hello everybody, i've followed the steps for installing emc and rtai in debian lenny and worked, but i was wondering if it's easy to make a bootable iso image with the new kernel patched and if there are instructions or a guide because i couldn't find anything helpful yet.. thanks in advance for your help :)
no, it's not easy. the process is done manually, following procedures you'll find with web searches such as: remaster ubuntu cd
well, if you're talking about debian, I know even less about their practices -- what I mention above is for ubuntu only
so it's a matter of practice
thank you jeff, i just needed to know if was possible to do it in a easy way, but i will have to investigate it
i will check the ubuntu remaster cd in google and see if i can learn something to use it in debian
the thing is that i'm new with linux so i'm not used to the commands and packages so in the beggining is a little complicated but thanks for you answer :)
3 answers the same.. Must be right
Yeah, the world is flat
cradek: Sounds like somebody is getting homesick =)
cradek: missing your fuzzy bunny slippers?
nope, I'm back now.
cradek: Ah, okey... clipped your heals together huh?
something like that
[08:03:26] <EbiDK|School> http://www.cnczone.com/forums/showthread.php?t=81836
Is this guy with the motors(second to last post) to be taken serious?
I updated my budget last night so it covers all of next year, looks like I'll be able to buy a mill in about a year :/
backlash is dead important but you can spring load the nuts and bearings
motors, if you gear them and accept lower speeds then any size
Speed is not so important
also looking at your current gantry design I might mod a bit to improve rigidity
Yeah it needs some improvement like more guide rails and such
Still can't make up my mind if I should build it or not
doing a self build is a good learning process
I'm mostly doing it for getting a good cheap PCB mill though so might not be so sensible
BTW, what price range would I be looking at for a CNC mill that'd do both SMD PCBs and aluminum work? Or should I really get two mills?
a mill that can do aluminum will do pcb's
and will also be more expensive
I would expect to pay 600-800 to build a barebones machine that was just barely able to cut aluminum
And if I'm buying instead of building?
a lot more
And you're speaking USD again today, right?
all my prices are usd
Okies :) no problem
EUR is more expensive anyway :p
the Tormach, which is a really popular desktop mill, is $6k
I think there are others available for 3000
if you are willing to get a used machine
Used is good as long as it's in good condition
you can find cnc'ed iron with a dead control, and those can go anywhere from $1k to 5k
for a FLOOR standing mill
usually a bridgeport, or similar
Yeah, I'd need a desktop one, I don't have room for bigger.
and that gets you something like a 30x10 table, 6" of Z axis under control, and 20" of W axis manual
since desktop machines are in such demand, used equipment is hard to find for a low price
I can at most find a 1x1 meter space anywhere in my room for one.
personally I'd look at building if you are budget limited and need a small machine
horror freight, et. al. sell machines that have plenty of reviews on the hobby forums to aid in picking the right one
You'd think they'd make cheaper desktop machines so they could sell more of them if demand is so high
machine tools are not easy to make well
and people who buy them tend to want to mill things like metal, which means everything has to be at least a minimum size
market not large enough to get economies of scale
Too bad the school doesn't have any used desktop mills laying around for sale.
[08:48:53] <toastydeath> http://www.youtube.com/watch?v=KachkbSG3_I
there ya go
haas mini mill
hm, that looks like the bigger one
Looks a lot like a mini mill yes, we have 3 of those here, but it's a bit smaller but still too big for em.
We also have a VF1 they're installing a VF3 over the summer break and we have two SL10 and one SL20 lathes too
And a bunch of manual mills and lathes
Wish I had money and space for a mini mill or VF3
tomp is now known as tom3p
where can i find info on gecko 540? ( geckdrive.com search returns nothing)
i think its a 4 axis drive but hard to verify when mfctr doesnt think it exists ;)
tom3p: you can only navigate geckodrive.com with flash plugin installed. let me see if I can find you an internal link ..
and a servo drive not stepper ( i hopes )
G540 is stepper only
it is 4 axes
jepler: woof, thx
ok, now off to mesa's 4 axis servo drive :)
they've got several brush servo drives
anyway, the geckodrive links for 540: http://geckodrive.com/product.aspx?c=3&i=14469
and 250: http://geckodrive.com/product.aspx?c=3&i=14472
(the guts of the g540 are "essentially" 4 250s
the 7I30 will mate with my 5i20 nicely
7i30 is a nice choice as long as your servos fit its specs -- 1A or 3A current limit (only 2 choices) and 36V drive voltage
the servo steve stallings brought to the fest was a nice and torquey and a nice size.
now to find 4 100W drives (3A 36V oughtta fit ) ( i got a machine coming in EMS, its past customs already, a sable-2015 off ebay )
I think that boy is confused... asking abut G540 then emntioning servos
* jepler goes on waiting for someone to find him the right servos for zenbot
only link I find is your website
Jymmm: zenbot is my little cnc router
jepler: those little pittmans might work? with proper belt reduction
raid some plotters for small motors (HP 74xx)
jepler: I thought you needed a new driver board, not motors
where I found some pittman motors :)
some guy sells them off and on through ebay
jepler: is your website havngissues?
Jymmm: yes, it's on dsl and the link keeps bouncing
plotter motors are nice, but very tiny
ng on my machine, this week it's wanting servos instead of steppers.
and so is my ssh, which also goes over the dsl :-/
Jymmm: yes, they are very small motors
I have some Yaskawa nema 17 brushless
jepler: 47 oz/in do ya?
1.8 deg/step, 12V@400mA/ph
I'm looking for servos, not steppers
jepler: why are you switching? just for fun?
jepler: Have you considered taking apart a cordeless carvac or three?
* Jymmm ducks
he has a mesa card - might as well do feedback.. ;)
and I think I can talk cradek out of the 7i30 he owns but has never used
that violates my "no more than $50 for a motor" rule
that's a brushless motor, so it can only be driven with more expensive interfaces. The type I need is a DC or brush servo, either dual shaft or with encoder
(I went over ebay pretty closely yesterday, so I dunno if you'll find anything I missed .. don't spend a lot of time, anyway)
brushed motors are a little more difficult to find nowadays
[16:01:03] <Valen> http://secure.oatleyelectronics.com//product_info.php?cPath=53&products_id=189&osCsid=70dd5de3af179d7bcf037b95db028091
thats what i'm looking at for our conversion
it's tempting to try the 3-phase motor drivers intended for RC planes
they wont work at 0 speed
they use the back emf from the coils to work out their commutation
jepler, actually if you want I do have a qty of a canon motor similar size to the pittman but they dont have encoders
If you get a sensored one it'll work
I'm still interested in motors archivist ;->
jepler: what stepper motors do you have now?
dey small not for big machines
how would they compare to the ones i posted?
Jymmm: 75oz-in, 2A
i have no idea what the specs on those motors are :-<
jepler: any issues with them?
other than a gut feel they should be fairly torquey
Jymmm: none that aren't my own damn fault
jepler: Why not just get one of these? http://www.xylotex.com/1axBoard.htm
Jymmm: what does that board have to do with anything?
your 3 axis xylotex go poof?
Jymmm, he wants servo are you deaf
I replaced the xylotex. I'm not looking for a replacement stepper driver at the moment
HI Jeff, did you and Chris make it back to Lincoln safely? Was anyone else riding with you?
I have also wanted to try out the three phase RC motors.
KimK: I was the only one in my car for the ride home. it was uneventful
chris made it home last night as well
OK, good. It was a fun trip, John (jackland) and I were glad we came. I know John was for sure, he came with half a mill and left with two! Well, sort of, anyway.
did you get to the point of pluggin in the robot arm? what about the linear scale on the big mill? Was that the G&L?
you would have to do something else for encoders... I wonder how it would work..
skunkworks: i dont see a torque rating
seb_kuzminsky: just going by size.. - it would be an experiment ;)
good evening. i want convert a signal of float at s32 and use the conv_float.s32 component, but have one error.is possible please help thanks?
if you want help with an error, you should say what the error is and show specifically what lines of hal caused it
ok jepler one moment i write
skunkworks300: might end up doing something terrible like mounting the encoder at the opposite end of the screw
but with no encoder and no dual shaft there aren't a huge number of options
i think it's worth a couple of bucks to get a motor with integrated encoder
i find them for a pittance at the local surplus store every once in a while
i write in my hal file : loadrt conv_float_s32 count=1 and after addf conv_float_s32.0 servo-thread.when run emc the terminal have error: fuction 'conv_float_s32.0' not found and after addf failed. because?
halcmd: loadrt conv_float_s32 count=1
halcmd: list funct
read the documentation or use halcmd interactively to find the correct name
ok i understand very thanks jepler
Seb when you start the project for spi bus of mesa , i free for betatester if you want
ok thanks motioncontrol
i have the 6-m5i20 and 4-7i64 and 2-7i46 and 10-7i33 ecc.. for test if you want
Shouldn't brushless motors cost less than DC motors? Cheaper to make...
better magnets cost more
That is the only thing that cost more... Well I suppose you could argue that BLDC motors are generally higher quality in the first place.
[16:42:09] <skunkworks300> http://www.futurebots.com/motor.htm
one question.Know some vendor for rotative switch for override with gray code conversion?
motioncontrol: like for a feed overide?
yes for feed overidde
i want use 4 digital input and mux 4 component for change the override
on the 4 bit , with gray code have 13 possibili override value
mux4 selects from 4 inputs using 2 bits, not 16 inputs using 4 bits.
4 bit gray code should have 16 possible values. Why not just make a quick lookup table in comp?
my co-worker decided to put on a very impressive plasma physics display
in the microwave
wrapped in paper, didn't realize wife had wrapped it in alum foil INSIDE the paper
sheesh, wives these days
I did that recently, the kids missed it so I turned it back on again to show them
very rapidly demonstrated the problem of magnetic confinement and z-pinch
[17:00:02] <jepler> http://emergent.unpy.net/files/sandbox/mux16.comp
Stays Crunchy in Hal! (tm)
melts on your endmill, not in your hand
tomp is now known as tom3p
Hi skunkworks300, seb_kuzminsky, and everyone. Sorry, I was away for a bit, someone came in. On the robot arm...
thats a pretty big arm to carry someone ;)
[17:07:26] <seb_kuzminsky> http://www.youtube.com/watch?v=wOCmoYU6h1Q
We got it to the point where the motor drive power would come on and the brakes released and the motors would lock up (power up), but we could not get it to do any stepping. First I tried to introduce noise to the step pin with a jumper clip. Then...
good use of a rotary axis at the end of a robot arm: http://www.youtube.com/watch?v=B4Uvlmay02c
I tried to connect a function generator using a TTL output (defective?) and using the analog output. No steps.
KimK: did you try stepping from your 5i22?
There was an Intecolor terminal keyboard on top of the control cabinet, so I imagine there was an Intecolor terminal connected once (RS-232, special codes to draw color text, graphics, etc.) I imagine someone has written an emulator for Linux, but I haven't bothered to look. The control was looking for 5.25" floppies that we don't have, so I wouldn't expect too much from the terminal anyway: "OS NOT FOUND", perhaps?
seb_kuzminsky: I did not try connecting the 5i22 due both to lack of time and because SWP thought that I should buffer the 5i22 card with something, and that seemed like a really, really good idea to me, since that's the only Mesa PCI card I have. Thus endeth the experiment.
makes sense :-)
maybe we'll play with it again next year
Maybe I (or someone else) will have a more suitable Mesa stepper interface card then too.
what about the linear scale?
On the linear scale on the G&L, First Jon did "microsurgery" on the reader head, then Chris found a break inside one of the reader heads leads. This was sufficient to get the alarm LED to go out on the Sony converter PCB, and I believe it stopped "wandering away", but it was still not working correctly at the time John (jackland) and I had to leave. I have not heard the final disposition from Chris and Stuart and the others.
We also decided during this that we were not moving the reader head with enough "finesse", so Stuart brought over a Bridgeport to clamp the scale and head to, to use as a more sensitive scale and head mover. That helped too. But it was in the aisle, so will have to go pretty soon.
cradek was still there when John and I left, so he will at least have a later report, if not a final one.
Oh, one more item on the robot arm. It is wired and driven for six axes, but there seems to be only 4 axes of mechanical motion. But there are two DB-25 sockets up near the last joint, so perhaps two more smaller, finer, optional joints can be attached? I don't know what they might be.
Oops, I missed breakfast, and now I'm missing lunchtime, back in a bit.
* KimK heads for the lunchroom
travis_ is now known as trav
trav is now known as tlab
skunkworks300 is now known as skunkdesk
skunkdesk is now known as skunkworks_
when I boot with my rtai-smp kernel, it only shows 884MiB of memory, but my generic kernel shows the whole 2gig, is this normal?
the emc realtime kernel is set to only see 1gb of ram
my see 2gb
can I change it to see all of it?
yes, you have to rebuild
shouldn't be too hard, there is a source package
just need to turn on high memory support?
someone went out of their way to make that change, I've never seen it
32 bit computers should support 4 gig
[17:52:20] <eric_unterhausen> http://kerneltrap.org/node/2450
I guess I never ran into this problem because my first linux computer with more than a gig just uses 64bit linux
travis is now known as tlab
so you can not use high memory support on rtai kernel?
some RTAI kernels don't work with > 1G RAM
I believe there are patches in later revisions to fix the problem
SWPLinux: how did the show go?
the one that's this Sunday?
ok, guess I'll stick with what I have then
don't know yet :)
My weekend got screwed up ;)
we're on at 8:30 or so EST this Saturday for the "behind the scenes" show
the one time it might be nice to have cable at the house - I'm not sure we know anyone my wife can visit to watch it :)
I suppose I should go meet up with the rest of the gang. see you later
skunkworks, how are you with kine mods ?
namely the puma kine
colin_: I looked a little at it about a year ago. so not good. I was going to try to make the vismach match the kins/hal but never got around to it.
i have an arm and the puma kins will work out of the box for the most part
i just need to change the code
because the puma arm is offset to the side of the base rotation
on my arm its ofset forward
so im assuming its an X translation instead of a Y
heh - that might be above my paygrade.. ;)
(steve stalling rubbed off on me)
that seems like a really good deal...
skunkworks_: yes, that's a very similar model to ones I've been finding new at $200
I can't read the full part number, though
the part after 24 VDC?
the part after 9236
I have a hint for you :P
it's on the page too
(found it after I googled it)
I looked but didn't find it
it's hidden top-right
yes I see it now
nothing else online for it though
too bad the part number for the encoder isn't visable.
I wonder if you could ask
as low as 50 and as high as 512 line
or if it is film - 1000 or 1024
jepler: how many threads per inch on the zenbot?
2:1 would give you close to 140ipm.
yeah, I just have to get cradek to build the mounts and turn the pulleys for me
lots of the pullman models there
I don't want a gearmotor, though
can you say backlash?
page 6.. 9234, no gear
alex can say backlash in a few languages..
yeah, 9236S009 is one I was looking at
the part after 9236 must specify options, but I haven't found a master table of options yet
(e.g., 9236S008 is without encoder, S009 is with 500CPR encoder)
You can see from the photo that the encoder is 500 CPR, The CPR is cut off by the power cord. At least that is usually what the label says there :)
ah thanks geo01005
500cpr would work great
so that only leaves guessing whether there's index or not, but I get "good enough" homing with switch only
32000 edges per inch - 2:1
I think I might buy a couple too :)
(there are 5 pins either way, but for no-index one of them is simply not driven..)
jepler: I forget - how fast where you getting it to go with the geckos?
skunkworks_: I don't think those numbers mean much, because we were running at 150% current ..
that would be .00003 per edge
I have a new Idea for a bot that uses plastic compliant hinges and linear actuators. I was going to use steppers, but an aything I/O card and a 7i30 is about the same cost.
what's a compliant hinge?
if it where me I would also be running it above the 24v limit ;)
[19:11:29] <geo01005> http://www.monroeengineering.com/hinges/living-hinge-plastic-butt-hinge.htm
skunkworks_: right now (with xylotex 1/8 stepping), one microstep is .00004in
non-trivial kinematics, but only 2d, so I can do that :)
jepler: you are only acurite to about 1/2 step though..
I think I was called a living butt-henge once in junior high
they are wonder why they just sold out of that servo
Think SCARA, but those living hinges instead of bearings and linear actuators instead of rotary.
it's a tiny little servo
unlikely to be of use to anybody
about how big is cradek's servos here http://timeguy.com/cradek-files/emc/DSCN6290.JPG
I don't remember size
I bet they're in the same ballpark
another picture shows they're pittman 9413, but I can't find specs on that
[19:28:13] <geo01005> http://astromechbuilder.com/mechloader/d/27067-1/GM9413_1.pdf?g2_GALLERYSID=372cb7ef5850ae3827c613a18ff91c71
less the gearbox
yes, I think the 9xxx motors all have the same diameter
with the 923x motors, the last digit indicates increasing length and torque
pittman specs (including the 9x36): http://highlab.com/~seb/bzr/fabrication/doc/vendor-docs/pittman/lcm_bulletin.pdf
continuous torque: 9.5 oz*in
stall torque: 61.8 oz*in
stall current: 9.6 A (at 24V)
* skunkworks_ thinks seb is going servo also..
i'd sure like to :-)
but the 9xxx pittmans are too small for my machine
i'm in the market for some 14xxx's ;-)
I've got one, but it is running my reprap extruder :)
what are you extruding, aluminum???? ;-)
I ordered one motor off ebay, a different one came in the package...
I ordered a small high gear reduction motor, but got a 14000 pittman with a low backlash planetary gearbox.
Can't complaint too much.
seb_kuzminsky: there are no flats on the 9xxx pittman motor shafts, right?
i'm not sure
I don't see any indication of it in the drawing
jepler: usually you get to cut them out if you really need them
that's why you keep cradek around, isn't it? :-P
don't let him know that :P
I dunno how you put a good flat on a motor shaft
with a good file :P
clamp it carefully in a V block and run over it with an end mill?
just make sure it's flat
with a dial gauge
travis is now known as tlab
I've lost my cnc icons on my menu, anyone know how to get them back?
tlab: what did you do?
maybe you uninstalled emc2?
I had to compile by source for my custom kernel
tlab: add them like you add icons normally
then you need to put the menu there by yourself
um how would I add them normally?
do you use gnome?
I just need stepconfig
where is it?
simply type "stepconf" like you do with "emc"
if you have compiled with run-in-place, then you need to . scripts/emc-environment first
any of you program microcontrollers on linux?
I do avrs some
ugh, my box locked up
travis is now known as tlab
so any of you program microcontrollers in linux? microchip, avr, freescale?
what do you use for ide?
then how do u load target?
lately I've been using arduino-style boards, and run the avrdude that is bundled with the arduino environment
serial or usb interface?
avrdude pick up the usb device ok ?
lately I've done a little bit with at90usb-based boards. those program over usb with a different program -- dfu-programmer or something like that
yes, avrdude has no trouble using /dev/ttyUSB0
will avrdude work with the avr dragon?
I am not familiar with avr dragon
[20:18:37] <tlab> http://www.atmel.com/dyn/Products/tools_card.asp?tool_id=3891
multiple types of dragon programming interfaces are listed in the /etc/avrdude.conf of avrdude 5.5-1 on ubuntu hardy
dragon_dw, _jtag, _isp, _pp, _hvsp
ok, ya dragon has all those interfaces
this looks informative: http://www.homebuilthardware.com/index.php/avr/linux-avrdragon-tutorial-1/
I need to test the realtime threads on a computer before getting the mill and the io board. What should I do to .hal archives?
boot the live cd, run the latency test
I won't run under ubuntu
still need to test its the easiest way
can you ask a more specific question?
I just received 4 of those motors... going to try and add a digikey encoder to the back of them... very nice motor at first inspection
jepler, they have a bunch of bigger motors as well for very nice prices,
yea... the scooter/electric bike craze has brought a whole bunch of new stuff to the market
brush design may be unidirectional, check
the motor I listed, has a 3mm pitch HTD 16 tooth pulley on it already
archivist, I havent put the taco on it yet, but it seems to run smoothly in both directions
LawrenceG: I don't see a provision to add an encoder, do you have some ideas how to do that?
I decided to get some of the motors skunkworks pointed out
Well, if i try to run emc right now, by choosing any machine on the config picker, emc will crash trying to load inexistent hal connections. What lines i need to remove from any .hal in order to run emc without the io?
you'll need new mounts
is anyone here suitably familliar with kine to help me edit pumakin ?
rbertoche: run one that's under the sim menu (sim/axis)
iv been reading the code and working out what i need to mod
and most of it is actually good already
the only issue i have is that instead of the arm being to the left/right as on puma
mine is in front of the base
so the A2 translation needs to be in the X instead of the Y i think
jepler, I am hoping to drill the rear shaft of the motor and add a stub for the encoder
colin_: Have a picture of it?
[21:06:47] <colin_> http://www.kawasakirobotics.com/PDFs/JS10.pdf
Well, if thats suposed to work, i guess my problem isn't not having hardware t
that hal is asking for
insmod: error inserting '/usr/realtime/modules/emc2/motmod.ko': -1 Unknown symbol in module core_sim.hal:7: exit value: 1 core_sim.hal:7: insmod failed, returned -1
seb_kuzminsky: making motor mounts is what I keep cradek around for, as I think sam noted earlier
that's when i tried to run something from the sim folder
rbertoche: for that kind of error, more information will be shown when you run the command 'dmesg' in the terminal and look at the last dozen or so lines.
The only strange line in dmesg is " motmod: Unknown symbol ceil "
yes, that's the real error
did you build your own kernel and/or rtai?
you must enable rtai math
Where, on rtai compilation?
I think this is --enable-fpu when you ./configure in rtai
geo01005, that any help?
I don't see how it is different from your regular old PUMA kinematics... I'm still looking though.
(an emc2 configure test that checks for this configuration problem would be nice, but I'm not sure how to check..)
[21:13:25] <colin_> http://prt.fernuni-hagen.de/lehre/KURSE/PRT001/EXAMPLES/eqs2/puma_dh.gif
thats the puma arm
you can see the arm is to the side of the base centerline
where the js10 is forward of it
Oh, ok, i'll compile it again
its the A2 offset that defines the sideways offset between joint 1-2
i need to change that to forwards
./configure --enable-math ..?
--enable-fpu according to my notes
maybe look through ./configure --help?
hmm crap looking at the code the A2 offset is actually between the base and the joint 3
oh, there actually is a --enable-fpu, ok. But why? what is motmod ?
colin_:I'm not seeing what you are saying... To me it looks like d2 is zero, and it is happy.
yes d2 will be 0
im going to need an A1 ?
if you look on the left hand diagram with the arm view from the side
there is a 100mm offset from the rotation of the base and the arm
Why not loading motmod module means having fpu disabled?
colin_: Ahh, I see that now.
Well That is a very simple fix.
it is ?
jepler: Have to yell when you get them in. Let me know how they run.
only changes the x-y position.
are you going to try to direct drive for the first try?
i thought that
skunkworks_: that's sure easiest
heck I could just hold the motor in about the right spot
but it changes the sideways bit between the base joint and joint 3 on the sideways for the puma
skunkworks: even direct drive, those motors are torquier(?) than the steppers on zenbot now
colin_:The forward kinematics are very easy
on my arm it will have to be between J1-2
Lerman______ is now known as Lerman
Really? I guess I didn't look at the specs.
60+ oz*in :-)
seb_kuzminsky: what size steppers are you going to use on your x2?
i'm planning on servos for the x2
ah - nice
aiming for 250-300 oz*in at the screw
mostly what I know about servos and steppers is that directly comparing stall torque and holding torque doesn't tell the whole story..
probably with some belt reduction
colin_: just add 100mm*sin(theta1) to x and 100*cos(theta1) to the y.
geo01005, i dont really know how
Jepler, how did you guessed about fpu being disabled? not seeing it
colin_ The kinematics are fairly strait forward, but It has been quite a while since I looked at the source for the kinematics...
rbertoche: "ceil" is a function provided by the math library. in rtai, this comes from rtai_math.ko which is not enabled by default
good night all
see you alex_joni
/usr/src/rtai$ ./configure --help | grep fpu
--enable-fpu Enable FPU support
Oh, thank you!
Yeah, i got it.
jepler: is this broken? http://unpy.net/cgi-bin/viewcvs.cgi/eagle/ulp/
skunkworks_: yeah, I shut down that system a month ago
oh - ok
I should probably get it online again, but it'll take a reminder
Does anyone know of a parallel port SPI interface for EMC? I found one for Linux, but don't know how it would work with EMC.
in general, no. if emc and another program tried to control the parport at the same time, you wouldn't like the results .
That's what i thought.
Seb_kuzminsky: With the work you're doing on SPI, If I don't use all of the functions of a Mesa servo card, would I be able to use the unused lines for a SPI interface?
I realize I would have to break them out to get away from the servo card circuitry.
Reset of hm2 stepgen possible?
i'm having a pretty tuff time with this HAL stuff
somethings just not clicking with it
whats the problem?
i just dont understand how it all works... i get part way into the hal user manual (pasically through the introduction) then i just get lost
well, imagine some electrical parts in front of you on your table
this are the hal-components, like stepgen, encoders and so on
like in real world, to get a machine you need some wires between them
ie. a comp like stepgen has input & outputs
yea, i kinda understand that whole part
ok, lets continue
take a stepgen
this has one important input: position-cmd
and outputs like step & direction
a step to wire the stuff together is the NET command
so: net xstep <= stepgen.x.step will create a wire called xstep connected to the output step (for example)
so i just dont get the relationship between emc2 and the component
gtom: you may be able to use the raw-write facility in the HM2 driver to reset the stepgens position register
This is not ideal because it may generate a step, better would be a hardware input that cleared or
(better) latched the position register
The only problem i have is the pos-fb... that remains in the old position when the index is hit
how do i calculate the counter (subcounter) from a given position???
see i'm already lost... net xstep <= stepgen.x.step is greek to me.
@Danimal: EMC is also a "component" on your table, with inputs & outputs
pos-fb comes from the hardware position count reg...
you might need a custom firmware that includes both the servo stuff and the spi ports you need
gtom: hi there :-)
from the 16:16 register?
Hi sebastian, I the SVSP8_6 config I sent you is appropriate for servo&SPI
gtom: i briefly discussed your machine with cradek and jepler
Had an idea today... why not set fb-position to cmd-position on enable=false ???
[22:18:57] <seb_kuzminsky> http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-05-28.txt
PCW: oh yeah, i bet you're right
i feel like i'm missing a crucial piece to understanding this stuff
three conversations at once cant help :-)
cradek and jepler get into it around 20:11
me too, 6months that i spend now trying to understand this stuff... :_)
i've been at it nearly two years and i still have no idea what's going on :-)
i got my machine working, kinda. it moves in xyz fairly well
now i'm trying to hook up my vfd
thats all you need xyz...
i wish... i need the vfd and the toolchanger too
Ok. What I'm trying to find out is if I need to purchase just the Mesa I/O card that has SPI on it, or go with another FPGA card. It sounds like I don't need another FPGA card.
Goslowjimbo: i think any of the AnyIO boards can do SPI just fine
it will depend on the fpga firmware and on the hostmot2 hal driver, not on the fpga card
gtom: your machine doesnt use the stepgen position feedback at all, does it? it relies exclusively on encoder position feedback as i understand it
Yes, pos-fb is ignored
i dont need it.. linear encoder fb is much better...
also dont want a pid /closed loop
machine-table is about -think- 350 pounds
what would happen if (for example) the stepgen lost a step? the stepgen position feedback would disagree with the encoder position feedback, and the system would be out of sync
in what order should i be reading this stuff... emc2 user manual, emc2 integrator manual, hal user manual, ect...
no, i wrote a rt-comp to correct the position...
if the pos-fb is >fmin_error away emc stops...
what does your position-correcting component do?
Well, if the stepgen accels an is below velocity xxx i.e. 0.5 the difference betw. encoder-pos and cmd-pos
is corrected (only to the direction where the axis moves)
when the stepgen decellerates and comes below velociy xxx the end-position is corrected also (just stepping!)
This works absolutely PERFECT!
Seb: didnt you just add a velocity mode to stepgen?
why not use the setpper as a velocity mode servo?
PCW: i did, and i suggested that to gtom, but he said he doesnt want pid or closed loop
Well, linear encoders on heavy machines do not work...
Well even the stepgen is closed lopp
Haidenhain is using a similar setup for their new hispeed milling machines
i mean controls...
with that component i got tolerances down to 1/100of a mm
And: Backlash is nearly eliminated!
see you later Goslowjimbo
gtom: i'll look into the enable behavior of sw stepgen vs hm2 stepgen in the next couple of days
it's reasonable for hm2 to do whatever the sw stepgen does in this regard
a reset-pin of position-fb would be good... :-)
sw stepgen doesnt have one
i know, for security reasons we need one... just to kill any remaining cmd/fb difference on estop errors and so on..
normally the stepgen is less than ferror away from position-cmd, say .010 inch (.25 mm) or so, but yeah...
i told you that it is running FAR away... even with 2.4.0pre Trunk...
sure, because your machine isnt set up right ;-)
normally position-cmd and position-fb are close to each other
i was playing with the vels & accels, now i can catch up the axis after ~ 5mm
That was my intention: Joint-follow errors cause emc to stop and stepgen is disables.
Think that the SW stepgen sets the pos-fb to pos-cmd in that case...
the hm stepgen remains with the difference of pos-fb and pos-cmd
if you turn the machine on stepgen wants to move to its cmd-position
the hm2 stepgen stops if .enable goes low
but it doesnt set its position-fb to position-cmd
Yes, thats the problem i thikn
that's what i'm going to look at, and update it if needed to behave like sw stepgen
ive put some code inside the enable stuff today...
didnt work :-)
tried to set *hm.... position_fb = *hm... position_cmd, ok that was wrong, pos_fb is a calculated value
next i wanted to calculate the desired counter-position from the position_cmd???
need to understand the stepgen code...
the current .enable stuff is not correct!
1. if you turn the machine ON after an error it should not MOVE!
If I am typing in commands line-by-line in semi-manual mode,am I right in thinking that I might as well just use a G43 rather than an M6?
oh look, stuart is already trying to get us to come back to wichita
Points at query above
man, would life be alot easier for me if i ditched my mesa 5i20 and just used the parport so i can use the stepper config?
i got the steppers working with the supplied config for the 5i20, but i have no clue how to set up the spindle control for my vfd