[00:01:11] <SkinnYPuP> http://www.metalsontheweb.co.uk/
now we're talking
I never think of ebay for metals
I get some boxes of 12l14 sometimes
brass "weights" sometimes sell dirt cheap
Thanks to all the input I have gotten from you fine folks on here, I finished drawing up the schematics for the control panel on my minilathe cnc conversion and can start building it tomorrow ....wooohooo
I have realtime monitoring for my limit and home switches, and plc control through the modbus connection to provide (spindle start/stop, vacuum, worklight,oil for cutting,cooling fan control for controller card)
Does EMC2 still let you design your own interface layouts for your machine?
I don't think it ever did in the way I think you mean
There was an editor, for adding indicator lights, and readouts, and push buttons to click on and such
Designing your on screen for your controller kinda stuff
Own* not on
that sounds mach-ish
there is pyvcp
yes it was VCP that I had read about , and pyvcp is what is recommended to use but not until v 2.2 released and 2.1 is only one available just found it in docs
With it I can add on screen controls to operate all the items being controlled through modbus connection to my PLC
and then tie all the logic together in hal and classicladder
I just couldn't remember what it was called , thanks for jogging the old memory hehe
good thing I didn't mention vi then :)
I though vi was just a test editor?
yeah, you can use it to change anything in emc :)
including buttons and readouts and such on scren?
wow good to know
just have to know what you are doing ...
I have been learning loads on this project
and lots left to learn
I just found out a few weeks ago that I could get modbus support in classicladder, and just learned how do do modbus with a PLC...and was given one from going to the class.... so I decided to add it to my cnc project
hmmm... internet went black for a couple of hours....
probably one of those terrorist Canadian Geese nesting on the microwave tower
hey, too bad the coastal ones taste like mud!
the grain feed ones from the prairies are quite tasty
somebody needs to install a CIWS near those microwave towers
jmkasunich, do you have a lot down by you?.... they take over the golf courses here... you need cleats (not because of the wet gras)
[01:49:38] <LawrenceG> http://www.youtube.com/watch?v=cP6GpAnmAPU
my Dad used to shot scotch mints at them with a sling shot from his deck!
yes, exactly those
also apparently they're fully autonomous
nobody has to clear them to open fire, they just do
and sometimes, they mistake other ships for things like missiles
if they modify the software for slower moving targets it might fix the pirate problem off Somalia
that IS adjustable
the minimum inbound velocity
nothing like 1000 rounds through the hull to let you know you are not welcome
"hey that thing is shooting at us"
"we must be hauling ass"
i wonder if they make one for garbage cans too
raccoons or possums are always getting in them
could be useful for preventing red light runners
the funniest thing I saw was a skunk raiding a garbage can..... I so wanted to sneak out and slam the lid back on
it might fire at the light itself
i do like the racoon idea though.
i have my eye one one of these for the problem: http://www.airgundepot.com/gamo-whisper.html
i shot a possum with my bow and arrow out of the 2nd story window last year, but wound up putting a hole in the trash can and garage door as well
then motion tracking and a little platform
and pew pew pew
they have a laser that would work great
i wonder how many watts are required to fry through a thick skinned o'possum
more importantly, i wonder if the co2 tank i got for tapping my keg can also be used for cooling a laser??
i propose we don't need to know how many watts are required, because any respectable death ray constructor will rather as, "how irresponsibly powered can i make this" instead
i have a 400A feed to my garage, is that enough??
run extension cables
to your neighbors
or i guess a chemical laser
pew pew pew
maybe just run a 6-0 cable back from the pole pig
there you go
they have 4-0 for like 1.50 a foot
at home depot
hm according to my reference, 4-0 is only good for 475 amps.
for an air-cooled cable
well i mean at 1.50 a foot we can afford a lot of it
compared to some of the smaller gauges which cost even more
yeah run one per phase i guess
100kw is a big laser
i bet it would take a big chiller to keep it from melting, too
1.50 a foot's not bad for that size cable I guess
the most powerful laser known to civillians right now is 25kw, made by boeing
according to this completely unchecked internet fly-by-night source
As the press release says, it "will damage, disable or destroy targets at the speed of light, with little to no collateral damage, supporting missions on the battlefield and in urban operations." Hit the jump for the full release.
no, there's a 80kw laser for combat
but it's hard mounted
it's not mobile
and i don't think they fire it anymore because it costs so much
i guess the power density's just not there
it was there, it was just too damn big
to make it really practical, i mean
"cool, we can burn a damn hole in damn anything"
100kw is hard to come by on a battlefield
"great, how do we get it there"
big lasers are so cool
i'm a little disappointed they don't make bright flashes and stuff
like you see in Freespace 2
speed of light and all
what's up with that, raining on all our collective parades
no scattering atmosphere in space
if they had big lasers, we would probably have to chrome plate our tanks
you dont shoot a tank with a laser
toastyde1th: plasma cannons would create the sort of fancy effect you're thinking of
A big enough laser could disable a tank, but you'd probably have to aim at the tracks
yeah, we need to get on that
er, plasmak i mean
i wonder how far that guy with the open source cruise missle got
i think he got shut down or scared underground
i think he underestimated the complexities of guided flight somewhat though
can you build a missile-like object that flies in the direction you want? yah that's cake
given time and money
i can supply the bearings for the gyro!
those mems gyros would probably be sufficient
actually in a twist of irony I have built bearings for a replacement guidence system for an ICBM
only once though
[02:23:39] <SWPadnos> http://www.popsci.com/bacon
I used prosciutto (Italian for "expensive bacon") because it is a superior engineering grade of meat.
obviously this guy is just a dilettante
but he did cut a pan in half with bacon and oxygen
what a great word
toasty: how big is the gyro in an icbm?
this was a friggin large gyro
i've seen two, one was tiny, a 3" sphere in the peacekeeper
this was a 15" sphere with 32 bearings
i have also made bearings for 3" spheres, that came out much better
at that point you might as well just use a laser gyro, (or even better an acoustic gyro)
nobody would let me do the 15" bearings properly, and we didn't get the contract because they couldn't handle the weight
[02:33:16] <fenn> http://www.abominablefirebug.com/AcousticGyro.html
fenn: they wouldn't tell us why they were using this
but the bearing had to have a specific surface capacetence, the real gyro had instrumentation IN it
because they didn't want to pay to redesign
"because it looks cool" isnt good enough?
so i assume they couldn't use a laser ring gyro because there's some other system that goes inside the gyro that has to remain static
15" hollow sphere, weighs about 150 lbs
you may be assuming too much intelligence in the decisionmaking
that is entirely plausable
I probably met some of the people involved now that I think of it
it was really cool
i was very upset that nobody wanted to do it right
we'd have made a TON of money on those things
and there's a lot of astrophysics applications for that style bearing, and very few people make them
probably fewer right now
i met a couple guys who used to make gyros like that, they're all retired/in nursing homes
good news is my friend that got laid off got a new job, bad news is not making bearings
what's he doing now?
something to do with ocean measurements
what type of bearing did the gyro use? ball bearings??
|dareposte|: air bearings
spherical air bearings
i want to see one..
i wish i had taken pictures
Zygo bought some to calibrate white light interferometers
Lockheed Martin bought some for... god knows what
and then there was that set of 32 for the guidence gyro
if they ever come up again i will
spherical air bearings... i have a hard time picturing that
it's just a cup
we use porous media, so there's nothing to look at, just a polished surface
how does it constrain the bearing?
[02:52:29] <toastyde1th> http://www.youtube.com/watch?v=0ELZcDLtlLI
the ball just floats
that thingy is supported on the spherical bearing in the center of the rig, obviously
but that's all it is, it's just a cup, there's no constraint
so if this was used in a fighter jet that did all kinds of weird maneuvers, it wouldn't really hold it so well
unless the bearing is preloaded with another opposing bearing, or with vacuum
well in missles and stuff you have a set of these bearings, all opposed
that makes sense
or under vacuum, which is freaky
because you can put the test ball in the bearing and turn it on
and move it any way you please
and it maintains orientation
even upside down
and when you spin it, it just goes and goes and goes
when the bearing is clean there's only a little bit of drag
and with a good push it'll go for about 20 minutes
I'm tearing apart a dover air bearing for fun
I believe they dont use the FOG's because they weren't around when most of the ICBM stuff was made
and here is a big damn bearing that is at our shop right now
[02:58:10] <toastyde1th> http://www.youtube.com/watch?v=jSMMiy41jSw
also for long term integration mechanical have fewer integration and quantisation issues
this goes in a siemens CT scanner, at 400 rpm with a 1/2 ton ring
the MEMS gyros are only OK for about 10 minutes of integration, at the end of that they will be out by a few degrees
mix MEMS with GPS updates and your set though
i wondered about those mems
they are great but suffer from a fair bit of drift
ROFL that bearing is insane
it goes SO FAST
like, holy crap
whats the air consumption like on one of those?
the really big ring i don't know
but more reasonable bearings are 5-15 cubic feet per hour
at 60 psi
not too bad
i guess there is some sort of seal on it then?
that's the no-load, open air flow
whats the $ like?
the cheapest bearings we have are like 25mm diameter plane bearings, 160 bucks or so
most customers buy something in the 300-700 range
they an air bearing still?
that's all we make
whats the application for them?
depends on the type of bearing, we have no idea what people use our plane bearings for
CMMs and other such
coordinate measuring machines?
high precision robots for semiconductor handling
semicon is a lot of our customer base
we used to make a ring bearing for a company that made wafer inspection machines
we also make full slide assemblies for machine tools
diamond turning lathes primarily
what sort of load and sizes
mega mega $ i assume though
a full assembly for a 21" slide is like 20k
we made a pair of 4 foot slides once, they were 125k each
ok lol a bit out of my price range ;->
well that's because we make the stuff?
we'd RATHER customers make their own slides
and just buy the 500 dollar bearings
to put in it
thats more plausable
Hard to make precision slides though
it's annoying as balls
so much grinding
but they do work on scraped ways
so if someone wanted to, they could scrape a slideway and drop air bearings on top of it
how do the air bearings perform under load, as far as accuracy
pretty well, the gap does close down
CMM's don't see much load variation
what's a typical gap though?
but the variance in gap is insigifigant compared to the bending of the machine frame
I think what we are asking is, would they work on a mill ;->
oh, of course they'd work on a mill
the gap changes by millionths
so then i presume the slide has to be accurate to that level too?
nope, that's just what the gap is
the bearing will follow the average height of the slide
mmm hover craft
they work on glass panels and stuff
so could you throw down some cheap extruded i-beam accurate to 0.002, and set a bearing on it and get it to function well?
if the beam really was flat to .002 all over it, it would work
it's stuff like sharp edges, blending problems, etc that will cause problems
a slow undulation in the surface is not a problem
It'd have to be aligned in all the other dimensions as well
hmm I wonder about making a "way grinder"
so why not make the slides out of super easy to machine material then, since there is probably very little wear
that's why we make out slides mostly out of aluminum.
downside: very thermally sensitive.
so we also have to offer steel slides.
and when you consider that most of the cost is in grinding, and that aluminum is harder to grind than steel
the price difference is not so great
I wonder if you could machine it rather than grind
not without a very expensive planer
well you don't need hardned bedways ;->
the issue isn't hardness
so you should be able to machine it to at least as good an accuracy as you can mill stuff?
It'd be a B grade slide but you could do it cheap cheap
it's ensuring that five different surfaces are all within .0002 parallelism and .00005 flatness
who is going to buy a cheap air bearing slide?
"cheap" is still going to be 10-15k
I was thinking of just making a cutter with all 5 surfaces driven
so 5 cutters
all inside one big cast iron "doovey"
where are you going to hold the part from?
mount it on the slide its making
the only surface that isn't getting ground are the ends
articulate the head and laser guide it
what's the slide mounted to?
Hadn't got that far
by doing one side at at time we can make adjustments in angle and taper
I presume some kind of table, It isn't that important
what adjustments do you need to make?
say one end of the machine is hotter than the other.
for whatever reason
maybe the sun was beating on it
that side is now going to be .0003-.0004 bigger than the other end of the machine
and will put a taper in the part in that direction.
so the part has to be taken out of alignment by .0003-.0004 to compensate.
then the day progresses and the ambient temperature changes, and the edges of the machine start to shrink but the center has enough thermal mass to lag
so now it's barreled
you would have to know that before doing it though
right, and that's why the machines that could do that cost a couple million dollars
ahh I see what your saying, you are adjusting to take out variances in the machine
not in the part
but also in the part!
because we have to handle the part
so if the part is hot when you cut it you still have a problem
well the part tends to settle thermally faster than the machine does
but you still have to dust it and see where it is
and on ad nauseum
four hours later, you have one surface done.
What if you just grabbed the end of the slide
then pushed the laser guided cutter down it
doing all surfaces in one pass?
what happens when one end of the interferometer heats up, and bends?
and now there's a bow in the part we can't take out because the interferometer thinks it's straight?
I wasn't using an interferomiter sideways
run your laser from the end along the path
but that's no better than just having a blind machine?
use a CCD sensor and some fancy maths to determine the position
so you can measure the deviation from the beam and correct for it
i'm not aware of any scanner of that type that will get the accuracy we need, but even if there was - the ccd and laser still have to mount in some way
so even if you have a laser scanner that can measure ten millionths of an inch
it's just resolution, not necessarily accuracy
If you wanted to you could mount the sensor itself onto the same bearing block as the cutter
where's the external reference?
I came up with the idea for our gantry mill to line up the ways
this is the "budget" version
get a laser pointer, shine it down the way
get a web cam and take the lens off
mount the web cam to the slide
shine the laser at the bare CCD chip
as the slide moves down the way look at where the laser is striking the chip, because its going to be going over lots of edges you will be able to do maths on it and very accuratly determine the center of the beam
adjust the mounts for the slide such that the laser strikes the same part of the CCD at each point
when your done you have a slide thats as straight as a laser
i like your optimism, but i assure you the slides are not going to be as straight as you think they are
what are the error sources?
the air, vibration, misalignment in device, the resolution of the device
air motion is canceled out through integration same for vibration
uh, air defocuses the laser
which is the limit when you're using an autocollomater or a tilt interferometer
not a problem because the laser is going to hit a whole bunch of cells in the CCD, then the magic software looks for the center of that hit point
i'm sure you can get a decent alignment using your method, but I'd really suggest you try it and get someone to verify your result
i've seen a lot of people's ideas of "high precision" in diy stuff, and it's almost always crap
hence my apprehension to the idea.
provided the laser spot is all inside the CCD face you should be able to get an accurate center
well It seems to work with astronomy stuff ;->
um. is the laser supposed to mounted on the slide as well?
what in astronomy uses that system?
SWPadnos, could be, like i said its just something i have been thinking about
oh, I see. you attempt to align the laser beam with the way, and move the sensor along the way
not the system but integrating noisy signals to get decent photos
photography and trying to align a bedway to .0002" are very different things, i might point out
.0002" is many pixels on an image sensor
sure, but even tilt interferometers, which measure angular displacement to tenths of a second of arc
have problems doing it
they are using a different method though
yes, they are
but you see the accuracy they can get
they are sensitive to loosing phase lock due to vibration etc
okay, but they're getting that accuracy
and it's STILL wrong
versus a method of unknown accuracy
how do they come up with the accuracy in the first place?
or the way, or what
of the reference beam
how do you make a "straight beam" from first principles?
the laser wavelength is stabalized in a very narrow band of frequencies
there's no reference beam in an interferometer?
or are we talking about the bedway
on earth, a beam of light is probably straight enough for anyone but CERN and the like
oh, that kind of beam :)
without any external references, then you're going to have to make a set of straightedges
you grind them, three at a time
See that just seemed like a pain in the ass to me in this day of $5 laser pointers ;->
label them A B and C, you test A:B, B:C, and C:A with prussian blue or similar
you'll get flatness to the resolution of your marking solution
i'm not saying it doesn't or won't work
It also seems like a very difficult thing to do for large beams
yes, it's difficult
well they don't do that for large beams, they only use the surface to provide spotting
not as a flatness reference
thinking about the webcam idea, I think I wouldn't remove the lens
I might zoom it as far toward telephoto as possible though
The thing with the lens though is its going to make the spot size smaller
yes, it will
at that point you are three steps and a jump away from an autocollomator
which is a bad thing,
you find the center of the beam by looking at pixels it crosses and the ratio between them
yes and no. you're actually testing several things when you move the slide around
height and two rotations come to mind
also skew (tilt)
actually two tilts
I was planning on using multiple sensors to check those things
err, I guess those are the rotations :)
one sensor is multiple sensors, that's why you use a webcam instead of a photodiode
The objective is for a 5mm x 5mm sensor with 300K CCD pixels in it is to have the laser occupy about a 4mm circle
so you then have about 1000 or so pixels that are "half covered" by the beam
IE not saturated
you they give you accuracy to about .1 of a pixel size
integrate that over 30 frames or so to null out random noise and you get a position
what removing the lens does is make it so the beam can come from more or less any angle and still illuminate the same pixels
your going to get a better position than looking at 20 or 30 pixels in an image
leaving the lens on will give you better localization of where the beam came from
hence using multiple sensors to measure that rotation
uh. the laser beam hits exactly one of the sensors (most likely), how would you plan to use more than one?
at $5 a pop its not a big investment ;->
probably need some better optics in front of them though
or better software to ignore the noise
so you'd need more edge space around the expected beam area, since you need to measure relative displacement
1MM should suffice i'd imagine
assuming you start off with all your lasers lined up
how does it reference the other way
yeah. that's a trick too :)
I hereby patent this method as "the Anderson method" ;->
toastyde1th: how do you mean?
if you have two ways, this only makes each straight. it doesn't make them parallel
how bad of a flat is a frozen pool of soapy water?
Ahh that is also taken care of
off your now perfect slide (see how easy that was)
ds3: pretty bad, there's a way to level a whole machine using a trough of water
but it's from like, 1930
so unfreeze the water
and you're set =)
toastyde1th: what makes it so bad?
I was hoping the soap would reduce the surface tension
you've got an unknown reference.
you shine a laser (or set thereof) across the bed, bounce it off mirrors on the other slide and back to sensors on the origional way
it doesn't freeze all at once
valen: how are you aligning the mirrors
you dont need to
people have been saying to use float glass as a poor man's surface table
you have your lasers such that they bounce at angles off the other side
any arbitrary angel?
you just put the sensor where it hits
or more likley adjust the mirror so it hits the sensor
you don't have to use the mirrors, you can just laser straight across to sensors, its just the mirror will double the error you measure so its a somewhat free accuracy gain
i'm not seeing this at all
but that's okay.
go do it, check it for accuracy using an accepted (and cheap) method, and if it works, sell it.
ok constrain the error you wish to measure to make it simple
lets say we want to make the 2 slides the same distance apart
the slides a ~1m apart and running from left to right
from the left hand end of the carrage on the reference slide you shine a laser across to the unknown and bounce it back such that it hits the right hand end of the carrage
your sensor is at the point where its hitting the right hand end of the carrage
as the carrage on the other side moves closer to the reference carrage the spot will move towards the left hand end of the sensor
and vice versa
i don't get it man, don't worry about it
ok you have a laser beam going like /\ that
as the from one carrage to the other, as the carrages get closer you see how the distance between where the laser leaves on one side and where it comes back gets closer
but where are these carriages
one is on the "reference" slide
the other is the unknown slide
oh, the one you're trying to match
the one we made with our first laser system,
how are you moving these in sync
I came up with this for a gantry style mill so you just drive the gantry down the way
or you drive both sides independantly
doesn't really matter
and how is this isolating the one error you want from all the other errors
in the "real" one you use other lasers to measure all 6 degrees of freedom
basically the same system in other axes
I'm interested to hear if anybody can see where errors will be introduced?
I can see heating of the carrage moving the sensors causing a problem
any time you move anything, you have the potential to introduce errors
i think you are being overoptimistic about the resolution of your sensor
but its a pretty dense lump of iron
the initial alignment
seperation of different errors
I don't see how reflecting a beam off one spot on one beam will hemp you align another beam
for instance, when i put a level on a bedway, i know exactly what it is measuring and i only get one component of error
your system is measuring multiple errors simultaniously, which is usually considered A Bad Thing (tm) in the ultraprecision community
when i put an autocollomator on a bedway, it measures just one thing
a kingway alignment rig, just one thing
a refrence flat - just one thing.
what is an autocollomatar?
yes but all those things are running off an external reference
autocollomators are defocusing telescopes that watch a collomated beam of light
as it reflects off a mirror
it measures tilt
I am trying to use light as the reference
it being considered generally "straight" ;->
right, autocollomators don't rely on ANY reference
they're gauges, they don't have an absolute reference
and they only measure ONE thing
the beam of light will be straight enough, but the relationship of the sensor to the beam of light is dependent on several things
well actually when you get right down to it they are using light as the reference still
valen: they'ren ot
wheels/balls may introduce a cyclic error
they are measuring the distance between one side and another of the mirror surface?
if there's a bump on the otherwise flat way, the carriage will tilt
so how do they work?
they watch the change in angle of the reflected light
you set it up, call one spot "zero" arbitrarily, and then as you move the mirror around and sight it with the scope, you get different tilts with repsect to your initial spot
ahh so basially the same as my "alignment" rig
what happens if i move your rig straight up?
with an autocollomator, nothing happens
the other sensor/laser combo picks it up
which is what you want
because you're not measuring vertical displacement
nothing will happen with the straight across distance meter
unless you make it such that it does
you wouldn't use an autocollomator for that application
it's for making individual ways or surfaces flat
you can though
no, you can't, because autocollmators don't measure distances
no displacement, no distance, only angle
what happens with your Autocollimator as you move it closer or further from the target surface?
(given a flat surface)
I'm trying to find out how they work now
it's a telescope with an internal, collimated light source that bounces off a mirror at some distance, and reflects back
so you get this dot at zero on your crosshairs
that's the beam the collimator is making
it bounces off a reference mirror that you're free to move around
Valen: we were going to do more or less the same laser beam method to check for straightness and coplanarity of ways on a CNC router
when the beam comes back, it makes a second dot on the telescope crosshairs
the vertical displacement of that second dot relative to the first is your angluar error
horizontal displacement is just error in how accurately the mirror is aligned back up with the scope
Yeah i think i see how it works, its beacuse its using a lens that the distance error is cancled out
the lens is a collomating lens, it's not a lens for optics.
you look through it dont you?
you do, but yout don't get an image
because the light is focused at infinity
you just get the two dots
so it is completely unresponsive to distance
because there's no focal point
the only problem that comes in is air currents
which are the major source of error.
If you had a camera hooked up to it though you could see it averaging out the position over time no?
fenn got any more info on your app?
valen: that's the "auto" part, yeah
uses a photodiode to give you a readout on a gauge
Valen: no, not really
it a commercial or hobby thing?
uh, does it matter?
it's a DIY thing
not super duper precision like toasty is after
pew pew pew laser accuracy
must use green lasers
they are 4x cooler than red ones
oh yeah toastyde1th you mentioned getting it checked somehow by a "cheap" method?
what was that?
gotta find a machine level and build a kingway alignment tool
the level will tell you the height errors of the ways, and also the relative height differences between the two ways
then the kingway alignment thing tells you the distance between the two ways.
I meant the cheap way to see if you had made a straight way
that is how you do it?
a level and a homemade rig + indicator
that is as cheaply a way i know how to align ways
i meant to make one "straight" in the first place?
beware that iron surface plate, is it really a lapping plate ? lotsa grooves in rectangular grid ( handy to have but not suited to inspection )
and scraper and straightedge, if you plan on modifying how straight it is
level tells you straight from side to side?
straight is not level, straight is not bendy
use a reference to see straight, like a scrapers edge and 2 dowel pins
tomp: that doesn't work over long distances
which is why they use levels and shorter straightedges
lay laser if long
sorry, my boss does large machine tool alignment, and he uses levels for long machine ways
hmm, mechanically? how long
long as in 200-600" boring mills
ok, a few meters, my 5meter ways
all you have to do to measure the height error is just note the starting position of the level
move it 8"
still id use a laser and retroreflector mounted on a bearing tractor
move it, 8"
move it again, 8"
until it's down the whole way
if the level read the same inclination the whole way down, you've got a straight way
if it changed inclination with reference to the initial position, the way has errors in it
how do you correct for errors perpindicular to gravity though?
valen: like Z?
use ways shorter than kilometers ;)
you're talking about Z axis alignment?
perpindicular means at right angles
right which had me confused because the ways are perpindicular to gravity
which is inherently what a level is measuring?
i think he means the sides of the way (if it's horizontal)
right, how do you measure the error in the non up and down axis
'down is where your feet are " ( skateboarder truism)
Valen: the 'wiggly' ?
valen: i have no idea what error you're referencing
the technical term for it
toastyde1th: say you have a box way
there's a top and two sides
how to check the straightness of the sides
they're never straight
need a reference ( a laser or a straight edge) its comarative ( there is no difference between a duck , you need 2 )
that's dumb, the whole point is for them to be straight
they're fitted to the carriage
so how do you make a straight cut?
oh, you meant all the way down
i mean along the axis
that's where the kingway thing i mentioned earlier comes in
and what is a kingway
so like, you've checked the differential level down the axis, and across
that just makes sure they are aligned to each other no?
so you know both ways are straight, and the same height
you know the top surface is flat
the kingway alignment tool is a 3-point gantry
two points sit on one of the ways, and the other is on a hinge sitting on the other way surface.
there's an indicator measuring the hinge's position, just a normal DTI
as you push the 2 point side down the way, the hinge follows the other way surface and the variation shows up on your indicator.
yeah thats still only making the ways parallel not nesicarily straight?
top parallel,motion along it not necc straight
hmm, you're right
i can drive down trhe street with my head travelling parallel to the ground, but weaving like a sob
i'm forgetting one of the measurements
the order... flat straight othagonal
I spose you could perhaps do something where it looks at its own way
and if it measures a diflection from infront to behind itself then its wonky
* fenn mumbles something about magnetic fields and compasses
the suspected thing is not a good reference
i've got machine tool reconditioning out, give me a second to see if i can find the part about that
It seems ok if you are adjusting it to match a null
you use the tool to give you zero error
stretched wire + capacitive probe :)
I like my laser system better ;-P
i like checking with multiple systems
connelly's saying to just use the straight edges
look at it with a reference, hold a stright edge up along side and look for daylight, thats how you were taught to check straight, now expand that to a long problem
toastyde1th: that'd be a long straight edge
for a 5m boring mill
they call it a laser
they use 4 and 8 foot straight edges
16ft straight edge would weigh 8 times what an 8 foot straightedge would
it's relying on locally straight segments
to give the other indication methods that work at long distances a reference
seems pretty labour intensive
reconditioning a machine is very labor intensive
no matter what tools you use
if you do it the hard way
fenn: strange how the only people who think it's the hard way are people who have never reconditioned a machine.
buy new machines! ( and dont bring me any of that old wine !)
i refuse to believe that "the hard way is the only way"
you'd think with a 25,000 dollar spindle analyzer, differential electronic levels, autcollomators, 8 foot straightedges and the appropriate scraping gear to maintain them, portable grinders, and interferometers
If you leave the "Anderson laser system" in place you could actually use it while you were running the machine, to correct it back to accuracy continiously
they'd be willing to spend money on a better system
if one existed
people do lots of difficult things, either from ignorance or stubbornness
I think they probably are happy with how their stuff works and are willing to pay for what they know when their business is at risk
and many completely inexperienced people think they know better than someone with 40 years of experience
new, better things do come along all the time, and many people with 40 years plus of experience refuse to adapt
the laser could actually watch where the tool was instead of extrapolating "it oughta be there" from the scales which aint near the tool
tomp I thaught about that, its really really hard to do
LODTM does it
too much crap floating about
large object diamond turning machine?
oh, optic, cool beans
actually the interesting thing with that problem is the control
since emc has kins, i imagine you'd be able to adapt kins to a metrology frame like the lodtm has
[04:58:49] <tomp> http://books.google.com/books?id=1CmQgcT72awC&pg=PA545&lpg=PA545&dq=LODTM&source=bl&ots=YKUn8lMAf8&sig=8j7RqChJIkNirKJmy_rKwRY9m0s&hl=en&ei=JgzoSZXyGIuZkQWunfmKBw&sa=X&oi=book_result&ct=result&resnum=8
[04:59:10] <Valen> https://www.llnl.gov/str/April01/Klingmann.html
we almost got that machine, by the way.
wow, thats a weldment too
After the tool is set up, machining does not begin for at least 12 hours to allow the effect of the machinist's body heat to dissipate
but it needs a new control
and since it's custom, the work estimate is about 5 million
haha, the old inspector would make us wait while had a coffee and and another stub of cigar, letting the grindings sit on the surface plate
you guys work on some cool stuff, gotta go
valen: it's actually a cluster of 4-5 controls
it would be cool to have the stuff in emc, that would probably be very useful to some of the emc users
b4 i go, let emc talk to other emc's :)
k, back to orcad
and back down again ;-P
is there a difference between an ac and a dc relay coil? i see the parts bins here are labled 'ac/dc' for small 'ice-cube' relays
ooh. looks like i'll pass on these interchangable relays, they are not the same
JustinXJS2_ is now known as JustinXJS2
toastyde1th is now known as toastydeath
wont AC change the magnetic field each time it changes? :)
so at one time it will pull the pin the next it will push it ?
its not a short answer , sometimes you can make it work, but its not a good idea http://www.google.com/url?sa=t&source=web&ct=res&cd=3&url=http%3A%2F%2Fcr4.globalspec.com%2Fthread%2F8100%2FAC-relays-versus-DC-relays&ei=hjLoScy1Hpnu6wP4i4zFAw&usg=AFQjCNHGK8jru93ZChhOEn73fWojGmJflw
and an ac relay doesnt buzz like you tendered
AN AC contact will have a MUCH longer life than a DC one
add a diode to a relay to prevent bemf iirc
diode across coil of dc relay, yes, to shunt the discharge when the coil field collapses ( just like old timey car ignition), but you suggest across contact that handles ac?
no, just coil like you mentioned
new pcbs arein, the guys soldered in the dip swxs. the dip swxs are crap!, the print sez " dont put in switches, use jumper wires on component side", doh, assembler is chinese
bu dui bu dui!
as one who has played with CSS, thats a good one
Is the msg "/dev/hda1 has been mounted 30 times without being checked, check forced" that appears occassionly when booting, normal?
u can change it so it never checks
but its better to let it check the state of the disk now and then
is that a good idea?
Yep you're right
[12:11:07] <pjm__> http://ubuntuforums.org/showthread.php?t=300477
that explains how to do it just for info
Do you know where I can find a good treatise on the use of the sub and call functions for program control in g-code?
Iv'e had problems understanding their usage
I've researched the wiki
ah i personally dont know, i can do g-code basics, someone here will know the answer tho!
I have never seen sensible docs, just struggled till it worked
the sub and calls were not designed by a real programmer
hmm quarter past dinner time!
Im about to g code a brocott excape wheel
[12:23:44] <archivist> http://www.orologeria.com/webart/foto/pendoleria/brocotruota.jpg
[12:26:06] <tomp> http://www.geocities.com/CapeCanaveral/Hall/3934/beb112.html
cool some of the math for the brocot
archivist, I've also had to struggle untill it worked!
looks awfully delicate, what diameter might that be?
tomp I dont think I shall be toooo fussy bout the real maths, I saw that page when looking for a link to a pic
23.3 mm dia
delicate, hah, i suppose the missing teeth arent part of the design
this one here has missing hence the job
.8mm thick brass to make it out of
pretty smalll at 23.3, that scales those teeth down to 'skinny', you need a small wedm . no force on the part during machining
pjm__, that made on your machine?
tomp Im going to profile cut with a slitting saw
I get used to backing a sheet for support during the cut
i dont see how you can use a slitting saw to cut the profile, nless you got <1mm dia slitting saw, even then, a helluva kerf (not familiar with profile cut, is that 'contour'?)
i had no idea it was on edge of a short cylinder
are there any details on your a axis? looks about 4 to 8" dia?
well that one was but the brocot is on sheet / disk
ok, even flat i can see it
a axis is just a hacket Vertex 4 inch rotary
there is a nother vid showing the flat version
Hmmm That xfer didn't work anything like I hoped it would. Guess I need to read the FAQ.
the vertex tilt rotary looks interesting, abot 330$us
really needs another proper cnc axis, the setting up of fixed angles is a pain
what a lovely morning to ride the bike to work :)
I bet - it has been 60deg and sunny here this week
55 and sunny here going up to 65 I think
[13:48:12] <skunkworks> http://imagebin.org/index.php?mode=image&id=45896
archivist - no not made on my machine, it was something me dad made last week on their 5 axis m/c
I did find local ups batteries about 1/2 the cost of apc or digikey.
which supprised me.
skunkworks: What is the plan for the new computer?
pjm__, I thought it was a bit big for yours!
none. Just playing. I have a goal3 board that I am running the gantry with.
archivist it is 316 stainless, all out of a solid billet
its part of a sonar apparently
skunkworks, is that an atom?
have you run emc on it?
I have not run emc yet. But the latency test is running while it is downloading the updates
what kind of numbers are you getting?
Highest I saw yesterday was about 18000
not stellar but not bad
I haven't tried all of my candidate computers, but that's better than the one I have in there now
I think that is pretty close to the numbers jmkasunich was seeing
are you going to rebuild the computer for SMP?
could have saved $15 then :)
heh - yes.
These are acutally going to run xp at some point.
I'd like to get some for our robots
now running around with laptops
hmm 6 tons for retrofit fleabay 170321680609
that looks a lot like the old cnc we have at work
retrofit! nice jog wheel atleast.
I can dream cant I?
how much does a jog wheel cost?
HNC is the toy of the moment :)
and it just makes encoder-style pulses?
how about pyvcp jog wheel anyone us that
yes, it's just an encoder
wow, they have two models now
do usb parallel ports work with emc?
that's the answer I was expecting :)
probably not realtime enough
SWPadnos: did you see the cnczone linke I posted yesterday?
till someone gets their head around the synchronous mode
[14:08:59] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=78514
would you have to run the pyvcp jog wheel through the sim_encoder?
what pyvcp jog wheel?
skunkworks does that have the compiz desktop/3d stuff turned on?
[14:10:23] <skunkworks> http://linuxcnc.org/docs/2.2/html/hal_pyvcp.html#r1_5_10
pyvcp you can create a jog wheel and it outputs a float
Valen: no - it is installed from the livecd which has the effects turned off.
ahh thats a shame (i was hoping it could nerf some latency ;->)
it runs opengl very good. (compared to other video cards I have used) it will do glxgears full screen pretty smooth
that don't look like a jogwheel :)
skunkworks, yeah i have the same board
can't rember if its single or dual core though
using it for a digital sinage thing
you don't want to run compiz :(
less than $100 though :)
cheapest jog wheel on cnc4pc is $70
my only concern was the single PCI slot
glxgears does about 80fps full screen.
too bad the pci slot isn't right angle
don't know what i'd want another one for just yet but its always a real pain when you find you want it
you can get right angle adaptors for PCI if you want it
any good sources?
I believe they come with cases that can use them
that means the case costs x times the mobo
something like that
If you aren't after super small just stick it in a regular case
where x is an integer between 2 and 5
is there some ngc files for 5 axis? ( to run with the cvs versions of emc with max5kins? )
or use low profile cards in a low profile case
m5i20 not low profile
bandsaw == sparks
your thinking grinder
ok, I found a riser card on ebay that is not right angle
2 slot pci mobo with pci cpu card and m5i20 = small
that's just about the most useless thing I can think of
if bandsaw = sparks then blade = junk
if you cut your pci card in half
or is that not the purpose of the bandsaw :)
if you want small do PC104 stack
PC104 would be great if it worked
mesa have PC104 cards
why would it not work?
have you used it?
I've wasted months of my life getting things to work
no but it seems fairly prevalant
somethings simply don't work
why do they not work?
[14:21:19] <skunkworks> http://cgi.ebay.com/Rightward-Right-Angle-32-Bit-32Bit-1U-PCI-Riser-Card_W0QQitemZ310136061136QQcmdZViewItemQQptZUK_Computing_ComputerComponents_InterfaceCards?hash=item310136061136&_trksid=p3286.c0.m14&_trkparms=72%3A1205%7C66%3A2%7C65%3A12%7C39%3A1%7C240%3A1318%7C301%3A1%7C293%3A1%7C294%3A50
anybody know some good suppliers of ballscrews
both screw and nut?
where good is ~= cheap
skunkworks, I'll just import directly from hong kong and save a buck
somebody sells a ribbon cable riser card
[14:22:50] <skunkworks> http://cgi.ebay.com/PCI-Right-Angle-Riser-Card_W0QQitemZ360009325440QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item360009325440&_trksid=p3286.m20.l1116
I must use a wireless xbox controller for the jog control when we get the mill going
skunkworks, the only problem is finding cases to put them in
that's not a problem if you have a bandsaw
[14:23:54] <skunkworks> http://www.youtube.com/watch?v=IgGxaNq46Pk
tis, because they need to have the card slot in the right position, a few mm out and it just wont fit
heh that the xbox controller?
for me it's going in a big cabinet
but a computer case takes up too much space
what is he using for motors on that?
[14:25:48] <skunkworks> http://www.electronicsam.com/images/KandT/cncworkshop/mazake2.JPG
where can i find the the ngc file behind this http://www.youtube.com/watch?v=fWKYQUj5AOs
you ping chris radek
thx ;) and i remmeber that panel
who needs a case ;)
skunkworks, is that a full sized power supply?
which? in the mazak?
Looks like it to me..
your last link
looks flatter than that
skunkworks is there any more info on the machine at http://www.youtube.com/watch?v=IgGxaNq46Pk&feature=channel_page
the video makes it look about 2x normal size
valen - That is 'theball' when he is on here.. I don't have any specs
looks like NMEA 34 mount motors at least though
I'm guessing not that big
some pretty big box section
mounting the motors
[14:34:02] <skunkworks> http://www.electronicsam.com/video/MVC-341W.MPG
[14:34:48] <skunkworks> http://www.electronicsam.com/video/VIDEO1.mpeg
when I run the pyvcp float into sim_encoder and move the jog wheel one increment off zero it acts like a button sim_encoder keeps making pulses "net velX Index_pyvcp.x-jog-wheel => sim-encoder.0.speed
Do you have to run it into the sim encoder? If it outputs a float - can you just hook it into the jog inputs?
found the 5axis demo file, its intuitively named 'cone'
* skunkworks has not done that before so I may be wrong.
I'd like at least 4 axis, but 3 axis would be a good start :(
i'll give that a shot
is the re-curring output desired? ( did you want the output to continue stepping due the the 'error' dialed in ?)
skunkworks, no, I hadn't looked. I'll reply now (if I can remember my password)
oh yay. Firefox remembers my password
the 5 axis demo is nice but i had to choke the speeds and overrides way down to stop 'joint 0 following error' (>way< down)
somebody showed me it's saved in plain text
SWPadnos: the thumb drive is partially persistant.. it remembered i had 4 desktops, but didnt remember the tcpip setup ( tho i saved it ). i moved stuff on desktop, and made bookmarks in ffox. only the 4 desktops persisted
I think the persistence only applies to personal settings
the casper filesystem may get mounted over the /home directory or something
hmmm. actually, the ff history should persist also
tomp: no not the desired stepping
does it seem like, emc sees the dialed in value as an error, moves to get rid of it, and sees it again and again? ( and how fast does it occurr ?)
when I move the dial one poss it jogs as a button would one poss opposite way moves in op dir
im still getting the idea..., and what did you want to happen? ( move one unit in the direction expected )?
like a jog wheel
pyvcp.jog wheel outputs a float and not a quad signal like a real jog wheel
ok, but sending the value to 'speed' will cause and endless motion. setting a new position would be finite.
speed != 0 = motion forever
and the 'event' wont retrigger cuz the data would be the same
dave hello how are you
goint to the workshop?
doing rather well, but not getting enough done ;-)
no I'm not but wish I was. Will pour equivalent $$ into my shop this year. Last years trip approached $2K... ouch!
can I i take the float and conv_float_s32 and send it to axis.0.jog-counts and bypass sim and encoder all together
tomp you home now or still in asia?
oh, in taichung, leaving on 30th
sometime when you get back email me with your address and I'll try to send that servo amp
alSMT_: i dont know but i'd try, it really seems like speed is not what you wanna tickle
dave, dont worry about it, can you use it?
probably not. but appreciate the offer
alSMT_: does that make any sense?
dave_1: ok, when i get back, but send cheapo, no hurry, put in a pc of paper and label it books ;)
probably use one of the PO single price boxes
alSMT_: that sending a value to speed will move forever
help! I'm having trouble with ssh. To one emc machne it works and to the other I get a 'connection refused' .. ideas to fix?
ps ping works to both machines
trying to connect to the same machine, or to each other?
oh, to the other ...
* SWPadnos had to share his coffee this morning
do you have ssh installed on the second emc machine?
ssh server, that is
don't know ... standard 8.04 emc Ubuntu
it's not installed by default
guess that is the first task.
will try that and get back to you. tnx
alSMT_: then experiment with sending offsets, or setting the jog counts like you said
dave_1, it's called openssh-server
isnt there a permitted machine list for ssh? (some /etc/blah file)?
tomp: rodger that
there are probably ways of restricting which machines you can connect to as well, but I've never messed with that
alSMT_: let us know the progress, sounds cool
usually I'm just happy to get it to work.
it is a little surprising that openssh-server and openssh-client aren't installed by default (client may be)
wow i had a skype to land line call here from USA, really good quality
i remember that FISH used it inside Midnight Commander, very handy
Any of you used SolveSpace before?
It's a little CAD package.
[15:19:21] <Jon_geo01005> http://solvespace.com/index.pl
SWP fixed that problem. ssh now works. tnx
Jon_geo01005: perl plugin? whats required?
No, it's a standalone application.
does parametric sketching, my favorite part of NX and solidworks.
Jon_geo01005: this group is about linux based cnc control, dont worry, but most would need to reboot to try a windows app.
i'll dload it and try out the demo
Well I don't mean to be tool pessimistic, but you may have to reboot to use any worthwhile CAD....
people are now developing linux capable cad
heekscad looks good
Don't bother rebooting though, the video demos on their website show all you can do.
video doesnt work here
a way do go to beat solidworks though but getting there
sorry jon, the vids say they require m$ codes
hese videos are encoded with the Windows Media Video 9 Screen codec. If you are unable to view them, then ensure that you are using a recent version of Windows Media Player.
I like linux, and I have used NX on linux and it's great. Sorry the videos don't work.
np, we can look after reboot
thx for the link
Im also waiting on the slamdunk gui on top of brlcad
but grass grows faster
i'd like to learn more of apt, tho its not cad as much as cutter path
a couple in #cam know apt
and im still putzing with the 5 aqxis vismach demo, lotsa joint erros till i uped the accel, now i'm stuck with axis 2 on a limit
archivist: yep, cluey people
apt would be a lot easier to use if the error messages were text and not just a number.
Heekscad would be much much better in my eyes if it had a sketcher that used a solver to enforce constraints.
Like solidworks, NX, ect.
since Stuart is an apt expert or has experts in his shop then emc workshop could be a good time to pick his brains :-)
joint 2 on limit swithc error
Jon_geo01005, add an issue to heekscad or comments to issue 49
dave_1: didnt know stuart used apt
Jon_geo01005, I was discussing with the writer yesterday
I'm looking around the web to see if there is an opensource constraints solver that could be more or less plugged in.
whole shop is a commercial version of apt
Jon_geo01005, by the way we lurk in #cam
Thanks for the note.
this is odd... if I drive my van from shop to shop it is 8 miles... if I drive the bike it is 36 miles
good thing you get 4.5x the fuel economy on the bike
only twice as much
huh - I just fdisked my keychain drive.
i bought GetDataBack a long time ago - looks like my activation key still works for the latest version. (looks like I recovered everything)
damn - jmkasunich sure knows how to explain things.
LawrenceG: how did it go last night?
got distracted.... no progress
eric_unterhausen: if I am reading that right - looks like maybe 30000 period is pushing it.. http://imagebin.org/45923
interesting. that makes me think that the base thread is getting interrupted by something
which is supposed to be impossible
impossible he says
I'd reset tmax to 0 and let it run another day or two, see if it happens again
if it were impossible, the machine wouldn't have latency :P
well, supposed to be impossible ;)
yes, it still would
there's a difference between interrupt response and re-interrupting the interrupt
also .. the Time and Max-Time are in machine counts.. right? not usec like the Period
CPU cycles I mean
hal_lib.c: /* capture execution time */
end_time = rtapi_get_clocks();
It seems to happen right at startup.
skunkworks: what kind of CPU?
(as fast as I can get into hal config - the tmax is alread > 20000
it is that atom 330
how many MHz I mean
skunkworks: if the numbers are in CPU cycles, then it's not really a lot
that's the Atom board?
if so, it's a 1.6GHz CPU I think, so still 20 uS max time
well, maybe 18 uS
still a lot more than I'd expect for a parport read/write/make-pulses
I think that is ns.
no, tmax and time are in CPU cycles
As posted by Alex:
[14:21:48]hal_lib.c: /* capture execution time */
[14:21:49]end_time = rtapi_get_clocks();
it shows you the base thread of 30000 - and if I set the base thread to 20000 - I instantly get a realtime error
28000 / 1.6 is pretty close to 20000
ah - ok
it's around 16 usec
so you should get some realtime erros, I'd think
1/1.6Ghz * 26952 = 0.000016845
that makes sense.
SWPadnos: realtime errors are calculated in the servo-thread
heh - I was reading that as 28000-29000 for some reason :)
aren't there two different possible errors?
one is missed deadline or something
one is an error reported by RTAI
but it happens in dubious cases :D
skunkworks: you should be able to see each of those functions
how long they took
is this the correct way to convert float to s32 "net conv_float_s32.0.in <= Index_pyvcp.x-jog-wheel"?
you need a name for the net
net mysig conv_float_s32.0.in <= ....
do I need to addf to conv_float_s32 also
[18:38:06] <skunkworks> http://pastebin.ca/1395161
I get this "ndex_pyvcp.hal:3: pin 'conv_float_s32.0.in' does not exist" when I use this "net mysigxf conv_float_s32.0.in <= Index_pyvcp.x-jog-wheel
you also need a signal name
err, nevermind me
alSMT_: I assumed you looked up the pin name
try using - instead of _
for a pin called conv_float_s32.* to exist you probably need to loadrt such a component first
it doesn't complain with out the name but it doesn't update the float either
and check the man page
also, you do need to add the function to a thread, or the output will never update
skunkworks: crap, I thought each function has a t-max
275. 4 s32 RW 49392 motion-command-handler.tmax
i load the component and it complains when I addf failed
alSMT_, are you editing a hal file and reloading emc every time, or doing this at a halcmd prompt?
addf conv_float_s32.0 servo-thread
edit and save reload
I think you'll find that all the names have "-" instead of "_" in them
you can easily see this if you use halcmd to load and then inspect the names
skunkworks: seems the parport access is the one that takes a long time
everyone says the onboard printer ports suck usually.
that chipset is a dog, in my experience
even for PCI
(much faster than a parport, but very slow for PCI)
SWPadnos: yep the name game it works
why the naming diff ?
SWPadnos: thanks you to Alex
alSMT_, is it spelled with underscores somewhere?
more specifically, if anything in the manpage is wrong http://linuxcnc.org/docs/2.3/html/man/man9/conv_float_s32.9.html
please point it out so it can be corrected
jepler: no I just didn't get it right the docs show the right thing "_' "-" I just didn't notice the diff in the naming
[20:55:16] <skunkworks> http://www.onel.se/all_files/gecko_g540.html
hmmm. that's unlikely to work right
the step outputs should be inverted
also the timings are wrong (that "gecko" selection is for the G20x/G21x drives I think)
[21:01:58] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=78514
didn't you get the email ? ;)
no latency errors running emc/axis all day with 30us base period
on the atom
hmmm. is that a DoubleStep config?
hmm - I don't know. It is the stepper_inc
I don't remember seeing parport.0.reset in there
ok, probably not then
would that help or hurt?
it could explain some of the thread max time
but the explanation was there anyway, the read and write functions both had numbers >20000, and even make-pulses had >11000 IIRC
good thing we have screenshot software otherwise we'd be forced to share .ini files
and you know what happens when you share unprotected .ini files
good night all
justa is now known as _justa_
I'm willing to use emc2 with TI UC3717 stepper drivers
In half step mode I need to feed the integrated chopper circuit
limiting current pins (2) with a High/low sequence.
Do I need to write a new hal component for that or
does something like that already exists?
the types of output available in stepgen are listed in the manual. I am not familiar with UC3137 (googling datasheet), but this doesn't sound like one of the standard sequences to me.
step types are documented in the following places: "man 9 stepgen" or http://linuxcnc.org/docs/html/hal_rtcomps.html#sub:Stepgen-Step-Types
if this requires 5 or fewer outputs per axis, then you can just change one of the "master_lut" tables in stepgen.c and recompile. If it requires 6 or more (from the datasheet it looks like 6 are required), then you will have to modify stepgen to accomodate a 6-phase output type.
if 6 outputs are required per motor, I don't think you can control 3 axes with a single parport; there are only 12 outputs and you'd need 18.
you might also consider adding a microcontroller which reads emc's step&direction or quadrature outputs and drives the 6 inputs of the UC3137 chips..
yes it's a 6 wires per axis
url for the data sheet
ah, there's a 4 I/O solution shown on figure 12
[23:28:47] <invite> http://focus.ti.com/docs/prod/folders/print/uc3717.html
[23:28:47] <jepler> http://focus.ti.com/docs/prod/folders/print/uc3717.html
that doesn't look like any of the stepgen tables, but you should be able to modify one of the 4-phase step types to give that waveform
to do this you'd have to install from source. there are full instructions for this on our wiki: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_6_06_or_8_04_from_source
so I'll have to implement the waveform shown in figure 14
stepgen.c is in src/hal/components/stepgen.c, and for each step type there's a master_lut entry which defines which outputs are active for each step in the stepping sequence. You could modify the one for step type 9 or 10 (half-step, 4-phase configurations) to match the one in the datasheet
or you could add a new step type which would be type 15
will see what my if c skills will fit the needs :)
or use a more modern chip :)
like what ? (I'm not looking for a whole board)
and isn't the duo L297/298 more oldish ?
Can you ever connect a pin directly to a pin in HAL?
I want to use the and2 to and 3 different pins to get essentially a 3 input and gate, then output pin to an output pin.
I got this working after looking at it for 3 evenings, but just by using dummy signals. (Actually, I never got the final output pin routed to another pin.
_Poincare is now known as Poincare