A compare and the f-error pin.
you can just use encoders for feedback, and if it loses position, you'll get a following error
that's how emc naturally works - you don't have to play any tricks
usually stepgen provides feedback - you can provide your own
I'm currently using the stepgen feedback, and I could very easily use the encoder instead
yes that's all you need to do
But I'm not sure I want it to trip a following error, since that would stop the machine
In some cases I don't actually want it to stop
but that's the whole point
not stopping ruins your work
Well, I guess what I mean is, I don't want it to go into an estop state
because that trips my contactors and cuts power to the stepper drivers
well it doesn't, it goes to machine off
theorb is now known as theorbtwo
Yeah, I guess I mean machine off
could just have a PAUSE button
so? turn it back on, rehome, and resume your work
In some cases that could be dangerous, for example when using a keyseat cutter
actually you don't have to rehome - the encoders maintain the position for you
But not if the encoders' electronics get their power killed
a pause, or stopping stepper gen, seems more universally acceptable to me
no sane setup kills the encoders when there's a ferror
Well, our stepper drivers are also our encoders
your encoder card powers the encoders
* Jymmm mumbles something about EMERGENCY Stop != Oh Shit button.
Our stepper drivers have opto outputs, which we have going into a mesa board with stepper encoders
sorry I'm still lost
If we cut power to our contactor, thus cutting power to the stepper drivers, the encoder loses power
I'm actually not sure if an f-error cuts the contactor
I'm gonna go find out
what do encoders have to do with stepper drives?
(I'm finishing up on EMC conversion, so some details like this are still getting sorted out)
This is a stepper machine with encoder outputs on the stepper drivers
But right now it's all open loop
And will continue to be
what do you mean by the stepper drivers have encoder outputs?
can I see docs online? I've never seen this
Just that! Stepper drivers that have position encoders
you mean the MOTORS have encoders?
driver = electric amplifier thingy that runs motor
Yes, I know
hold on, I'll get you a datasheet
Here's a pdf link to the system manual: http://www.orientalmotor.com/products/pdfs/opmanuals/HM-6205-6E.pdf
The stepper drivers also do closed loop control to make prevent missed steps
I'm guessing they're closing the loop on position error and coil current
lepton: I think you meant that the drives have encoder INPUTS, not outputs.
um RS232? what the heck is this
It's a big fancy stepper driver
The RS232 allows it to be reprogrammed such that it's IO doesn't match the factory specs any more, which led to some surprises yesterday
Jymmm: I meant outputs ;)
Check out that link, if you're curious
lepton: No, INPUT from the encoder to the driver
these aren't the droids, er, stepper drivers you're looking for
Jymmm: it's some weirdo thing - rtfm
cradek: Yeah, no thanks if that's the case.
It's certainly a funky system
they have their own control built in, commanded over rs232, and they do have encoder type outputs or something
When I make a new machine it'll all be servos
I bet these are quite useless for any coordinated motion
unless you can seriously dumb them down
cradek: Ah, so minimal cabling, and chained serial control.
They do a good job when treated like normal open loop stepper drivers
Sonds much like the OEM750X's
how do you command them to step?
step and dir signals
Fairly normal in that regard
I've only used the RS232 to program them
oh ok, I wasn't seeing that in here anywhere
and I only did that today for the first time
We've been running this machine since December quite a lot, started the EMC2 conversion about a month ago
I thought I'd be done within a week
...how foolish I was
Fortunately I could do it all over again in a week
Actually, I shouldn't say that, since I haven't cut anything yet, and I still don't have modbus up for spindle control
Hmm... any ideas why I'm getting following errors at high speeds?
Somewhere between 200 and 300 ipm I get following errors, shortly after it starts moving when I jog it in axis
this applies to all three axes
Or better yet, suggestions for how I can track down the issue
How old is your machine? What has it been used to cut in the past?
Built in Novemeber of 2009, probably has produced about 200 lbs of aluminum shavings since then
Along with a fair amount of wood and plastics
mostly aluminum, though
So, I'm guessing you can program two of those general-purpose inputs to be step & direction?
There are already pins permanently designated for it
In either step CW / step CCW, or step / dir configuration
Oh, OK, I must not have gotten to that part yet.
It's part of the 36 pin mini centronix cable plug
Stupid expensive cable...
Are you microstepping? How many?
1625 microsteps per motor rev, pre gear box
And how many motor steps/rev (pre gear box)? Have you got all the internal values (limits) set wide open? (Acc/Dec, max vel, etc.) I think you'd do better to let EMC2 manage those.
As far as I know the stepper drivers do have some internal limits on acceleration and velocity, but they're above those set in EMC
I was thinking it might be better if they were a lot above rather than a little above. Current and similar items can stay as usual.
Well, I made one pass thru the manual, halfway looking for that part about programming step & direction inputs, but must have missed it.
What size motors are these? Is it reasonable to expect 200-300 IPM from them?
And what's the screw pitch?
Yeah, we've moved at those speeds without trouble before
On our old control system
It's also rack and pinion
2482.8171122335 steps per inch
which everything is said and done
What was the old control system?
I'm noticing that the following error comes on doing decceleration
Shopbot control software
I've ranted on here about it before
I'll refrain today
Not familiar with it, sorry.
You're better off for it ;)
(Or maybe I'm not sorry, if it's as bad as you say it is!)
Can you pastebin the internal program these drives are loaded with?
Unfortunately not easily, because it's just a terminal that you interact with over rs232
So I don't have a file that I can upload
I've hardly even done anything meaningful programming for it
OK. But you're running the same program on the drives as when the old control system was in charge?
Do you recall if the program was dirt simple, like "START", "connect pin 1 to STEP", "connect pin 2 to DIR", "END"? Or was it doing something to "help"?
I don't think it's possible to reprogram which pins are step / dir
I'm pretty sure those are fixed, and going through opto inputs
OK. So was the program more like "set standard step & dir mode", "end"?
Sorry to keep bugging you about this, but I'm wondering why the old control guys specified such a smart (more expensive) drive, and why they needed to.
Has anyone here ever CNCd a mould for casting polyester?
I don't think very highly of the company that did the previous control system, and selected this hardware. I believe their motivation for going with these steppers / drivers was to use it's internal closed loop control for preventing missed steps
They advertise it very heavily, and it didn't require changes to their own windows based software
Why not just use a stepper/encoder?
Encoders aren't so expensive.
That said, I don't know jack about EMC.
Or are these steppers with integrated driver circuitry?
....sorry, it's really hot in here and my brain is melting out my ears.
Oh, OK, so they were using it in stepper/servo mode? (Ala Gecko ???, forgot which model)
it's really hot here, too, my brain is also melted
Yeah, they're giving it step / dir signals, and expecting IT to perform like a servo (make sure it is where it needs to be, varying coil current accordingly)
It's a nice idea.
I can't speak for the efficiency of steppers, but the net result is basically the same as a brushless servo with no gearbox.
One moving part and no brushes to go wrong.
except these steppers have built in gearing :p
So....doesn't that defeat the point?
Brushless motors generally are happy at high speeds.
Ah, but steppers are generally not. Are all axis motors and drives the same size? How do the axes differ in weight/mass and is gravity involved in any?
Lepton: following error with HostMot2 stepgen?
KimK: Motor theory is WAY beyond me, but from what I remember, configuring a brushless motor for low RPMs/hight torque generally kills your efficiency and, by extension, peak power output.
A good R/C airplane motor can produce more power than the largest of steppers despite fitting in the palm of your hand....at about 30,000 RPM.
Do you have a base thread enabled in your hal file?
No, only servo threads
I do define a base period, but as far as I know that's not starting a base thread
Theres a bug somewhere either in EMC or HostMot2 driver that causes trouble if an (maybe only empty) basethread is started
I saw some talk about that on message board archives
for example : loadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIOD num_joints=[TRAJ]AXES
will case trouble
I've noticed I get the following above a certain speed threshold (somewhere above 300 ipm) whenever I move, but when I glow lower (upper 200's), I only get the following errors on deceleration
Here are my files: http://pastebin.com/BVZqbAdQ
what is the maximum number of steps per second emc can produce through the parallel port?
I do have this line in my HAL: loadrt [EMCMOT]EMCMOT base_period_nsec=[EMCMOT]BASE_PERIOD servo_period_nsec=[EMCMOT]SERVO_PERIOD num_joints=[TRAJ]AXES
looks like that might be troublesome?
So just ommit the base period altogether?
leading to: loadrt [EMCMOT]EMCMOT servo_period_nsec=[EMCMOT]SERVO_PERIOD num_joints=[TRAJ]AXES
I took it out, tried it on the machine, same problem :/
following error on deceleration
What is you maxaccel?
We're trying to get a more sensible value for that right now. Attempted to calculate it based on the mass of our moving parts, and max torque of the steppers (including gearing)
That lead to a very small value, so I'm skeptical of our math / thinking
try setting it to 0 (and keep the previous patch ie no basethred)
trying that now
You know, Seb had recently halscope-captured a case like that, where the decel had issues, but much stronger and longer accel did not. You might try that on your machine since it's simple to try. I might still have his screenshot.
Did Seb ever resolve that case?
pcw_home: Setting max accel to zero solved it
Though it of course leaves our system operating pretty rough
Any guidance on tuning the acceleration value to something appropriate?
Since you're running a stepper/servo, I might have to take back what I said before. You might have to tinker with the internal tuning values on the drives, since hal/emc doesn't really have control of it.
yeah, you're (unfortunately :p) probably right
You can coarse-tune classic servo drives with a square wave (not too big at first) and the motor decoupled. You might be able to do something like that by running (and re-running) simple gcode programs while making adjustments to the drive?
Hmm, so you mean basically trial and error moving the axes around?
Sure, adjusting one axis at a time. First the drive, then EMC. (Lather, rinse, repeat?) General guidance? Heavier axes (assuming same size motors & drives) will need more time to accel/decel without losing steps. Do the lightest axis first, and then estimate the weight/mass and expect proportionally slower acc/dec. Maybe max_vel a little bit lower too. (Proportional? Or is there a square in there? Oh, nevermind.)
That makes sense
I was thinking about calculating the kinetic energy of the moving parts of the machine at max velocity, and then dividing by the max electrical power of the steppers
Which would yield time
and than apply a saftey factor, and use that as a starting point
Sure, if that helps. I'm not sure I'd bother though. A dial indicator and a start/end point will be more useful, I'll bet. And you can write little gcode programs (two lines, out and back) so you get a repeatable input from EMC to the drive.
Oh, you have encoders, maybe you don't need a dial indicator?
Just open up the following errors while you're fooling around.
And watch your position on halscope.
Maxaccell of 0 should be fine
There you go. Thanks, PCW.
You might keep an eye on your following error on halscope too, to make sure it looks "reasonable".
( this is assuming you are running in pure step mode)
Stepgen Maxaccel of 0 and no base thread should get you a working HostMot2 stepgen system
(of course the trajectory generators accel must be set to a reasonable value for the mechanics)
And speaking of mechanics, I seem to recall that using backlash comp alters the acc/dec in some way, so if you can live without backlash comp for now, that might help too.
If you are not using the encoders and only a stepgen, there are only 2 ways I know that following errors can happen:
1. Insufficient headroom for the stepgen
This can be because the stepgens maxaccel is too low (if used (non 0) it should be a little higher than the axis accel)
or because maxvel is too low ((if used (non 0) it should be a little higher than the axis maximum velocity)
2. Because of the empty basethread bug
* alex_chally has tapered thrust bearings
for cheap on ebay too
[02:26:06] <alex_chally> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=150392422221
i had following errors on my first emc setup - that had open loop steppers, i just had to increase stepgen_max.... values for the headroom and that solved it
Thanks for those explainations, pcw_home
We've been doing accel tuning for the past 30 mineuts, and are much much happier
The machine is moving fast and smooth :)
Indeed! EMC is lots of fun sometimes
Glad to hear it's better.
especially when its finally up and running :)
I expect to be cutting within the next two days
you planning to use those for your xy screws alex_chally? or they for something else?
Still need to get modbus running via classic ladder, which I'll hopefully accomplish tomorrow
what type of machine is it lepton?
A 3 axis router with a 5hp spindle
Started life as a shopbot
We've done a lot of mechanical changes / upgrades to it over the course of this year
and of course now we're running EMC, rather than the incredible awful shopbot windows based control software
i have the mount drawn out and everything
tomorrow I am going to do some careful measuring of the bearings and then rough out the stock for the bearing holder so I can put it in the CNC and not take up too much time
machine that, fit everything together and then do the Y axis ball screw end
and I *think* that I can do the same design on the X axis for the bearing block, but I have to extract the X axis screw and bearings from it's bearing block
it would be totally boss if i could do both parts exactly the same
are you in canada by any chance, lepton?
then the last big hurtle with the physical parts is my X axis nut holder, which I think I am going to manufacture from scratch
it has a split holder for the X axis leadnut, but the holder is too small for the ballnut and half of it is integral to the casting
as opposed to trying some XY interpolation bullshit on the CNC I am going to face the entire thing off and built a new one from scratch out of aluminium and secure it with screws and some dowel pins
i guess that way you can even shim it to height if it happens to be off
madsci44, yeah, I figure I can tolerance it to +0 -.001 or so and do really well
it might be worth it to make it out of steel so I can surface grind it really square..
probably won't matter fitting it on the first time, but when it is taken apart some day and has to be set up again I bet it would make life easier
i need to get off my behind and do my mill, i am still arguing with myself over how to orient the x-motor
I can see the end of the rainbow with mine, it is nice
well, at least the 2 axis rainbow
after that is finsihed and working I am going to start on the Z axis mount and stuff
which is something I have not even started considering how to do really
madsci44: I'm in Colorado
*goes back to control box wiring
oh god i forgot the z axis - im arguing with myself about that too - it would be ok if one of me won decisively
madsci44, yeah, there are big questions like gas strut counterbalance on the knee vs moving the quill
and then little questions like if I go for the quill, how the fuck do I hold the entire thing onto the head?
i think im going to skip the knee until i see how well the quill turns out
is your head a standard type with the feed stop screw in front?
well, close to standard type, but that is where the feed stop screw is
there are holes tapped on the other side that are the covers for the feed gearbox
I could use those, and then do some 1/4-20s or something in a bolt hole pattern around the bottom of the head
yeah thats very similar to mine
how do you deal with quill v knee in emc2? do you have 2 .ini files?
assuming you've got both motorised of course
I'd make an XYZW config where W = knee
i dont - yet - so not sure
you could put your tool lengths on W and just move it when you change tools - that would be awesome
cradek: Is W like a bastard step child sorta thing?
oh wow - mechanical offset heh
T1 M6 G43, G0 W0
(knee moves down because this tool is longer)
I'd make positive be knee-moves-down
so like an extension of Z ?
well they're separate axes - they just happen to be parallel
so Z++ =)
you'd use it just like a knee (move it when you have to for more clearance)
I wonder if you could get a cam program to work well with a setup like that
so the knee isn't used with a short quill for heavy jobs?
so the position of w ads/subtracts from Z ?
cradek, it makes perfect sense to have Z up on the knee be in the negative direction, but that breaks my brain just a bit
it would be easy enough with my CAM program ...blender
It be slick if you could setup W to operate in conjunction with tool table
Jymmm: W tool offsets are just as easy as Z tool offsets in EMC2.4
or maybe if you hit the limit of Z travel it starts moving the knee
yeah you could do funny stuff in kinematics - it might be fun to experiment with that
you could have them add
so the first 4.5 inches of Z negative happen with the knee all the way up, extending the quill down, then have it switch motors for the rest of the move
you could never interpolate like that though..
that hueristic should probally be tuneable in case someone wants to avoid wear on parts of the screw
maybe i will do my knee from the start :-)
but that might make constant whatever contouring "interesting" to do right
knee mills are super flexible and take very little space - but nobody enjoys cranking a knee up and down
my elbow agrees
ill never attempt it im sure but i have always imagined what it would be like to do everything, ram, tilt pitch, turret
I'd do the lot from the start if the axes are there
I'm finding 3 axes rather limiting already
I don't think tilt would be any use
madsci44, I don't think the standard knuckle design could be easily converted
it would be easier to make it from scratch
it would probably be a nightmare mechanically - youd have to use some sort of strut setup - and yeah im not sure what it would add for capability - would be fun to watch tho :)
yeah I suppose a trunnion that adds 2 axes to the workpiece is more useful in reality
accuracy holds better with all that stuff locked down and stay that way after tram
MattyCNC, not to mention a more simple build, with less weight being shifted around
yep, and servicable/replacable
are there any known setups in EMC with XYZW cradek?
there was a nice ancient lathe on ebay that would have fit on my bed nicely
MattyCNC, I keep thinking that finding a little 9x20 lathe with scrap ways would be perfect
I've seen video of a XZW (two turret) lathe
I've never personally seen a cnc-controlled knee
chop the ways off and mount the headstock on my table, build a trunion that uses a MT4 taper or whatever
ah without the ways, a homemade headstock would do the job
there wouldn't be much room for tapers on my machine. I need flat rotaries
my problem is that this looks so sketchy
[03:26:57] <alex_chally> http://www.youtube.com/watch?v=M_AKhV9-eIk
it annoys me tha thte screw needs to be that far away from the quill...
geo01005_home is now known as geo01005
thats about the layout i was contemplating - there is some good solid casting there to support the bearings - despite it being that far away i am leaning to think i would get better accuracy that way than any other means
theres too much slop in all the other existing feed mechanisms
madsci44, I can't think of anything either, but I still don't like it :-p
no matter what something is gonig to be "way out there" i figure
for my x axis motor im down to either leaving it straight out - adding to the overhang, or moving it down lower and having the back face the knee where it can tuck slightly under the saddle to avoid reducing x or y travel, i dont want it at the front because sometimes that surface is handy too
piezo gibbs in the slides :)
piezo gibbs? lemme guess they vibrate to stay tight but not wear?
nope, vibrate to move the axis
how do you control the direction?
madsci44, I would tend to have it stick out in and use the extra floorspace
there's a good vid on YT, I'll try to find it tho I can't play it
madsci44,it would be super annoying if it was in the way some day...
also would probably lead to a smaller belt length, which is mope bettah
yeah thats the only reason i didnt go for the underneath - yet - maybe :)
Y on mine is really easy - motor below back hanging under the edge of the front of the knee
you are doing your knee ? are you keeping the bevel gear in the drive train?
nah, I am going to do the quill
actually - hah youtube is a handy place to see setups - i never considered looking there - this mill is one size down from mine http://www.youtube.com/watch?v=78KUDWnz-WU&feature=related
he did the knee too on it
SWPadnos: cradek jepler jmkasunich alex_joni http://www.warp9td.com/
ries_ is now known as ries
the SS has been around a couple years at least
Fox_M|afk is now known as Fox_Muldr
Fox_Muldr is now known as Fox_M|afk
alex_joni: No USB love?
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2010-07-20.txt
Not called the Universal Serial Botch for nothin
pcw_home: did you see the board?
Didnt look closely
* skunkworks still has to order one more servo interface board (7i33)
Jymmm, you have to realise that board does not work the emc way
archivist: Besides the emc doesn't support USB?
no because that has the motion on board because you cannot run that over usb
Right. Is that such a bad thing?
this argument has been gone over **** times in here
The argument should be referred to by number to save time (which reminds me of an old joke)
archivist: I'm not debating anything, just asking
w00t, found a 512Mb stick of RAM for my box :) That should help with latency, no?
Jymmm: flexablility and realtime control of things like FO.
elmo40: probably not....
The basic problem is that USB is not very good at realtime so for example a step generator like the SS
has to buffer motion itself meaning that EMC no long knows the actual axis positions anymore.
This makes operations that must know real time position like threading or rigid tapping
(or anything that requires smart reaction to real time events) no longer possible without moving
the locus of control to the external hardware device. (which has many bad side effects)
most could be dealt with in the usb2 synchronous mode except for an idiot plugging in a new device while running
Yes the multiplexed nature of USB make it unsuitable as well
Well, I underside the USB polling is such a pita, I thought it was just being used here for comms, not so much control
should be fine for non-realtime but no-go for motion
Well, it must work at least in part
for systems without realtime control sure
So it's on a delay (so to speak) ?
yes, to keep the step motors running smoothly the SS must buffer some amount of motion
Is that such a bad thing?
No its all fine until you need to interface with realtime I/O like a index pulse or limit switch
Isn't that what the FPGA is for?
unless all the other io and motion is on the usb board
and then its not emc anymore
So not much different than a mesa board?
just use a pcit has motion so is different
it has motion so is different
Well I dont think I would like to move EMC o the FPGA (Much less developer friendly environment)
mesa board doesn't queue motion, or know about homing or probing or threading or feed override or arcs or ...
We have buffered motion configs (USB as well) but they are not a good match to EMC
the mesa hardware when used with emc is basically encoder counters and output control (pwm i/o or +/-10v)
Well if EMC was moved to FPGA, there would be less demands on PC requirements, no?
with mesa hardware - there is less demand on the pc.
(without moving emc to it)
9 axis motion is not trivial to stuff in an fpga
yes that's pick-your-poison setup. if you have externally queued motion you don't need realtime in the pc, but you lose the benefits of realtime in the pc
Yes any hardware that has step generation and encoder counting will reduce the latency requiremants to the 100s of uSec region
suddenly can you do lathe threading with mesa? with galil? with ppmc? with stg? with motenc? with parport? are all separate questions and the answers depend on ten manufacturers getting it right in their own products
history tells us those manufacturers have a hell of a time getting it right - read web forums to see that :-)
well you could keep the real time benefits of a PC and have EMC in both the FPGA and PC of you have a very low latency IO
but it's not built into PC chipsets so it's expensive
Can someoen tell me what the USC board is doing? I know it's treating the steppers as servos, but I still don't get the benefit of the board.
we use ~1uS response IO in HPC nodes but the FPGA's are a few hundred $$
Jymmm: mainly - it moves the step generators into hardware.
skunkworks: Ok, so doesn't that have some cacheing too?
no, it has a step pulse generator whose stepping rate gets reprogrammed every servo cycle (millisecond or less)
So EMC sends the "command" to the USC, and the USC carries it out, instead of the PC generating the pulses?
external hardware is good at generating nice even step pulses at whatever programmed rate
Couldn't that be done with USB ?
because you can't talk to something reliably every millisecond over usb
if you look at it as a servo setup - (like with mesa hardware) emc sends out how fast the servos should go - and reads back in the actual encoder postion every ms. Then adjusts accordingly. emc doesn't have to do the grunt work of counting encoders or generating the pwm to run the servos (or +/-10v)
Even with USB voodoo?
I have no idea what that means
Non-polling mode (?)
nice even pulses, yes, knowing where the motor is at any given servo period, no
yes the USC looks exactly like velocity mode servos to EMC
Jymmm, usb is dead for indeterminate periods when other devices are around
Jymmm: on paper - you would have better luck with rt over network. (no one has done anything with it yet though)
PC IRQ response is like a German or Japanese schedule, on the other side of USB it's like a US Amtrak actual time
skunkworks: Yeah, I've been bugging SWPadnos about that for a couple of years
the door to the station is a revolving one and its too small
plus only 1 door works at a time...
other train loads stop your information
I guess what I don't get is how everyone else is using USB and getting away with it.
we have such awesome products available that work beautifully - I really don't understand the whining about usb
they dont have 9 axix realtime control
and THATS what I'm trying to understand.
archivist: Do you?
Does anyone ?
I have 5
you could have a FPGA off of Hypertransport, we could strip down Extoll http://www.hypertransport.org/docs/wp/UoM_HTX_NIC_02-25-08.pdf
Jymmm: it works badly. read the mach forums for more details.
this would respond fast enough
Ive been told of the problems over here in the UK as well
cradek: Well, ok. But what about things like a printer or document scanner? (the only usb things I could thinkg of off the top of my head)
but now you're into AMD server hardware with multi CPU sockets
skunkworks: really? won't alter any latency? I find that hard to believe... it will put more programs into RAM now. That should improve latency. doesn't it? LOL one way to find out... boot it up :)
what about them? nobody cares if their printout comes 10 milliseconds late
Jymmm: you're just not paying attention if you think machine control is like printing or scanning documents
usb is great for those applications
cradek: Like I said, it's the only USB device I could think of with motors.
the computer doesn't control the motors in a printer.
you see synchronous mode in webcams but do you notice the steppyness of the image
try and stop your USB printer in under 1ms :)
I DO get it that USB sucks. But I'm not thinking of usb for "direct drive" as much as "those vacuum tubes at the bank" it just transports the commands.
Kinda like Remote Desktop
the communication in EMC is realtime
any news on the EMC port to the beagleboard? Did the RTAI guys ever get ARM smoothed out?
RTAI and ARM are not already working?
ARM cortex8/9 smoothed out
elmo40: last I heard they had some earlier ARM cores working with RTAI ARM7, strongARM
Does anyone have a spare GPIO card they want to let go for cheap?
the only ARM SOC's I was considering for EMC are the Marvell devices since they have PCIe, openRD is a possibility http://www.globalscaletechnologies.com/t-openrdcdetails.aspx
L84Supper: how much is that?
Jymmm: the openRD's are ~$149 for the board and ~249 for the client but there are some low cost tablets and boards coming out of Taiwan soon
I had demos of ARM 7-10" tablets with PCIe that were ~$90 at Computex
[16:51:39] <L84Supper> http://chicago.craigslist.org/chc/bfs/1849040014.html
how old is this?
[17:11:47] <L84Supper> http://cgi.ebay.com/WEBB-VERTICAL-MILLING-MACHINE-3HP-TABLE-10-X-50-/320561450814?cmd=ViewItem&pt=BI_Mills&hash=item4aa2f38f3e
<--- needs another mill fast
hi all someone can help me to update from 2.3.5 to 2.4.1
[17:40:18] <IchGuckLive> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.4
Click "Mark All Upgrades" this will update the heaader the image and 415 packets
can i only update the EMC2.4
does anyone know what the difference between normal and orthogonal is?
normal all degree allowd ortogonal only 90-180-270 and 0
i wish my brain could remember maths
if i could just remember it i may be able to understand it
i thought normal to a vector was the vector perpendicular to it
in CAD its only alowd to draw horizontal or vertical
and I thought orthoganal was the same, but some now starts telling me that orthogonal means that the product of the two is zero...
my eyes start to glaze over
dot product of perpendicular vectors is zero
in vector space there are different kinds of 'products'
so dot product of 90-180-270 and 0 is also zero?
so two vectors can be orthogonal but not normal?
I don't understand that question
usually you say 'normal' when you're talking about a vector vs a plane
I think the problem is i dont understand the math well enough
so my question is pointless any ways
i managerd the upgrade to 2.4.2
yes mashine is up and running after new Stepconf
also Sf(x)f(x) = 0 for orthogonal functions (like sin(x),cos(x))
the normal vector is orthogonal to any vector in the plane
I am trying to work throught some robot arm kinematics, and I am begining to think its beyond me
its 10 years since i done any math
and i was pretty poor even then
orhogonal is always 90° to normal
moop did you look at Wipipedia
does anyone want to write some robot arm kinematics for me?
if mulitply to orthogonal vectors yo always get a zero
wrong: 20:09 < cpresser> orhogonal is always 90° to normal
i wanted too write: 'tangential is always 90degrees to the normal'
moop how many Pivot's
I am going to check wikipedia
dot product multiply...
6 pivots, but strange control with only 5 motors
6 in the arm or only 3 to position
4 with the main rotat
its robot arm but there are chain drive such that the angle of some linkages stay the same relative to the base regardless of what happens at intermediate links and joints
its sort of looks like a puma arm
I think I am totally confused
[18:16:13] <IchGuckLive> http://www.wescottdesign.com/articles.html
that site was of no help to me IchGuck
I think I need to go away for two weeks of intensive study, maybe then I will know enough to realise I should give up
moop do you got it ?
sometimes i begin to think there is an ubergeek that dwarfs my intelligence, writing the code behind this software
I run emc-2.4.0, but the program not runing correctly because the emc2 require that the my system it is overloaded of programs in execution , in this manner the promgram emc2 run correctly without axis
salvarane: I don't understand, can you write it other way?
I rewrite this
(I'm also not native english speaker)
when I run the emc2-2.4.0 , the programma not runing if my system is idle state, but the the program runing if the my system is very busy
because thi situation
the message is the tipical : waiting axis or some thing like
write in your native language and use google translate
salvarane: off of the top of my head... Remark out (#) the nlm line of your ini file. http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UPDATING#emc_nml_changes
thanks , I read
sorry , but into my configure file stepper_mm.ini , not isn't the option " NML_FILE = emc.nml "
I think this problem is due to , comunication between the sub system realtime and emc2
but, I can mistake
I remeber tha I compiled the kernel 188.8.131.52 patched with rtai 3.8
ries_ is now known as ries
[20:30:54] <skunkworks> http://www.youtube.com/watch?v=LJYhz5aTMnA
my LJ4 was faster :)
not as cool though... ;)
that vismach model is pretty useless ;)
I saw that ;)
very cool overall though
that guy has a few different configs..
Tripod and such
seen some of the other vids
this one is also a tripod, even if there are 6 motors
Fox_M|afk is now known as Fox_Muldr
When using PWM for spindle control, is that typically 0-10VDC?
pwm is typically 0-100%, voltage being irrelevant
except when PWM is used to drive analog drives that expect 0-10V
in that case some hardware circuit is used to construct the 0-10V out of the 0-100% PWM
whee.. first delta robot run by emc2: http://www.youtube.com/watch?v=2JbOUqE2eFI
I have just finished some parts (of my own design, nobody else to blame) that are positively begging to seize together. 80mm x 1mm pitch thread, aluminium into aluminium.
With EMC screw threads have now become just about the easiest way to join turned parts. Especially as it can do any size and any pitch that takes your fancy
i been researching about my idea the other night and seems someone has already done it!
Which idea was that?
[21:56:31] <moop> http://www3.interscience.wiley.com/journal/114194947/abstract?CRETRY=1&SRETRY=0
now someone just needs to implement it in emc
That session only seems to work for you
It works without the stuff after the ?
Genserkins is pretty close to what is described ther.
I will have a look at genserkins
what would be really nice would be a cad app to layout the joints and geometry of any cnc machine
which of the sample configs uses genserkins?
I am not sure any do.
Is anyone using running an arduino for IO in EMC 2.4.2, a 'la: http://axis.unpy.net/01198594294
i try "grep -re genserkins configs/"
I suspect that once you have abstracted as far as the D-H parameters that a CAD package would only confuse matters.
I suspect that the only way to understand genserkins is to look through the source with a stack of textbooks.
looks like the puma560_sim_6.hal uses genserkins
moop: in master there is puma 560 config that use it
i found that already micges
I started writing a shell script to do the generation, but i think it is probably all beyond me
I would like to find a copy of the pascal source code referenced here: http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V4P-47XNPNF-S&_user=10&_coverDate=12/31/1991&_rdoc=1&_fmt=high&_orig=search&_sort=d&_docanchor=&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=c0f3d0bf5ac5ba2577d7017541573733
moop: only $31.50
Hmm, so it seems that the Arduino code for HAL doesn't work in current versions, based on a forum posting, and Jeff posted a link in response, but the link is dead
Anyone have any leads on what the status of that is?
Try this link?
[22:15:34] <andypugh> http://firstname.lastname@example.org/msg02827.html
That is what the link in the first message should be, it seems to have been oddly mangled. (I looked through my mail archive)
Sorry for the drop out, my neighborhood doesn't have any broadband and is stuck in 1990
It seems that the Arduino HAL interface is non-functional right now
You could an email to the mailing list to see if jepler has any ideas there.
That's a good idea, I'll try to motivate myself to do that after I try to solve it myself some more :)
a "grep -re rduino" in the source tree shows no mention of and arduino interface??
is there one available?
I would think a pic would be a better chip to use?
Fox_Muldr is now known as Fox_M|afk
It's a "contributed component", something you can download and use with EMC2, but not part of the distribution.
is it used as a usb interface?
And the advantage of the Arduino is that it is all there on the board, power supply, programming interface, break-outs to wires etc etc.
[22:34:03] <andypugh> http://wiki.linuxcnc.org/emcinfo.pl?ContributedComponents#Arduino_ADC_PWM_and_digital_I_O
are arduinos available with more io lines?
Yes, the Mega has several more
[22:35:44] <andypugh> http://arduino.cc/en/Main/ArduinoBoardMega
a while back I was considering using a pic18f as a usb interface, but never got further that looking at the pic usb code
That is rather my point. You can buy an arduino, download the GUI, and be twidding LEDs inside 10 minutes
It uses a usb > serial convertor on the board
Yeah, it's a great idea
whats the price?
I think the HAL module communicates via serial.
$20 all ready to go with no extra components?
Sorry, $30 (all ready to go)
[22:38:25] <andypugh> http://www.adafruit.com/index.php?main_page=index&cPath=17&zenid=588411c8a70e828f7a67223aa30be3a7
the pic i was looking at was $4 but needed quite a few more components and required programming to use usb
Arduino is open-source, so there are many legitimate copies like Freeduino and Seeduino. They seem a little cheaper
[22:39:41] <andypugh> http://www.nkcelectronics.com/arduino.html
anyone ever try one of these retrofit kits? http://www.servosource.com/Servo_II_Brochure-2pg_070604.pdf
I'm getting hungry, I have to go get some dinner
L84Supper: Looks good. Doesn't look cheap.
somebody has one for $500, never installed
moop: I am using an Arduino to convert resolver to quadrature as my servos are resolver.
probably worth it for the motors and mounts, I'd just use Mesa cards instead of it's controller
moop: In fact, I have used the Arduino to do the resolver conversion and the phase signal generation: http://www.youtube.com/watch?v=oyeJfNg3NfQ
ey cradek got mah probing thing working
looks like our switch is +-~.002mm at the moment
thanks for your help with the probing
nice switch, what type is it?
Incase anyone is interested, it turns out there's a pretty well developed Modbus Master library for the Arduino
[23:03:44] <lepton> http://code.google.com/p/modbusmaster/
I might try to use that instead of modbus in classic ladder or gs2_vfd