anyone know why the pluto-p board is having problems to work with 2.2.2???
[01:14:06] <jmkasunich> http://blog.makezine.com/archive/2008/01/make_your_own_vaccum_tube.html
Tzak, was pluto working for you before and stopped with 2.2.2, or is this your first try with 2.2.2?
also, are you using pluto-servo or pluto-step?
pluto-servo .. i was talking to you a few days ago SwPadnos
oh, ok. :)
I have my other test computer running i think 2.1.2 and am making progress with it
SWPadnos u still there?
is there a circuit online for the pluto servo?
yeah - just watching that guy make his own tubes o_O
that is pretty impressive
yeah i saw that too really interesting
eric_U, the schematic for the pluto is only available to people who buy it I think
I dunno how much of the equipment he made himself
lots of it looked homeade
I like the spotwelder
the FPGA code and driver source are included in the EMC repository
what do you hook to it?
i'm not really sure what the problem as it is working on my other system
uh - a servo drive?
it outputs PDM or PWM to an H-bridge, I belileve
Tzak, can you try the liveCD with 2.2 on the system where it works? it could be a parallel port problem
so Tzak, you have an L298 circuit hooked up tothe pluto?
i havent tried that yet but i can...
I guess before you do that, it would make sense for me (or someone who really knows about pluto) to ask what the actual problem is
I definitely don't remember that, since I don't remember talking about it before ;)
was the oscillation problem you were reporting on cnczone on 2.2?
on my other setup i have everything sort of working.. but i can't get the motor to servo correctly
so oscillation on 2.1
no response on 2.2
it could be the paralle port so i will check that but on 2.1 i would still like to figure out why its not servoed?
too much integral
yeah i thin my numbers are like P=25 I=1 D=.1
really hard to say, my first controls professor said to never use a gain over 20 :)
with an op amp
forgot that part
"These lasers are fascinating and they emit a very powerful fun field in addition to their ultraviolet beam."
plus they make certain clothing glow, like bras
err - and white starched shirts
what is a "fun field"?
i have gone as low as 5 on the gain and have a huge deadband even though the deadband is set
it's like "Field of Dreams", only fun
its a field emitted from hacked gadgets
deadband probably isn't that big, the gain is too low to make it pay attention
ive made gas cans emit a fun field then
I knew someone who used his research laser to get free lunches for a couple of months
well, if there's a "flat zone' in motor torque/speed near zero output, that makes a "fun" discontinuity in the tuning
one of those scratch and sniff games
yeah so im not really sure whats going on ?
I'm not sure if there's a way to tune the PID around a driver dead zone
why would there be such a large deadzone?
I don't know how well the L298 performs at low PWM duty cycles
(or PDM, as the case may be)
how bad is the oscillation? It's fairly typical to sit there and hum
im running PDM
no its noticable the whole motor shakes
eric_U, I see this as a similar problem to using a linear scale - the PID ramps up a bit because there's no feedback for a little while, and when the motor finally starts to move, it overshoots
and so we repeat the process in the other direction
Tzak, another possible cause would be friction in the drive system
no the motor is not even hooked up to anthing
or in the motor bearings themselves ...
by all means, hook it up to something :)
try it without the integral term just for giggles
ok i could do that
heh - true. the added mass acts like a damping filter - it's much easier to tune (from what I've heard) when there's a load
i just dont want to have things going all over the place
what kind of motors do you have?
step 1: hook up e-stop button. step 2: hook up motors ;)
sorry i have to clarify actually im running an high speed discrete h-bridge ..
reminds me of a science fair project: me to daughter: "here, hold this motor"
daughter: "uh-uh - I remember what happened last time you said that" :)
she would say that now :)
i have tried with a I=0 but nothing changes
if it oscillates with P only, P is too high
hmm, seems you dont have anything to slow the motor's own inertia
debugging the circuit and the control at the same time is a difficult task
adding mass really will help
the motor should be stable with nothing hooked to it
my shaper cutter has decided to give up with good cuts, sigh
not in a feedback control system with multiple sources of hysteresis
and the delay using a computer can't help, don't know how much it will hurt, depends on the system
there is no friction.. the only friction now is in the motor
there is no friction.. only the motor has friction cause its not connected to anything
is it a servo motor, or a motor made for something else
no DC servo
minute variations in the rectified dc waveform or even in the ac wave form are enough to cause havok on an un loaded system, typically those slight varaitions are absorbed into the total mass being turned/moved/displaced, you have nothing to displace so your fighting how stable all your signals are
then you are relying on the mass of the system to make it become stable ... that is un-predictable
Tzak, what size are the motors?
no, it's necessary
tiny motors can accelerate at amazing rates. If you leave the EMC2 servo rate at 1 KHz, then the motor can get moving much faster than EMC2 can keep up
too bad the pluto doesn't cost $25, $60 is too much for a throw away project
if you increase the servo rate, then you might be able to get it working unloaded
i am running fairly large motors rated up to 400w
but that's not necessary for a real system, because the load works against the rapid acceleration of the motor, slowing down the system
with an L298?
he said he has an h bridge
you aqsked about an L298 and he said "yeah"
high power, unknown provenance
yes he did
a discrete h-bridge
so i was mistake with the l-298
he just wanted to get me to shut up, didn't work :)
is there any info on the output wave form signal of the pluto???
I think it's like the PDM mode of stepgen, only faster
so you could look at the stepgen source code
did you get the circuit from somewhere Tzak?
it's also possible that that's in the manual, but I don't know
do you know what speed its running at?
I believe the base clock is 40 MHz
I know skunkworks published one somewhere
I don't know what the minimum pulse width is
yeah do you know what the min and max shape of the pule width is? 0-100%
it's possible that the FETs (assuming that's what your bridge is) aren't being turned on fully with the very narrow pulses at low duty cycles (??)
dunno - there should be a min and max parameter, but I don't know how they're used
any idea where i could find that out?
in the pluto-servo docs I hope
man pluto-servo might shed some light
the source code is available
also the sample config
and the EMC2 manual
and the source
"The PWM period is approximately 19.5kHz (40MHz / 2047). A PDM-like mode is also available." http://linuxcnc.org/docs/2.2/html/hal_drivers.html#sec:pluto-servo
thanks - I was about to go look for the online docs
jepler ..do you kow what the pwm width and deadband are on it?
skunkworks is hiding
You would configure a "dead time" by setting max-dc. e.g., a max-dc of .98 would mean that 2% of each cycle (about 1uS) is "off".
(this also means that values of .98*scale and above do not give any additional duty cycle)
ah - and there's a corresponding min-dc parameter for setting the "base PWM width that will actually cause some motor motion" ??
value = value / scale + offset;
if(value < -max_dc) value = -max_dc;
else if(value > -min_dc && value < 0) value = -min_dc;
else if(value > 0 && value < min_dc) value = min_dc;
else if(value > max_dc) value = max_dc;
here's the logic that the driver uses
ok - cool
that should help :)
after applying scale and offset, it ensures that the value is not outside +- max_dc
then if the duty cycle is positive, it ensures that it's at least as big as min-dc
hmmm. what happens if the value = 0?
and if it's negative, it ensures that it's at least as big as -min-dc
duh - nevermind
it doesn't get changed, that's what happens
so you can get zero duty cycle even if you set min-dc
sometimes I amaze even myself with dumb questions
right - useful when you have DEADBAND for the PIN and TP
which is confusing, but so is deciding whether to give -min_dc or +min_dc for a duty cycle of 0
no - zero should be zero
in fact, if fabs(value) < epsilon, the output should also be zero
but choosing epsilon (unless it's a parameter) would be a pain
and it would be irrelevant if it's <1/2047-ish, due to the PWM/PDM resolution
another thing should be noted: if you're using pdm mode, the 'interleaving' means you won't get your entire 1uS of guaranteed off time at once
so if you need dead time don't use pdm mode
ok well no i have allot of testing to do... I will get back to you guys and let you know where i end up.. thanks for all the help
good luck - I hope you figure it out :)
so here's what the transfer function should look like: http://emergent.unpy.net/index.cgi-files/sandbox/Screenshot-Gnuplot.png
for scale=1, offset=0
that link didnt load??
can your browser display PNG files?
(it loaded fine here)
ok never mind
thanks for all the help
was i wrong about the motor needing somthing atatched to it?
no - the tuning is totally different anyway, and an unloaded motor probably responds too fast for a 1ms servo loop
can small voltage variations be enough to imbalance a motor?
you mean "make it work funny"? :)
ok :) (imbalance does havea meaning for 3-phase motors, but not so much for DC)
if you have a drive -- motor combo, and the drive is holding the motor steady, and the motor gets very small voltage change between the coils, can it sorta rotate on a level that the encoder would sense and then the drive try and react to?
as in the diference between being hooked to the wall, and a car battery, where the wall varries ever so slightly in both voltage, and frequency
more load on an ac system results in a drop in the frequency
that would depend on the type of drive
im talking in terms of a whole town here
a torque mode drive controls current (torque), so it doesn't care much if the supply voltage changes
a voltage-mode driver might care more, but in essence, a servo drive is a switching power supply
even in a fixed state?
the output is just regulated by tach feedback or something
well, that's why you have feedback - if a position error is detected then you you can correct it
I think you'd need some pretty major variations to have a problem, but that's just a hunch
so a load on the end of the shaft would be an object at rest, resisting slight issues with drive timming , but where no load exists, the motor does what it wants to do in its "frictionless " state right?
err - no, I don't think so
there's more friction, but possibly more important is more inertia that resists acceleration
im just trying to learn, I just dont see the purpose of a non connected motor, even a spindle motor gets enough friction from the drive belt to have the motor in a slight bind of sorts
an unloaded motor still has some friction (more at rest than when moving, as usual), but once it breaks free of static friction, it just zings up to high speed at high accel
oh sure - the drive gears/sheaves on a Bridgeport use up something like 10-20% of the motor power
ive had to make small adjustments on fanuc drives that when the spindle would go in for tool change, it wouldn't stop, it would rotate a bit, every sumer and winter we had to do it pretty much
I don't know a lot of specifics - I'm talking mostly from background information in electronics, physics, and mechanics
(and I'm not a major expert in all those fields)
oh you know what, all that changed when we got the electric company to change our tranformers from 2 to 3 or something
im just picking your brains here
sure - just don't take what I say as gospel - there are people with actual experience that really know the answers :)
open delta to a y delta i think is the change we had them make, they left basicaly a scope hooked up to our lines for like a month, cause we told them we had power issues, and they woudlnt belive us, till they tested things
well, its wierd for me righ tnow, normally this time of night im deep in a dungeon with my friends tryiing to defeat a monster
I think this is when I say that my least favorite electronics class in college was "power transmission"
I hope they are beating a monster right now,
glad to see someone else hated that power xmission stuff
some of it is handy though
plus, the professor alwas reminded me of Alfred E. Neuman
so it was hard to keep a straight face
did the prof wanted to teach that stuff?
oh yes - it was his specialty I think
the one I had didn't want to teach it but I think it was his turn to pay the piper
heh - we had a prof like that, but instead of being an ass, he was really fun
it can be a prof's speciality but that doesn't mean he want to teach a required class to a bunch of EEs who only cared if they can get a C to fulfill the requirements
we had exam questions like "a planet of xx size has a crust of yy thickness, and is filled like a giant jelly donut, with dielectric yy - what is the capacitance from the north to the south pole" :)
ack E&M nightmares
I liked a lot of that - we had a great professor for a lot of those
Ken Golden - a physicist who ended up being the EE department head
he'd say funny stuff sometimes as well
EE's had to take it twice in different depts using different units and it was not any easier the second time around :(
one day, he was filling the blackboard with "interesting" equations. this happened to be during exam review time, and of course we had never seen what he was doing
they never seem to explain wtf you pick a given path for the integral other then... by XYZ law, the term is zero and...
he turned around to see 50 slack jaws, and said "oh, don't run up your laundry bill - this won't be on the exam"
CGS vs MKS
the constants are different and there are these factors of PI all over the place
though it's only a few decimals
shouldn't be any extra pis there
there is if some are done in radians and others are not
or something along those lines
one thing my mother taught me was "always balance your units" - that served me well on many exams
balancing units works great in chem
she was a physical chemist at the time (PhD - both parents)
oh then there is the fun about the coordinate systems... cartesian, spherical, cylindrical
those weren't too bad for me - I have (had) a good spatial sense, and had learned about coordinate systems as a kid
except when you have to do those integrals
terms drop out in certain systems and in others you get all these sins/cos or complex exponentials :?
one thing that was almost always true (on engineering tests anyway) was that you could solve all the problems with a simple calculator, or in your head someties
for the "in your head" ones you were esxpected to be able to multiply/divide by 10, 5, and 2
you didn't have to get a numerical answer - 23*PI would suffice
yep, or 2 * sqrt(2)
so that was a good check on your work - if you were looking at sqrt(437*ln(X)), you had probably used the wrong formula
what irks me now is, HS students are not really taught to solve problems and be able to present them in these exact forms. they are told to use a calc and give a numerical approximating to so many digits
also, all the constants and formulas you'd need were given on the first page (in addition to some you wouldn't need, just to keep things interesting :) )
right - that's bad bad bad
I was lucky enough to have a calculator (still have them actually) that had a catalog of units, given in base units
(e.g. 1 watt = 1Kg*m^2/s^3)
I guess it is the audience but the G code class instructor just say use a calculator and punch in stuff blindly... it is useful to know 45 triangles are X, X, Xsqrt2 and 30deg tri are X, Xsqrt(3) and 2X
so balancing units was a very useful tool, when I could decompose W, V, and ohms into base units
older versions of mathcad are great for that
well, mathcad didn't exist at the time, I think
it was around since around 1990
well I went to college before then :)
whoot, another bad cdrom drive, I get another stepper :)
why am I not surprised to find that the RPM nameplate on my shoptask is wrong
it says 460 RPM for this belt setting
spindle encoder says 580ish
and my antique Hasler hand-tach agrees with the encoder
is the motor turning at the right speed, or does it seem that the ratio is wrong?
its an induction motor, can't be going too far off speed
I wonder if china uses 50 Hz and they calculated the speeds accordingly?
thats not too far off
50 yes, but not 100 rpm
the plate says 460
that would be 550 - still off by 5%
the diff between 460 and 483 I can put downt to belts - its just v-belts, they're never exact
60/50 * 460 = 552
I'm running unloaded, so the motor is going to be turning faster than nameplate RPM
jmkasunich: You did say this was CHINA label, right? =)
no, the label is made in the USA ;)
there you go then
at least the chinese can count
I wonder if we ever checked out shoptask
Whenever I see "shoptask" I think "shopbot" - those all0in-one thingies
the shoptask is a mill/lathe
the drill press, table saw thing
but it has separate spindle and lathe motors
lathe/mill/drill they claim (three in one)
table saw too
no, the shoptask
its actually a half-decent lathe, a rather weak mill, and a POS drillpress
shopbot we don't care about
compared to a drill press (tabletop), it's not bad
though it does lack depth
3.3" of Z travel and no ability to move the table up/down = shitty drill press
no, no, what's that thing for woodworkers that has a saw too
shopsmith I think
yeah, thats it!
well I'm happy my encoder is mounted, wired, and working
[04:36:38] <Jymmm> http://www.shopsmith.com/
threading fun tomorrow
That's what I alwasy think of where I hear that.
the encoder mount: http://jmkasunich.com/pics/spindle-encoder-mount-1832.jpg
my machine is much older than the ones on that website
no head lift
smaller table (only two t-slots)
hmmm. I don't know the table on the one we got
it was around 1994 or so though, I think
doesn't have that brace thing on the right (an attempt to stiffen up the floppy millhead)
I don't have the weird-ass combo tailstock/steadyrest
we got it when they were still off-white instead of yellow :)
I got mine in 1998, its yellow
anybody used 80-20 or Bosch aluminum extrusions? opinions?
this one may be the one we got: http://www.shoptask.com/17-20xmtc.jpg
tomp2, yes, but not for a machine
SWPadnos: very little difference between that one and mine
in fact, I currently have an open-frame LCD monitor mounted on 80/20 on my desk :)
yep - the color seems to be the main difference
mine has buttons to control the motors instead of knobs
I think the older ones had 1HP motors, but I could be wrong about that
mine has a barrel switch ( Furnas )
no they went to the pub
ah crud :( I can't recall what these are called in english; http://i5.ebayimg.com/01/i/00/f1/f5/68_1.JPG
flange mount self-aligning bearing
title in catalogue is "two bolt flange unit"
hmm, isn't there a general term for them?
self-aligning bearing maybe?
or would that include spherical bearings?
my first reply as it has a flang mount
ok, what about a general term that includes all manners of mounts?
self-aligning bearing does also include other types perhaps self-aligning bearing ball bearing
hmm ok, I see
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-01-09.txt
hmm - wasn't there a discution - maybe on the emc-user list about h-bridges. someone had a huge deadband in their hbridge - petew had said if your not shorting the motor windings out during the off cycle - you get some odd effects..
I think it was sebastian and the thread started on 10/8/07
At 0% duty cycle the motor is stopped, and at 100% it runs fast,
deadband? you mean temporal space between activating the different parts?
Duty cycles from 0 to about 55% give me increasing whining from the
motor but no movement at all. 55 to 100% gives increasing motor speed
where is the link to the emc-users history?
the mailing list?
skunkworks: you putting enough volts to the motor?
hmm, IIRC there was a link on the linuxcnc.org page near the link to the IRC archives
yep there are links to archives here: http://www.linuxcnc.org/content/view/4/8/lang,en/
this is what I was trying to get across.. http://www.cnczone.com/forums/showthread.php?p=390997#post390997
(atleast that is a little of what I got from the discussion last night)
my bridge seemed very linear.
basic friction in the motor?
I would guess his h-bridge is setup using 'fast decay mode'
cradek_ is now known as cradek
Hey, new pc for EMC http://newspaperads.mercurynews.com/ROP/ads.aspx?advid=32664&adid=5606843&subid=19392257&type=
[14:51:55] <JymmmEMC> http://store.shopfujitsu.com/fpc/Ecommerce/buildseriesbean.do?series=U810
[14:52:19] <SWPadnos> http://laptop.org/en/laptop/hardware/specs.shtml
SWPadnos: no touch screen
heh - and the keys are so small, I can't really type on it
SWPadnos: which the fuji or kid?
the kid one
SWPadnos: you should try the fuji.... it's even worse.
oh - I didn't realize those pads next to the touchpad are stylus areas
the OLPC is $200 ($400 if you donate one to a kid somewhere)
SWPadnos: Someone brought one in last night and let me play with it...
so you can get 2 and donate 2 for the price of the Fuji
hmmm. wsvga - is that 1280x768?
the Fijitsu needs a touch screen because it has no trackpad
it is touch
it does have a track pad
doesn't look it
it does and is sensative as hell too
top right corner
ooooh - and a fingerprint scanner too
I don't see the trackpad, but oh well
[14:57:16] <JymmmEMC> http://www.computers.us.fujitsu.com/www/products_notebooks.shtml?products/notebooks/photo_gal/gal_u810#
the blck sq
the whole thing is a black square ;)
next to the silver buttons
very top of pic
ah - that's a joystick/J-key kind of thing
not a trackpad (the larger rectangle you move your finger around on)
i hate those
I hate the j-sticks
or j-keys or whatever IBM calls them
I love em, works great
SWPadnos: Well, tonight's (basically) my last night
Couldn't get ahold of my boss yesterday
a lil of both
when do you start?
sent him an email, no reply yet.
see - if you'd do this next week, you could have a few extra days off while I'm there ...
SWPadnos: I'm taking two weeks off for vacation
gotta readjust to the daytime schedule too, I imagine
a lil bit, but I think time off between jobs will be a good thing, haven't had any for 15 months
worked turkey xmas and new years last year, xmas and new years this year
did you ever sign up for MacWorld?
SWPadnos: url and code?
guh - hold on a sec
[15:08:26] <SWPadnos> http://www.macworldexpo.com/registration
may still work
did you watch the drobo video?
SWPadnos: yeah, was cool, but I'd really like a NAS solution
[15:09:55] <alex_joni> http://www.acvt.com.au/research/videotrace/
I think they did a great job with the automatic management though
SWPadnos: Yeah, but says no RAID, so not sure what they're doing
it is RAID internally
"no RAID hassles" ...
alex_joni; yeah, quite nice :)
SWPadnos: not valid
here's the one DR sent me: 08-G-PC73
08-G-PC260 worked (googled)
wait, $10 wtf
Exhibits Only $10
I think it's probably easiest to take CalTrain to SF - do you agree?
try my code isntead
it's getting close to the show though, everything may be $10 now
your code is $45
hmmm. it was free for me
[16:17:55] <skunkworks> http://www.electronicsam.com/images/KandT/conversion/mainelectricalbox.JPG
notice the yellow aligator jumper near the bottom of the box ;)
(I think that was to override the hydraulic level sensor.. (we didn't need 200 gallons of oil in the tank anymore when we got rid of the hydraulic spindle)
*hydraulic oil level
i thought #emc was a little too quiet..
* fenn procrastinates disassembling the electronics bench
its the most modular "chunk" of my activities
so its the first to get put on the wagon
i'm moving to a new residence slowly
more space to play?
more space? or on your own?
in my old age i've come to realize that i hate change..
fenn: are you going to try and make it to the cnc workshop this year?
If I have two different locations on my plasma cutter for tools what is the best way to handle this? For example if I have the plasma in one location and a spotting punch next to it...
BigJohnT, I don't know that there is a good way to deal with that
you might be able to use some offset coordinate system
like g55 for one and g56 for the other?
something like that, yes
is it possible to use the tool table for that?
you can't use tool comp because that's always tangent to the path
skunkworks: yeah i think so
no, you can't
unless you hack the lathe tool shape stuff to work for "normal" milling
fenn: cool :)
sounds like fun
BigJohnT: if you have a lathe format tool table, you can have tool "length" offsets in X and Z both. If your two tools are different in X (but same Y coordinate), this will work great
yes they would be on the same Y plane
then it'll work fine
just use G43 like usual
"The way to learn, apparently, is to assert something with authority and then wait for someone to correct you." :)
make your "lathe" tools orientation 0 or 9, which is a circle with the controlled point in the center
SWPadnos: i believe that's known as the scientific method
quoted from a post to a Yahoo group that I happened to receive today
though I was 17% correct. all you have to do is use (not hack) a lathe-type tool table :)
[17:02:51] <cradek> http://www.linuxcnc.org/docs/html//gcode_main.html#r1_4_2
make your angles zero and orientation 9
crradek: thanks for the link I was looking for it but couln't find it...
so if one tool is +1X on the Y plane from the other I put a 1 in the Z offset?
no Z offset is still height/length
ok the diagram kinda had me confused
the diagram probably makes little sense unless you're looking at a lathe
Looking at my lathe and the diagram now makes sense if I stood my lathe on its head
tool on top?
Z is always parallel to the spindle
yeah that's why it says +X on the diagram - lathes can't agree where the tool goes :-)
got it thanks
I just wanna say that we just finished building sixth machine based on EMC :)
It works :)
EMC is REALLY good piece of software !!
hard to obtain but great
kinda confusing to configure it if your not box standard but easy once someone shows you how...
tomorrow we clean everything and then I get pictures and movies :)
just must remember :P
we dont mind the mess, want pictures
if it's too clean nobody will believe it's real
micges: what kind of machine is it?
I am already after work
in Poland it is 19:00 :)
table mill machine
lunch time here, fried duck and chicken wings
enjoy meal :)
you should add the pictures to emc wiki site...
thanks a nice ale will go good with it...
I must first know how :)
[17:30:11] <BigJohnT> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BasicSteps
machine table is 1125x1650
follow the above page
ok thanks :)
*BigJohnT went to get calculator from mill
44" x 65" pretty big
we builded also 1500x2500 :)
I can't wait to see the pictures...
present machine has analog correction of tool deep in material
tool height measuring and correction
table has vacuum hold of material
to what section can I add description of this machine ?
and vacuum clean of shaving chips
I dont know if I proper translated :P
it is worth showing you all :)
you can add a wiki page, and link to it from the page that lists people's machines
you can also link to an external site, so if you make your own site, just add the link to the wiki
my simple question: what is reasolution of measuring with G38.2 ?
it's the resolution of your feedback
except that there may be uncertainty due to backlash and also the feed
(or your step resolution if you use steppers
speed of reaction ?
no we have servos
speed affects accuracy, but not resolution
feedback - the encoder resolution
sorry I mean accuracy :P
the accuracy will depend on (a) the accuracy of your feedback, (b) any backlash (related to feedback accuracy), (c) the servo thread period and (d) the speed you do the probing move
a and b are basically the uncertainty in the feedback mechanism - backlash and encoder errors will make the numbers wrong
c and d are the response of EMC to the probe input
if you have a 1ms servo period, and you're moving 600mm/min, then the machine can move up to 0.01 mm during a servo period
I actaully did a microswitch tool touch off with the spindle running.. but the carbide mill was dull enough and the micro switch pushed so easy that it didn't effect anything.
if you move 60mm/min, then the uncertainty due to timing is 0.001mm
ok I will calculate this
if you increase the servo rate or decrease the feed rate, those inaccuracies will be reduced
but the accuracy will never be better than the resolution of your encoder feedback
I want accuracy of 0.01 mm
now I have 0.02
what is your encoder resolution?
Im not sure but about 5000/mm
ok, that's more than enough :)
how repeatable is your tool lenght sensor?
600mm/min = 10mm/sec, or 0.01mm/millisecond
so that's just barely good enough for the accuracy you want
it would probably be better to usea lower feed rate, like 300mm/min
you can also use a relatively fast probing move to approach the part, then use a slow probe move when retracting, and only log the retract move
the probe cycle would be reasonably fast, but more accurate
(assuming your probe is just as accurate breaking contact as it is making contact)
how does one do that with emc?
use the newer G38. probe codes
a simple subroutine for depth probing would be fast probe down, slow probe up and log (if you can choose to log only certain probed points), rapid up to safety height
"The tool in the spindle must be a probe." what's this supposed to mean?
it means you'll break stuff if it isn't ??
I first lower speed
do you know on what things in EMC SEVO_PERIOD has meaning ?
function linked is named motion-controller
the motion controller and usually PID (and classic ladder) runs in the servo thread
that mean that every pin in module MOTION has delay of 1ms ?
when you're dealing with the "real world", such as a probe input, the input may change at any time during that 1ms period
but the motion controller will only "see" it the next time it runs
that delay can be anywhere from "almost zero" to "a little less than the servo period"
(or even longer, depending on your max RT latency)
I must write this down to check tommorow :)
ok writed (pictures, check measuring :)
I have laser
and throttle response to M04/M05 command vary from 0.01 to 0.2 seconds
any suggeesttions ?
changing CYCLE_TIME in IO section has no effect
time between M04 and start burning
hmmm. I'd expect M3/M4 to be in RT code, but I'm not sure
someone else may have to answer this one
we measured this many times and it is about 0.1 sec
micges: do you wait for a response from the laser before starting a move?
delays of 0.1 before and after a move
spindle on/off is not controlled in realtime
I think it is IO module
cradek: is there something else to use for laser/plasma to control the arc on?
does the machine control Z or is it only XY?
some people have used a comparator on Z (<0 is on, >0 is off) for some digital control like a pen plotter
even the digital outputs with M64 are not fully in realtime
if you need exact realtime response I think all you have to work with is motion
M64 has also SERVO_PERIOD delay ?
which is about 1ms ?
not realtime so the delay is not predictable
only motion is realtime
does it make a difference in the cut quality if the laser is on 0.1 sec before starting motion?
Its possible to IO also be realtime somehow ?
Its make difference
600 W source of beam
and in future more power
ok didn't know I building a plasma cutter and they seem to be more forgiving
laser to cut die cuts
rows burned must be very precise
cradek: have some time to talk about AXIS ?
I cannot commit to any particular length of time, but there's no reason to require that in order to talk about something
what was page to share images ?
there's motion synched digital output available, (m66) but it has not been tested very thoroughly
fenn: it's not realtime
(i thought io was realtime ever since all the separate io modules went away)
wtf does "motion synched" mean then
"a good dancer"
the io changes at the right time between the motions
instead of what?
changing whenever the interpreter happens to feel like it?
after motion starts or before it ends..
well that was one "feature" we originally had, yes
if you think the docs are bad, please fix
cradek: you also wroted python modules used by AXIS ?? emc, gcode, rs274, hershey...
yes I probably wrote some of all those
my name is probably in most of the files I've helped write
the hershey fonts are some very old public domain vector fonts
AXIS uses them to draw the dimensions and XYZ labels
they were also used in old autocad, hpgl (hp) plotters, etc
wow - is that for your laser? creating a dieboard?
looks like your making a cardbaord box.
of some kind
this is an AXIS from MILL and gcode form laser
my programs are exchangable between machines :)
when I was writing this I have thousands of problems
for first: any direction to improve/add functionality to AXIS ?
all submitted patches will be thoughtfully considered
I think we don't have any immediate big plans if that's what you are asking
it would be very nice if we had a pl translation, but we don't, that's a place you could definitely help
looks like you have done some or all of the work already
almost all are translated
where to find how to translate ?
that you can add to officila
jepler: can you answer that?
[20:21:12] <skunkworks> http://axis.unpythonic.net/translations
gettext and ".po files" are a standard way of translating applications on linux; you should find lots of information about it (including GUI programs you can use to help you) on the internet or with google.
any clearyfing/cleanup pathes are also appreciated ?
clearyfing/cleanup of axis.py code
translation seems to be quite easy I hope
did you get your program compiled
not yet. I am in the middle of putting the GPL2 license on it.
I connected from the shop today. we will see how this works.
uh-oh we'll never be rid of you now
I don't know. It's possible
I will be in and out.
cradek: did you see the gpl2 note in the files?
yes in the new ones you sent me
I will put them on www.mpm1.com:8080
if anyone has downloaded them they can downloaded them again if they want.
they are there
maybe someone will be able to use them
you have a lot of nice pictures there too
how is the G&L by now?
Randy M. was here until 9:00 PM last night. He is here today. He says we will be able to turn the G&L on today.
If that is the case I will be installing stuff tomorrow.
I can hardly wait
man that will be scary to get tuned
It has scales.
but motor encoders too right?
do you have both encoders and scales coming into hal?
it would be nice to try jmk's recommendation of using the scales only for the "I" term
that way it will always settle to the right position, but not have the instability that you get when you use only scales
we could try that
I think it would only take a change in the HAL PID component
the other option is just use the scales to map the screws
when we get it going we will try to use both.
what a waste to use the scales for only error mapping.
is it hard to get 8 encoder inputs with ppmc?
no - you only need two encoder cards
seems like a lot of people have asked about scales recently. it would be nice for us to have good support for dual feedback
there has been a lot of discussion lately
most of negative
it seems as if it is not as nice as it sounds
I think people try to use only scales, and that works badly
our viper has two encoders per axis. almost like scales. one encoder on each motor for motor control
it works fine if the system has 0 backlash..
and one encoder on each axis for position
cradek; (my apologies if you've already said this), but what are the bad effects of using scales? backlash that causes jittery motor movement?
lerneaen_hydra: yes I understand that any slack anywhere in the system will make it untunable
or at best, the performance will be bad because it has to be tuned "loose"
oh, that bad?
but I think jmk's idea is very promising and I'd like to see it tried
was that on the list - or on here?
even better would be to have (analog) tach input to use for the "D" term
skunkworks: on the list I think
Jon E. made the comment about the resolution of scales not being fine enough for good tuning. I think he is correct.
yes but they could be used to drive the axis to position (the I term) - they have plenty of resolution for that
analog for D sounds interesting
cradek; hmm that sounds smart
you'll get rid of most of the leadscrew/gearbox/other stuff irregularity
BUT - you can't get much better than EMC2 will tune now.
when does the law of diminishing returns overshadow everything?
if you try to use the existing pid with scales for feedback, I think it wouldn't work very well
from the discussion I have seen I would agree but with encoders the tuning is very good and smooth
how much precision do you need?
not many projects need less than .001
these are the scales I want to get working :) http://www.electronicsam.com/images/KandT/conversion/accupins.JPG
stustev: what do you think is the accuracy of the screws on the G&L without using the scales or screw comp?
seems like it could be a lot worse than within .001 everywhere
skunkworks: don't be silly. that's a hairbrush.
the cincinnatis we set up had approx .003 inaccuracy in 120 inches. Each one was short.
I don't know about the G&L. It has always had scales.
my guess is the ball screws would be similar to the cinci's
you will be able to measure it directly by plotting both encoder positions. that will be very interesting I think.
I questioned some old machine rebuilders about the pitch being short. They said the old controls only had positive compensation. therefore all screw were ground with the pitch a few tenths short every inch.
If the control could only add error compensation (for whatever reason) then the screws would have to be sure to not have the pitch long or the control couldn't compensate for the error. That was what was explained to me.
it makes sense in a scary sort of way..
with the control capability today it seems crazy but then they were dealing with hardware issues not just software
if you'd heat your shop a little more maybe all the screws would be the right length
have a heater for each machine
lol @ cradek
hmm homebrew laser intereferometry
that's nice, hal pid has a maxoutput setting, that sounds useful for initial setup of big machines
Anyone want to hear my latest score?
* archivist just scored a rotary table
Qty 2: http://minicom.com/kvm_smartipaccess.htm
lewin1 is now known as lewing
hey everyone whats up
man. it is not easy to find a good rolling travel case for a 17" widescreen laptop
SWPadnos: Check out my latest score, qty 2: http://www.kvm-switches-online.com/0su51068.html
saw that - looks cool (though I didn't read the whole product description)
SWPadnos: Well, it's just a IP-KVM, wanted one for years, just couldn't justify it and now I got two for the cost of getting them RMA'ed
$8 shipping =)
whats the definition of climb cutting?
it's when the cutter rolls along the cut surface like a tire on the road
Gamma-X: the rotation of the cutter causes it to pull on the work
vs. the moonwalking (moon-rolling?) way - conventional milling :)
where each cutting portion of the tool, walks down the path, taking a step forward kicking the chip behind it, aka unconventional milling, the oppsite is where the cutter kicks the chip in front of it as it moves forward
and howdy yall, i just made it home
counterclockwise on an inside cut, or clockwise on an outside cut is climb milling
JymmmEMC, ok - that is a cool device
(assuming right-handed cutters)
SWPadnos: Yeah it is.... you can now have headless units in your home and access them anywhere
I was climb milling today with backlash and got away with it!
it sure can jerk a machine around thats for sure
spindle has very low torque and stalls
SWPadnos: or, in a DC, you can do complete OS installs, even update the BIOS remotely.
I use it a lot when using a slotting cutter
exactly what I was doing gezar
JymmmEMC, sure - I assume you just have to tell it the key-combo to switch the KVM, and you'r egolden
but were talking billions of teeth though
SWPadnos: It's a single machine IPKVM, just remote console if you want to think about it that way.
SWPadnos: there are multiple machine ipkvm's, but come with a comma in the price
sure - but to change a real KVM attached to it, you need to be able to send the right magic key sequence (for many KVMs anyway)
SWPadnos: or you hook a regular kvm multiple port to a ipkvm
that's the setup they have on the page you linked
SWPadnos: ah, I only searched for the model#
option 1 on the diagram
SWPadnos: there's a diagram?
speaking of which, have you madea diagram of what's supposed to go in your control box?
SWPadnos: No, still trying to get ahold of my boss.
SWPadnos: Then have my other boss in the other window chatting with what I want for tyos
oh - can I add some? :)
SWPadnos: I think I'm pushing my luck as it is... somewhere around $4k
SWPadnos: I'm trying to be gentle
SWPadnos: not sure on a 15 or 17" mb pro
[22:10:54] <SWPadnos> http://gizmodo.com/341413/alienware-curved-monitor-looks-like-its-from-another-planet
15 would be easier to carry around, but I like the larger display
me too, just make sure you get a good carrying case for it
the one I have is kinda crappy, and I'm looking for a good replacement right now
does it have to be a mac?
SWPadnos: Well I need one of each.... Just thought the mac laptop would be better than a Dell.
ewww to either
you'll need Linux, I assume
man, those are relatively expensive
SWPadnos: well, it' be MBPRO + vmware + XP/vista, then a desktop box
ok, and you need the macbook because OSX doesn't run on a non-mac?
right, but it could be a mac desktop
and a dell laptop, my choice.
does it have to be a Dell?
bummer. that's pretty stupid
well, I could get a ThinkPad I think
I dont mind the mbpro too much... gives me more time to play.
SWPadnos: I got PC's at home, just not a Mac (running)
SWPadnos: Having the MBPro will force me to learn the details
SWPadnos: how do you like your 17" ?
it's not as bad as I thought - the mac is only $500 more than a similarly configured PC
(like the one I have)
I like it quite a bit
SWPadnos: for carrying around and such
I have 3 pet peeves with it: 1) home, end, pgup, pgdown are FN+cursor keys
SWPadnos: worth the extra 2" for the inconvience it adds
it's heavy, like my old one ;)
yes - I got the high res screen, it's 1920x1200 instead of 1680x1050
[22:19:44] <JymmmEMC> http://www.apple.com/macbookpro/specs.html
peeve 2: the mouse buttons are a little weird - you have to press them kind of near the center or they don't work well
sure, I'm looking at the apple store right now
the 17" is a 1.5lb heavier
mine is heavier than that
hmmm - you can get mie with a 2x blu-ray drive for about the same price as the MBP
br is dead jim
heh - I think the recent stories have been the other way around - that HD-DVD is dead
they both are
ok, the video on mine is better - 8800M vs. 8600M
SWPadnos: I never do anything, just like large enough res for two-page up on screen
actually, for you I'd say the 15" is enough. you have computers at home, and you'll have one at work, so the amount of time you'll have to spend looking at the laptop should be minimial
oh, in that case the 17" is better :)
and 1920x1200, so you can actually read the text
SWPadnos: I'll get a 24" LCD at work, connected to the desktop PC, then use http://synergy2.sourceforge.net/
SWPadnos: That's what I'm thinking anyway
the 24" will be 1920x1200 I imagine
I'm not sure if it's better to match the dot pitch or the resolution with the laptop monitor
SWPadnos: then I can just set the MB on my desk and use one kybd/mouse for both
yes, I'm familiar with synergy
though I've never used t
I use it all the top with my laptop, love it.
can even copy and paste between machines
I say splurge on the laptop and go weasy on the desktop - desktops are easier to upgrade
is there an adapter to convert a pci to a pcmicia like use a pci card on a laptop?
the other way around
you can use cardbus/pcmcia cards on a desktop
i want to use a pci card on a laptop lol
I don't think you can do that
but most of the times you can find the same card in cardbus/pcmcia format
you can get PCI card cages that attach to PCMCIA/cardbus
also AGP and I think PCIe
SWPadnos i tried lookin got a link?
i wanna use those mesa net cards with my laptop
it will be better
in my situation
its a fast laptop and i already have it'
better than the laptop itself, but you'd be better off buying a cheap PC instead of a card cage and PCMCIA adapter
they're not cheap - minimum $300-400
[22:53:42] <SWPadnos> http://www.magma.com/products/pci/index.html
i gotta find some computer aprts damn
every business has a dozen old computers they don't want
many households have 1-2
well crap, that cdrom I just finished taking apart, had an elaborate zero back lash rack and pinion drive
I like the term "had"
how would I change the color of some of the components in the 5axisgui file?
I think jepler added color support recently but I don't know for sure how to use it
also, I bet it's not on the w-tool branch
too bad there is no driver for the mesa pport card
but there is
does it owrk?
with IO only, so far
I think someone is working on a better one
that don't do me no goood
eric__u: why not?
you can attach stepgens to mesa io's
is the servo drive working?
and it will run way faster than parport
the servo stuff works with the 5i20
stepper stuff doesn't work inside the 5i20 yet
i might just buy another 5i20
eric_U, if you have any FPGA experience, putting the servo / stepgen code into the 7i43 should be easy
but the 7iwhatever would save quite a bit of cash
I haven't looked at it, but I imagine peteW has made a 32-bit local bus that's similar to what the PLX chips do for the 5i2x
(with USB / parport on the other side, instead of PCI)
if i/o is working then that part should be easy to adapt?
yes - that was kind of my point :)
the HAL driver is probably the hard part
then there's the driver.. another nuisance.. but it should work eventually
all the code is more or less written - you'd probably want to combine some aspects of the PPMC driver (bulk reads/writes and shadow registers) with the mesa driver
SWPadnos: that somehow gave me the impression of a 100 monkeys snipping & stitching code together till it works right
I resemble that remark
most code is written that was, in my experience
how many windows programs contained generic.c?
it reminded me somehow of the monkeys & shakespear works paradigm
that's how most software programs are written, you only feed the monkeys that are working on stuff you want
err.. codemonkeys :P
that's a difference without a distinction
[23:21:06] <alex_joni> http://en.wikipedia.org/wiki/Infinite_monkey_theorem
what do you get when you put 1000 monkeys in a room with typewriters for a month?
1000 dead monkeys
I thought HAL drivers were easier to write than the old drivers
so is JMKasunich the only person that's ever written a driver for HAL?
is a hardware driver also a hal component?
sometimes it's even a RT component
good night all
is it always a kernel module?
eric__u: mostly it's a RT component / kernel module
yes all realtime components are loaded as kernel modules
there are also userspace drivers which are not realtime
mostly for misc IO
(like jeff's driver for the arduino board, or other drivers for modbus & co)
* alex_joni is off to bed