/Z plane radius
you could do it with a ball end mill.. it would just take forever
as in like I were contouring?
yuck, that is forever and a moment
problem with small mills is I can't even use a radius EM even if I were to want to (unless they are made w/0.25" shanks)
on several occasions we've used large, long endmills
and put parts on edge
but there's no "easy" way
ds2 whhy not just get a router bit
you probably already have one
do router bits work well at around 3600RPM on aluminum or plastic?
no issues with clearance/rubbing ?
they have tons of rake
and plenty of clearance
lots of gullet volume
then what makes then router bits and not EM's?
endmills are helical (usually), are made of tougher grades of HSS, there's less room in the gullet
I'm sure this is not true of all router bits, but endmills I use are sharper than the router bits we buy
router bits tend to have an enormous amount of rake ground in by radially offsetting the tooth, rather than by grinding the gullet to a form
fenn_ is now known as fenn
So, when I'm running latency-test on the *live* ubuntu 6 cd; everything is going pretty well until I start firefox and I take a big hit (jitter was around 10,000 , jumps to 157,000) -- is that an issue - it seems because of the CD access? If I don't have CD access during CNC - I'm ok right? Second question - do you always take the worst case jitter number? for the 1ms/25us threads?
jumping to 157000 is a bad thing
how much memory is in the computer you're testing with?
hmmm. that may be too little for running off the CD
But the hit seems to come with cd access... as I'm running the live cd.
right. the CD image is compressed, and it's possible that decompression (most likely run in the kernel) is causing the delays
just a hunch though
So I guess I'll just need to try to install to hdd. So the point is - no matter what - CD access or xyz - the RT kernel should have < #### jitter given the system.
its's OK for the number on the servo thread to be higher than the number on the base thread
the base thread number is the one to enter into stepconf, if you use that
or the one to use when doing step rate calculations, if you do them byhand
fenn_ is now known as fenn
btw - I'm wanting to make a CNC machine to use with a punch tool - something I'll make (unless one exists?) to take a razor blade to cut out cardboard. So basically a programmable die cut machine. Have you heard of anyone doing such a thing?
yes, cnc turret punches
not with a razor blade
look up steel rule diecutting and turret punches
a thin blade like a ruler is bedded on a die block, and surrounded by neoprene or cork
it stamps down, cuts whatever
the turret punch has a programmable table and multiple tools, so if you have a 6" long rule, you can move the table and punch every 6" to cut a line.
my thought is to basically use the razor blade as a single steel rule die, then move the head and then give it a down command. I'm thinking to put another stepper to rotate the head.
toastydeath: how is g-code generated for that?
i've never seen turret press g-code
what i do know about it is that it's a bunch of coordinates and tool changes.
call a tool, go to a place, the machine punches.
next location, next location, etc.
I guess for my project (cutting out hexes or squares from paper) it should be simple enough to do by hand...
it is done mostly by hand.
definitely look into loops and subroutines
but you've got a more complex machine than the ones in use
Say I had a 1 inch blade, I'd move 1.010 inches over to leave a little sprue(?) to hold the peice in place until punched.
punch machines are dangerous
I figure I can do pneuamtic (control PSI for pressure) or a servo on a screw to push the blade down.
york: the actual machining strategy is dead simple
the complex part is your rotary head
Out of curiosity -- would you move the tool or the table (is any one common for a CNC machine?)
both are common.
for a punch that's going to cut cardboard, i'd probably go with a router-type arrangement
but with a much beefier gantry.
so a table that moves in X, a stationary gantry, and the head moves on Y and Z.
and C, obviously.
but that's just my opinion and any configuration you can come up with will work.
and you also may want a punch+die kind of setup - you need to go through the material, but not attempt to cut the table itself
Hmm. I'd never thought of X table, Y/Z head.
or you'll dull the blade awfully quickly
if you're willing to do it, you might want to have a Z axis move the stop
and the "punch" command just has a ram set at 90 psi
and bludgeons down on the stop.
save your blade that way.
C isn't ideal for tool rotation, since C moves would be coordinated with XYZ
oh, nevermind - you were talking about orienting the tool
SWPadnos, by punch+die - do you mean using shearing force vs. (ehm, I guess) cutting/splitting force?
swpadnos: i don't follow
york: no, he means having a block set up to recieve the blade
toastydeath, I initially thought you were talking about using C to select tools, but now I realize it's for orient
a slot in it, or some other arrangement so the blade can make the cut but not hit the table.
and can you elaborate on C, I'm afraid I don't follow. Rotational (of the head?)
XY = moving around the plane of cardboard, C is rotation of the blade in that plane
Z or something is the "punch" command, which cuts that line segment
with a head that can orient the blade, it would become more complicated to make a die block
for the punch
it might be worthwhile though
but you don't want to mash your cutter down on a steel table all the time, you want it to go through a slot behind the cardboard
it would just have to be slaved - not a real problem
it also makes the table more complex
other than tool shape changes of course :)
you'd have to make it like a real turret punch press, instead of like a mill or router table
yeah, I'm too tired to think of how to do it, actually
the table on a punch press is just an open frame, like an EDM
so you have this flat stationary thing to put the work on
I'm not sure how I'd mount the cardboard I'm cutting... I guess it could be free floating... I'd just always envisioned it taped down.
and the "table" is just edge clamps.
I did see a system that used EMC for positioning end stops for a punch press, but the material moved over the punch location
so you clamp the edge of the part, and it just drags it around on the stationary table.
I guess if I have steppers, technically I could have two steppers one rotating the block and one the head. Sounds ineffecient, but may be the simplest thing...
yeah, with cardboard, you could use rollers and a couple of capstans (almost) for motion, and have a stationary punch
you need a LOT of space though
because you can no longer use a gantry.
punch presses are very large machines.
Well, I figure my biggest media will be 8.5x11in
oh, then not so big.
i kinda figured you'd be ripping apart UPS boxes or something
So I should be able to use EMC to control whatever contraption I dream up?
my plan in cereal boxes..
if you can dream up a contraption, you can dream up the configuration to control it with EMC2 :)
i would suggest not flying by the seat of your pants though, sit down and say step by step how you're going to put the material on the table
and what the machine will have to do
Just to clarify - gantry is the term for the mechisim which moves the tool...
since you are kind of moving into uncharted territory
no, a gantry is a bridge over the table
york the latency is probably caused by the actual disk access itself, that can often throw a big lump of CPU out the window while its spinning waiting for the access
think like a big crane at a manufacturing plant
a beam moves along tracks (usually made of two other large beams, which don't move)
right, but doesn't it normally move the tool?
a gantry machine CAN have a mobile head, or the whole gantry can move, or it can be completely stationary
and the winch (tool) moves along the moving beam
the contrasting term is a C-frame machine
where the head is not connected to both sides of the machine, but only on one, making a C shape.
Valen: so you wouldn't sweat it? (or is the theory for the 1GB of RAM that all disk access should be cached whilest EMC is running?)
more ram is never a bad thing
disk access shouldn't cause a 157ms spike
It might be a bad driver
CD-ROM disk access though?
but the CD is significanty slower, and probably has to spin up
If its accessing the CD
I reckon try it
Now it sounds like stepper motors are the simplest implementation.
oh, here's a nice gantry example: http://www.oneoceankayaks.com/madvac/cnc_images/wholemachine1lg.jpg
I got a price on my big servo's
ignore the guy
that is a nice gantry machine
$460 + 70 for 500 line encoders
what power and type?
think 500 would be enough? or should i run 1000
that tells me nothing
AC, DC ...
[05:04:36] <Valen> http://www.magmotor.com/brushed/brushed.html
down the bottom
what size machine?
Have you guys ever tried automobile power-window motors or windshield wiper motors?
and servoing them?
swpadnos: additionally i like your capstan idea for a punch machine
since the feed loads are so low
but i'm planning on direct driving the ballscrews
Valen, those are probably big enough. a good Bridgeport retrofit usees ~30 in-lb motors, an dthe machine is ~2000 pounds
nope, you won't want to do that
york, I was going to use car fan motors in a cheapy one
toastydeath, yeah, though it would get tough once you cut pieces out :)
ohh you meant to drive the material DIRECTLY
there would be gaps in the cardboard eventually, so the capstans would have nothing to push on
that would be cool as a concept, but yeah
i was just thinking of driving clamping rails
Valen, what screw pitch and diameter?
SWPadnos, that one is only 15 pound inches though?
yeah, more or less
i can easily stop the commercial retrofit machines at work with one hand
half size motor, half size machine ...
or cause them to fault, anyway
25mm x 5mm threads and about 750mm long max length (500mm travel)
do they have 30 in-lb cont. servos on them?
Now you could gear any given motor down to get the power you need, the only con would be machining time would go up?
my maths says it should have a peak force of 1300kf (2600 pounds)
Valen, you'll be lucky to get more than a couple hundred RPM on those screws
before they bend like a jump rope
25mm diamiter ballscrews? thats an inch
We were also thinking of putting a few hundred kg worth of preload on them in tension
hmmm. maybe 500 RPM or so
Valen: what is your CNC for? Is it using EMC?
why not just buy a better ballscrew
instead of hanging weight on an axis
but even if you cvould get to 1000, that's still 1/4 the rated speed of the motor
you usually want to design for 80% of max speed for a servo
toastydeath, its preload, its just loading up the bearings
no actual mass
toastydeath what would you use for a screw then?
how fast do you want the machine to move>
dual nut ballscrews with the right amount of preload when they showed up in the box at my door?
fast enough ;>
it's hard to buy a screw with "fast enough" preload
you usually need a speed and a mass
by preload i mean the bearings that the screw turns in
note that the motors on the BP are 30 in-lb, and geared 2:1 to the 5TPI screws
(which are 1.25" thick, I think)
yeah so this is just a little bit smaller in terms of diamiter and about half the mass
yeah. I think stiffness is a square law thing though
1" on a ~1000lb machine
but then again, the BP screws are twice as long
stiffness is measured in in^4
close to 3x actually, but also held in the middle by the yoke
a 25% increase in diameter is twice as stiff.
If we have ~2000 pounds of tension on the screw (ie the tensile force is higher than the maximum force that can be applied) it should be fairly straight i would imagine?
it will be straight, the problem is "whipping"
the preload doesn't affect the straightness of anything
ok i think you are misunderstanding what i mean
you can only spin it so fast
our plan is to hold the bearings on the left side, then on the right side of the screw torque the other bearings such that it applies a tensile force to the rod
in any case, you won't be spinning it anywhere near 4000 RPM (or 3600 RPM), so you'll want to use some gearing between the motor and the screw if you want to use the power of the servo wisely
err, 3200 is 80% of 4000, isn't it? :)
the winding I'm looking at should be able to hit ~2800 RPM at 160 volts or so
valen: i don't understand why you want tensile load on the screw.
if you geared 3.6:1 or 4:1, you'd still have a top speed of ~5M/min, but with much better accel
to stop it from "whipping"
considering manufactuers of machine tools go to completely bizarre lengths to NOT do just that
exactly how fast are you going here?
the "theoretical" top speed is 400 IPM i believe
okay, that's really easy
I'll be surprised if the screw can be safely spun at 2800 RPM, so make sure you have a camera running when you try it :)
figure out the rpm of the screw at 400 ipm
then go find the shaft whip calculations
and it will tell you how big of a rod you need
SWPadnos, the calculator from the ballscrew manufacturer says in its low end bearing configuration its good to 10000RPM
no preload necessary then
over what length?
like I said, make sure you've got a camera running ;)
time for bed. night folks
the max RPM we can hit before running out of volts is going to be 2500 or so anyway
putting a screw in tension or compression does Really Bad Things (tm) to your accuracy
if you care about such things, anyway
yah, it'll bend the bed as well
Thanks so much for your help/input. One last question - I have some steppers (can't find a datasheet Minebea 23LM-C006-T2) which are rated at 24V/phase and 0.3A/phase -- that's 7.2 Watts. Power = Torque * Angular speed. How can I find how many oz-in these steppers can exert by math? By experimentation I'm thinking pump 24V into a phase, then attach a 6 inch bar (l) and then start adding weight until the motor turns, then I can know o
york: you won't be able to find the continuous rating.
It did seem for the lighter routers, they were doing stuff in oz-in.
you will find the stall torque, which is bad.
i have no idea how to calculate the maximum continuous torque of a motor
stall torque bad? Wouldn't that be a good thing to know though?
slick-: hi :)
you'll burn the motor out
if you design around the stall torque.
If you know the max cont wattage
Stall torque >= continous torque though, right?
and the amps per ounce inch you can work out the max cont torque
usually much less than.
york are those ratings max or cont ratings?
I don't know - it's just what's printed on the stepper.
I think minebea no longer makes these steppers.
I couldn't find datasheets on the web.
how physically big are they?
anybody heard of "rollerscrews" ?
I think they are NEMA-23
yeh, i have heard of them
I'd guess that those ratings are cont ratings then
better or worse or what compared to a ballscrew?
2 in diameter...2in lenght.
i suppose it depends on the application
but i'm not aware of any particular pros
york yeah that should be able to disipate 7 watts
of roller screws over ballscrews, and ballscrews are more common.
i'm not going to make any guesses as to which is really superior
On the order of oz-in if you had to guess. greater or less than 100 oz-in?
but I'd put .3A through it and see
hang onto the case and make sure its not getting too warm
But if go slow enough -- I'll have better luck moving the tool/table though...
guess so, its your time ;->
ok thanks again all.
speed has suprisingly little to do with point to point accuracy
oh yeah, by slow do you mean just turning slow or do you mean more gearing
even with gearing, you add a whole tolerance stack
between the encoder and you
yeah, (thats why i want direct drive)
for really high feed rates, you drive the ballnut instead
be hard to get the power to it though i'd imagine,
at least on smaller machines
on industrial machines, even small ones are "big"
so they've got space for the wire looms
its space for the motor thats my problem lol
or they fix the ballscrew to the table, and the ballnut to the frame
and drive it that way, but it takes up lots of space.
I thaught they would have typically just stuck the motor next to the ballscrew and belt driven it or something
Although that does make an interesting application for a motor we are developing
It could actually be integrated around the ballscrew itself
it is on fast machines
the ballnut IS the motor
just like the spindle on direct drive lathes IS the motor
toastydeath, think people would like self driven ballnuts? brushless dc servo motor style
you just said they did
toastydeath: the ballnut IS the motor
no, i said big, fast commerical machines have that
how many mega$?
which is not a very big market
what if it could be done for less than mega$?
the sort of thing that would work on inch shaft
if you can make a machine move a ton of metal at 6000 ipm
for less than 200k
you probably would have a very large customer base beyond machine tools.
heh that might be our automotive sized ones perhaps ;->
the issue is those machines have to move 2000 lbs, accurately, at 6000 ipm.
Kitamura has repeatabilities under a tenth
and geometric accuracy in the millionths
the prototypes we are developing at the moment are around 100mm diamiter
thats all down to the ballscrew
we just have a motor
that could couple coaxially rather easily onto a ballnut
i mean if you want to market it and see who bites go for it
it's just not a product that has been popular with smaller machines
mainly because of the price?
perhaps due to marketing reasons more than technical difficulty
but there's more technical difficulty in designing a ballnut that's driven
into a machine, that is
not the ballnut itself.
then there's price
needs more room and wiring looms etc
if you're not moving fast, why do it?
you get worse surface finishes
If you can make the price similar though
a heat source right under the table.
I wouldn't have thaught it would have mattered that much at the smaller scale
well again, i'm making some assumptions about target market
someone who wants to move fast.
we are planning a water cooling option for our motors anyway so bonus there
although that was more the automotive scale ones again
fast machines have cooled screws and cooled nuts
looking at around 100Kw or so output power at this stage for a smallish automotive scale one
as in, machines designed to machine automotive components?
We were origionally planning to make motors for electric cars
wheel motors for robots and the like
thats what our prototype motors are, 100mm diamiter wheel motors for robots
well 80 before tyres and the like
they seem to be matching what the theory says they should do
the motor is not the issue though
in what we're talking about
It is if thats the bit I can make lol
you could connect tiny hamsters in wheels if you wanted
right but that's why i'm arguing with you so much
you are coming at it from a motor design and technical standpoint
versus from the perspective that the machine will have to meet a goal
why do something new if there aren't any benefits?
There would be benifits i'd imagine
you can't just imagine benifits
they not only have to exist and outweigh the economic problems associated with changing technology, but they have to do something to further the goal of the machine
you get a LOT of issues when you mount a motor to the machine table
If the cost of the motor + ballnut is comparable to the cost of a traditional external motor and ballnut system then there are savings to be made in terms of end fixities and the like
Other than wiring and clearance issues what are they?
it is VERY simple to mount a ballscrew the wayi t mounts this way
not paticularly cheap though
it is much more difficult to mount a driven ballnut
decent end fixities are ~$300
why do you think you can use cheaper parts on your product?
I'm thinking you can use fewer parts
because you arent spinning as much
if you are putting a driven ballnut on a machine
you need to have it floating
you only need one precise spinning to fixed mount
either with fixtures
or with hinges
the nut on machine tools is gimpy for a reason, so that it flexes and movement and twist in the screw doesn't move the table in an undesired way
Heh, fixed mount - famous last words.
and all the big mounting crap can be out at the edges
where it matters less
if you have a motor with heavier mounting you have to deal with that extra, and unwanted, rigidity
so what would it need to do in order to be a good seller in the "somewhat smaller" end of the scale?
it needs to be almost entirely vibration free, it needs to run cool, it needs to be easy to wire, sealed from the chips and oil that will get to it,
and it must be mechanically isolated in five of six degrees of freedom.
I was thinking of using our motors in place of the magmotors when they are fully functional externally but this is an interesting idea
how do you mean mechanically isolated?
Five out of six dentists agree...
valen: if the ballnut twists in any way, it cannot impart that motion to the table
if it shifts in either of the non-driven directions, it cannot impart that motion to the table
the only motion you want is axial.
we are hitting 95% efficency now and there is no iron in the motor so its pretty smooth
so theres your cool and vibration free part ;->
needs to be "an order of magnetude smoother than a servo that's mounted at the end of an axis"
ok its as smooth as the bearings, happy now?
lol it cant get much better than that
you're putting the bearings in it
there is 0 cogging torque
there are bearings in our ones now because we are using them as wheels
er, what i meant by that
is that you're the one deciding the smoothness of the bearings
obviously in the new application there would be changes
and that the surface finish capability of a machine is very easily destroyed by a noisy driven ballnut
It should be about as close to perfect as its possible to get
then that won't be an issue
so if you have a cool and smooth motor, the motor design is done
mechanically there's nothing in it to cause a first order departure from "snooooth"
steppers and servos all have big lumps of iron in them that cause cogging and the like, thats why they have to have skewed windings and the like, that hurt electrical efficency
right, but the other 90% of the problems with the ballnut
have nothing to do with your motor
that one is out of my hands
I presume somebody else would have solved that problem before though
they haven't, not economically, which is why you don't see that drive layout in all but the most expensive machines
and then they're going to linear motors
which are faster and have fewer problems with the mounting
are they induction or permanent magnet typically?
i don't know.
How would you like a linear motor in the smaller scale size?
very poorly, i'd imagine
i don't do anything where i could see an advantage to them.
plenty of room to get enough torque
but there are pleny of dudes who need tiny, fast robots.
semicon and pick/place
high accuracy pick/place, not like, lego bricks
I imagine that they would put some sort of DRO next to it rather than running open loop in an aplication like that?
dunno, i don't deal with that stuff
don't know a bloody thing about robot control stuff
Hmmm I might look at that a bit closer
if i had to invest in a company, i'd invest in the linear motor company over the one trying to drive a ballnut.
If we can make a retrofit kit for an X3 size with linear motors it might be interesting
zip zip zip
people like the "latest and greatest"
and no pesky ballscrews to muck with
but also speed limited
people like cheap and easy
due to bearing quality
your the air bearing guy arent you? or was that somebody else?
yeah that's me
well there ya go match made in heaven l->
the bearing set would probably cost 1500 per machine
to float a table.
and that's like, skimping.
thats still in the ballpark for most of the conversions
well like, i'm saying that just the bearings
they are typically 4K or so
you could run on the existing slides i'd imagine
i'd have to look at them
you would have to machine the carriage though
Probably easier just to make a mill from first principles
because even slim bearings are like, .5"
now theres an idea
or .4" i forget
just float the table on a 2 axis bearing
and use a 2 axis linear motor underneath it
yeah, you could do that.
positivley cheap ;->
you can run on flycut surfaces, if you took aluminum U chanell
just need one really bigass flat thats made of plastic
they're just not very accurate
yes glass would be best
glass isn't very flat.
it can be made to be though
sure, but then so can aluminum.
and its a nice bearing surface
aluminium is conductive though, it will muck with the motor
i was thinking more of a stacked axis thing again
rather than the flat
so that makes more sense
now that i've caught up again.
actually if you wanted to you could make the flat out of steel or al
then use and induction system to shunt the "table" around
only thing is it will need some sort of DRO system on it to get postion
cos its really really open loop ;->
oh hey, how about this.
L shaped plates
one long leg, one short
mount a linear motor on the short leg, sticking up
and drop another L shaped plate on top
the air bearings can just fly on the big surface
i am tired, and going to bed, goodnight sir
I was up till 5:00AM talking to suppliers in the USA
its 4:00 PM here now
time to take the dogs for a walk then go get sushi
good luck with your adventures in canis familiaris and dead fish
craigw_ is now known as cw76
cw76 is now known as craigw_
craigw_ is now known as craigw__
craigw__ is now known as cw76
cw76 is now known as KB5UEJ-EM50
can emc handle input from DRO type scales and use that to close the loop on a servo type system?
i guess it can, the outputs of those are normally quadrature, so it should be possible to simply interface them etc
Pondering the practicalitys of a 2D floating linear actuator type design
so basically you start off with a "big flat" perhaps glass that you polish down to being actually flat
or Al or steel
stick an air bearing on it
actually definatly steel or Al, needs to be conductive
then stick a set of linear induction motors on that air bearing carrage
and you can push the carrage around in any direction you feel like
for extra credit you could add rotation to it as well just for laughs
IE a static cutting tip, like some kind of demented lathe
interesting. is this something u are building?
Good morning. I have a small problem compiling EMC: ./configure --enable-run-in-place works but when I do a ./configure; make; sudo make setuid; sudo make install then emc bails out: "_tkinter.TclError: couldn't load file "/usr/local/share/emc/tcl/emc.so": /usr/local/share/emc/tcl/emc.so: cannot open shared object file: No such file or directory" - any ideas?
This is with the 2.3.0 cvs version I got from the server last thursday
i am wondering what setup you would recomend for driving a 6 axis CNc machine
ashaw: Depends partly on the budget. I'd prefer servo motors but steppers are also possible. I am just setting up a 4 axis stepper system with Mesa 5i20 card: lots of I/O and stepper/servo control is configurable.
i was thinking of $700US for control electronics plus another $500 for motors etc.
My config: Stepper motors are 6.4 Nm (92 EUR), stepper controller is an MSD-50-5.6 128 microstep driver (70 EUR) and a 40V power supply (110 EUR). I use 4 motors+drivers and 2 supplies (one was not enough). Totals 868 EUR
how much was the mesa 5i20?
But then, the only thing to add are the Mesa 5i20 card $ 199 and a bit of motor mounting hardware.
and 6.4 Nm on a 5mm ball screw spindle is strong enough (for me) if my arm gets stuck in the gantry it will break (my arm) ...
have you looked at the 5i22? is it better, worse?
Similar. I use the hostmot2 software and that fits the 5i20. Only advantage of the 5i22 would be 92 instead of 72 I/O (which already is a lot) but it is + $170 ...
an aditional $170, wow!
what is hotsmot2?
hostmot2 is the FPGA software to control stepper motors and/or servo motors.
how many steppers/servos can it controll?
8 steppers or (as far as I know) also 8 servo motors
Look at http://www.mesanet.com
there you'll find the card under the "parallel I/O - misc I/O - FPGA" section on the menu in the left
They have specs/datasheet and all software downloadable - before you buy the card.
i am at that page.
meanwhile I am trying to see if I can get you some more info on the I/Os by configuring a 6 stepper config.
i am thinking that it might be worth the extra money to get a 5i23.
there are different software (firmware) version for the 5i20 and 5i23 delivered with the new 2.3.0 (beta) version of EMC. The 5i20 supports 4 steppers + 8 encoders or 8 steppers + 2 encoders. The 5i23 (as far as I can see) 8 steppers + 4 encoders and possibly even 8 encoders.
so the 5i23 is worth the extra $30?
if you need/want the extra encoders Yes.
I am not sure if EMC supports stepper motors + quadrature encoders on the screw spindles
So if you are not using different pendants on each axis (and do not use encoders on the axes) the you don't _need_ the 5i23. But with the extra 200k gates and spartan 3 it looks a good option to be future proof
i lost the connection, did you say anything?
last msg was: So if you are not using different pendants on each axis (and do not use encoders on the axes) the you don't _need_ the 5i23. But with the extra 200k gates and spartan 3 it looks a good option to be future proof
One of these things to jog the axis instead of using the keyboard. Kind of a hand-wheel
what do i need if i have want pendants and encoders on the axes?
a lot of time to browse the internet for parts and more time to configure everything :-)
without joking ...
yes, but would the card be able to handle it?
you'll need encoders on the axis and a pendant. Configurations - or at least help with configuring - is in the EMC2 knowledge base on linuxcnc.org
i mean a pendant on each axis.
The card will be able to handle it, but I am not sure about EMC2. I do not know if EMC2 handles steppers with encoders on the axis.
who would know?
That should be possible, but I am not sure how to do this. Maybe best is to ask this via the mailing list - there may be some persons who tried this.
so a pendant is a quaderature encoder with a knob attached?
Yes - I also want to add a pendant, but just one with a switch to switch between the axes - that is the most standard configuration found.
something like this http://markusmechatronics.com/Handrad_HowTo.html
i think 2 pendants would suit me, that way i can use both hands
Y - problem is how to know which pendant controls which axis. I still only have a keyboard where arrow keys control XY and pg up/down control the Z and [ ] the A axis - But I like to add an MPG pendant to get finer control
one thing that might be cool to add to the software is the ability to use a spacenavigator to controll the movement of the cnc machine
[08:43:43] <ashaw> http://www.3dconnexion.com/3dmouse/spacenavigator.php
Cool thing to use indeed. I have a their navigator and the pilot to control my CAD software. Works great but I am not sure if I am willing to risk my CNC machine on this.
Nice thing on the MPG pendant is that the dial has clicks and each click corresponds to exactly one step - so I know exactly how much the axis moves by feeling and counting the clicks
what do you mean by risk your cnc machiene.
If the spindle runs into one of the clamps or a workpiece I risk damaging spindle, gantry or at least the mill cutter (my 'cheap' cutters are around 20 EUR). Control with the space navigator is not good enough - there is no fine control or feedback
what about the limit switches?
You do need those on all axes. I am still operating without any; I just set the home at the middle of the table and have the soft-limits set such that EMC will never run to the end of the table. But I've already set the home position wrong a few times ... So _must_ add home and limit switches
I damaged a motor mount. 15 mm aluminum sheet bent due to running the gantry into it :-(
mechanically what is your CNC machine
[08:56:25] <RobJansen> http://www.myvoice.nl/freesmachine
and there is a photo album at http://www.myvoice.nl/freesmachine/album.php
- but these are all old photos made while building. It looks a bit more 'done' nowadays
Size is X = 2000, Y = 1000, = 360mm
cool, that is similar to the size i want.
how much were the ballscrews?
Pfff - don't remember exact prices. I'm checking now - these came from http://www.metallstore.de/
stepper motors, drivers and linear guides were from http://www.stappenmotor.nl
and some other stuff (wheels and belts) from http://www.cnc-plus.de/
ball screws are 86, 133 and 183 EUR each (for Z, Y and X). X has 2 spindles and 2 motors. The nuts were 91 EUR. T-nut plating was 1700 EUR alone ... Complete machine was about 8000 EUR
what accuracy were teh ballscrews?
two spindles? how do you keep them synchronised?
best thing is to search in the mailing archive, search for "2 motors on 1 axis" or similar.
RobJansen: what are you driving?
whats the wattage on the motors your using?
Valen: the spindle is a 2.2 kW ELTE spindle with 2 kW Toshiba inverter
I'm wrestling with the inverter just now. I like to control the 0-10 V analog input to control the spindle speed using a PWM from the 5i20
whats the drive motors?
If its a VFD it may have some sort of digital input perhaps?
drive motors? that drive the axes? Those are 6.4 Nm steppers
what is the cont wattage on them though?
yes, VFD does have a digital input as an option. The connector is there but I can't get it to work so now I'm going for analog
Steppers don't have cont wattage. current goes high only during moves. Max 5A per phase, max. voltage is 42 V but wattage is hard to measure due to micro-stepper config.
It should still have a maximum continious rating though i would imagine?
at some point it will catch fire ;->
if you know the current and the winding resistance of a stepper, calculate maximum rated power dissipation according to ohm's law: P=I^2*R.
depending on the stepper driver, the power when an axis is not moving may be reduced, but I personally think it should remain at full power. One problem with reducing power is that the axis is less resistant to being pushed out of position by cutting forces. It seems like a good way to loose position.
argh, "lose position"
emc on boingboing, in an abstract way:
[12:13:32] <Optic> http://www.boingboing.net/2009/04/17/laser-cutter-motors.html
I guess someone needs to send them daisy
Hello from the EMC booth at NAMES.
you guys get up early
its 10:30 PM here
dalegrover, where's the webcam? :)
well that's annoying. I actually want to answer a question on CNCZone, but my password doesn't work today
The show just opened the doors. One question we'll get for sure--how does EMC compare w/ other software. Is there a page or list of main points to hit? Besides initial cost?
in the hobby market, there's no doubt in my mind that it's the most reliable control out there
it's probably comparable to most things in the pro market ($5000 and up) as well
I don't know how you want to say that though. I always have a hard time wording that kind of thing so as to not put other products down
(sell yourself, don't trash the other guy mentality)
EMC is also very adaptable, so you can run just about any machine you want if you're willing to take the time to figure out how
oh, and it's capable of driving servos in a completely closed-loop fashion, which no other inexpensive (<$2500 or so) control can do
Thanks--this is the kind of thing I was looking for.
May be asking more specific questions as folks drop by.
I'm sure there are other things, like rigid tapping and threading support that actually use feedback from the spindle :)
ask away. someone is usually here to help
Dale: some advantages are: Good documentation (both user docs and installation/configuration), easy to start with (configure parport setup using stepconf) for beginners but can also handle very advanced configurations, regular updates, very stable. These are some of the reasons why I am very happy with EMC
A wide support base of users...
the 'cone' demo for 5axiskins, gives joint 1 following errors, after lloking at it , reducing accel and max vel and allowing huge fe's, i think its really just stuck on a limit 'joint 1 on limit switch error'... uh how do i fix non physical limit switch?
override limits and then jog
(kind of like a real machine)
the limit switch is a comparison between the joint position and some number, so to EMC it looks just as real as a physical switch connected to an input
i forgot i had tried to override, un-estop, turn servo on, and you get the error right away which sets estop ( rinse lather repeat )
like a real machine, leave it in estop, turn the screw till away from switch, then enable machine and drives ;)
Just set limits 10x what they really are =)
Limits, We don't need no stickin limits!
in halscope the vert gain goes from 500>u< to 1>m< . What does that mean? micro to milli? microns to meters?
tomp: p=pico, n=nano, u=micro, m=milli, K=kilo, M=mega, G=giga, T=tera
yep, thx, was just rescaling till it was logical m=milli
(standard SI prefixes)
yep, nver saw si on a tek/hp scope ;)
just bitchin never mind
i'm lookin at the 5axiskins running 'cone'. joint 1 ( Y?) gets fe's that trip,
but I dont see why.
the ini values for accel max vel min fe, and fe look same for many axis that dont give fe..
might i ask what CAM you would recomend?
cam the apps you love to hate ( the apps most worked around , not with ) , i'm a sadist, i use Acad MDT
apart from Acad MDT what else?
paper pencil , the ITW trig book ;)
some people here like SOlidworks, some NX5
or 6 i spose
i mean to generate a gcode file from cad data that allready exists
nod CAD but CAM
what kind of machine?
[13:53:08] <JymmmEMC> http://www.dakeng.com/ace.html
dunno, i thought CAMs came with posts
the code that produces gcode from path created by cam
yes but some is much better than others, or so i have heard.
usually its very flexible, you tell it your special mcodes and gcode formattin and number of axis and maybe the kinematics, and it writes gcode from the path you make on the surface of your model
Ace Converter - utility to convert two dimensional DXF files to g-code
better, well some is pretty complex and soem is not adequate for some machine tool configurations ( like 5 axis wedm or a puma robot vs a lathe )
shows how unknowledgeable I am, I have been working with a pen and vellumn
and illinois tool works trig book ?
or carr lane ...
no i have a slide rule.
thats slip stick!
actually i do use a calculator,
but i do find my slide rule useful often.
well like JymmmEMC said, you got a 'post' with Ace Converter, and yo can get 2D cad with Qcad , a freebie under linux
so, you can get gcode for a contour
cool, i have heard of qcad.
tomp: max5kins includes simulation of joint limits, but emc doesn't yet obey joint limits (only axis limits)
tomp: that is why you can get a following error
jepler: anyway to elim for a demo?
net table-pos-cmd axis.0.motor-pos-cmd => limit3.0.in
net table-pos-fb limit3.0.out => axis.0.motor-pos-fb max5gui.table wcomp.0.in
there are a pair of lines for each joint
or should i use ... whoa... thanks
link cmd back to motor-pos-fb (bypassing the limit3 block) and the errors will go away
uh, easily lost, but will try, thx
net table-pos-cmd axis.0.motor-pos-cmd => axis.0.motor-pos-fb max5gui.table wcomp.0.in
something like the above ^^^^
and so on for each joint
that's in max5kins.hal btw
might i ask if EMC supports auto tool changing?
jepler: understood, a set for each axis
ashaw: EMC works with tool changers, various types, again very configurable ( with the power comes the responsibility.... )
ashaw: ashaw: but i do find my slide rule useful often - For what, a back scratcher? ;)
a harmonica for the deaf
ashaw: yes, and it's pretty good for turret-style toolchangers, but less great for toolchangers that require several steps of coordinated axis motion during the toolchange
no sometimes it is quicker to use logs than to type into a calculator for multiplication.
cool. about the tool changes.
Mazak toolchange http://www.linuxcnc.org/movies/mazak_300kbps.avi
(from the wiki)
ashaw: So, where do you have your punch-cards made up at? Or do you just have a surplus? lol
not a great video :-P
anyway, all that motion of the changer is controlled by classicladder
I actually am to young to remember punch cards, though i have seen some and they are quite quaint.
all "emc" does is move to a predefined location, and put the desired tool number on a HAL pin. the rest is up to your abilities with classicladder or a custom component to sequence the other stuff required to change the tool
oh. so you cannot have it pick up a tool by itself? from a rack?
you can, but all the ways that are currently available feel like hacks
one way is to just have the steps of the toolchange and perform them inside an O- sub instead of just using TxM6 to switch tools.
what are they?
yes that is a sub.
another way is to have a custom component that performs the moves by adding an offset from emc's toolchange position to get the motor position, and then subtracts that offset to fake the position feedback to emc
we'd love to find somebody who is a programmer and wants to support that kind of toolchanger -- we recognize that this is very far from ideal, but none of the developers have a need for that style of changer right now.
would it not be hard for you to use gcode to describe the toolchange, so sub the TxM6 instruction for arbitrary gcode, like an O- sub, yet on the fly?
(there's also disagreement about how to do it -- for instance, maybe the solution is to allow TxM6 to actually call a gcode subroutine that is defined by the integrator, or maybe it is to allow a HAL component to make emc move to a specified location during a toolchange)
and then there's the added complexity of moving to the old tool pocket first, so you can drop the old tool where it's supposed to be before picking up the new one
is this type of tool holder like a linear rack or something?
i am planing but that seems the easiest sort to build.
Interesting, I never ran across a machine with a drop-into rack of tools/pockets. Is yours linear or circular?
(or some odd mix?)
it will be an array.
Clearance holes in the bottom of the pockets, exept for a little lip?
in your place I'd probably go for writing an O- sub and writing calls to it in place of TxM6. If your favorite post isn't flexible enough to write O1001 call [x] instead of TxM6, your favorite post sucks :-P
Clearance in the side, too, I'd guess.
More like two fingers?
no, clearence hole at the side and top, with a cone shaped lip.
or just a flat one, yet to decide.
this is one I saw at last year's cnc workshop: http://www.youtube.com/watch?v=RzR6L3lEXdM
not running emc2 though :-/
similar idea, but my spindle has a pnumatic collet.
I recall a commercial retrofit where they gave you, what, six moves associated with the tool change. This was on a mill package. So you got a table of entries: X1,Y1,Z1 then X2,Y2,Z2 etc. up to X6,Y6,Z6.
Any entry in the table (other than zero) causes a move of that axis (or those axes). The table entries "toward the end" had to somehow cancel out the moves mad at the beginning, so your net move would be zero. (Otherwise you come out of the tool change with a new offset.)
so there was one set of entries per tool?
that seems doable.
No, this one was intended for conventional tool changers where the machine had to make moves to cooperate with the tool changer. But I suppose you could have multiple tables, one per tool. But the video that jepler showed wasn't like that, it was all table moves, so...
...it looks like it lends itself to subroutines or frequent macros or some such.
yes, that is what i was wanting.
Maybe no need to get involved with hal, tool changers, etc.? Just make a move, and viola, there's your new tool. Kind of like a Wasino lathe, with all the tools arrayed at once on a flat plate.
but that sounsd what i was thinking about.
It's the name of a company known for building lathes that change tools that way. There are many others too, I'm sure. To change tools, you make a large X move (several inches) and a new tool comes into play.
no not like that.
You usually get a couple feet worth of tools, LOL.
5 axis CNC, i do not think it will work.
Isn't your situation not too different from jepler's video?
yes, i thought i had said that.
Then it's two sides of the same coin.
i do not quite understand.
Get a new tool into position (by moving around) and change offsets (tell the control what you did).
and do so by running a O- sub ?
I'm assuming that your tool changes are done in a safe area, etc. Although in jepler's video, I could see trouble coming with long tools and tall workpieces.
i am just wanting to build autochange into my DIY CNC mill build.
Yes, it seems like an O-sub should work. Would anybody else like to comment? Autochange? Yes, that's always nice. Maybe you can have your cake and eat it too?
Maybe there's a way to have your O-sub look at the T number, and make appropriate moves?
yes, that would be cool.
If T=1, then move like this; if T=2 then move like this, etc.
but i think g-code is not turig complete.
Turing complete? You mean if T=(unrecognized) then move to safe corner and trip overtravel?
no, that is a programing term.
maybe get a hold of this guy and ask him also.. http://www.youtube.com/watch?v=MV6ogckf3n8
You're going to make me go to Wikipedia, aren't you. OK, BRB.
it basically means that g-code is not a real programming language.
Clearly you have not looked at the O-sub section of the manual, LOL
you can do calculations, conditionals, and subroutines in emc's gcode. http://www.linuxcnc.org/docview/html/gcode_main.html#cha:O-Codes http://www.linuxcnc.org/docview/html/gcode_main.html#sub:Expressions
jepler: I wonder if you can solve for pi by making mechanical moves, babbage-style?
Steampunk fans should be delighted, LOL
i do not know, but it looks like you can do general computations, if one can find out how to store memory.
tomp pictures a guy on express way ramp , ol dclothe, cardboad sign "will solve for pi"
ha ha ha
I like it
oh yeh, slip stick in hip pocket
no he used trig tables, tat he made.
in shirt pocket, right
on his computer
that was what his computer wasa for.
you can store info into vars - like #1 = 0
they can be named now also.
can you store into an arbatrary var like #a[#1]
although if you want to save them from run to run - I think they have to be numbered and you have to add them to the var file.
Yes, I think you have variables you can use in O-subs, with some minor restrictions. Here's the link: http://www.linuxcnc.org/docs/2.3/html/gcode_main.html#r3_1
named? like 'detail1depth" ? 'keywidth" ?
but I have not played with it yet.
To webmaster(s): I tried to give the link to #r3 only, (Chapter 3.0) but it was defective. This link doesn't work right: http://www.linuxcnc.org/docs/2.3/html/gcode_main.html#r3
[15:12:25] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Oword
[15:12:27] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?AlternativeNamedParameterSyntax
To webmaster(s): Boy, that was fast. Thanks.
ashaw: What's your machine, a tabletop mill? Steppers?
will be a stand lalone mill, with steppers, table size 0.8m x 1.5m x 0.5m
That's pretty good sized. Based on anything, or built from scratch?
will mainly be used to make wing sections for model airplanes.
might i ask what a U axis is.
i still do not get what U is for.
Some controls also use U as a relative mode for X. Same for V-Y and W-Z.
is there a diagram that i can see?
In a knee mill, if Z is the quill, W is the knee.
If you move Z (quill) down one inch, and move W (knee/table) down one inch, nothing changes.
ok. so why have it?
Have to accomodate tall and short workpieces.
(If quill travel is short, which is often the case. You seldom see a 20" quill.)
How does your machine get its Z (W?) travel?
Half a meter is pretty good
in emc2, UVW are three additional linear axes. They are never used in emc2 as relative coordinates.
stationary workpiece, except for A axis, B/C axis off Z axis, single motion for Z, one ballscrew, 2 linear slides.
jepler: I've never seen EMC in a 3-axis mill with a knee DRO. Does the W have to stay there all the time, taking up space and not changing, or can you just pop it up when you need to look at it?
(OK, I guess that would be 4-axis, but anyway, same Q.)
right now all axes are displayed all the time
OK, thanks. Just wondering what to expect.
ashaw: That's a pretty big machine. Sounds like you're going above Bridgeport size there. What does it weigh, or do you know?
we will only be milling foam, light wood.
Oh, OK. That helps.
What is your max spindle RPM?
i want a smaller mill for steel, but this one is not it.
i have not decided on the spindle.
i have started buying parts, but and have got rawings for the mill, but the design is still in the conceptual stage.
Well, if you're doing wings, you're likely to have long CAM-generated g-code, I'd guess. Unlike us more simple machinists who commonly make things with a bunch of convenient flat sides.
yes, that i will.
jepler: i keep getting "joint 0 on limit switch error" with X at -101, tho i changed the .hal wcomp.0.min to -105, and the .ini MIN_LIMIT to -500 ( i dont know where it gets the -101 from)
ashaw: Do you know how you'll make your g-code yet? Or is that still undecided for now?
i was asking above, i am not sure what CAM program i will use.
Oh, sorry, I must have missed that
i think you can do airfoils in APT, and APTos360 has an emc post ( join #cam for more info )
ashaw: on #cam he might want to talk to crotchet about apt360 and 5 axis
what is the link to APT.
jepler the fixes you suggested got rid of the fe's
setp wcomp.0.min -101
tomp: in the hal files
You might want to join #cam, the HeeksCAD project is showing promise. Lots of work needed still.
heekscad does not do 5 axis at this time
i have 6 axis.
but only 5 will interpolate simoultaniously
ashaw: Nice! There you go.
ashaw: Why only 5 simultaneously? (In for a penny, in for a pound, etc.)
i am not sure, this airfoil has some unky curves
i did not say only 5,
Ah, I see, sorry
even a siemens 840D will only do 5 at 1 time
what control is it??
i dont think it will do more than 5 at a time either
plus microstepped 400/rev steppers
then a 5i22
and if that will not i will get my viltltex 505 prototype board and hook that up.
ok good luck
jepler: "setp wcomp.0.min -101" was original value, error is at -101, had expanded to -105, but now back at -101, result is..joint 0 on limit switch error
emc/task/taskintf.cc 607: Error on axis 0, command number 329
why will not EMC2 do 6 axis, might i ask?
i thought expanding the limit would help, but err occurs at same place ( these wcomps and the ini file limits made to agree but larger than the err position )
ashaw, emc can interp 9 axis simultaneous
then why dmess say it cannot?
i think the contect was some cad did not
bcz i was anaware of that tidbit of information
not extensively tested ;)
where does it say that?
i got 6 units here all 7 axis, very very not extensively tested ( not emc, very off brand )
ashaw? extensively tested? wher << that is said?
it is open source, nothing clames to be fit for any purpose, implied or no, and no warrenty is given.
ashaw: you got it :)
but that's probably the same you get from a "commercial" control aswell
in the zGPL
in the GPL
or where 9 axis is stated? ( answer: in the cvs code )
ashaw: I dare to say you get better support with emc2 than with any vendor
because you get to talk to the people who actually write/design the code
tomp: I dunno what else would be leading to that message but the window comparators. did you increase all the values? you could also just unhook the limit switch inputs to the motion controller
and really undestand what is going on
e.g., remove the lim-sw-in items from net table-limits-tripped not.0.out => axis.0.neg-lim-sw-in axis.0.pos-lim-sw-in axis.0.home-sw-in
(of course that support isn't guaranteed, and it depends how you ask for help :)
yes often Open source software is more fit tha non open source
jepler: i may try unhooking them for the dem, i even searched for 101 in the src... its like a bad penny. thx
ashaw: doesn't mean we can compete with a control that has been developed by 50-100 programmers for the last 10 years (feature wise they might have a lot of things which customers asked, which we don't need/have atm)
but since it's open source, you're welcomed to add things yourself
that, my friend is the point
Hi. Has anyone here configured emc2 to work with a tormach pcnc?
And if so, do you have a HAL file for it?
seems the configs are part of the next release (2.3.0 - pretty soon)
you can either grab the beta2 to test it out, or use those configs (with minor edits) on 2.2.8
Martyn: basiclyt [EMCMOT]SHMEM_KEY has been removed in 2.3.x, you still have to add it to the configs in 2.2.8
I have 2.2.8, so I'll recycle those configs
[16:06:26] <alex_joni> http://cvs.linuxcnc.org/cvs/emc2/configs/tormach/pcnc-1100.hal.diff?r1=1.2;r2=1.3;f=h
<- that's a diff, so you can easily see what's changed
I'm trying to do adaptive clearing using HeeksCad to generate the gcode...
hmm.. if I'm reading it right, it should be in 2.2.8 too
look under sample configs/tormach
but the tormach's native gcode interpreter is doing some strange things with tight arcs
so I'm going to switch to emc2 to make -sure- it's the interpreter, and not some fundamental limitation in the machine
define "strange things" ..
alex : The arcs that are being cut, look like the PID control is bad... the arcs are too long, and too shallow
which is indicative of a problem with the way either the g-code is being interpreted, or the fundamental PID control
yeah, but it's a stepper machine.. right?
so PID shouldn't be _that_ present
PID comes into play, pretty seriously, when doing arcs
and the tormach's interpreter and control system does have some arc acceleration
things like PID should be independent of the type of move, but ...
By handing it off to emc2, I will be taking those optimizations out of the loop
Martyn: let me know how it goes
I will. I've also submitted a bug to tormach
but I'll have to wait until Monday to get an answer from them.
even if I'm not around, if you put an alex_joni: in front, I'll notice it when I get back :)
catch you later :) I need to reboot this machine.
we are talking mach - right?
skunkworks: who <strike>cares</strike>/knows?
but yeah, I think they use mach by default
jepler last suggestion ran entire 'cone' no errs :)
now where can i get a virgin set of max5kins.hal max5kins.ini ( so i can post clean patched vrsns )
did somebody say virgins?
old mr make a backup outsmarted hisself
oh, and i;'ll need to adjust the values in cone, the table comes clear off the saddle
but very cool, thx jepler
tomp: CVS is safest for "virgin" sets
it looks like there are NO 6 axis cam programs in existance
hmm.. is there ever a need to interpolate 6 axes
ashaw: I think there are
but probably not ones that output g-code
the CAM I use at work interpolates up to 18 axes
but it's for welding robots, not CNC & g-code :)
whats the name of the 18 axis CAM program?
I reckon when they were machining those engine blocks it'd come close to 6 axis
hugomatic1: it's probably useless for anything else
there are a couple: like RobotStudio from ABB, RoboPlan from CLOOS (which I use)
basicly all robot manufacturers have their version
I'm just amazed that such wonders exist...
there is another "smarter" program out there called 3DCreate
[16:24:09] <tomp> http://imagebin.org/46011
some guys in sweden build it, but it's very nicely done that you can extend almost anything by python calls
it looks like there are NO 6 axis cam programs in existance
well ashaw.. the way i see it with 5 axes of freedom.. your free... any othere axes would be co-linear or non-orthagonal... so the question i ask is WHY bother??
ashaw: for regular machining 5 axes are usually enough
as the rotation of the tool is along the 6th DOF, so not really usefull
you could use it perhaps to machine *inside* bits that arent straight accessable from outside perhaps?
think robot arm
im planing on using it on the new toshiba boring mill i have coming in... d'andrea boring and facing head
Valen: even then, usually the 6th DOF is the rotation of the spindle
that's true even for tools (spindles) attached to robots most of the time
alex you ever seen a contouring head?
dmess: the term doesn't sound familiar.. but english ain't my mother tongue, so I still might have :)
ashaw: anyways, I suspect it's in the <xx>k USD range
it turns the spindle of the boring mill into a cnc lathe / trepanning/contouring machine
i have this landing in 13 sleeps http://translate.google.ca/translate?hl=en&sl=ja&u=http://www.toshiba-machine.co.jp/machinery/lineup/btd200qh/index.html&ei=0ArqSbvKJ5HGM9LfuOQF&sa=X&oi=translate&resnum=2&ct=result&prev=/search%3Fq%3Dtoshiba%2Bbtd200qh%26hl%3Den%26client%3Dfirefox%26rls%3Dorg.mozilla:en-US:unofficial%26hs%3DHTE
NAMES is closing up for the day in half an hour. Handed out many flyers (couple hundred?), talked to many folks, and handed out about 70 CDs.
Almost always had folks in the booth. A few had had bad experiences with Linux and/or EMC, but lots of folks who had had positive experiences too.
Ron Ginger sent a lot of folks our way during his seminar on CNC.
Greg's photos really went over well. Had the video of the 5-axis machine from Belgium running in a loop--lots of folks were mesmerized.
Nice work! Did chris show up?
I think cradek is planning to be there on Sunday, not sure though.
jepler: decided against going?
martyn: how goes it?
I just about got the tormach working
I'm taking it slowly.
is there something wrong with the Tormach sample config?
downloading and installing emc2 took the longest time. I can only download at 150k/sec
No, it looks good.
ok, phew! :)
I am milling some example objects now, and checking the dimensions with my handy micrometer
So far, straight cuts are accurate to within 0.001 tolerance
(well, actually they are accurate to within 0.00085xx which is acceptable
I'm going to mill the first circular 1 inch piece now, milling slowly.
and I'm using aluminum bar stock, nice and soft material with a new sharp endmill
their machines seem able to take some pretty heavy cuts
especially for their size
I've also locked down the 4th axis
SWPadnos: did the shuttle get figured out? I remember someone getting the details figured out.
no, the patch to hal_input didn't work
I also patched my kernel with a USB quirk I thought might help, but that didn't work either
though it could have been me
Okay, I'm happy
1" circle has cut to within the same tolerances
ballscrews are a wonderful thing
Well, now I'm going to move from 2D to a full 3D part
SWPadnos : Good PID is also a wonderful thing
shouldn't be any different
there's no PID on a Tormach, unless you added your own servos
SWPadnos : I'm not talking about the tormach .. just PID in general
oh, well in that case I agree :)
SWPadnos : However, the vertical tolerances in a tormach /are/ different than the two horizontal axis
So, I'm hoping I get the same standard devs and tolerances
mechanical tolerances, yep.
It's also why I'm not using the 8" fourth axis for these tests, yet.
hmmm. the resolution is the same. do they use a different ballnut on Z?
actually, a different screw (larger diameter) entirely
because the vertical axis has the entire weight of the tool
yeah, the head is somewhat heavier than the average quill
still, I don't /expect/ it to be much different.
the motor that runs that axis is correspondingly much MUCH beefier
I'd expect a little better if anything, since gravity assists with anti-backlash
and the tormach is the only 'hobby grade' mill that can do steel
Well, the part is milling now. It's my favorite test part .. a hemisphere on the end of a piece of square stock
it's just pretty to watch it get cleared and cut
After this, I'm going to work on learning how to get this panasonic IP web cam to work
and point it at the machine :)
error .. damn
The command to pump some lubricant was not executed, and the machining stopped
this cut requires regular lubrication, since there is a lot of contact
Good morning.i have one question for spindle and drive spindle with analog command 0-10 volt.i use now drive +/-10volt and not ahev the problem whith m3 and m4 because the analog change signal positive or negative automaticaly.But whith drive 0-10 volt the analog negative value is not ok for drives.is possible resolve this problem?
SWPadnos : Fixed. Error existed between user and keyboard.
G-code is so frigging picky :)
SWPadnos : Should I rezero after an error, or can I trust the zero calibration to not have drifted after an error in emc2?
motioncontrol, you need to add an "abs" HAL block between motion.spindle-speed-out and the PWM
you mean machine zero?
did you lose power to the motor drivers?
then you should re-home to be sure
heh, at least the zero position has not yet been milled away :)
if you had encoders, and they stay powered even when the motors get shut off, then you wouldn't have to
rezeroed, and off it goes...
ok thanks i undertsnad
I love the sheer POWER of the Tormach.
they're nice machines
For a 'hobby' machine, it's got all the power of any production line tool I've ever maintained. It powers through aluminum like it was butter
I'd consider one if (a) I needed it that badly and (b) I didn't have a Bridgeport waiting for retrofit already :)
I'm up to 10 inches a minute at the moment, and I know I can kick it up higher than that
you know, emc2 vs sprutCAM ... compares nicely.
do you have a post for SprutCAM that seems to work well?
if so, please add it (or directions to set it up) to the wiki
unless they're already there :)
I wrote one in python, I'll think about open sourcing it.
Z tolerance is well within one thousandth
I think I've found my new CAM control ... emc2
that's pretty good for a machine with resolution of a tenth
did you have any trouble with the sample config?
or anything that was unclear as to how to use it
not at all
it was pretty straightforward
the hardest part was installing emc2
I didn't like that I had to partition the system to install the whole ubuntu-based OS
I want to install a program, not a whole operating system.
its machine control not just a program
I still don't understand why there are so many dependencies, that you have to tie it to a given revision of the OS (8.04) rather than just have it install on any linux distribution
the realtime kernel makes it a little harder
you could use any flavor of Linux, if you want to patch and build (and test) your own kernel, as well as EMC2
it absolutely requires a hard realtime kernel, eh?
yes, of course
I wonder how windows based machine control systems do it, since they don't have that fine grained a scheduler...
they suck more
or hand off to external realtime
I think archivist has the right of it
I've looked at step pulse output from EMC2 and a Windows based control, and the pulse timing was much better with EMC2
they use drivers that tie into a hard realtime controller that is plugged into the machine...
that makes sense
sure, that's another option, and one which we can't do at the moment
and they cost buckets
sure, if you want to spend a few $$k, you can do that
I think I'd rather have a micro hard realtime controller on an SBC (say, one of those wonderful ARM linux based Cortex A8 boards)...
one thing to remember though - EMC2 is acpable of doing things that the "hobby" controls can't do at all
I haz no bukkets so I stick with EMC
SWPadnos : utterly agreed
unless the hard realtime controller can do them
like rigid tapping, actual spindle-synchronized threading, realtime feed override, realtime jogging ...
you *can* jog EMC2 with a USB controller, but if something would happen to make user-space applications pause, so would your jogging
SWPadnos : One impressive thing though .. after I booted the emc2 partition on the dell machine Tormach configured for the P1100, and did the configuration, it was happy. Machine control worked with no hassles.
that's the only option with the Windows controls
EMC2 also has the ability to have the hard realtime code read a jog-wheel (encoder), which won't get screwed up by user applications
SWPadnos : Whomever did the original Tormach configuration has my respect.
thank you :)
(that's me :) )
Can I use my spaceorb with EMC2?
I assume it shows up as a mouse device under Linux...
well, not necessarily
there's a driver called "hal_input" which works with just about every "Human Interface Device" or HID
you can test it:
I have a space navigator by 3d connexion
with EMC2 running, plug in the SpaceOrb
It's the best thing I've ever found to do 3D motion
ok, I have one of those as well
So it works?
dunno - let's find out ;)
What did you have to do to configure it as an input device?
got yours handy?
but I need to wait until this part finishes milling
I have one for PCB CAD, I don't use it on a Linux machine
ok, no ptroblem
I use mine with SolidWorks ...
actually you can do this stuff with the machine running
I don't want to take the chance :)
no, that's truw
one thing that can cause a glitch in the realtime timing is a USB re-enumeration
Plus I'd have to do downstairs and pull it off the machine running Solidworks ...
(ie, don't plug or unplug things when the machine is running)
well, I can give you the instructions and you can try it later
Well here's the test about to happen
once you've plugged in the orb, open a terminal hmmm. one sec. I can try mine on my sim system
I'm using the HeeksCad Adaptive Clearing Path generated arcs to do the clearing .. it's using every possible edge on the tool ... almost constant contact with the piece
oops, bad paste there
This is awesome .. I'm watching it take little "bites" out of a pocket. cut-arc-cut-arc-cut
so much more efficient!
Have you guys been able to use the 3dconnexion as a mouse yet?
Space Navigator that is
martyn: are you using axis?
Martyn: video? ;)
skunkworks : After I mill these pieces, I'm going to start learning how this panasonic IP webcam works
I still have to read the manual, and figure out how to make videos available
<-- NOT a techie
I just know enough to get myself in constant trouble.
not a techie - but using heekscad and emc - I think your doing just fine.
Well, sure .. I'm not stupid either. I can read a manual :)
I'm going to have to use the rotary table soon, and moving to a 4th axis in emc2 will be interesting
well, the Spacenavigator works
I did not see any support for the rotary attachment in the sample configuration. Did you have a chance to use one SWPandos?
you can even control the cool blue LED in it :)
what did you do to get it work work?
I loaded the driver
it doesn't automatically make EMC2 do anything
when you load the driver, you get access to all the buttons, lights, and "joystick axes" on the device
you have to connect those inputs (or outputs, in the case of the LED) to other things to make anything happen
does Ubuntu load the driver automatically after connecting the device?
and is there a gui utility to do the mapping of functions?
there are (at least) two kinds of driver
Ubuntu does recognize the device, but it didn't seem to do anything by default
you also need an EMC2-related "HAL" driver if you want the device to do things specific to EMC2
ie, to not act like a mouse
as I said though, there's a generic driver that works with almost all kinds of input device
you can attach a second keyboard and get loads of "input switches" plus a few LEDs
Well, lets see if I can say what I mean as far as what I would like
hal_input driver can take control, such that the rest of the system doesn't get the keystrokes from that keyboard any more
I would like the X and Y axis to jog when the paddle is pushed right/left forward/backwards
I'd like the Z axis to jog when the paddle is lifted/pushed down
and the rotary attachement to turn clockwise/counterclock when the paddle is twisted
you know how you might buy a button and some wire and a doorbell, and hook the button to the doorbell with the wire?
that's how you customize EMC2
the SpaceNavigator looks like a bunch of buttons and sliders (plus an LED)
there are places where you can connect "virtual buttons" within the Hardware Abstraction Layer (HAL)
These inputs do certain things
The HAL files just describe a connection between an external device and a function call inside of emc2?
like jogging each joint, starting the program, turning lube on/off ...
that's more or less it
not quite a function call
it's more like having chips on a PC board and wiring them together
hardware abstraction layer ...
HAL "components" have functions that run in threads, and also have "pins" that are connected together with "signals"
I'm familiar (kind of) with hal thanks to the way Mitsubishi lathes work
do you have to modify the code directly, or is there a HAL description file somewhere I can tie those functions together with?
take a look here: http://cvs.linuxcnc.org/cvs/emc2/configs/tormach/pcnc-1100.hal?rev=1.3
or look at the local copy you have :)
the configuration is just text files
That means you didn't have access to a 4th axis rotary bed, eh?
the Tormach file is set up for 4 axes
I notice that only 3 axis are defined in the HAL file.
does your AXIS display show XYZA?
off by one there
the numbers start at zero ;)
This whole idea of counting from 0 will one day make sense to me, I swear.
if you take a look at the manuals, there is a lot of information on using, exploring, configuring, and manupulating HAL
I will. this is all of day <one> for me and emc2
taking it slow, I hear :)
I think I'm doing well to have the machine fully operational with a default configuration
speaks volumes about emc2 though
yeah. it's good all around
I'm milling my favorite thing out of barstock aluminum right now.
A micro engine. 1cc of one cylinder power.
good morning.i have the question for tool table.if i write in mdi g43 h1 the emc load lengh h1.after modification in gedit the lengh at new value.write g49 at affter g43h1 and emc load old value.For load new value i restart emc because ?
Right now the cyllinder head is being machined, and it's coming out quite nicely.
It's a very good test of the machine tolerances :)
too bad I can't get the tormach to tap the threads of the cyllinder head or the main engine block
all you need is a spindle encoder... :)
(for emc anyways)
skunkworks : To have a spindle encoder .. there's the small problem of not having any hardware for the Tormach as well, after all.
well, this is one thing I disagree with Greg about - I don't think the machine should be limited to "what can be connected via the parallel port"
but he wants all their machines to be compatible, and it's his company :)
there are interfaces, some of which run off the parallel port (others are PCI), which can give you mich better performance than software step generation out the parport
much, that is
there's one that costs $90 or so, and has 48 I/Os
okay. Cylinder head complete, now moving on to machining an engine block. I already have a piston machined from a different machine, as well as the rod and crankshaft..
it also has onboard step generators and encoder counters, which relieves the PC of that very time-constrained task
and thus reduced the strain on the kernel, reducing the need for realtime
and can generate the analog reference needed for spindle speed control
well, sort of
you still need hard realtime, but you have much more relaxed deadlines
Well, while the block is machined, I can tap the the threads by hand. It's not exactly an onerous task
since you only need to run code every millisecond or so, you can tolerate 50 microseconds of jitter (or more)
the threads are no more accurate if a machine mills them, than if I tap them by hand.
when you're trying to generate steps on a timescale of 20-40 microseconds each, you can only tolerate 10 to 15 microseconds of jitter
it's a heck of a lot easier to have the machine do it though, if you're doing many holes
that's not many instructions, is it. Even with a modern processor
it's several thousand instructions
but machine tapping is just cool :)
but in this case it's a thread at the top of the cylinder head (for the glow plug), and one at the bottom to attach it to the block
skunkworks : Machine tapping -is- cool. Doing inside threads though, is a pain
in any case, there are lots of things you can do with more than the 17 I/Os of a parallel port and without the timing limitations of a PC
um, with a tap?
or with a threadmill?
or on a lathe? :)
(all of which are supported by EMC2, if you have a spindle encoder)
[23:03:31] <skunkworks> http://www.youtube.com/watch?v=C740zS9R9kk&feature=channel
I'll even bet that the tormach has the power to do straight tapping :)
^stepper printer port machine.
uh-oh. I sense more machine porn :)
sure, it's a reasonably beefy machine
that machine is a lot smaller than a tormach
That is straight up machine porn
But is it QUALITY machine pr0n?
On the other hand, it's tapping a small hole :)
these cylinder heads are 1.2cm wide
okay, block is now machined. I love how fast this 'hobby' machine is
at least with aluminum
Tomorrow, I'll have to try steel
SWPadnos : You said that the accuracy was a _tenth?_
tenth of a mm, I assume...
resolution, not accuracy
"tentth" = 0.0001 inch
(ie, a tenth ofa thou :) )
Martyn: pictures or it didnt happen
contribute to the EMc2 Youtube Pr0n channel
anonimasu : :p
I'll ask my son later tonight how to upload to youtube
he's been trying to show me how to use firewire to copy the videos from the camera to the computer
good morning for reload in mdi the tool table the command is g10 L1 ? if i write g10 L1 have error because?
I wish the tormach had a tool changer :(
Martyn: It does... see end of your wrists
JymmmEMC : /automated/ tool changer :)
JymmmEMC : I'm quite happy changing tools.
Martyn: You automatically change the tools when it needs it, right?
JymmmEMC : *chuckle* You're as bad as the machinist that showed me how to use my first CNC machine in the late 70s
Sometimes I'm worried that as I get older, I'll never be able to keep up with the changes in technology .. that said, I will _always_ know how to run a bridgeport mill, and will never forget how to manually do anything.
That said ... three cheers for a home CNC machine.
Martyn: I bet you haven't even figured out you're it's biotch =) "Yo Biotch, tool me!" "Biotch, feed me" "Biotch, lube me baby" "Biotch gimme more power" LOL
SWPadnos: wanna bet I can top your machine pr0n ?
[23:16:57] <alex_joni> http://www.youtube.com/watch?v=-CKNajrRIeI
nice hot wire cutter
must be someone who makes model airplanes
is that 2 tripods?
Guest958 is now known as skunkworks
2 x bipods
right - what you said..
and emc2 of course
but it's a very interesting concept to tension the cutting wire
SWPadnos : For the record, emc2 did a _better_ job at interpreting the paths than the native software that comes with the tormach
The final pieces are absolutely wonderful.
let them know that :)
I think we knew that already.. :p
"A free piece of software did a better job than the $995 piece of crap you sell"
they sell it?
alex_joni: Oh, the U bow under the table?
err, I mean - excellent
hmm.. seems maybe it's not emc2
SWPadnos: you smooth talker
another user linked to it, that wants to do it with emc2
alex_joni: some windows thing running on a laptop
delphi program (seems open source)
SWPadnos : heh.
isn't it $995 for the PC with the software?
(ie, not just the software)
SprutCAM 2007 w/USB Dongle (30691) $992.00
oh, that's different :)
That's the line item
yes, I know :)
Sorry, wrong item
Configured PC w/Monitor (30282) $922.00
That's the right one
they just throw it all in that cost
it's a nice Dell, by the way
sure, I've seen one before ;)
Looks very professional next to it.
I may still be able to log into their EMC2/Ubuntu test system
I bought this from a company that went ka-put, so I paid $6500 for the complete kit
that's a good deal
bummer for them, good for you
very. Although they kicked the _crap_ out of the tools. Every single one was badly damaged
"damned thing doesn't work, get rid of it" :)
It was used for some TV program in San Francisco.
on public television of all things.
I wish I got the one Jim Newton managed to get a hold of (TechShop.ws)
\\he\\ got the one from Prototype This .. it was barely used
Also from the SF area
the stuff at the TechShop gets beaten a fair amount
it's like rental cars
yeah, he got it even cheaper than I got mine... some kind of behind-the-scenes deal
Yes, but the TechShop equipment is all pretty durable .. two bridgeports, two dont-know-whats
yeah. been there :)
at least to one of them
There's just one for now. Menlo Park, CA
near that bridge with the funny name
I helped with the initial setup
The next one will be not far from where I am now, in Austin TX
I was very impressed with the fact that they have -three- Epilog laser cutters
Those are expensive toys
I talked to several of the guys there about setting up a CNC "work cell" with EMC2
but ones that anyone can use in a heartbeat
sadly, I lost the card of the person who I needed to contact again
SWPadnos : That would be nice.
in a work cell how would emc's talk to each other ( or is that the architecture) ?
SWPadnos: Contact - Jim Newton (Founder) : (650) 521-9027
autin texas, try ernie applegate at applegate eng
oh, thanks. I'll write that somewhere I can't lose it
tomp : ernie applegate is involved with the new startup of TechShop in Austin?
tomp, EMC was designed from the beginning for having multiple layers of contril
no, justa cluey guy runs a shop there
SWPadnos : What's a work cell?
I'm trying to wiki it
one of the concepts was to have controllers for each machine and robot/shuttle, controlled by a cell controller, controlled/monitored by a "line manager", controlled/monitored by a plant controller ...
oh, just several machines that work together to make stuff
That's kind of counter-indicated in techshop though.. since people tend to do things one thing at a time. It's not set up for production
work cell, like a robot taking a part from a lathe and handing it to a mill and 3 more sets of lathe robot mill down the row
one plant I was in had a big-ass EDM-type machine which was fed with cast parts, then those were taken out and machined in a row of VMCs, all moved around by robot
once the part was put in the EDM, no human would touch it until it was done and deposited in the output tray (by robot)
hmm cell controller, line manager, nice
I don't think they called them that, but it's meant to be hierarchical like that
that would certainly bring emc2 into the realm of professioal line control
just like EMC and EMC2 are internally
I think a nice ui would start to become a neccessity at that point
the task planner tells the IO controller and motion controller and ladder controller and ... what to do, and monitors their progress
Very nice, the parts are now complete
I have a little finishing work to do, but everything looks good
Martyn: what are you making? on emc?
I have an all aluminum 1cc single cylinder engine block, cylinder head, and glow plug cap
it will never run, of course...
but, it's very well machined.
i like motors :)
tomp : I'm just testing this tormach and getting it working on emc2
SWPadnos : How long did you work for them?
ah, i 'member yesterday, a tormach, nice solid looking machine
Martyn, I didn't. I was designing part of the power system for the EDM machine
I've seen some plants where they use big-ass robots as tool-changers
there were a robot in a booth, with lots of shelves around it
yes, those are bigger-assed than the ones I'm talking about :)
it would get a new tray with tools, scan them and put them back to their places
then it got a new command for tools to put in the tray, which it did
the tray then moved to a CNC machine, where it was used with the ATC
I need to put in an order with McMaster-Carr
I need more small endmills ...
all of the above happened without user intervention
a mazak integra? like 4x6 x1 meter cube pallet rack
robot looks like a hyster fork lift?
tomp: nope, the robot was a 6-dof Kuka
about 2.5m tall
thats more fun
[23:45:23] <Martyn> http://www.hossmachine.info/projects_7.html
<--- a really REALLY upgraded mill
maybe 4.5m reach
For the record, I achieved 13 ICM with a .25" endmill , .1" DOC on the tormach into aluminum. Just fantastic. This thing is a monster.
Martyn: yes - but he uses mach.. ;)
yeah yeah yeah
but .. it has a quick change :)
That's one modification I will be making to this tormach.
hossmachine mods are nice
Even if I can't do automated tool changes, being able to quickchange the tooling will cut down wasted time
Allright, dinner (and wife) call.
Good evening gentlemen. If anyone here knows about web cameras, and how to set up streaming video .. I'll be back on later tonight.
emc has a manual tool change feature, pops up a dialog saying put in the 1/2" c'bore and squirt some lube below the chuck" ( whatever ). could be tied to one of the tree lights
Because frankly the manual of this thing reads like half japanese, half crazy talk
'make for removing the lens cap'
hmmm. I guess I should pack or eat or something also