anyone do a mach 3 to EMC2 benchmark comparison?
[00:51:16] <Ziegler> http://www.youtube.com/watch?v=lqVKOilZO2A
<< that nice looking lady can dance
re comparison: interesting... lotsa tiny variables to make it an apples to apples comparison, then arguably a crippled version of one player. (and i no got mach or steppers )
sorry guys that was the wrong window
Reason I ask about comparison...
the computer I just picked up for my new set-up
has it already installed
before I wipe it out, I wondered if the info would be of use to anyone
Ziegler: you might image the drive before wiping
in streamer, is the depth option the number of lines? ( how many data sets of input )?
tomp: "depth" is the buffer length, for example you'd need more depth for a fast thread because there's more data going in before you have a chance to write
fenn: is that measured in 'lines'? if i cfg streamer with 'ffbs' and want 100 lines of 'ffbs', is depth = 100 ( or do i calc the bytes or what )?
a buffer is usually some reserved bytes right?
depth appears to be number of samples
but i think in hal everything is 32 bits anyway
oops, samples times number of channels
max_depth = MAX_SHMEM / (sizeof(shmem_data_t) * tmp_fifo[n].num_pins)
user_ is now known as skinnypuppy1334
32 bit floats? single prec?
yes, HAL uses 32-bit floats for floating point data
64-bit would require extra locking to have atomic updates
it's important because i want just 1 line of buffer, and for streamer to report when its done, an outside app determines next motion ( this may allow real edm and avoid pid )
SWPadnos: thanks, no atomic update needed :)
just sounded funny
you can't rely on a userspace app to get the signal in a known amount of time
right noit real time
it sounds like you're using the wrong tool for the job
are you sure you shouldnt be using classicladder instead?
i want to send a single positon to hal, and wait for that to be achieved and then send another.
i dont see a reason to use ladder
stepgen or pid?
actually, it probably doesn't matter
neither, servo values to servo amplifiers made from the proportions of each vector asked for. the desired motion is broken into 3 analog voltages by the direction wanted and by the velocity desired.
streamer won't tell you when the motors are in position, only when it has output the next set of coordinates
so you still need to rig up some modules to look at ferror
right i ask for a very small amount and hope i can tell streamer 'foreget about it' if i want to change direction before it gets done
you may be able to make some configuration that enables streamer once it gets to the next point
but that's probably a nightmare
i dont have ferror, thats a feature of velocity tracking.
i have perror ( position error )
not velocity error
whichever error it is that tells you the machine is in position when it drops to near-zero ;)
back to the main idea, i hope to be able to ask hal to move a single position, than ask another. ( all non-realtime, all non-pid, because thats what i've been aiming for and building )
if you dont use PID what makes hal turn the motors?
how do you expect HAL to move from position to position?
right - what he said
its not like you're positioning an electron beam
even if you were, there's got to be an interface from the number representing the requested position and the actuator moving something to the requested position
send velocity voltage to amplifiers thru hal, monitor position thru hal
and how do you decide what voltage is correct?
you really really want that to be in realtime, probably with PID
the thing that calculates positions doesn't need to be RT or PID, but the thing that actually positions the motors does
you could just switch on the motors from across the room, and then walk over to the machine to shut them off when you hear they are in position
external program uses the known velocity/voltage ratio of the amplifier * the proportion of joint travel (eg 1mm/minute * the part of the vector for each joint concerned in the motion )
so you're essentially doing open loop control
i think it might be necessary for him because encoders aren't high enough resolution
hal sees it as open loop. the loop is closed by the external program making the request
then you don't need streamer
your external program just needs to connect to the DAC pins on the hardware driver
true, but streamer & halstreamer are already there
but they aren't useful - they just add complexity
i just want a 1 depth on streamer
what language is your external program in?
python maybe, parts in C ( gap voltage vs desired voltage to desired velocity )
just make your program a component (user space, no need for RT)
export a few pins, connect them to the DACs, and write to them whenever you want
yah - use comp, it'll be the easiest
speed where speed is needed, but modeled in scripting languages like python to begin with, ok comp. i'm trying to save the dro and basic machine operator panel as much as possible ( flush machine on.... )
comp just makes it very very easy to create a HAL interface. the code is more or less C after the preamble
actually, it is C
ok, thanks... not entirely a new direction so maybe i can grasp it
Does putting 2 caps in series double the voltage capacity?
capacitance acts like resistors in parallel
hyperphysics is good for this kind of stuff http://hyperphysics.phy-astr.gsu.edu/hbase/electric/capac.html
but is it considered safe in common practice?
well you'd be doubling the uncertainty
and there's more curent being dissipated because the ESR adds up
in general it's a bad idea, but sometimes necessary
ESR - my last study of electronics was in the 70's
if you series caps, you need to use the same capacitance, and you probably need balancing resistors to ensure equal voltage across each cap
two in series is standard practice for 480V AC drives - the DC bus is 480*sqrt(2) = ~650V, and electrolytic caps are only available at 450 and sometimes 500V
saw someone selling a stepper pws - used 4 caps - 2 parallel sets of 2 in series. Caps were 25V - Pws was rated at 48 open circuit
thats just dumb
50, 63, 75, and 100V caps are readily available
yeah - no margin for back emf
and probably would result in lower cost
I can't imagine a good reason to series caps at that voltage
these were snap in on a PCB which also had a 25A Bridge rect. - Size was nice - but I didn't trust it
maybe somebody pulled a big box of 25V caps out of a dumpster ;-)
so series caps increases WVDC?
and decreases uF
so how much % of the rated MFD is lost?
two identical caps in series have half the uF and twice the voltage of a single cap
(you only get twice the voltage if you ensure sharing, usually by balancing resistors)
yeah that would mean you loose alot of your bang for the buck
for anything under 400V, series caps are a waste of money - less performance for more cash
newcapacitance =1/[ (1/c1) + (1/c2)] http://hyperphysics.phy-astr.gsu.edu/hbase/electric/capac.html
(that 400V rule of thumb is handy )
the formula for caps in series is the same as the one for resistors in parallel
the 400V rule applies to electrolytic caps - 450V is the upper end of the "mainstream" products
500V ones are available, usually the cost/performance isn't as good for 500V ones
but if you need 1000V DC, 2x500V might be better than 3x400V
for film and ceramic caps, series almost never makes sense, since single caps are available up to 1000V volts or more
fun with electrolytics: http://jmkasunich.dyndns.org/cgi-bin/blosxom/capbang.html
now i get to figure out what to put on the budget-request list for my local univeristy robotics club
totally open-ended, and we've been pretty much guaranteed at least $1k
thing is, nobody really has any idea what they want to do
nite jmk, thanks
how do i find the distance traveled for an xyza motion?
if Xmoves 1", Y 2" Z 3" and C moves 45degrees at 2" radius, then XYZ moved 3.74...inches and C moved 1.47... inches.
but im stumbling on the total travel. is it just the sum?
I want to shoot some pics of our old GMF robot. I been thinking about making one like it - maybe 1/3 scale.
you have to use kinematics with rotary axes
unless everything pivots around the tool-tip (unlikely)
Skullworks-PGAB: so far we have, a big tank-thing, a robotic saxophone, and a gantry of some sort
and a very well equipped dorm-room-sized workshop
sherline mill and lathe (not cnc)
nobody seemed terribly interested in cnc'ing them
I'm thinking I will have about 24" vertical travel using 36" drill rod for the colunms
its not an arm?
old GMF CARROL
well that doesnt tell me anything
I'll shoot some pics
it moves a ram assy on 2 vertical shafts
ram telescopes straight out
tip: when a lathe makes rumbling noises, stop and inspect the cog belt
fenn, try for a pick&place machine
or at least a robot that will pick things up from somewhere and place them somewhere else
a slipping cog belt makes a loud rumbling noise!
i think that's what the gantry-robot was originally for
has a double gripper head which can rotate 180 degrees around the centerlint of the ram
it would be cool to do some stuff with image processing though, if i have access to people who know about that stuff
a robotic tracker would be cool
a scara controlled by emc would be cool too.. i was thinking about making one anyway
this is O L D FANUC yellow cap 4 brush DC servos with air counterbalance on the vertical ram lift
I was thinking of dumping the air counter balance and using a counter weight system
ok - that would just be tooo
build the robot - to act as the ATC for my mini mill
I would find myself in Classic ladder hell for sure...
toast is now known as toastydeath
fenn: the machine is like this http://imagebin.org/10741,
what distance does the tool tip travel? the sum of [ sqrt(x^2 +y^2 +z^2) + ((deg/360)*(2*pi*Radius)) ] ?
im guessing it >is< just the sum
I don't think so tomp
your formula doesn't give a different answer if C is going with (say) X or against it
i don't understand the question
the distance the c moves doesnt depend on X or Y or Z does it?
with X -> C <- the tool might barely move
it's also different if you mean "total distance traveled" or "net offset at the end of the move"
with X -> C -> the same amounts, the tool moves a lot
right, I assumed the former
i dont get the notation X -> C ->
if you mean dist(end - start) it sure might be zero (imagine C turning 360)
x moving to the right, C moving to the right also
is this a drag cutter
like adrag cutter, just a poser really
in sink edm you dont turn tool or work, its like punching
tomp, are you looking for the distance between the start and end points, or the length of the path traveled?
length of path traveled
so C moving 360 = 2*pi*radius, not 0 :)
the xyz is easy, the arc travel is easy, is it just the sum?
(good to know )
tomp: no :-)
you need the integral of the sine and cosine components of the arc travel
(which luckily are also sine and cosine funcs)
hey I bet that's right
but it's also dependent on the other axes, and I think you need to intrgrate the overall motion, not the components
the sin & cos of the angle are known, but the rest is vague, integrate form where to where ? from begin angle to end angle?
so it's integ(x+ [X component of C])
from start of move to end of move
so that would be X from x0 to x1, y from y0 to y1, etc
oh, are you looking at the total X motion? the part of X caused by Xaxis and the part caused by Caxis?
it may end up being a simple formula when you're done, but I think that's the approach to take
so you need to integrate the sqrt of the three components along the path
sqrt of sum of the components
on a good day, when it's not 1:30 AM, I might be able to do that :)
sorry,i'm looking for the total distance traveled in space not in any axis
understood, that's the dqrt of the sum
this is the basic calculus way of doing it
sqrt( 1^2 + 2^2 + 3^ + 1.57079^2 ) ??? cool
no, you need to integrate that along the entire path
which will hopefully result in some relatively simple equations, but it may not
are you saying that the tool travel distance is affected by the c rotation and the x axis travel? (to simplify it to just 2 axis )
consider that there is no travel if X+Y are moving in a circle, but C is rotating to keep the tool stationary
there's clearly travel in X, Y and C for that case, but they sum to zero
(the displacement of the tool is always zero in that case)
that may be a bad example, since I think you're looking for a linear move in XYZ plus some motion in C
i'm not sure what the C axis is for in this situation
so you can maintain orientation of the tool, I think
err - maybe not
oh, right, the drag thing
it's for cutting vinyl, right?
no, it's EDM
c creates a pose , a sinking edm sinks a form at 12" radus from center at 30 20 and 10 degrees
then i'm really lost.
that didn't help! but nevermind.
its simpler than other motions, neither the tool or work spins
like a rubber stamp
oh, so you're just orienting
well i'm trying to calc the velocity of each axis by finding what proportion each axis moves of the total motion.
and even if the net result is null, the velocity depends on the units moved not the distance between start and finish.
SWPadnos: what am i looking for if i want the work done, not the net effect? regarding "consider that there is no travel if X+Y are moving in a circle, but C is rotating to keep the tool stationary"
if this were a stepper, i'd just add all the steps and ignore the direction bit , that would be the work done. yes, this is just semantic. work is a better word than travel.
I'm not sure I can think it through at the moment. maybe after a god night's sleep I could
hey, thanks for talking about it :)
A'God night' sleep huh... I wonder how long that is???
all of Sunday
lol, very good =)
OH MY GAWD!!!
tomp think you need something like (cos(C)*mumble + x )^2 + (sin(C)*grumble + y)^2
toast_ is now known as toastydeath
Hello All, what do I need to change to speed up step pulses in the ini file?
if you lower it too much, it will lock up the machine
if that isnt enough there's a new double step rate driver in CVS
but i'm not here, so good luck
Hey alex_joni, r u there ?
just dropping in to say thanks for helping me out, got my pluto reading the encoders today, now just have to build some drivers and I am in business :)
Hello .. I have a question regarding datum of a mill that has travel of X=160mm Y=90 Z=115mm. The mill (A Denford Starmill) has three micro switches which I need to travel (one at a time as they are looped together) too to set the home possition. But I then need to travel X-160 and Y-90 then set home again as the micro switches are on the opposite side to standard cordinate setup. I wish to...
...do this automaticaly has anyone got any ideas?
Paragon37, that can be done by setting HOME_OFFSET. I don't know the exact setup you need, but homing should be well described in the user manual
SWPadnos:has anyone ever used EMC with a Galil?
don't think so, but I don't know
I think the Galil cards like to do their own PID, and that doesn't fit well with EMC
that may be why its running in a dos program
I think their cards are good, but there's no EMC driver. I'm not sure they release their programming specs, and if not, there won't be a driver ... :)
i heard they are good also
I have just set HOME_SEARCH_VEL = 1 ie non zero does this turn homing functionality on ie via switch .. or do I need other directives, I am using the stepper_inch ini file.
Paragon:HOME_SEARCH_VEL is how fast it travels to the switch
the latch vel is how fast it moves after it hits the switch
mine is .5 and -.1
i can pastebin my .ini file if you want to use it as an example
[14:17:20] <jlmjvm> http://pastebin.ca/720598
alex_joni:this is what I was gonna put on the wiki,if you think its ok
jlmjvm: it's ok with me
but you might want to make it a bit more readable (more & shorter paragraphs)
* alex_joni is a bit tired.. hard to read the stuff when lacking concentration
perfect.. maybe some pictures / schematics later
03alex_joni 07TRUNK * 10emc2/docs/src/ladder/ (5 files): initial ladder introduction
jmkasunich__ is now known as jmkasunich
is that one smileyface beating the other (dead) smileyface with a folding chair?
umm - yah
I see why people like web bbses - you just can't get that kind of thing on a normal mailing list.
fiscal year end here. yeck
well, he closed half of the loop
yep - he is making some drives based on modified H-bridge that jmk and I had made.
those motors look quite a bit smaller than yours
dare i say it might be overkill
I think he is scaling it down and making some sort of current limit
I wonder if I have the smallest fully-functioning servo emc machine
cradek: any new pictures of the vacumm table?
nope not yet
I hooked it up yesterday - I need more vacuum (as I think we all predicted)
what are you using for vacumme?
that little 15"-of-water pump
it's more like a vacuum cleaner
ah - we have a 1hp vane pump :)
no clue what it gets down to - but it has enough volume
did I mention I hate doing docs?
I have two? other vacuum pump shaped things in the barn, I don't know anything about them
they may both require 220 though
left over from milking?
so I'll worry about it in another week or two
alex_joni: Heh, it's not so bad
anyone remember the cli tool for tweaking services?
something like that
sysctl - configure kernel parameters at runtime
let me try that
there's sysv-rc-conf in ubuntu
yeah... that's it!!! ty
If I disba;e cups, will that prevent ghostscript printing too?
no idea :D
LET's FIND OUT!!! =)
I think it shouldn't
03alex_joni 07TRUNK * 10emc2/docs/src/ladder/ (5 files): more classicladder docs
oh gawd... festival is running (text to speech) and I don't even have a sound card! Eeeeesh
This is almost as bad as M$ and running crap that you really don't need
at least you have the option to turn things off :D
you have the source JymmmEMC ... you can remove it!
use the force Luke
alex_joni: If I knew they were even running.... there's no gui that has all the services, just a limited subset
use the source luke
JymmmEMC: ps -aux has 'em
use the source JymmmEMC
alex_joni: how do you think I found them running in the first place, but that doens't let me shut em down.
hi guys... working on a lathe config
SYSTEM > ADMIN > SERVICE doesn't even show hp in the list
LawrenceG: Use the soruce luke!
rebuilding latest cvs as we speak
LawrenceG: apt-get or nuttin!
oops... make install fails...... something about ../bin/stepconf : no such file or dir.... rats
ok ... who broke it?
03alex_joni 07TRUNK * 10emc2/docs/src/ladder/classic_ladder.lyx: fix line too long for pdf
did you make clean/make first?
cvs up -dPA
in the emc2/ dir
trying that now
LawrenceG: following these directions?
[19:27:25] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_5_10_and_6_06_from_source
yes..... have built many times..... just todays is acting up.... make clean might help if some things have been rearranged for the 2.2 release
jymmm you whiny whindows hyuser
I want pre 2.2 for the named varaiables in gcode....
with owords, gcode becomes a full programming language
fenn: If I wanted to work harder and be all stressed out, I'd just hire oyu instead ;)
03alex_joni 07TRUNK * 10emc2/docs/src/Master_Integrator.lyx: include classicladder docs in the integrator manual
JymmmEMC: yep i'd keep ya busy
fenn: Yeah, probably cleaning my gun and burrying the bodies
Is there any significance to a dollar-sign trailing on a service name?
pay per view?
lol, tha'ts pretty good
03alex_joni 07TRUNK * 10emc2/docs/src/Submakefile: add ladder docs
03alex_joni 07TRUNK * 10emc2/docs/src/index.tmpl: add ladder docs
03alex_joni 07TRUNK * 10emc2/docs/src/config/.cvsignore: silence
build works yea...
03alex_joni 07TRUNK * 10emc2/docs/src/ladder/.cvsignore: silence
a paper on preliminary tuning of servo motors, the 'step response' test to determine the system acceleraton http://gram.eng.uci.edu/~dreinken/MAE106/lab5.pdf
anyone know about this 'autotune' script that's supposedly in cvs?
fenn: it's not a script
it's another PID component
[19:51:39] <alex_joni> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/hal/components/at_pid.c?rev=1.9;content-type=text%2Fplain
The baby cockatiel isn't feeling too good today, not a good sign at all =(
any swedes around?
you're a bit too north
* alex_joni was wondering about the weather over there
it's starting to get cold..
we had 27-30 these days
nice and sunny
we've had more like 8degreed
Am I really connected to EMC IRC?
indeed you are
WOW it works
indeed it does
BigJohnT: got any emc questions?
First let me say thanks to the EMC community for thier work
I'm building a portable plasma table
I've got Gecko 203v drivers for steppers
would it be John Thornton?
what do I need to get inputs back into EMC
yes it is
hi .. hope you got my answer about THC
yes I've been looking at the references
seems like all of them use a commerical THC
you can probably do it yourself, but filtering noise & such is not going to be easy
I was looking at using something home made
noise from the voltage signal?
I have a Hypertherm 1250 plasma cutter
from the cutting
I have two outputs and one input
sounds like a nice plasma.. (kinda smallish for my taste :D)
biggest I could carry LOL
what kind of outputs/inputs?
the outputs are arc stable, and arc voltage.
the input is start arc
the voltage is true arc voltage and the arc stable is binary dry contact
BigJohnT: yeah, but the height conrtoller isnt..
the problem with arc voltage is ignition
I suppose it does HF sparking?
I think so
the problem with that is that it pulses something like 1-2kV
so I'm not sure what the true arc voltage outputs
let me look
BigJohnT: you should talk to rugdallur about it
Dallur might be around
he tried building/buying/both
and two different commercial ones I think
BigJohnT: either Dallur or Rugludallur (they both come online from time to time)
I ran my plasma without and it worked just fine
I plan on running it without to start
this is a small one with 48" x 24" inch
thanks gotta go for a bit
good night all
hi all... who would know how to compile a peeload of fortran source???
how do you pick a tattoo artist
watch your friend get one first
toast: ask someone with plenty of tattoos.
someone who looks healthy :-)
I bought a low-dollar edge finder from enco - it's the sorriest excuse for an edge finder I can imagine
the two parts slide against each other ok, but they're not even close to parallel
if I roll it on the surface plate, the end moves around and there are big triangles of light showing under it
hmm, name brand are only $12 - that was sure false economy
has anyone used the clicking kind?
I use a electric one
do you like it?
I thought about trying one, but the offset wiggling kind work very well
i use a guage blonk and indicator .0001 every time
you mean something like read, turn the spindle 180, read gauge block, split the difference?
ive learned to dived by 2 in my head to 7 decimal places..
I think for me that would be more error-prone
do you know how many times ive seen parts out by 0.100" precicly
too many hundreds
I always lift up, then go to 0 and sanity-check it, since I make that mistake too
i stopped using an edgefinder yrs ago
fortunately .1 or .25 is very easy to see
but ppl miss it tooooo often.. trust me.. ive been in hundreds of shops.. done thousands of setups.. and seen scrap to high heaven
none mine thankfully
use indicator.. swap for c/drill.. dbl chk.. rock and roll..
note to self... use finishing eye
assembler brings back flasshbacks to the Commodore "SUPER-PET"
why does this still look familiar.... ahhh im losing it
once you get over conditionals without using the words 'if', 'else', 'while', 'for', asm isnt so bad
it works great..
only bad thing is when you have painted/nonconductive surfaces
and whats wrong with edge finders?
I like my electrical one..
it's damn fast..
its still offset, tho, no?
the ball at the end are 10mm..
and you can see when theyre lines up with all the ones ive used
jog until it lits up..
okay so its 5mm offset
then you have your centre..
I centre on holes all the time with it..
for reaming and stuff..
jog to one side zero the machine jog to the other one.. and divide the distance by 2
then repeat for y..
well, if I need very very much precision I usually do it twice..
though I'm going to buy a probe.
i'd be faster on any hole
and as fast on any edge... experience
probes are nice if maintained..
if 1 guy bonks it its yeuchered
dmess: _any_ edge takes about 3 seconds..
renesis: are you on #cam??
need your help pls??
i take 5 to be sure..