Some emc2 questions ....
What happened to FF2 ?
2.1 has FF2, 2.0 does not
Hopefully, it will be a minor effect until 2.1 is ready for public consumtion...
Have there been people who have had to retune , or have the EMC1 PID values worked right off in EMC2 ?
I think retuning is always needed because EMC1 used arbitrary units
the proportions of P,I,D might be the same though
Hmm...might be an exciting night then...hi...
Do not have any auxiliary I/O figured out...have just disconnected my "bridgeport" aux i/o...
Might have to figure out the Estop to get things moving...
is there a good 'how to' on installing ubuntu on a xp machine. (partitioning and such?)
is that on parport or stg?
skunkworks: is there space or do you have to shrink the existing partition?
skunkworks: I think ray shrank a partition with the dapper install BUT when I tried that I porked the old install so I can't suggest it without a good backup
I am starting with a fresh drive
going to put it on my portable.
oh, just install windows first, leaving disk space, then install dapper
it'll work really easy
(nothing to lose)
Its pretty painless skunk, once you get a CD that works...
I really don't care if it has latency prblems... I doubt if I will run a machine but it would be nice to have ubuntu at my fingertips to play
you might get lucky, but you probably won't :-)
(backing up right now)
Ubuntu seems pretty response...on the 800 MHz PIII here...
have a 160gb drive to put in it
USB drives even work...That is what I put into my machine here 160 GB...
wb9mjn: I have been running it on a 450 and 600 and it runs pretty snappy
emc2 and axis running dapper
Working on setting up my Motenc-100 servo machine with it tonite...probably will take all weekend, and part of
next week to get it where I want it...
Off work next week...
Edited the .ini file...
Need to get some Estop stuff figured out....might just try it, as the Estop will inhibit the amplifiers on an independant
circut, the way its setup ...
Nope...the Estop out from EMC has to be working, to enable the amps...gotta figure that out...
you had emc1 working before?
was that the same motenc?
One thing...the EMC 1 had a kludge config for spindle speed, using channel 4...
ok I thought emc1 had only stg and ppmc drivers for servos, shows what I know
is your speed control a dac output?
Is that setup in EMC 2 ? the 4 th DAC output for the spindle speed signal ?
commanded spindle speed is a hal pin - you can do anything you want with it - including hooking it to an external dac
I have an analog opto-isolator board that I ran channel 4 into , and the output of the board went into
the SCR drive to a 90 v DC spindle motor...
so it wants pwm?
No, it wants a analog voltage...
ok so it's just a regular dac output
you just hook them together in hal then
yep...that is why i have the analog opto-isolator board...feeds right into the pot input on the SCR drive...
I can figure that out later.....
it's trivial then
can it reverse? switch gears?
No...its a vocational machine...unidirectional motor and drive...although the spindle is mechanically capable of running
If I put a reversing drive on it some day...
is there a relay for on/off too?
There is a relay hooked to the E-stop, that depowers the spindle, and motion PS...
ok so to stop the spindle normally you just set the speed/dac to 0
The Spindle also has a latching enable and manual selector (comp control, off, manual control) switch...
that'll just work, spindle speed out is nonzero only when the spindle is turned on
No, I would switch the selector to off, or do a M5 ...
There are amp gain settings in EMC1 that made a very non-linear function, that was basically 0 v, when a CCW command
was sent to it...
It could eventually be servo speed controlled, maybe...
gotta run, yell if you have troubles
The drive already has the DC feedback speed compensation, though...
well i made some more headway on the edm.....
i messed with the adc gain and offset, and i got the machine to feedhold? before zero volts....
have been cutting tool steel this week with mixed results as far as finish is concerned....
i have lines in my part.....
compared to the aluminum....
what causes that?
or is that the hard question :-)
okay, today when i was cutting in the y axis i noticed that the x axis was moving .ooo1 kinda like bouncing back and forth, and when i cut on a 45 degree angle there were quite a few more lines....
almost like it is slightly(and i mean very slighty zig zagging thru the angle.....
could it be the tuning of the servo loop? seems for speeds that slow the tuning would have to be very good
very hard to tell on the 45 cut what is going on, but the straight cut in y, the x was moving ever so slightly + and - by .0001
skunk it could very well be....
seems like it is bouncing between encoder pulses....
instead of just sitting there...
maybe your deadband is set wrong?
you wanna take a peek at my configs? don't laugh too hard......
gee, now cradek get's to see how much of a moron i really am......
I feel the same way about my hal files
well there is only one way to learn, nad that is to expose yourself.....
altough these past 2 years of emc has been great....
sort of the swiss army knives of machine control
If I could I would use it to control my car engine ;)
i think it control my life.....
mac, can you observe any x motion when only Y is commanded and NOT cutting?
(is the observation connected to cutting?)
just the .0001 motion but not all the time......kinda like it is bouncing (searching) between encoder pulses.....
what is the scale? is 0.0001 really one count?
ok, so it's tuning ( stuff you didnt ask for :-)
scale is 40000
deadband .000015 seems right
cradek encoder is 2000 ppr, and leadscrew is 5 turns per inch
.2"/rev and 2000ppr full step?
output scale is 10? I don't even know what that does
would that be the scale to the servo? 10 volt
cradek nor do i, but that is what it took to get 1" = 1"
(if output scale is hooked to something in the hal file ;))
1 commanded step is .0001", the jitter you see is 1 command unit
setp motenc.0.dac-00-gain [AXIS_0]OUTPUT_SCALE
i think you might get further away from the error by having smaller steps
mdynac: I'm worried about your P-only tuning. You might have oscillation.
if scale is 40000, how is one commanded step = 0.0001
it should be 0.000025
1/4 step k
this is a servo machine, right?
no steps, just encoder counts
dead band should be around .00003 then? still don't quite get that.
theres are 4 somewhere missing if encoder is 2000 and pitch is .2"... gearing!
tomp: 2000 lines is 4000 counts
tomp: 2000 lines is 8000 counts
if the encoder is 2000 ppr (pulses per rev) then the counter will count 8000 times per rev
jmkasunich: if you're interested, http://timeguy.com/cradek-files/emc/motenc.zip
encoder is 2000ppr quadrature, diff output....if that helps
mdynac: your following error limit of 3" is pretty bogus, but that isn't the problem
then an observed .0001 motion is 4 command units, lots
it could be just one, if you're in the right place
mdynac: still could be a simple tuning problem (oscillation)
cradek ferror means? (in dumb guy english please)
mdynac: you could experiment with increasing P until you get sustained oscillation (you may have to give it a kick to start it - maybe a fast jog)
mdynac: the ferror setting tells emc when to abort motion because the machine isn't tracking properly. you're allowing a position error up to 3 inches
you probably want more like a dozen counts or so
i c now....but it cut the maple leaf just like that
mdynac: does it ever move (singl;e bump/tick) at idle ( no command? )
however my feedrate is like 3 inches per hour....
tomp yes, but not always....
maybe the leaf cut faster?
cradek yes it cut fine....as it was aluminum...
I recommend finding the P setting that will just sustain oscillation
cradek: make it as tight as poss
cradek, do i increase P util my axis just sits there and moves back and forth??
mdynac: yes. you may have to kick it with a jog or something to start the oscillation, but what you're looking for is sustained oscillation once it starts.
then you should be able to back off P a bit at a time until it damps out
moving back & forth... more like buzzes
yes it might just buzz, you'll see it on the pulley/belt/motor shaft
okay, i tuned the copleys via there set up procedure, which meant a square wave into them for osc...but i had no idea that i had to tune emc PID also, it confused me so i left well enough alone...
see how dumb i am...
so your encoder feedback goes to your external amps and also to emc?
no, encoder goes onlt to the motenc board, the amps are copley 412ce's check the website for datasheet...
yeh, no enc to amp cnonnection
ok I don't understand what you tuned then...
i bring a tach feedback and the plus/minus 10v to the copley
ok a velocity loop
i tuned the copley amp....
ok, you do have to tune emc too - it is the position loop
with a reference 1 hz square wave from my awesome tm500 tek rack....
cradek, yes velocity setup....
so i take it i need to decrease my ferror .....
after you tune the positionn loop
you might want to decrease it a bit to prevent any runaway
maybe set it to a half turn of the screw or something
model 412, "set s2 on, apply 5khz swq wave to REF"... if this is ok, then WITHOUT emc, it should never jitter (separate amp tuning from emc tuning )
(i use 423's
screws are 40 turn
tomp i tweaked the coply without emc connected
and would it ever jitter as much when not connected?
well, er, hmmm, i never checked that.....
(this is a tiny error... takes carefull action to eliminate)
cradek: he can choose torque (current) mode, and remove the dual velocity loops... opinion?
tomp: I don't know
right now 2 guys are driving the car ( emc & copley are trying to control speed )
however, when i connected emc to the tuned amp, i got a bit of drift, i tried to adjust in emc "output offset" but i still adjusted the balance pot on the copley to eliminate it.....probably wrong....i think i should have nulled it in just one place......
null while REF is jumpered ( no command input )
that's for tuning the amp (not emc )
tomp, gee why didn't i think of that................
no, i didnt either, i just read it
tomp, please don't tell me it is in the 412ce manual.......
the copley maintains velocity based on tacho feedback
(its in the 412/413/422/423 maNUAL in front of me )
so when you move slow, the copley onl;y has NOISE on the feedback loop
esp at .000zippo feeds
tomp is that available? i only could get the 412ce portion.......
i got copies :-) gonna show me the machine :)
i can send it
you got 5 pots & 2 swx?
tomp when are you available?
and a place to insert diff caps and resistors for diff inductance etc?
wait 4 pots i think...
yes 40 pin dip socket
yeh same thing, email me at tomp-tag at sbcglobal dot net for address to send to
k, tom and yes 5 pots....
the 423 can dump 40 amps at 150V dc fast... i like 'em but tachos are a bad idea in the long run
well i am only looking for 10 ipm max......
tomp email sent thx in advance....
speed and torque are diff, you can use a big amp to really hold position on a small motor... which is what you need to avoid lines in wedm ( imo)
i thought i overkilled them with the copleys.....
mail tomorrow... great stuff mac... i wrote analog pots for halvcp... maybe you can use them with your wedm for inputs ( desired gap, or wire speed )
tomp, thanks....stop by the shop next week.....let me know....
the 412 may be overkill ( the good kind, you want to be able to grab the shaft & have it fight you :-)
I'm feeding 60 volts to the copleys (unregulated)
sounds right for 412.. 20A pk 10A cont 24-90 V= brushed
i gotta go put in the hscale & knob & a meter (I hope )...
for the meantime http://shoutcast.omroep.nl:8022
i had to reconfigure the andrew power supply as it used a +30v and - 30v supplies to feed the motors....
sounds like jmk's bridge... ( or an ONA pump :-)
Andrew engineered their own propietary motor drive system....just like most other manufacturers did back in the '70's and 80's.....
cradek, so my ferror is quite ridiculous, and my deadband is set too low?
no I think the deadband is about right
it should be a little more than half an encoder count, which it is
why does the machine bounce back and forth .0001 then?
* skunkworks might finally understand that.
because it sounds like it isn't tuned very well :)
an old and respected edm guy told me that when a machine does that it is sitting between encoder pulses.....
he called it searching....
skunkworks, after this discussion tonight with the gurus, i know now that the machine is out of tune......
then my mahine is pretty well tuned.. even though I tuned it pretty randomly
my encoder pulses go -_-_______
hell the stock Andrew does that, and if i just jog the machine a bit it stops......
then both stop
(waiting build ) yeh, i call it hunting, and jogging usualy quiets it down.. but, doesnt solve it
yes, i occasionally get the "-" sigh flashing.....
I thought you should have a bit of jumping between 2 counts..
or is that wrong?
wrong is what makes bad parts ( like lines on wedm or cracking taps. ;-)
but with a quadtrature encoder, how can it "search" ? am i still sitting in between encoder pulses?
tomp: heh, yeah but how is it supposed to be?
mac searches for edges
well, hell i was a radar tech for 10 yeatrs, so a half a nanosecond is an eternity.....
anonimasu... depends on what you do,, it's ok to jitter an amount less than noticeable (again NAN not a number )
tomp: yeah, but is the PID loop supposed to stop jittering?
mac once the servo is stiff, THEN the control can work well
or will you always have jittering?
steves_logging is now known as steve_stallings
quickie quiz....what is a radar range mile (in microseconds) ????
anonimasu, pid doesnt stop jitter, it reduces vel err using the 3 principles, so it reduces, not elimin8
please, no googling allowed....
tomp: my servos do stop jittering between encoder counts.
i dunno speed of a radar wave
I would thinkit would be pretty close to the speed of light
6.18 to hit target and 6.18 to return.....
Servos will "hunt" if there is no deadband. Deadband specifies that no correction should be made if error is less than deadband. On machines where mechanical load is less than friction then deadband can make the motion compeletely stop when no motion is requested by trajectory planner.
hence we can then calculate the "doppler effect" to determine speed....
steve_stallings: is that a good or bad thing?
It is good for wire EDM because there is usually no mechanical load while at rest.
steve_stallings: dosent matte for milling right?
it's only at standstill..
sorry cannot find any info on the eltee wire.....
steve: this deadband has what unit of measure ? mV or um ? or ...
Milling usually has other influences that are greater than the hunting distance, so hunting is just an annoying audible noise, not a quality of cut issue.
ok, um or smaller ( this is AGie's tieffreeze .. frozen zone ... ok )
Not sure want internal unit in EMC is, must admit to being a looker, not a user. Probably in distance.
the eltee prints may not be so much an issue with mac's success...
In the old days resolvers were used for precise positioning. The had analog position output and old Hardinge lathes used them with great results to provide the slow motion needed for shallow tapers.
tomp the eltee prints are for steve.....
yep... steve: did you try any of the other edm user groups
steve_stallings, in the old days i serviced synchros, andresolvers in U.S. Navy radar systems....worked like a charm
not yet, long time until I have a chance to try to set it up so I only picked on you and Mac because you were here
ther was at least 1 other (non proprietary) and a german one... (funkerodieren )... lookin now
funny, I used an old radar tube as my first CRT tube on my first homebrew computer, vectors drawn in software by an 8008 microprocessor needed the long persistance phospor
gee, i thought all radar phospo was long persistence.....
modern radars use lots of digital signal processing and simulate the sweep on raster displays
looks like hscale widgets work now... back to the search ( btw: still got 'vectrix' game machine -- long persistance phosphor game machine )
true, digital is now king......
you are no longer "truly" looking at yor target on the display....it is a digital representation of what the antennna finds.....
and the antenna is a phased array that doesn't actually turn...
the apg-65 turned with servo motors.....and yes it was a planar array, but that is all i can tell you, or i must kill you.....!!!!
I worked on the other end, sigint looking for and identifying emitters....
yes, but private industry
the EA6-B was the most awesome piece of machinery on the USS Coral Sea.....
4 man crew to run countermeasures.....
stuff I worked on went on subs
ooohhhhh, quiet stuff, my dad worked for Bell Labs, Underwater sound project.......
are the black helicopters headed my way.....
direction finding, ultra-precise frequency measurement to differentiate between the quartz master osc frequencies of target ships, unique just like a sonar signature
correct, my man, i think we've said enough about that.......
don't ya just love how water retains information????
no groups yet, but found something on primitive electrode/ no wear http://www.mtm.kuleuven.ac.be/Research/GBOU-IWT/SPARK.html
nature's delay line memory
Flemish government !!!!
research on EDM
well after i get my obviuos screw ups fixed in my axis tuning, whatcha think about incorporating a backup function for edm?
becuase the adaptive feedrrate works like a charm.......
after i adjusted the gain and offset of the adc.....
backup by playing trajectory output backwards seems feasible, but is limited to memory depth, true reverse run of trajectory is much more difficult
uni leuven has always been an edm research site
she is runnng just like an old Japax.....
steve ; i gotta dig the groups outta old notes
are the typical carbides used for lathe inserts workable with wire EDM, i.e. homemade form cutters?
ooh, some new piezo head for senkerodieren http://www.f1.fhtw-berlin.de/labor/adk/tech/Laborversuche/Senkerodieren.pdf
mac dont make the system too stabile, here they add back in jitter to keep the process going :-)
sorry, I don't read German, much less technical German
i dont read german, just look at it
i am just using feedhold to not short......
what triggers feedhold
and, gee, it works.....
my adc input
maybe i should,nt call it feedhold......
do you use steve's deadband? (Verror < Limit goto feedhold?)
the emc adaptive feedrate function is basically....0 volts into my adc = no movement and 5 volts in = max movement.....
my description of deadband was about regular motion, I assume that the EDM feedhold is just that, a feedhold when electrode voltage is below threshold, deadband is bi-directional about zero
you might make a bit of deadband there... 0.005 v in = no motion
or more or less...
my axis movement speeds up and slows down according to gap voltage.......
yep, this keeps ypu away from the 'wall'
tomp - could what you are proposing be better described as a hystersis about the feedhold threshold?
(on the positive side only... if it goes neg,,, then run away!
by setting the adc offset, i can never allow movement below say 1 volt applied to the adc.....which = about 30 volts gap at this time.....
if there is not a "hystersis" module in HAL, it would certainly be possible to make one
my gap voltage can never reach a negative voltage....i see that as physically impossible......
another tuning issue is ... can the system respond to a tiny input ( one bit of the dac output )
mac, as gap voltage rifght... but as error ( diff between what yopu want and what you got... how a lot of control systems work )
it is responding to less than 1 volt right now, almost too sensetive.....
steve: deadband sound good for this? (velocity hysterisis)
mac: does a dac output to the amp?
I was thinking hysterisis on the comparison at the input to the feedhold request. Stop when Vin goes below X, resume when Vin goes above X+hysterisis
btw: the right side of the pdf is in englisch
tomp i take the sampled gap voltage (analog) and input it to the motenc's adc input, massage it a bit via the "gain" and "offset" lines and the via HAL input it tomachine motion.....via "adaptive-feedrate" function written for me by the emc guys...only available in the HEAD 2.2 version of emc
mac, i was thinking of the least significant bit of the dac....
can the system actually respond to 00000001 vs 00000000?
so the end result is feedrate controlled by a varying gap voltage.......just like an old Japax.....
i doubt it... and that's already hysteris... somewhere maybe 00000011 the system will begin to move
it is a 14 bit MAX135 chip........
excuse me MAX125
and whats thje full range? 5 V ( divided by 2^14 = a value the amp will ignore )
but at some value the system will move....
just get ther datasheet from analog devices.....
I have no EDM experience and defer to you two guys on whether it is better to have the feedrate stop and go, or attempt to slow down in a linear fashion
it does do sample and hold on the input.....
but you say it seems very (too) responsive now, so LSB is not an issue
and steve's clipping sounds good
you MIGHT be able to do this by lifting the LSB and sisters off the chip
assume the wire process does not use the "jump flush" technique like plunge EDM
this would get rid of the 'fiddly bits'
steve_stallings, yes the feedrate MUST be controlled by the gap voltage.......in a perfect world it should vary according to voltage.....high volts = fast motion and low volts = slow motion, and when it detects short, it should backup up out of preset short condition, just that simple....
hell, the entire OS of the Andrew machine was written in less than 8K of code, including many external funtions not directly used for the machine to run......8K of 12 bit memory.....
sounds like gap voltage needs to be evaluated into 3 regions, too low = halt feed entirely (later add backup when motion support exists), medium=scale feed to react to gap voltage, too high=stop on fault?
it backs up, edge finds hole centers........and stores and executes gcode....
steve....wire edm is all about gasp voltage.......
steve: yes, many system used 'curves' or ranges of vel for diff gap voltages
backup and edge find require motion features not yet implemented, hold and scale velocity in response to gap voltage should be possible in HAL
the physical distance between the wire and the work.....
steve_stallings, edge find has not too much to do with gap voltage....the probe function already implemented in emc should work fine....
scale of velocity could be linear interpolated or could be a table lookup or a more compex math function, still in HAL
steve: try linear ranges for now... but it ends up wanting to be 's' curved, yes 'hal-able'
at the moment i scale my adc to make 30 volts gap voltage look like 0 volts to emc.....hence no movement util the gap voltag rises above 30 volts then the axis move forward....
I think next magic step needing help from cradek, jmk, et all will be some way to back up to clear shorts, do you think a short memory buffer could work?
what would the buffer contain? gcode? tp segments? recorded encoder pulses?
samples at the servo rate
simple, but memory hungary
samples of what? position?
steve_stallings, correct....i do know that the pdp used a seperate discreet inut to sense backup, where it just ran a subroutine to move the axis .050 until the discreet input went high again as to move on with the g code.....
did it restrict the backup to a straight vector?
.050 which way? along the gear tooth flank?
no it reversed the direction of all axis to a preset .050, hell the new Chmer macines i can program the length of backup......
does that work on a .01" circle?
0.050" could be lots of servo samples, how much time would this typically represent?
could be minutes... so use motion ticks not temporal ticks
(chgs on encoder, edges )
my problem is when it is leaving one gcode line to another, and at that precise time it decides to backup.....
I still don't understand how it can back up in a straight line when you're not on a straight path
it doesnt , else it scraps dies
cradek, it reverses all axis that were being moved by gcode....
that is why I recommended position sample buffer at servo rate, play backwards at 100X speed
mac: you mean it reverses the gcode and travels .050" along thet reversed path?
you "just" have to keep enough gcode segments to represent that distance
reversing G code is VERY comples
not gcode segments really, motion segments (plain old lines, arcs in world coordinates)
um, i wrote trim cut stuff that reversed wire paths, moved off & chgd comp & power & went back... so i'm no genius :)
i can take a jumper and short my wire to the table and watch the axis move back and forth until i remove the jumper......
you don't want to reverse gcode because it has all sorts of awful modal stuff
Well, we got the keys today. It doens't feel like home, and the anxiety and turmoil of all that has happened over the last 4 months still remains (actually more to come as I'll be moving all this week and thru Christmas). But time will tell.
so buffer canonical commands
Even looked for a small Christmas tree tonight... nada.
worst case is a "simple" arc
motion segments really
the stuff tp spits out
motion segments have start and end, canon commands just have the new destination
congrats Jymmmm, I know the anxiety of moving well, just moved me, now moving company...
Jymmmm: glad you're on the right track again
steve_stallings: The moving itself is not a big thing (other than it being over Christmas), just the whole ordeal in actually getting a place to live *DEEP heavy sigh*
Jymmm benn there done that....twice.....
But Thanks, I Appreciate it, though I'd appreciate it more if you guy wanna carty a few hundred boxes - I'll supply the beer and pizza =)
yep, my business landlord is trying to force us to move two months sooner than expected, scrambling for temp storage...
something steve said a few mins ago gave me an idea
how big might the motion segment be ( at edm velocities) might that be >too< small?
he said "save positions at the servo rate, then play them back at 100x speed"
mdynac you had to move thru Christmas twice?
if you're gonna play them back at 100x speed, you only need to save every 100th one...
if you play them back any faster than normal, you can violate machine constraints
so maybe the motion gurus can choose where it is easier to buffer and replay motion....
Jymmm yes due to divorce and leaving the Navy....now you know why i build cnc machinery......it is much more forgiving....
not likely on a wire EDM, but you could compensate in ini files
jml: what about recording the last 100 encoder 'events'
I know edm guys don't care, but it would be nice if it were "right" instead of expedient for just EDM
tomp: at this level you don't know anything but the commanded position every cycle
tomp: encoder counts are way out on the other side of hal
mdynac: OUCH and DOUBLE OUCH! Well, at least I'll have a garage... not sure if I'll have enough power out there, but we'll find out.
cradek raises an interesting issue, wire EDM is so slow that motion would not likely exceed machine capabilities, but how should we handle reverse (or for that matter other, like tool change) motion that is not generated by G code?
Jymmmm you got heat out there?
I'm thinking of a separate motion queue that always starts at the current point and contains the reversed path for at least .05" or whatever distance we need
'events' may be diff from counts... the last 100 'occurances' of encoder edges... still unknown to hal?
mdynac: Only the heat from the water heater and the clothes dryer.
mdynac unvented clothes dryer at that (330VAC)
Jymmm i hope you live in Florida....
mdynac San Jose California
Jymmmm I got married at the La Veunta Inn in Palos Verdes.....oh such sad memories......
mdynac Well, at least there's still Vegas =)
or as I like to call it... VIVA LOST WAGES!
steve_stallings: generic interleaved coordinated motion and IO like some tool changes is a can of worms
Jymmm my ex's uncle founded a little company called "Body Glove"
steve_stallings: there's no existing infrastructure for it (except writing C)
yes, but we seem to have opened the can containing wire EDM, sooner or later the tool change will be unavoidable too...
Jymmm he just died a few months back.....
mdynac but Bofy Glove lives on
steve_stallings: tool changes that move to a particular spot, then change, then move back are supported now - that seems to get us by for a lot of machines
I need coffee and a shot or 80 of brandy...
steve_stallings: something like a grid of tools at the end of the table is pretty unworkable today
Jymmm Abosule and cranberry here.....
Jymmmm, were you at the fest?
cradek: by extension would this also be true for an extra axis to move tool holder tray or rotary disk?
you could move an independent motor using just hal. you could do that today.
if it takes coordinated motion (and unhooking from the coordinated planner) it's not so possible
so anything like a carousel that moves the desired tool to one coordinate in XYZ space can work today
well I hesitate to say "anything" :-)
cradek: you could actually use a hal hack to do the "grid of tools" version
cant the grid of tools be moved to without uncoupling ( inside gcode? )
sum a tool position onto the postion coming out of the motion controller
ugly, but i've done that kinda stuff
the tool pos would normaly be zero, emc would move to a certain place (perhaps tool 0,0 in the grid)
then the tool pos would be set to the proper value (with a limit3 to control accel and vel)
ok, all you have to do is be more clever/perverse than me and you can do it
after the change, set to zero again
in my limited knowledge of coding, i belive that backup is still a function of a subroutine called by an external input to the computer, it halts forward motion and the reversres what is in a preset buffer until the discrete input changes, then continues on with forward motion......
there is a neat grid style tool change shown at: http://home.comcast.net/~jlt313/Router/mov00429.mpg
wow, that is neat
interesting collar setup
and the holder system is commercially available
jmk: "sum ... onto" is interesting
[04:43:15] <steve_stallings> http://www.hightechsystemsllc.com/index_files/Page837.htm
steve_stallings: if you watch closely I think you can see radius comp turn on when he's moving into the cut - I wonder what controller this is
when it reaches said forward motion it checks to see whether said in put is toggled, if so it repeats its reversrse and forward motion, until the time it hits the originalforward motion and there is no longer the the backup toggle, so it moves forward as the gcode commands, until it sensenses the next backup signal.....
jmk: how would you get back to the planned posn w/o suddenly having a huge error ( i think 'jerk' )
tomp: limit3 hal block
note to self: gotta learn limit2 hal block
tomp: there's limit2 too - just the number of derivatives you can limit is different
steve_stallings: that's really cool
well if i can add my 2 cents worth, Andrew Engineering did it with a pdp8 and the entire code was less than 8K, and for that matter, BRAND NEW edm machines may have a kewl pc "front end" but the number cruncher is nothing more than a 286, or perhaps a 386 cpu to crunch all the motion control, the fancy pentium "front end" is just used for user i/o and pretty graphics....
I think they mentioned it was done in Mach, but should easily be possible in EMC
8 bits is plenty for edm....
jmk: hscale tests ok now
mdynac: be careful, if you say "it's easy" someone else will say "be my guest"
mac: there is no concept of reverse in emc, the whole idea is to keep the fwd velocity right and the arrived position on time
sorry folks, i never said it was easy.....just ststing facts....we are all here for the same reason....to improve emc as we know it.....
I guess our approach, which is to try to make a machine controller that can control any kind of machine, is quite a bit different from a dedicated edm controller
yeh mac: i'm saying Andrews began with the idea, not add in to a mill control later
it might even be the wrong approach for edm.
so, back to embedding motion in HAL, yes it is possible but we lose all the elegant motion control of the trajectory planner
maybe we could allow switching tp's and leave the one we got alone
ewww... did I miss someond suggesting that hal be used for backing up?
no, nobody did that
no, HAL for tool change
the tool changer conversation was a tangent
cradek correct...we can go on and on about specific edm stuff....but let me say that in my opinion, after having test cut material on my edm machine, using emc i think it is a feasable product.....
I suggested that ;-)
jmkasunich: then you're insane
um, we were looing at recording the last bit o motion for some reason
jmkasunich: backing up in hal? seriously?
no, toolchange in hal
jmkasunich: you don't know the path, you don't have data structures to save positions
let's back up 2 minutes and start over :-)
the prob is that 2 or 3 mins ago I wasn't online
(having accidentally hit the switch on my powerstrip
OK, is it really so crazy to support two world views, run G code in one, switch to magic mode while saving old world view, run magic code, switch back
)-; t'nsaw I oga...
heh I would have blamed that on a cat
in fact, I'm about to go offline again, ubuntu just did some updates and wants a restart
and I am feeling ready for bed
jmk: will do dial & meter before bother u with widgets ( hscale works now )
tomp wanna stop by the shop on Monday afternoon?
got training on that beast in the vids monday... thanks, we'll get together someday
i think you're making great progress
we're always open....
hell, even Ron Vogel knows you....
tomp Crook never showed up today, he had to test an Agie board (cutting) that always screws up on the AC100 series.....
AC100 is way past my days... i was solid into sinking before that
i cant get dec_points to work ( finer resolution on the dac sliders ) it gets ignored
tomp some kind of high voltage fine finish circuit that is used only in certain "modes"
oh, yeah, i know it.... hochspannungsomethingorother
you got it!!!!!
hochspannung verblungen keit
the old machine used thes square backlit buttons
tomp Larry always wants my first born male child as hostage before he will give me schematics.......
and we had the square kitchen magnets
so there were buttons
onttime offtime beertime lunchtime :-)
you want schema?
tomp and being a component level repair tech, that gets old....
i have the schemos, i just gotta choke them out of Larry....
ask Truty ( that'll po larry :-)
he thinks i am gonna run down the street, naked with Agie schematics or something.....
i picture it
if i could shovel the useless edm crap outta my head,,,,
well he did steal them from Agie, when he worked there, however that machine is no longer in production....nor is Agie capable of servicing that machine....
bte: yesterday saw a guy w a nice sodick inova hopop,,, ran crappy, suggested you guys to fix it
gee a sordick?
cnc and a cartidge tube exchanger ( pneumatic)
does it use the WHP and NRG cutting section?
yes big green machine, i watched it go nowhere on a 1mm dia for near 15 minutes
WHP & NRG not familiar
yeah, his cutting amps are fried....as usual....
uses silly water ( expensive long molecule flushing agent )
kinda like shocks on a car, you don't notice them until you replace them......
i used to fix trx power supplies by buying 100 trx, then sorting out a matching set of 36 , returning rest
considering you are driving them into a short at all times.....
idnt bother to find what was badm, just replaced with a good set
well tom i get to repair all of those boards....
scope sez not short ...... near 30V you said so
ok, expect a call on it
unfortunatly they will talk to Craig first....
they were using 2 hole copper... is copper std on sodick? or just a feeble attempt at low wear
unless they call Ron Vogel direct...then they will talk to me....
i think i used Ron's name
is that a hole popper?
was that Lester McFarlen I saw at your IMTS booth?
i have no idea, i didn't go to the show....i am an independant contractor, who just happens to work for edm network.....12 years...
k, back to the gtkdial widget!
i c you must get back to the task at hand....
tomp is that some kind of rotary jog thingy?
uh... gui widget output a floatying point number ( 123.456 ) and used for analog stuff like OpenVoltage or SpindleSpeed or Gain
not digital = kewl......
like for desired gap voltage or flow rate or ... i did this cuz i saw a need for edm
i can achieve my desired gap voltage, by sensning it......
[05:25:25] <tomp> http://188.8.131.52:8000
tomp i do enjoy jazz, but what exactly is Gypsy Eastern Jazz World.....your shoutcat link????
no, just nize muzik i find sometimes
i plug it into mplayer like this:
i c , i just happen to be an audiophile, all tube amps.....and awesome analog playback stuff, however i must listen to digital every now and then.....
make that executable and you got gypsy musik coming outta computer
and i'm near deef from years in factories
it was indian bungra before ( thier version of hiphop)
not really, cut it off before i heard it, listening to all tube audiov here tonight......no way can a digital stream compare to this....
make a tube edm generator ( great for hi freq & carbide)
you should know, tubes is faster 'n fets
there was a tube ram EDM at the CNC workshop, must have had over 100 tubes in parallel
tomp there was one at the fest....all tube power amp.....
it was... >still< there :-)
yeah, a tube was 3a each... and an old trx was 3 a each....
you cannot beat an tube audio amp....and yes they sound better than fets.........
oh, someday a EDMophile will snap it up.... 8-)
and uv course they are faster, prpoerly biased.....
i went to univ texas, austin
the cheap digs were the old ROTC barracks
and behinf\d them on a hill was the Naval research center ( go figger)
they threw away yards of that kind of stuff
i'd drag it down and cover the walls with tubes gauages dials.... just looked like \flash gordon !
i am listening to a Sonic Frontiers Anthem 1 all tube integrated right now, not too bad, i may even keep it, it does justice to average speakers....
quad EL84 output, per channel
however i do not own "average" speakers...
Fulton Tempo's vintage 1986
Half-speed mastered Abbey Road is sounding quite nice right now....
wanna build yor own fantastic tube stuff??? check out bottlehead.com
that is a good start....
or possibly decware
the decware Zenplate preamp is quite the nice piece....
this is what is called "minimalist" amplification....
nothing in between the music but a capacitor (maybe) and a tube....
John Lennon is jumping out of the speaker at me right now......
and most americans think the japs build the best electronics....
well, night all i am rambling to myself.....
steve_stallings is now known as steves_logging
mdynac: If you're interested thinkgeek and this place has USB Turntables if you want to take your viynle and dub it CD... http://www.betterbuyelectronics.com/index.asp?PageAction=VIEWPROD&ProdID=300132
gotta wake up & goto bed... nite all, thanks
who wants a cheap bot?
[11:53:49] <alex_joni> http://cgi.ebay.com/Westinghouse-Unimate-Puma-500-Industrial-Robot-Arm-Auto_W0QQitemZ320061301596QQihZ011QQcategoryZ87079QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
puma 500 for 10$ so far ;)
cleaning up :-/
anonimasu: is this any good? http://cvs.linuxcnc.org/cvs/emc2/docs/src/common/whatstep1.png?rev=184.108.40.206
alex_joni: looks great
Hi All, looking at the Motenc_io.hal, and the inputs start at 0, but the motenc-100 manual has the input numbers start at 1 ?
might be a 1-0 mapping :)
Looks like it, will try that first...
CQ DX, de wb9mjn...
* alex_joni drops a needle
Making good progress here....good thing I got 50 wide color coded Ribbon Cable, hi!!
Ready to wire in the individual I/O's after lunch....Have the hardware Estop setup, and the grounds from the Motenc-100
Might be able to do the PID tuning tomorrow....
you should probably document that :)
If the Estop stuff works right off...
hi all, what's up ?
I've given up on the AC servos and I'm now going to build the mill with DC servos
oh.. too hard to drive?
sorry to hear that :/
hope you didn't invest too much into it
not too much... I'm now selling the motors on ebay
or maybe you can say you learned from it :P
the auctions end monday and tuesday - it will be interesting to see how much they are worth
yes I learned quite a lot about motor control and electronics
they probably are more woth than you'll get on ebay :)
what was the final problem?
I just don't have time to finalize my own drive. And Tero Kontkanen's drive isn't ready either.
The mechanical parts of our mill are coming together just now so I want to quickly have servos up and running
ok, that's reasonable
anything new and dazzling going on with EMC 2.1 ?
now only testing and packiging is left
but there's not the same drive force behind it, as there was for 2.0
because there is a good product already out there..
ok. did lathe threading make it to 2.1 ?
how about rigid tapping for a mill ?
chris was investigating that lately
it won't be in 2.1 though
I think if he does it in a reasonable time-span it will also be included
proves how much I know :D
alex_joni: well it's not there now
he doesn't have a reversible spindle
perhaps I'm just pessimistic but I don't think it will be in 2.1.
I don't have an encoder on the spindle yet, but I'm planning to do it
ok.. I think he mentioned it's fairly easy to do, and if it's as easy as he expects he's gonna backport it.. but I might be suffering from paranoia or other problems :D
I'm more likely to put the latest g76 into 2.1, but it's not well tested yet
(my lathe isn't quite working right now)
hi chris :)
I don't intend today to put rigid tapping in 2.1.
fine by me :)
anything new with vcp ?
did you see the new spindle my fried Jari built ?
[17:33:39] <etla> http://www.anderswallin.net/2006/12/new-spindle-installed/
anyone know a lot about the tool table ?
a lot ?
is the only way to reload the tooltable to exit EMC and restart EMC again ?
etla: there was a tool table reload message in NML
and I think there are hooks in tkemc to send it
what about AXIS ?
got me there ;)
in 2.1 there is a menu item to reload the tool table
it's a bit frustrating to always restart EMC for example when doing small adjustments to the tool diameter compensation
jepler: sounds good.
anyone tried Ubuntu+EMC on a Lenovo T60 laptop ?
I switched to one recently but only have winxp for now
ubuntu seems to work fairly well on modern laptops -- but I wouldn't depend on realtime working properly, laptops are worse than desktops in that respect.
is 2.1 going to have the ability to run in sim mode without realtime ?
I depend on it
(I usually develop on a laptop and it has realtime problems)
does that include HAL ?
cool, so you can do practically everything on a standard ubuntu installation ?
I use configs/sim/axis.ini on my no-realtime system
everything but load hardware drivers (e.g., there's no hal_parport)
how much diskspace would a small installation require ? I got a 80gb disk on the laptop, and it's filling up fast!
it needs about 2G
plus a swap drive if you don't have enough memory
I've installed to a 4GB disk and there's something like 1.5GB free after installing ubuntu, emc2, and everything needed to develop emc2.
that includes some swap space, I assume
sounds good, so I might set aside a 10G partition for ubuntu + emc. need to find some partitionmagic software or similar to do that...
on a 4GB disk, Ubuntu gave 3.8GB to /, 200M to swap. 1.2G is free.
10g is overkill
I'd go with 4
last time I partitioned I allocated only 2. that's a bit too less
assuming I want the dev tools, the doc tools, python tkinter wxpython
this doesn't have wxpython, gnome development packages, ...
4 might be tight if you want to develop other linux software besides emc2, since everything has its own -dev packages
anything in particular to take into account when installing/compiling 2.1 without realtime ?
you have to give --enable-simulator to configure, and I almost always use --enable-run-in-place too
what about working with HAL files, is that similar to a realtime system ?
yes it's pretty similar
instead of loading modules into the kernel, shared libraries are loaded into a process called "rtapi_app"
you can actually run the debugger on rtapi_app which can be a big help
is there something about sim mode in the dev docs I should read ?
I don't think there's much written specifically about 'sim' yet
I'll try to answer any questions you have, though
that's great, thanks. I think I'll try repartitioning and installing Ubuntu later today and we'll go from there. I'm a little bit worried about all the preinstalled stuff that comes with my Lenovo laptop. There's a hidden rescue partition etc. that I want to keep intact. Also I'm hoping and praying that grub will install correctly and I sill have access to XP after installing ubuntu.
thanks for the chat guys, gotta go now, I'll leave you with some shameless advertizing: 120064261408 and 120063932099 (hint: ebay)
good luck with the repartitioning
if you have any way of writing the recovery stuff to a CD/DVD you should do that before trying to repartition
etla: no shame in advertising your nice motors in here
etla: if I were you, I'd drop a message to the users list too :)
I know some people would have wanted some motors from surpluscenter? after they sold out
alex: OK, I might email emc-users...
so now I see there's Ubuntu 6.10. Should I download that, or use 6.06 as I did previously with EMC ?
6.06 is what the ubuntu people call a "LTS" version
long term support
they will be supporting the distro for 3 years or more, so we've chosen that version, instead of playing the update game every 6 months
now someone is recomending VMWare for using both xp and ubuntu... anyone tried that ?
what do you mean?
I've used ubuntu in vmware -- it works fine, a bit slower...
you can't do RT stuff in a VM
you could probably have an XP vm running on an ubuntu host, and use the host for a machine control
you can't run an ubuntu VM on an XP host and do machine control with it (you probably can run the sim version of emc on a vm)
jmk: I'm settling for nonrealtime and emc-sim since jepler tells me that is working now. I'm on a laptop.
etla: for that VM would be enough
and the laptop already has XP installed?
in that case, its certainly worth a try
it will probably work
is it a fast laptop?
T5600 core 2 duo
VMs tend to eat memory too
so where would I get the VM software ?
I'm running 3 VMs on this system for the compile farm, and I only have 1G
vmware server is the free version
and that's a piece of software that runs under XP ?
there are versions for doze and linux
the "host" is the the operating system on the main computer
"guests" are the systems running on the virtual machines
vmware can use either linux or windows for host, and any mix of guests
etla: it will be enough for you not to notice it's a VM
I'm not sure what the XP install process is like, I have a linux host
* alex_joni did a win install
OK... so I don't need to reinstall XP ?
it's just like any other XP programm
vmware is an application that runs under XP
download and installit
then you download the ubuntu cd, and burn it
what about disk use, can I just create a directory where the root of my ubuntu file system is ?
from the vmware software you will be able to select building a new "guest"
you do need to have significant disk space free, because part of it will be set aside as a "disk" for the virtual machine
it will use one single file for the whole disk
and one additional file for memory
there are some very good manuals on the vmware site
btw, you can (at least on a linux host) skip the "burn the CD" step
I have a image which you could use, but it's using a RT kernel.. and because it's 3.2GB it might take a while to upload :)
when you create the virtual machine, you can just point the virtual CD drive at the .iso file
jmkasunich: never tried that.. I had a CD in my drive, and it just used that
I had 3 distros to install.... using the iso files was easier (no burn) and faster (hard disk speed instead of cd speed)
will I need the "vmware server linux client package" ?
I think so
actually, I don't think so
so is the XP I have now installed also going to be run under VMWare when I install it ?
I got "client" mixed up with "guest"
your XP install will be completely unchanged
OK. I'm going agead with the Wmware server install now... let's see what happens
XP is the "host" operating system for your setup
all you need at the moment is "VMware Server for Windows Operating Systems"
actually I was annoyed that I couldn't find a program to burn an iso directly from the internet
without downloading it first
very few internet connections can keep up with even a 1x CD burner ;-)
jmk: I just got the ubuntu .iso from our local university server at about 40mbit...
jmkasunich: I download a full CD in about 2 minutes, and burn it in 10 (slow burner on the laptop)
so now the vmware asks me for a name and a serial number ?
etla: you need to register with them
what if I press "skip" ?
its free as in beer, not free as in speech
etla: what about that FON movement you mentioned?
I don't see any reference for the free hardware on their page
alex: I've signed up, but haven't received any hardware
I'm pondering if the 30$ are worth it :)
err.. 30 EUR
alex: you put the router in the shopping cart, and when you select a good country :) it will change the amount to 0 eur
0 eur might be valid just for finland, sweden, norway, denmark. I don't know...
so now I've installed VMware server. Now I should create a new virtual machine ?
number of virtual processors ? I have a choice of either one or two ?
amount of memory for VM ? what does ubuntu+emc2 reasonably need ? it suggests 512mb and 1740mb is max
512 is ok
256 is minimum for dapper
boy re-isntalling xp is going to take forever. Plus all the service packs.
skunkworks: get vista ROFLMAO
vista is the bomb
virtual disk type: IDE / SCSI (recommended) ??
use the recommendation
(IOW, they know more about it than we do)
jmkasunich: how is yours running (vm) I
well, when somebody makes a commit and all three of the VMs start doing a build at one time, it gets kinda slow
anyone know what will happen as I did not register with vmware, I just pressed 'skip' at the serial number dialog
I have 1G memory, and a Sempron 2800+ CPU, not exactly as blazing fast box
etla: no clue
* alex_joni likes the FON idea
what is this FON?
Wifi community network kind of thing
you plug in their router at home and share your broadband, for that you get a username and password that allows you access to all other fon routers in the world
jmkasunich: share some of your wifi at home, and be able to use any of the others in the world
they have a promotion which I think runs only in the nordic countries now where they hand out the wifi access point for free. otherwise it costs 30eur
not sure how I feel about letting others use my IP address
wow. there's a lot near chicago
hmm... now I've created the virtual machine, but when pressing 'play' it won't start and complains about the serial number... I need to get one!
I forget exactly how thats done, but its not hard
etla: you need to register I think
jmkasunich: I don't think they'll be using your ip
they have to
they will be tunnelled through the main office I think
that seems like a major bottleneck
you have a site with access lists, where you can see who has been using your network
such a system is only interesting when it has a crapload of users, and sending all traffic thru a central point (or points) doesn't scale well
I might be wrong
if big brother is watching (who knows?) then I only want to be held responsible for the things I do, not the things some dood parked across the street is doing
big brother is probably closer over there.. I don't waste any breath over that
some people worry entirely too much about that
wow, now I'm running the virtual machine. It tries to boot from CD, network etc. Need to get it to boot from the .iso file now
mostly people that don't have a clue
skunkworks: its not that I'm particularly worried
the same ones that get sucked into fishing scams ;)
just less than happy about the state of privacy in general
yay, now I'm booting the ubuntu .iso in my virtual machine!
60gb should be enough for ubuntu, right? ;)
skunk: I allocated 6gb for the virtual disk
skunkworks: depends what you need
I finally upgraded my 18gb drive to a 120gb (laptop)
So I allocated 60gb for xp and the rest for whatever (ubuntu and stuff)
ve7it is now known as LawrenceG
should be interesting to see how the rt kernal plays with it. (have not tested it yet)
what laptop is it ?
does not matter - I probably won't use it to run a machine but it will be nice to have emc at my finger tips to play
del insperon 2650 - I think it is a 1.7ghz pentium m
that's my thinking too. but rather than repartition my drive I'm now trying this VM thing which so far seems to work!
I have had it for years. it has been a real tropper. NO issues.
it is pretty beat up now ;)
xp is up
(guess that wasn't so bad)
you should probably install antivirus software right away !
avg all the way ;)
used to use inoculate until they where bought out.
seems avg will change their policy sometimes (just read that a while ago)
hm.. now I have ubuntu running in the VM, but it seems ubuntu doesn't have network access
Ethernet is set to NAT
in the past I've used Bridged .. but the host is always linux, not windows
you should have the same on the XP side
I have 2 VMWare specific adapters in my list
yes, I got two new adapters too
one gets an IP 220.127.116.11
the other is 192.168.50.1
on the network I am, the DHCP server will probably not allow me to get more than one IP address that works to the outside world
yeah, but it should use your existing IP
also, in full screen mode the VM uses a 1024x768 graphics mode, when my native resolution is 1400x1050. how do I change that ?
etla: we mostly know what you know about VM's now
there are no monitor settings for the VM, so I guess I should just add a modeline to Xconf ?
etla: you should install vmware tools now, before you start messing with video stuff
instructions are in the fine manual
and what do the tools give me ?
among other things, vmware tools will give you more video options
and some faster HW access for some things I forget
vmware tools is something installed on the guest OS ?
both I think
I only see docs for vmware tools on the guest os
ah, my firewall was blocking network access for the VM. now I can see webpages from ubuntu too
but I'm sure it recompiled some packages on the host too
there's an rpm file and a tar.gz file, which one should I choose ?
guess the rpm won't work on ubuntu
now I have vmware tools installed on ubuntu, trying a reboot to see if that will give me 1400x1050
yep, now it runs at 1400x1050 !
now I need to remember how to install EMC... I probably want HEAD, since the sim mode that works with a nonrealtime kernel is pretty recent I understand
40 of 60 updates downloaded (not including sp2 which is next):)
skunk: I expect my fresh ubuntu install will start to suggest something similar in a while...
etla: sudo apt-get update
there are some emc specific repositories I need to enable in sources.list ... I just can't remember the details...
etla: check www.linuxcnc.org
alex: so you're suggesting I run the install script for 2.0.5 and then start to mess with HEAD afterwards
depends what you want
you can add the repository (in that script file it's written out)
and apt-get build-dep emc2
and add a few more dependencies
then get emc2 from CVS as described in wiki.linuxcnc.org
I agree with alex: to be specific, install emc 2.0.5 first, from debs, even if you're not sure you'll use it
then get the build-deps (also from debs)
then get a cvs checkout of 2.1 or head (or both)
when you work with the cvs checkouts, use run-in-place so they don't stomp on the 2.0.5 install
and on my nonrealtime ubuntu running in a VM the standard 2.0.5 install will not mess things up ?
etla: if you would have downladed the emc2 live cd there would not have been as many updates :)
ok now I'm running the standard 2.0.5 installer... let's see what the VM thinks about the magma kernel
oh, I forgot about that....
we should have a wiki page for non-rt installation
well now I'm rebooting the VM after the 2.0.5 install..
etla: choose the non rt kernel on booting
even the RT one will work.. but be a bit slower
I think I did not get a choice..
you have to hit the escape key very early in the boot sequence
uname -a will tell you what to use
but you can always uninstall the linux-image-2.6.15-magma package
of course it will complain you are removing the currently active kernel :)
yeah! now I'm running 2.0.5 in the VM !
seems to be the RT kernel after all
jogging in AXIS works fine
is it useable?
it's OK. Maybe not quite as fast as the previous non-RT kernel, but definitely useable
for what its worth (maybe not much) here is my log of what I did to get a non-rt build on breezy for the compile farm
[20:14:31] <jmkasunich> http://pastebin.ca/281429
windows task manager shows the cpu % hovering at 30% when I jog AXIS
jmk: I'm now running the RT kernel under VM and it seems to work. So I don't think I need or want a non-RT kernel. I have the ubuntu primarily for emc testing anyway
you'll still need to fetch a few packages before you can compile cvs head or 2.1
yeah, but configure will bitch about those..
yeah, I've done that previously on another machine so I should manage...
the list on that pastebin might save you a few iterations with configure anyway
or a look inside emc2/debian/control
"unexpected realtime delay"
well of course
* alex_joni grins
./configure runs cleanly
read thru the output, make sure it will be building axis, etc
(some things don't stop it from running, they just print "no" or "will not build foo", etc
is it simple to convert a simple voltage regulator application into a voltage regulating application that also features adjustable current limit?
now it's compiled and I'm running pre 2.2 just fine in the VM ! nice !
[20:35:10] <jmkasunich> http://knitemare.org/cats/computercat.jpg
good night guys - thanks for all the help once again
I want to try the realtime benchmark, but the program complains with "cannot find run info: /home/user/.runinfo" any tips ?
[21:22:52] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
the latency tests on the VM aren't exactly great...
lat avg is around 93000
100 times a normal machine
overruns increase by about 8000 per line printed out
but it doesn't seem to stop emc from running just fine. I only get the realtime timing error once in a while when I do some strange thing
etla: if you're running in a VM, you don't need/want realtime - just build with --enable-simulator
cradek: he knows
I didn't read back much, sorry
alex_joni: I'm trying to figure out whether to backport the g76 changes I made after branch
alex, cradek: why would RT under VM be so bad ? it looks and feels like it works
it brings no advantage
well the advantage is that all commands, code etc. is exactly the same as on the machine that drives the mill
alex_joni: do you have time to play with g76 a little? I'd like someone else to try to break it
what is g76 ? threading ?
cradek: I probably need a lathe 101 for that
but I can try to break the g-code interface
alex_joni: have you cut threads on a manual lathe?
nope :) but I understand the concept
hmm, it lets you do some nonsense things
I can surely do that :D
or wait.. probably not, cause I don't know what "sense things" are :D
ouch, it doesn't really work for boring anymore, I thought it did
inside threads I mean - not really boring
no inside threads? how come?
because some things are specified as 'depths' but I didn't think of how to specify which way they point
I thought it used to work by sprinkling some minus signs around, but now it doesn't seem to
are those tasty?
I'll have some
not at all
I was afraid of that
the sweet ones are
* jmkasunich is having fun
the shoptask is starting to be a pile of pieces on the floor
this is all so you can move it away from the wall a bit right?
move it away, and turn it 90 degrees
also, while its apart I'll be cleaning 5 years of chips and oil and filth off of it, and studying the construction to plan the ballscrew installation
at least lefthand threads do seem be right
jmkasunich: you're not going to do ballscrews at first though?
not right away
at least I don't think so
I'll need the lathe working in order to machine the ballscrew ends
my plan is to take photos and measurements while its apart, draw up all the needed parts, then put it back together, make the parts, tear it down again enough to install them
sounds like encouragement to get the plans right the first time
its kinda nice - some parts aren't going to go back on it
like the power feed geartrain from spindle to X leadscrew
it has an X power feed?
isn't that pretty unusual?
oops, Z (in lathe mode)
for manual lathes, power feed in both directions is rather common
thats why the apron is usually full of gears and such
does your lathe (the big one) have power feed for the cross-slide?
we have a nice little monarch has that.
what kind of lathe is it? (make and size)
[22:24:34] <jmkasunich> http://www.lathes.co.uk/southbend/img10.gif
see the handle in the center of the apron (right of the large wheel)?
up is for power longitudinal feed, down is for power cross-feed, middle is neutral
(I might have up and down backwards)
the lever on the far right is for the half nuts (center lever in neutral for that of course)
better pic: http://www.lathes.co.uk/southbend/img12.gif
you can see the gear that drives the cross-slide screw
[22:28:37] <skunkworks> http://www.machineco.com/1655_Lathe_Monarch_Toolroom_10EE-newer.htm
that is close to it.
it's a craftsman (atlas?) 12"
fathers first lathe was a atlas/craftsman
I've seen some 10EEs go for reasonable money, it pained me to pass them up
Can not seem to get the Motenc outputs to change here...
but I don't have a heated place where I could put a 3000 lb lathe
Inputs are Ok....
wb9mjn: there's a thing in the wiki
oh.. sorry .. it's a m5i20 thing
[22:30:47] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?M5i20_Halvcp_Test_Panel
maybe you could make a similar one for motenc
only this hal file needs to change: http://wiki.linuxcnc.org/uploads/iotest.hal
cradek: are you _sure_ it doesn't have power crossfeed?
[22:31:39] <jmkasunich> http://www.lathes.co.uk/atlas/page5.html
0 pages found "moten outputs" ...
scroll down to the last pic
I lied, I think it's 6"
cradek: size matters? :-P
[22:32:36] <cradek> http://cgi.ebay.com/Craftsman-6-Lathe-Sears-Atlas-no-reserve_W0QQitemZ150066581020QQihZ005QQcategoryZ97230QQcmdZViewItem#ebayphotohosting
I think it's this one, but not all busted up
6"? that ain't much bigger than a sherline ;-)
i've never been really sure what its part number is
yes it is...
a friend has a nice new lathe, in the original crate, all oiled up
the ebay ad says 75 lbs
I might consider CNC-ing it one day :)
I can't believe this one sold for $325
sherline: pick up with one hand
craftsman 6": pick up with two hands
no motor, pulleys, all the knobs broken off
wb9mjn: think you can go from there?
real lathe: pick up with several friends
change gears mostly missing
there's a sucker born every minute I guess
and most of them want little lathes
the light weight was probably a driver - hobby folks often don't want to mess with something that is too heavy
I think I paid $250 for mine, bolted to a heavy table, with motor and pulleys, a couple chucks, all the change gears, etc
yeah, and found it locally
it's very worn but with care I can make parts the right size and shape.
hmm.. it seems there's a new frontend for mach3 in german nowadays
* skunkworks goes to dell to get the rest of the drivers I need
and it's supposed to be much easier to use
[22:38:20] <alex_joni> http://www.einfach-cnc.de/seiten/Mach_deutsch.html
all I can say is: ROFL
looks better to me
I thought I was going to click on the link and find a picuture of emc2 and axis ;)
I mean the general look is less random
yeah, way better than the original
but still .. a bit much for my taste
but still I don't understand the aversion to standard widgets that people know how to use already
it has the "everything is a rectangle" syndrome that I think makes it mysterious
cradek: they design with touchscreens in mind
that's why it's mostly buttons
that makes sense
but why it's those ugly looking, ugly coloured buttons.. I have nfc
a side effect of "everything is a rectangle" is that you have to use something like color to convey state information
to me, it makes it mysterious
if they could at least show as pressed in or not, you could skip a lot of the mystery color stuff
for 15 years now people have understood gui buttons that are in/out
but red/yellow/green/blue? not so much
can't find anything in newsgroup, or wiki about the enable outputs...the motenc-io.hal has one-line entries for these
wb9mjn: what's the problem again?
outputs, but all the other inputs/outputs have two or three lines...
cradek: I think it's fashion these days to change how it looks & feels
None of the outputs seem to be working, but specifically, Xenable , Yenable and Zenable are not coming on with
the machine-on state...
cradek: look at vista.. it completely renders somehow literates useless
alex_joni: I agree, but it's a foolish fashion at the expense of the user
wb9mjn: what does 'halcmd show pin' say?
I am getting limit inputs working just fine...
and please don't paste 100 lines in here
I ll go look...
Have I ever ?
especially look at the signals: halcmd show signal Xenable
wb9mjn: wasn't sure :)
use pastebin.ca for long stuff
Yea...hal show shows a working output...
But there nary a flicker in the output voltage...
It changes from red to yellow...
all three signals...
don't know enough about the board to tell you what to check
you're using default config?
Slightly modified...-not on the Limits...and changed the signals from the second output header to the first...
Rearranged the Spindle stuff to the second output header, and put the enable outputs on output 2/3/4 ...
did it work while it was on the second output header?
Do not know, I do not have any wires on the second output header...
ok, I see
There are 4 I/O headers...
I have one cable on the first one...
The inputs are on that same header...
what I would do first would be to: 1. load emc2 with that config, 2. unlink all signals from io pins, 3. create a dummy bit signal, 4. connect it to all outputs, 5. set it to true and see if that works
One thing that changed, was that the encoders stopped being read....
that's not a good sign.. probably something wrong there
If nobody has heard of this, I ll just go and hack on it some....thanks for the idea...
wb9mjn: I suspect you put the outputs over the inputs
or something like that
Do not think so...says out......All I did was change numbers....will double check...
I see the mazak uses a motenc card
the numbering is a bit odd
jmkasunich: do you have some filing system that uses the numbers on all those containers and drawers?
[23:03:23] <alex_joni> http://www.robcon.ro/emc/weaving.pdf
any idea who made that?
* alex_joni guesses JMK might
I remember the guy working on it, but not his name
he had some nice screenshots of the weave in AXIS too
I think it wasn't that guy that made that pdf
it was in the old dropbox on linuxcnc.org.. he didn't have access
anyways.. it could be in the integrator handbook
Roguish__ is now known as Roguish
well, I now have 6 axes of Siemens servo powered motion more than I did this morning :)
nah, got it to the factory at 21:00
locked it up and brought the manuals home
figured it was best to read stuff before hooking the power up and pressing "auto run" :)
alex_joni, so your "weaving" servo ... that looks like it will work for motions that are along one axis
yeah, it seems like that
although it won't be too hard to move that across 3 axes
do it 3 times, and adjust the scales accordingly
the weave needs to be related to the direction of travel
based on the proper projections
robin__sz: for welding it does
and related to TCP and TOV
so a simple HAL component can do that easily
tcp = tool center point
tov = tool orientation vector
tov is especially crucial for waving
because usually the waving plane is determined by wire and direction of travel
I dont think my bot has the arc software installed ..
if the torch is bent more.. tov changes :)
and you wouldn't want to wave along the wire direction .. would you :D
you may want to do circles along the direction of traavel .. is left to right an forward back motion ...
yup, it's closely related to the rotate-blad discussed on the list lately
blade.. what's wrong with me :)
for paper/fabric cutting
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-12-16.txt
steves_logging is now known as steve_stallings
robin__sz: off topic, when I asked earlier if anyone here was an Asterisk VOIP user, you were mentioned?
hmm im also an Asterisk user btw
OK, curious about putting extensions off site using public Internet with no QOS, does it work, echo badly, require special codecs?
steve: most of the time it works just fine, without any codecs or anything
good to know, I work from home a lot and would like to have a phone that appears on the office system (thinking of setting up Asterisk there)
steve: what I did is I put in land line cards and had public internet, if no prefix was used it used public internet but you could always prefix it with 1 in my case and use the land line
steve: I ran Asterisk for my last company as the main pbx that way, worked from home all the time, never had problems with that connection but then again we had our own dsl system
I will not be using VOIP for trunks, plan on Digium FXO ports to land lines
steve: I see, probably a smart move, but you can also implement international dialing via internet trunk though
will try putting a Grandstream 2000 at home and hoping version 1.4 gets the Shared Line Appearance stuff working
I'm a stick in the mud about old style keysystem features 8-)
steve: hmm , I would stay away from the Grandstream, I ordered like 20 phones and they were all really bad, about half failed in the first month
steve: the 2000 is the converter right ?
too late, I ordered two of them to play with
steve: I also ordered a bunch of the converters from directly from Digium, the blue boxes
no, 2000 is the higher end business phone
steve: Those are horrible to
steve: not the business phones, the boxes :P
steve: Don't know about the business phones, they might have pulled their act togeather
steve: I gave up in the end and ordered Cisco phones
comments I have seen state that the 100 series are junk, the 2000 acceptable but not great
steve: I always wanted to order dual phones for everyone, GSM and 802.11g so that when people were in the office the calls would go over Wifi, when they leave the calls would go over gsm
steve: something like the Nokia e61
wireless is not an issue for us, remote phones with key system like functions would be a blessing
steve: btw we had people working via Internet from the UK to our system (in Iceland) without any problems so unless your ISP is really bad you will get better than mobile quality on calls
well, actually the public Internet in the USA is often worse than in Europe
you only say that to make us feel better :P
actually I think South Korea is supposed to have one of the best
bet that's true
but then again it depends what you're accessing..
if it's still in the US.. no improvement :)
Rugludallur_: connection issues?
steve_stallings: 23 msec to Frankfurt, 40 to London, 150 to NYC
Alex: got disconnected, im way out of the city and technically not able to use dsl but I got friends to do help
our consumer providers have low standards for quality of service and often overload local circuits
steve_stallings: that's the same all over :)
alex_joni: nope, got friends at the telco and they did spectral analysis, boosted the signal and got me 6 megs :D
alex_joni: but I loose sync once every day or so here at home
I get 1.5Mbit for 30$ at home, .5 Mbit for 100$ at work
alex_joni adsl vs sdsl or ?
cable vs. fiber
alex_joni: hmm your in Rumenia right ?
fiber with 99.6% uptime guaranteed
romania.. but close enough :D
alex: err sorry about the spelling (in icelandic it is spelled Rumenia)
btw, that's .5 Mbit only for internet .. local nets aren't restricted at all.. quite common to get ~70Mbit
I get to pay US$89 month for about 2 mbit burst with a fixed IP, much cheaper with dynamic IP
Rugludallur_: good to know :)
alex_joni: well technically it's Rúmenía
Rugludallur_: so a VPN to the local college really helps :)
especially since they don't have much internet speed restrictions :D
155Mbit or so
alex_joni: do you guys have to pay for traffic at all ?
nope, only what I said above
residential fiber is coming to my area, but slowly, at least two more years to wait
alex_joni: over here most ISPs charge about 1 cent per mb international
hmm.. network's a bit busy :) (Current in: 497.27 Mbps)
are services regulated by government for you, or open market competition
alex_joni: im working on some really interesting stuff at work, embedded linux and I have been thinking it would be really cool to make an embedded kernel with X and EMC
steve_stallings: open market here, the problem is just that I live way out in the middle of the atlantic
can you really say X and embedded in the same sentence?
steve_stallings: sure you can
lots of distros smaller than 32MB
I thought X was a memory hog
steve_stallings: think about OLTP
Rugludallur_: what platform?
alex_joni: AMD Geode LX, x86
alex_joni: same basic hardware as OLTP
steve: One Laptop Per Child, Negroponte