as linux isnt windows it usaly works to just swap hd =)
Well, no *I* won't backup your data, you'll have todo that yourself =)
hold my beer so i can slap my knee
only thing that sucks is i'll have to buy ram
* Jymmmm drinks aa-danimal-shop beer
thats ok, it was mostly backwash at this point
aa-danimal-shop: like I said earlier, you can put (AXIS,stop) in your program to make AXIS stop generating preview
it's definitely the preview of your huge program causing you the pain
cradek: those htmlk docs were whatr i needed, i had oinly ever seen the pdfs
cradek: i put it in parenthasies
and as far as the preview, do you mean the actuall toolpath preview on the display, or something else?
doesnt seem to do anything, not sure if i'm entering it right
ugh and this is the second time z didnt retract and it rapided through the part
wtf is going on
didn't retract how? what did you do right before?
it's loosing position
it thought it retracted, but it didnt
i was running a program
and it didnt retract when it was supposed to in the program (this is definitely a good program)
crap i'm not sure what to do, i need to run these parts
Jymmmm is now known as Jymmm
wooo my machine does what it is suppose to now
only thing left is to get the limit switches set
i need to figure out how to make the limit switches being ignored during homing,
and i have to figure that out through stepconf
cradek: if you're around, it seems to screw up in about the same spot in the program, but not the same exact line everytime. Near the beginning. The program is right, just the z axis is having issues retracting or feeding. I cut the program in half and made it into 2 seperate programs and so far so good.
Ok, you're about to buy your first SET of resistors, what qty of each value should you get?
analog or digital or mixed tech? one-off prototyping or small production? hellwifit just buy a digikey resistor assortment of cheap 1/4 5% carbons and light a candle
I'd say get a set of 1206's and be done with
wow theyre < a penny a peice, whats the budget & expectations? yaego ok?
through holes are rapidly falling out of use; they are more useful as little probes then resistors
i didnt even think of surface mounts, too old school, how can you build bob pease hairballs with smt? ;) no broadway :(
same way you stack blocks together
you get little mountains with 2 resistors tacked on top on a pad setup for 1 resistor or double stacking them
huh. good to know. so your boards... the only holes are vias?
i bet the idea of top = component goes away too
I do protos on unetched boards
but for real PCBs, yeah... the only holes are vias
makes PCB milling somewhat practical as it cuts down on the pile of broken tiny bits
well for old skool, the vellman asst seems good bang for your buck at 11$ http://www.bgmicro.com/index.asp?PageAction=VIEWPROD&ProdID=9464
i finally found a reccomendation.. get the 'e12 series' and at 8$ for the vellman pkg http://www.mcmelectronics.com/product/80-6665&CAWELAID=220591396
[03:52:58] <Valen> http://www.geekologie.com/2010/01/want_tron_legacy_lightbike_in.php
aa-danimal-shop: that sounds serious - please email me the program and as much information as you can think of that might be relevant for reproducing it
aa-danimal-shop: I wonder if your overheating something
also how much ram is in your machine?
if you run sudo lshw it'll spit out more than you want to know about everything in the machine
and yeah swapping hdd's isn't a problem, the only thing you need to do generally is re-set up X
LawrenceG: soem serious lag there
sleeping... just junked 16 computers.... sorting the remains
LawrenceG: If you didn't use an angel grinder, you didn't do it right
now what to do with this pile of floppy cables?
LawrenceG: When you made your light bar, how did you setup the LED's?
only 2 50pin ribbon cables.... 4 in series with a resistor then switch with a fet across 13.8v buss
LawrenceG: remember what fet you used?
180 1/4w resistor... fets were ... some onsemi samples .. logic level gate drive... let me see if I put it on the schematic
18 amp, 60v logic level
60v@15A... Did you setup like 48 LEDs on one fet?
60 leds/fet (4x15 array)
LawrenceG: did that give you dimming too?
only about .375 amps/array segment .. did not due dimming... mostly just flashing on/off
375mA per 60 LED's?????
each series string of 4 is only 25ma * 15 rows
0.025*15 = 0.375A and there are 8 arrays, so all on, the bar draws 8*.375 = 3 amps
The WHOLE bar, not bad
LawrenceG: Did you make a PCB for the segment?
it does get warm, but not hot... it would be fun to make a bar with some of the 1w or 3w leds
no, the leds are press fit into 3/16 holes in a piece of plexi/lexan/??? that I drilled with emc (on topic)
5mm LED fits into a 3/16" hole?
hand drill would have looked horrible
yes.. very nice wedge fit
5mm == .1969, 3/16" = .1875 <----- Um
I had sore thumbs after pushing in all 480 led's
the led's are tapered
Alright, I guess I'll believe ya =)
about a 0.009" fit... but I think 3/16 drill drills a little large
LawrenceG: Heh, even better (cheaper and brighter) http://www.dealextreme.com/details.dx/sku.5752
LawrenceG: Oooops, read the coments on that last one.
[05:05:15] <LawrenceG> http://www.dealextreme.com/details.dx/sku.2394
I have seen these on commercial light bars
LawrenceG: how thick a plexi did you use? I'm thinking 1/8"
it was about 3/16.... caliper is out in the shop
hey.. got to sleep... early morning for me.... goodnight
g'night and thanks LawrenceG
[05:39:22] <frysteev> http://www.flickr.com/photos/34832969@N03/4386824944/
tomp is now known as tom3p
What do you all think of this http://imgur.com/ue2ay.png
Should be fast enough
looks great. what kind of machine?
Intel BOXD945GCLF2 Atom 330 Intel 945GC Mini ITX Motherboard
Not bad for under $200
yes, very nice
Note to anyone running SMP you really really need the isolcpus=1 in grub
and hyper threading off.
where's pin 1? if you find a single row connector with a latch. and you face the latch ( latch is seen behind the pins), is pin 1 to left or right?
( no chamfered corner on silkscreen, no pin 1 on silkscreen... and a dozen such connectors on pcb )
( no chamfered corner on silkscreen, no pin 1 on silkscreen... and a dozen such connectors on pcb )
cradek: what info do you need from me besides the program? ini and hal also?
I kind of doubt those would affect it, but may as well send the ini
the program is about 4,000-5,000 lines
worked fine once i split it in half
i tried the (AXIS,stop) thing, but i didnt see any change, what exactly is it supposed to do? It was still showing the toolpaths on the screen. maybe i entered it wrong into the program?
couldnt find any info on it on the wiki
maybe I spelled it wrong, one minute
[16:16:00] <cradek> http://www.linuxcnc.org/docs/devel/html/gui_axis.html#r1_11_7
hm, the docs are inconsistent about whether a space belongs there
maybe i should try the space
just sent you an email with the files
JT-Work: how's the build going?
aa-danimal-shop: I figured out what was wrong with the Hardinge!
the turret encoder has a short and was drawing the 24 buss down
took me and my other brother John a half an hour to trace it out this morning
did it damage anything?
no, the power supply just drops the voltage down
when it is shorted
I'm assuming that was the original problem with the lathe LOL
i had a similar issue with the original control on my hardinge, a fan was seized and would do the same thing
kept popping fuses actually
so did you fix the short?
cradek: the space didnt do anything.
then some goober snot took a bolt out of my tower so it fell down yesterday so no internet
I've not dug into the turret yet
I just unplugged that wire for now
cradek: how would it work in parenthasies? wouldnt it just ignore it like a comment?
JT-Work: so is it moving yet?
no, I am in the process of checking all the I/O first... the outputs work and now the inputs are ok except for the turret encoder
aa-danimal-shop: some comments are magic
aa-danimal-shop: maybe it's just broken - it should stop the preview
once I get past that ti's on to the drive enables
LOL, I can move the X and Z with an allen wrench on the end of the ball screw
the hnc is like that too. with the original control, you had to do that if it got on a limit switch.
(and there were no soft limits, of course)
yea they move easy
do you have any idea what turret encoder looks like?
magnet over a bunch of reed switches
it's not an encoder, it gives absolute position output in four bits of binary
(and yay for that)
yours might be different of course
yep mine is the same
cradek: i just realized the (AXIS,stop) is hiding some of the preview, but not all of it
it only hides the part that follows it
i put it at the top of the program
but it doesnt hide at least 3 tools
aa-danimal-shop: does it come back on at tool change?
i take that back, looks like it just doesnt hide 2 tools, the first tool and maybe the 4th or 5th
hi channel i finally managet a install of emc2 on xubuntu
now i'am up and reading all the infs to get startet with it
aa-danimal-shop: got your email, thanks. I will look at it tonight.
no problem, i finished the parts, so no rush
thanks for looking at it
aa-danimal-shop: I forgot you have steppers. when Z didn't retract did you see an incorrect path in the AXIS backplot, or was it just wrong machine motion?
i cant recall
i know at least one time the home positiom moved when it didnt rtetract, meaning it thought the stepper moved, but it didnt
unless you see it in the backplot (or reproduce it in sim) I think the chances are good it's just steppers doing what they sometimes do :-(
another time it didnt retract it did go to the proper home position
it just struck me as odd since i've never had this problem and it was only for that one tool
yeah that does sound odd
the rest of the program worked fine, including a peck drill
do you do something special with the tool? single step it? change feed override?
i peck drilled 72 holes without issues
it's the first tool in the program i sent you
all it's doing is milling circles
and it worked fine once i split the program in half
where in wiki is the list for Driver Type coudt not dind them
i gotta go rent a carpet cleaner, brb
found it http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Stepper_Drive_Timing
linux command to query executable file type?
the machine I just built says that a file doesn't exist, while I'm looking at it. 2 other machines run it fine.
one sets a permisstion to make executable
it isn't an execute issue.
and have your path search where you are as well
I'm in the directory. I type ./setup and it gives a "file not found error"
Not that exact error
what's the file
aa your cpu's are back in the mail headed your way
the xilinx 9.2 setup
clytle374, note linux is case sensitive
I need to know if it is an .out type, I don't have that in the kernel.
it's probably a script, not a binary
aa-danimal-shop: your cpu's are back in the mail headed your way
how can I change the jog speed if I have it turned all the way up and I want it to jog faster?
ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV)
No 32 bit support?
tlab: you need to add a max velocity line in your ini in the traj section
is it maxed at 60ipm?
yup ok.. I had one I just changed it
was maxed at 24ipm
youch, thats slow
Not to be lazy, but what is the package for 32 bit support in ubuntu 9.10? It is really cold where the computer is, so I'm working it all remote.
lytle is my last name!
maybe you're long lost twin brothers like Danny DaVito and arnold schwarzenegger in that movie
I'd be danny davito
I'm short and no muscle
ok so now when I try and move the machine faster I get a join 0 following error
I'm not short, but Schwarzenagger isn't a good description of me.
stepper or servo?
do you have your max velocity set correctly for each axis?
are they all suppose to be the same?
depends on the machine
but you need to set your max acceleration and max velocity for each axis
ya but I'm trying to figure that out now
because the stepconfig is limiting me to .3 or .4 max velocity
yea i dont know anything about stepconf, i have mesa boards so i cant use it
JT-Work: so do you think the poor latency was an issue with the 5i20, so i have at least one good processor coming back to me?
no, I think both cpu's are good
u tried the other one i sent too?
the speed is not correct that I marked on the package as after I marked them I realized that was the max that my board would do
did you check bios for a muliplier?
i know mine there was a way to increase it
ugh so looks like it's a limit of the parallel port... 0.4 velocity for that port low?
tlab: wish i can help, but i have no idea... i think it has something to do with your latency and the step length
so if it's that low, chances are you cant get much faster unless you decrease your microstepping, decrease your latency, or get a Mesa 5i20 or something
* jackc decreased microstepping to get faster rapids
but like i said, i never used a parport so i'm just making a slightly educated guess
hope you have a nice time(painless_
im running in problems configuring my encoders
anyone available for some help?
will be much apresciated
Just ask, maybe someone will know.
i have a industrial servo machine with incremental encoders and Mesa5i20 interface cards
in the process of retrofiting im finding dificulties in setting the right machine size
erjos: are you using the m5i20 driver or hostmot2? how many lines on your encoders?
with input_scale set to 6000 or even 60000 i have satisfactory results in axes movements
it is a 500 pulse encoder
which version of emc2 are you running?
the problem is that i got false readins from the encoders.
they say my machine is for example 1 mm long,but i have a 2000 mm long machine and min limit and maximum limit set to o and 2000 in the ini file
if i put the machine scale to 600 i got a for example 100 mm axis length
sory input scale for x axis
are the encoders on your servo shafts?
1:1 coupling from servos to leadscrews?
5 mm/thread leadscrews?
i got aprox 32mm of travel per one revolution
whoa, that's a lot
yeah,its a fast machine
distance thread to thread is 32mm for the x axis and 26mm for Y
cradek, it was the 32-bit compatibility. I really didn't expect that to lead to a file not found type error.
ok, so: 4 edges/line * 500 lines/rev * 1/32 rev/mm = 62.5 edges/mm
the problem is that when i put the inputscale to 62.5(which is my aproximative exact calculations and axis length matches) i have no way to move the axis with any PID combination
hm, why not?
it makes the joint follow error
i have tried all possible pid combinations but no way to move the axis
with 60 000 input scale i got perfect moving but the machine size is all wrong
there was a bug in i think you should set the input scale to the number we both calculated, and then tune the servos
that can be tricky, but it's the right way to go
i started with P 100,nothing P50 nothing,P10 Nothing P200 nothing P1000 nothing
tried all the calculations
all the combinations
no movement at all?
just stop and joint0 following error
axis try to make just a tinny move and stops
that sounds like too much movement. For PID tuning it's a good idea to set ferror pretty high, try 10mm
you can set it back down when the axis is tuned well
thanx,just 2 sec to try
i got now a terrible oscillation (axis shaking) with P100 I 0 D 0
smth better yeah!!
make sure I, D, FF0, FF1, & FF2 are all 0
then set P to 1
2 secs again
i got good signs
with P 0.5
no more joint0 follow error
anyone into electronics and up for a small check of a schematic? =)
now it's just a small matter of tuning... maybe here is some good info (i haven't read it): http://wiki.linuxcnc.org/emcinfo.pl?Tuning_EMC2/HAL_PID_Loops
i have read this
thanks a lot
no problem! glad it's starting to work for you :-)
any starting numbers ?
what you think about normal PID parameters that should work?
it really depends on your machine - you'll just have to experiment & play with it
There is no normal when it comes to PID
And just for an idea,the Ferror rate should be lower as possible?
once you've tuned it to where you want it, you can use the ferror you end up with, plus some small margin
but where to find the Ferror numbers,or you mean trying with ini editing?
part of tuning your servos should be to run halscope, and looking at the commanded position, the feedback position, and the following error
see how those curves relate & how then change as you tune the pid controller to minimize ferror
erjos: ferror numbers come from the design of the machine
its a 20 year old machine
and now trying,i got no more following error but no move at all from the axis
i cant go higher that 0.6-0.8 for P
at 1 it start trembling the ground
are your servo amps velocity mode?
are they tuned adequately? can you move the axes in a controlled fashion by giving them a small velocity command with a battery?
im runing them from mesa 7i33 analog servo card
yes with a 2 volt they move smoothly
even 3 volt or 4 is ok
at 2/10 volt do they move about 2/10 of the max velocity?
but higher they run like hell
how can i figure this out
well, it doesn't have to be perfect. does it seem right?
Cradek,if you read at first i explained that i can move the axis perfectly from emc with a inputscale grater than my encoder calculations
the problem is that this gives me erroneous reading in the machine size
now,with seb,i tried rising the ferror rate to 10 and i got no more joint0following error
then I think seb is right and your tuning is bad
im runing now everything with 0. margins
0.1 0.5 0.7 for P
but this seems anormal for me
and the machine dont move at all
your encoder feedback is very coarse
there should be a P setting between "doesn't move at all" and "oscillates wildly"
"doesn't move at all" sounds strange. I would expect some motion, maybe slow
it just tries to move and stops without any error
earlier you said 0.7 doesn't move at all. now you say between 0.5 and 1 moves?
nono i said at 0.5 - 0.8 it doesnt oscillate
at 1 it oscillates wild
what does - tries to move and stops without eny error mean?
but even at 0.5 or whatever preoscillating numbers are,i got no move at all
there should be a P setting where you have some (maybe sluggish) controlled motion.
before it was the joint0following error and it dosent moved
now its no error anymore but still id doesnt move again
im running at ferror =10
machine units are mm
now i will try to lower the max velocity
if the machine is not moving, you will not get a following error until you've commanded 10mm of motion
as i see from the emc when i jog in any direstion it asks for velocity 52
maybe this is too high
when tuning, you will want to move slowly at first, and as the tuning gets better, you can move faster
cradek,in fact with 0.7 P i got a tinny move
i cant see it by free eye,but the encoders read that the machine is moving in 0. mm
the axis position is incrementing by 10.1 10.2 10.3 mm when i keep the jog(continiousmode) pressed
so i got a tinny move
what does this means for you?
maybe you have it set to jog very slowly? what does the jog speed slider say?
its too slw i think
ok, 1mm/sec you should be able to see
yes,it moves o little bit
erjos: is your maxvel 52 mm/min?
i rised it to 120mm.min now
and i got the following error
Rised D=1 smth better
no follow error
you are going to have to follow seb's advice and use halscope to plot commanded vs feedback position, and following error
remember you can be 10mm away from your commanded position and not get the following error. not getting an error is only a very general start.
yeah,thanks guys,its starting to make sense now
good, you will get it
you have controlled motion already - that is a big start
thanks a lot SEb and Cradek
yeah its a lot of brainstorming in the process,i know
thanks for sharing your experience
let us know how it goes
lol, anyone need a truckload of bridgeports? http://cgi.ebay.com/TRUCK-LOAD-BRIDGEPORT-MILLING-MACHINE_W0QQitemZ360226252752QQcmdZViewItemQQptZBI_Mills?hash=item53df2867d0#ht_500wt_1182
YOU ARE BIDDING ON ONE MILL 1 HP MACHINE PULLY TYPE IN EXCELLENT SHAPE
wow, looks like they're new.
wonder if that's real.
you'll have some room in your garage soon, maybe you should buy one
I think that would be silly...
hey i used to work in Farmington, CT
Ah i heard of Acorn MFG
more than likely legit
not a bad price
too bad they're step pulley heads
yeah, you could probably have one delivered for $5k
true but they are so quiet...
i have a step pulley head on mine, and i just use a VFD
Vari speeds aren't too load when all the parts are still tight
works fine, if not better than a vari speed head
aa-danimal-shop: how's that working out? any low-speed torque issues?
seb_kuzminsky: works great, as for low torque, i can always just throw it in low gear if needed, but i can cut pretty healthy cuts with a 5/8" endmill no problem with it on the second slowest pulley and in high gear
in tool steel
400rpm's in high gear with a heavy cut are no problem
and i just run it up to 120hz if i need to go faster
i havent moved the belt once since i put the vfd on, and i only needed to actually use low gear once
[20:27:01] <frallzor> http://pici.se/p/large/TTneZowLF/
happy! a bit rough, but hey, its wired :P
I still laugh when I see how small the gecko drives are..
those are tiny
they sure are
I though they were at least twice that size
well you know what they say, it's not the side of the drive, it's the motion of the machine
but i've really only heard people with small drives saying that
just the exchange above heh
i have no clue what you're talking about
we were just discussing servo drives
i am looking for some help with emc
you've come to the right place
my friends and I are designing a 3 axis CNC controller with a USB port on it, that will look like a serial port
its going to have a PIC microprocessor controlling a trinamic controller that will driver trinamic motor drivers
how are you going to handle real time event
I'm not sure the best way to interface this to software such as EMC
emc generally doesnt like to interface to USB, because USB has so much latency
[20:58:27] <seb_kuzminsky> http://wiki.linuxcnc.org/emcinfo.pl?Emc2HardwareDesign#USB
[20:58:30] <skunkworks_> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Emc2HardwareDesign
well since the board will look like a serial port, we were thinking we could just send it move commands, i.e. X axis move +800 steps
i had suggested we use a better microcontroller with a big buffer on it
then just download a movement list
that would probably work, but that's not how emc works
but that would be more complex in circuit design, as well as programming
you can't get nearly as good motion as with emc
emc wants to be closer to the machine than that, and have the buffer of moves in itself instead of in a helper processor
plus it gets really messy quick with FO and such.
and emc does that because it's the best way to do it
emc doesn't want to be a fancy GUI - it wants to be the machine control. ;)
so you're saying there is too much/unpredictable latency with USB?
that's one problem, yes
we could add a breakout to the serial port for the PIC
but i see that the serial is just too slow
serial goes up to about 115 Kbits/second, but that's too slow - exactly
for PWM at least
we're saying you'd have to re-code so much of stuff that's already in emc that it's not worth it (and the mcu would probably be very overloaded)
is there a way to modify EMC to send serial commands?
pic could not handle the maths for all the loops
nmz787: jitter in usb is what youre worried about
nmz787: i bet you could handle the control on your uC and modify axis to just push the nc commands to the controller
but thats sidestepping EMC
jackc: jitter means what?
nmz787, emc has feedback and keeps all the axes within an error limit
error in loops and command timing
if you use an mcu, you should probably interface with it in the same way the mesa cards are interfaced
mesa = pci or epp
boy - we sound like a bunch of negative nancys.. ;)
well this is good criticism
skunkworks: well, serial really doesnt work too well for this application!
or whatever you call it
I know. :)
so EMC doesn't output MOV commands from a CAD file, etc?
is there a free program that does that?
it has a trajectory planner to get moves right
does EMC support plugins?
its seriously better than mcu attempts
for what definition of plugins, as that is a yes in certain areas
does anyone think it would be a bad idea to write a plugin that takes the trajectory plan from EMC and make conversion table to send out over serial to the uC?
(I don't think you really can do motion control in a pic, especially if you think the pic will generate step pulses)
tinyPICs do have PWM with their own clocks and pulse step counting
nmz787, you forget the feedback, its realtime
nmz787: jitter is the derivative of lag
do the mesa cards send feedback back to emc?
nmz787: what problem do you have that you think will be solved by this external device?
and emc tells them only the wanted speed of each axis?
not all computers have paralllel ports
USB is faster than parport too but its not reliable
ok, I see then.
currently parports are cheap and easy to get for any PC that has slots. EMC also has really great solutions based on PCI cards.
some of us hunt for pc that do or use a mesa card
what about laptops?
laptops are usually not good for realtime os's
nmz787: tbh my lathe uses a laptop and it works great. im not saying USB wouldnt be cool, but...
it's true that some laptops have neither parallel port nor pci slot.
nmz787, laptops have power problems that cause latency errors
you can find laptops that work, but not all do
yes, including the one im on
you have to find an another one, then
thanks, I just wondered what your goal was.
nmz787: honestly - and I hate to say it - your model fits better with mach. (there are already some similar solutions that are unfinished - like smoothstepper)
nmz787: were certainly not stopping you, just saying its not super easy
IMO it's not a good fit for EMC. But some other systems work with a motion queue setup like you're discussing. It would be cool if you could do that with/for free software, but you have a lot of work ahead of you.
one day someone will crack the synchonous mode of usb
is mach free?
new-ish laptops have ExpressCard, which is PCI-Express, so they should work with the 3x20 and hostmot2
nmz787: no, mach is neither no-cost nor open
for a very small mill, for milling pcb's or something like that, i actually think there should be a serial->mcu solution. someone would have to write a serial driver for emc which would output speed data and input position data from the mcu
nmz787: when i worked in a HAAS shop i set up serial control wherein the mills just received a single g-code line at a time and executed it, was kinda cool
celeron55: serial's not fast enough to do that - see that url
i'm 100% sure i could make a setup with serial ports that could engrave pcb's
if i wanted
cradek: but are there controllers out there that take cmds like "go 35 steps" ? sorry, i dont know how any of this works other than parport
and to think rs232 is going to be more common than parallel ports in the future is probably misguided
USB to serial is very common
usb to serial sure is very bad in latency
if i sent move commands
i.e. x move 35 steps y move 35 steps z move 0 steps
and not move until a GO
(USB to parallel is common too, but that doesn't help you with realtime)
then latency wont matter
you can't move fast that way because you'd have practically no controlled acceleration
itd be tuned on the controller
the machine would always need to be in a state where it can stop instantly or start instantly at full speed
yeah the controller can do speed control
i guess interfacing with emc can't work that way
the controller would have to do the full motion planning in order to move along the desired path.
you need coordinated motion, which is only barely like three motors going at their own speeds
nmz787: what about when there's a circle in the g-code?
can you do that with your mcu?
i bet not
well the mcu would have to follow the circular path, just like the straight path
also a motion planner has to blend adjacent moves together so the machine does not have to stop at each segment. it has to do this while staying along the programmed path.
does EMC have a command line mode?
there is a 66mhz atmel AVR that can run linux
if you want your device to support threading on a lathe, or tapping on a mill, you need to sense the spindle position and adjust the speed along the programmed path accordingly
nmz787: i guess anything that can run linux with the RTAI patches can run EMC
[21:27:37] <celeron55> https://www.rtai.org/
(you probably will have problems building it, but most of the code should work)
thanks for all the help so far
(the AVR is not on the list, though)
i had an inkling of thought that told me this would be a bit harder than I first thought
just get a computer with good realtime performance and a parallel port
and everything will be very easy
so in that case you would just have the parport driving the H-bridges?
or the stepper drivers rahter?
EMC can do either
stepper drivers, preferably
and i'm guessing EMC can do microstepping, or would the drivers do that?
that's a driver job
preferably my foot... ;)
and while you *can* drive h-bridges with emc, you might run out of parallel port pins
so just make your mcu to a stepper driver
celeron55, EMC depends on RTAI ?
yes - for realtime control.
isn't the control just 3d vector math?
i guess it's a boatload of complex stuff you don't want to care about 8)
EMC is a thousand things we're glad are already written, all smooshed together
(smooshed is a technical term)
and we're extremely glad they seem to have been written fairly well!
well most :-)
many enough :P
it can do things like this - http://www.youtube.com/watch?v=35tHYaDUmZQ&feature=channel
or this http://www.youtube.com/watch?v=C740zS9R9kk
so can you get to the trajectory planner easily with a CAD file?
is it like this basically, in text form?
[21:41:15] <nmz787> http://wiki.linuxcnc.org/emcinfo.pl?Trapezoidal_Velocity_Profile_Trajectory_Planner
how do I get to the trajectory list if I have a CAD file?
Probably you must calculate it...
And good evening to everyone
I mean from EMC
Ups, i dont kno
is there a "compile trajectory list" function?
i guess there is no such option
no, the path is planned in realtime. at each period the correct position is calculated.
there is an open loop mode though
if you plan ahead, you can't get things like feed override right
is there a simulation mode?
open loop mode just assumes the actual position is exactly the one that was asked
and that's how it is with steppers
(as long as you don't abuse them)
so is there a simulation mode>?
nmz787: yes there is
do i need to install the RTAI to use it?
if i'm to write a converter for the trajectory list, then I needn't use RTAI
cradek: would the stand alone interp be a good start?
maybe, if you wanted to do all the motion planning elsewhere
at least you wouldn't need to figure out how to parse gcode
i think that using any part of emc for something that's not emc is way too complicated
celeron55: if making free software, it's smart to reuse the parts of existing free software that fit your needs
well, yes, if they fit
and whether the payoff justifies the work is just a matter of opinion that's different for everyone...
seems that the trajectory list will give vectors and speed
so that seems like an easy way to control what I want to do
which is a laser CNC mill
arcs and helixes are not vectors, but are represented in gcode
a vector has a length and direction, and does not have enough information to represent an arc
bezier curves also cannot make arcs
you can approximate an arc with lots of vectors (or bezier curves)
but a device processing gcode ought to make regular old circular arcs as an exact native operation
(last statement is a matter of opinion - the rest are facts)
How is this not an arc? M 128.57143,578.07647 C 320,295.21933 391.42857,555.21933 391.42857,555.21933
it's just jibberish
alex_joni: do be expected from Jymmm :)
No, it's actually the xml output of a bezier curve
aa-danimal-shop: I'd be careful if I were you
* alex_joni noticed an order of trouts on Jymmm's latest cc excerpt
order of trouts?
ever been slapped with a wet trout?
* aa-danimal-shop slaps alex_joni around a bit with a large trout
* alex_joni is immune
msc has a 40% off metalworking sale today, but i'm not sure if i want to spend the $100 to get the discount
i need a coolant tank for my bandsaw, but that's not neccesarily a neccesity
[22:18:34] <Jymmm> http://codepad.org/g92WAFyD
Save as drawing.svg
anyone have ideas for a cheap coolant tank?
i have a little giant pump i can use
the tank needs some sort of screen
to keep chips from going in
thought about just using a heavy duty rubbermaid container
a parts cleaner would be nice...
oh yea good idea
[22:27:34] <aa-danimal-shop> http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=96952