I had INPUT_SCALE and OUTPUT_SCALE set to 8000 0 in emc1 - does anyone know how I should set them in emc2 (steppers)?
hmm, changing INPUT_SCALE doesn't seem to do anything
and... changing OUTPUT_SCALE doesn't do anything
emc runs on a 2.6.9
I mean emc1?
cradek: emc2 doesn't use INPUT_SCALE and OUTPUT_SCALE yet
so the scale is just fixed?
scaling is determined by lines near the end of core_stepper.hal
currently 5080 counts/inch (200/mm)
there is a commented out set of lines that are 8000 counts/inch
and of course you can make it anything you want
but you'll have to tweak the max_accel params as well
cool, the numbers in there explain my .636" inch
the max_accel params should be about 1.5x the emc.ini MAX_ACCEL values (converted from inches per seconds^2 to steps/sec^2)
in emc1, jogging updates status.motion.traj.position and .actualPosition. In emc2 it updates only actualPosition. Is this intentional?
traj is a trajectory planner thing, jogging takes place in free mode, the planner is completely uninvolved
in coord mode (MDI or AUTO), the traj planner is active, and you should see traj.position getting updated
yes, it is
but traj.Actualposition DOES get updated when jogging
thats because the actual position is changing... the TP has to know where it is even in free mode, so that when you go into coord mode, there isn't a big jump back to the last traj position
ok, got it
back to scaling for a moment...
currently MAX_ACCEL is 20 in/sec^2
ok I set the scale to 8000
times 5080 counts/inch, = 101600 counts/sec^2
[01:13:26] <paul_c> http://22.214.171.124/snapshot.png
I set the stepgen.x.max_accel to 150000 - it must always be somewhat higher than the emc limit
but if it's too high you will get the jitter
if you are using 8000, then the nominal value in counts is 160000, and you should set the max_accel to about 200000-250000
it's not too critical
hmm, my steppers are skipping...
highly reliable that logger is.... not
when do they skip? starting a jog, ending a jog, during the jog?
ok for the X axis emc1 has MAX_VELOCITY=.2 and MAX_ACCELERATION=2.0
starting a jog
why so low?
that's as fast as it goes
what kind of a machine is this?
it uses 1/4-20 leadscrews and way too many steps per rotation
8000 steps/inch you said? that means 400 step/rev - must be half-stepping
I'd set the emc.ini MAX_VEL and MAX_ACC the same as in your emc1 setup
I did that
and set the core_stepper.hal max_accel to 20000 or 25000
jmkasunich: Just added noreply@sf to emc-dev
cradek: did you ever try full-stepping it?
jepler: I'm pretty sure that made it less reliable (lower max speed)
cradek: was that before this controller board? Maybe you should double-check the step table...
jmkasunich: now it still skips and it goes to ESTOP RESET state after a jog
hey! who assigned that bug to me?
(but no error pops up)
jmkasunich: me - apparently you're the default
bet it's following error
it doesn't pop up like it should...
that's yet another thing that doesn't work yet
I'm at a serious disadvantage not knowing much about the GUI
if you guys already know all these things don't work, I think I'm only being a pest
oh, well, gee -- cradek knows lots about the GUI
you should ask him for help
not a pest - I need a few kicks in the ass to get me going
well you fixed one real roadblock already today
you want to go thru the jmk way of tuning emc2 to your machine?
when I set accels to 40000, it jogs
but I think it's too high to be reliable
does DEADBAND not work either? I can't get it to quit twitching
how twitchy is it? one every few seconds, or faster?
8-10 a second
it's like this... the good news is that stepgen contains a built in pretuned "PID" loop
the bad news is you can't change the tuning
and my machine is too slow for your tuning to work right?
and I think it has a stability problem (in fact I created a tracker for that some time ago)
Dang, I wa just about to pen a note to the U.P.
in emc1 I have X: VEL .233 P 160 Y: VEL .16 P 160 Z: VEL .167 P 30
What was it about to say?
don't ask me why it's set that way, but I tweaked for a long time to get it to be reliable
willing to take a stab at tuning it for emc2/hal?
sure, if you want
first step - lose emc2 ;-)
ok, done, now it works
emc is shutdown, and you are at the emc2 root directory?
you root, or a user?
I can be either
do whatever you want, but some steps will need su -c, or sudo, or you need to be root
I usually work as non-root, and use su when needed
first, load the realtime stuff: su -c "scripts/realtime start"
ok, it loaded some modules
bear with me, I'm makeing this up as I go along
su -c "bin/halcmd loadrt stepgen cfg=\"0 0 0\" period=5000 fp_period= 1000000"
su -c "bin/halcmd loadrt siggen num_chan=1"
where is your parport? 0278, 0378?
su -c "bin/halcmd loadrt hal_parport cfg=\"378\""
now we have all the modules loaded that are needed to move your motors
gotta connect them
do bin/halcmd show thread
does it list one thread or two?
stepgen.threadFP and stepgen.thread
ok, great - I made a type here, only got one... will fix
ok, fixed that
bin/halcmd addf siggen.0.update stepgen.threadFP
bin/halcmd addf stepgen.capture_position stepgen.threadFP
bin/halcmd addf stepgen.update_freq stepgen.threadFP
bin/halcmd addf stepgen.make_pulses stepgen.thread
bin/halcmd addf parport.0.write stepgen.thread
bin/halcmd show thread
should show two threads, with three functs in the first one and two in the second
ok, now to connect signals
bin/halcmd newsig Xpos bit
bin/halcmd newsig Xdir bit
do I replace `bit' with my parport pin?
bit is the type of the newly created signal
you using the emc standard pinout?
what pinout do you need then ?
step on 2, dir on 3?
yes that will work (it will move my Z which is fine)
well let's do X, which pins is that?
well X will fall off the mill if I jog too far
is Y or Z safer?
ok, Z it is
I just realized I misnamed a signal anyway, I didn't want Xpos, I wanted Xstep
bin/halcmd delsig Xpos
bin/halcmd delsig Xdir
bin/halcmd newsig Zstep bit
bin/halcmd newsig Zdir bit
bin/halcmd linksp Zstep parport.0.pin-02-out
bin/halcmd linksp Zdir parport.0.pin-03-out
bin/halcmd linksp Zdir stepgen.0.dir
bin/halcmd linksp Zstep stepgen.0.step
bin/halcmd newsig Zpos float
bin/halcmd linksp Zpos stepgen.0.position-cmd
bin/halcmd linksp Zpos siggen.0.square
lets see - you have 8000 steps per inch, right?
bin/halcmd setp stepgen.0.position-scale 8000
and you have run it before at 0.2 inches per second?
Z at .167 ips
ok, 0.167 * 8000 is 1336 Hz max step rate
but we'll start slow
bin/halcmd setp stepgen.0.maxfreq 500
you did 2.0 inch/sec^2 acceleration before?
6.0 on Z
that would be 48000 steps/sec^2, but again we'll start slow
bin/halcmd setp stepgen.0.maxaccel 500
bin/halcmd setp stepgen.0.maxaccel 5000
sorry - that's too slow
5000 is good
bin/halcmd setp siggen.0.amplitude 0.1
bin/halcmd setp siggen.0.frequency 0.2
bin/halcmd setp siggen.0.frequency 0.1
what we're doing here is commanding a square wave of position, 0.1", repeating every 10 seconds (so a 0.1" move every 5 seconds)
let's save this config to save hassles later:
bin/halcmd save >foobar
now at a later time, you can do the three loadrt commands, then do bin/halcmd -f foobar, and it will do all the others that we just did
ok, amps turned on, and axis centered (at least 0.2" from any limits)?
should get movement
very very very slow
500 counts per sec
2-3 counts per sec
* jmkasunich watches his motor turn nicely
hope I didn't typo
bin/halcmd show net
bin/halcmd show sig
should be three signals, each with two pins connected (using ==> and <== to show data flow)
the Zpos signal is at either + or - 1.00000e-01?
bin/halcmd show param | grep stepgen\.0
should give 11 lines
dirhold and dirsetup both 1
frequency probably + or - 5.000e+02?
if frequency is 500, then you should be getting 500 step/sec
oh, it just slowed down
that's so much better :-(
I was getting about 2.5 steps/sec when it was going "fast"
was it moving the same direction the whole time?
it had an accel and decel period
yes I think so
lets keep looking at parameters... stepgen.0.position-scale should be 8000 (8.00000e+03)
dirhold, dirsetup: 1
steplen, stepspace = 1
steplen, stepspace 1
I thought you said frequency was 500?
it was earlier
when it was going "fast"
bin/halcmd show param | grep siggen
should be five
amplitude = 0.1
frequency = 0.1
offset = 0
time 768 tmax 1888
ok - that's execution time, I have 1000 and 5000 here - slower box
click select, then signals, then Zpos, then OK
should change to +.1 after about 5 seconds, then back to -.1, etc
nope, not changing
select, pins, siggen.0.triangle
is that changing?
should be ramping from 0.1 to -0.1 and back again, one cycle every 10 seconds
it's much slower here
bin/halcmd show thread
changing something like .0002/sec
period of threadFP should be approx 1million (ns)
and thread should be about 50000
thread is 5000
bet that was the type
bet that was the typo
my fault, when I was cuttin and pastin I bet I dropped a digit
is your machine a little sluggish?
you did say 5000...
no, it's fine
(cause you're running at 200,000 interrupts/sec....
how can I change it?
unfortunately you can't without redoing the loadrt command
got a good bash history? ;-)
su -c "bin/halcmd unloadrt all"
yes I do, and I wrote it to a file
su -c "scripts/realtime restart"
then find the loadrt commands in your history - there should be three, one for stepgen, one for siggen, and one for parport
the stepgen one should have parameters called "period=50000" and "fp_period=1000000"
oh, you'll need to close the halmeter too
ok, I'm caught up
you ran all three loadrt commands?
bin/halcmd -f foobar
that should set up all the pins, signals, and params
now it's moving
back and forth every so often
should be moving +/- 0.1 inches, every 10 seconds (one move every 5)
yes, that looks about right
now we play
bin/halcmd setp stepgen.0.maxaccel 500
that should give you a quite noticable accel/decel ramp
we want to find the max speed first, then the max accel
how long is the axis?
so we can run much greater amplitude later - good
we'll leave accel low for now
bin/halcmd show param | grep maxfreq
should be 500
bin/halcmd setp stepgen.0.maxfreq 1000
so far so good
it hardly has time to get up to max freq since the accel is so slow
right - speed is 1000, accel is 500, so 2 second ramps at each end
lets kick up the accel some - try 2000
bin/halcmd setp stepgen.0.maxaccel 2000
now the ramp should only be 1/2 sec
now it ticks while stopped
mine does too - that's the stability thing
I need to work on that, but let's ignore it for now
we figured a max freq of 1336 steps before, right
try setting maxfreq to 1200
sounds very familiar
try loading it with your hand or something, make sure it can do it
then go faster a little bit at a time, see what it can do
yes it can still lift it when I'm pushing down somewhat
try 1300, 1350, see what it can do
it becomes pretty easy to stop at 1800
that's 0.225 inch/sec - quite a bit better than before
but you'll need some safety margin
try setting it to 1600 (0.2 ips)
then you can start increasing the accel limit
seems like it can still lift the dead weight of my arm somewhat reliably
still works at 32000
the accel rate you had before (emc1) was something like 48000, so that isn't too surprizing
64000 is iffy
maybe 48-50K is good?
you can do setp stepgen.0.frequency 0.2 now - since it's running faster
play around with maxfreq and maxaccel until you have numbers you are happy with, then write them down
stepgen.0.frequency is not writable
my mistake - siggen.0.frequency
to make the moves more frequent
ok, 1600 48000
you'll eventually want to do the same for each axis
those numbers represent the limits of the motors/screws/etc
note that this is completely independent of emc - we're only running a signal generator into the stepgen module
you can also do tests where you use sine (or cosine) instead of the square wave
that tests your motors at a variety of speeds, looking for resonances, etc
X = cos, Y = sin, makes cute little circles ;-)
I am half-stepping so I don't think there should be a problem with that
once you have numbers for all three axis, put them in your own copy of core_stepper.hal
there are already lines in there for maxaccel, you'll need to add ones for maxfreq
so can I change just the parport bits to change the axis?
how can I undo linksp?
although you have three stepgen's loaded, so you could just link another one for X and the third for Y
unlinkp <pinname> unlinks a pin
ok now x is jogging...
just a minute, be right back
* jmkasunich is away for a minute
ok, I have all my numbers and put them in core_stepper.hal
but I think emc.ini is overriding them (making them slower)
I strongly recommend that you keep a copy of core_stepper.hal and emc.ini under a different name - sooner or later, you're gonna do a cvs up and wipe them
that's the next step
take the numbers in core_stepper.hal, and convert to inches per sec
(divide by 8000)
the emc limits need to be a little below the hal limits
set emc's MAX_VEL to about 95-98% of the scaled stepgen.maxfreq
set emc's MAX_ACCEL to about 80% of the scaled stepgen.maxaccel
that should give you an emc that runs pretty close to the machine limits, but doesn't ferror out or have severe jitter problems
if your emc limits are too high, emc's commands can get ahead of what the stepgen can do (because of it's limits), and you'll get an ferror
the jogs sound good now
I need to write this procedure down sometime...
well I ran a program and it got only a few lines and estopped
again, no error popped up
(I often do tail -f in a shell window, and just leave it there)
feel free to post the lack of error popups as a bug - that will motivate me ;-)
billions of messages of the form xxxxxx: CMD xxxx, code 1 ABORT
where xxxx are increasing numbers
an abort storm...
also some code 5 DISABLE_AMP messages
code 3 DISABLE
I put those same values in emc1, let's see what it does
I hate abort storms - they come from situations where the handshaking between user space and rt is wonky - the motion controller is stopped, but the user space is trying to clear an error by issuing ABORT commands, and the error is already clear (or something like that)
emc1 seems to play nice with those numbers
don't know how sharp my corners will be with those numbers...
can you send me the part program that makes it fault?
I think it's quick reversals of direction that did it
(and your .ini and .hal files?)
did you ever find the actual error buried in the kernel log, or did the abort storm hide it?
(I'm guessing it was a following error)
I'd like to set up halscope to trigger on following error, and see what happens
ERROR: joint 0 following error
watch value 0C06 (after set_operating_mode())
that "watch value" stuff is leftover debugging printfs
the ERROR is what matters
let me make sure it errors in the same place again
jmkasunich AT att DOT net
I think it errors around line 42-45
very short segments in opposite directions
does it error for you too?
just starting it now
* jmkasunich loads halscope
how do you reset back to the beginning of the program?
cradek - you still there?
try changing the maxaccel numbers in core_stepper.hal from 48000 to 100000
actually it looks like even 64000 is enough
it's getting further
right - but it still faults
trying 100K now
faulted at 64000 too
right - that's what I triee
right - that's what I tried
what are we milling anyway?
it's a pcb
don't you have a viewer?
didn't turn it on right away, then the scaling was wrong
finally got it going, but the routing part was almost complete
it's drilling holes now
I think it's done
some kind of little SMT connector, and a 0.100 header?
working at your end?
it adapts a strange plug to .1 SIP
I didn't increase it past 64000
at 100000, it seems to work fine
I worry about what my corners will look like
I have a spring-loaded zero backlash setup
just because the motors can reverse that fast doesn't mean the table can...
then lower the max speed (in emc.ini - no need to change the maxfreq in stepper.hal)
you're running 0.2 ips, with 2ips^2 accel?
yes I guess so
so accel will take 0.2 seconds
I bet your spring can handle that
note that changing the maxfreq or maxaccel in .hal does not change what emc asks the motors to do
it only changes the limits that the HAL will apply
right, 0.1, my mistake
0.2 to reverse from +max to -max
you're afraid that the reversal will be too fast, and the spring will allow some backlash from table momentum?
so can a person find 1/4" fullthread with fewer than 20 tpi?
has to be 1/4"?
yeah, then my mods would be reversible
and I wouldn't have to machine the table
how about 10tpi?
I bet that would help
ugh, I wonder if I could find a tap to match it
eek, never mind.... $111 for a 6ft length
looking in mcmaster carr
that's acme thread, btw - lower friction than 60 deg threads
that would be nice
but I don't know how to make the nuts
precision screws, 0.0006"/inch, stainless
matching plastic springloaded nuts are $17
www.mcmaster.com ( I think - I'm reading in the paper catalog)
I think maxnc makes a model with those acme nuts and people say they have backlash of up to .008 or so
p/n 6350K128 (screw) 6350K204 (springloaded nut) 6350K166 (regular nut)
maybe that's if there's enough pressure to compress the spring
how about 4tpi ;-)
$43 for a 3ft screw, $22 for a bronze nut
on 1.8 degree motors ... I don't think so
if you could fit 3/8 diameter, you have lots more choices
1,2,4,5,6,8,10,12,16,20 tpi in 3/8 diameter Acme
I just don't think that would fit
maybe I should stick to what I have!
I just go get a snack while it's working...
16 tpi is available in 1/4 - not much better than 20 tho
is the existing screw acme, or 60 deg?
acme would have lower friction
so you could probably run it faster in RPM, as well has higher pitch (16 instead of 20)
thanks for looking
it's probably not worth $150 to me
(this is just a hobby machine)
I can make excellent circuit boards now, I don't know why I need it any faster!
I need to get busy making some chips here.... modifying a ballscrew to fit my Z axis
that sounds like what I need
let me send you my machine too
this is 5/8 dia 5TPI
only 6" long tho
oops, gotta run
thanks for the tuning help
time for me to go too... IRC is hazardous to productivity ;-)
* paul_c compiles smdromod & vitalmod for 2.6.9
I've seen your screenshot...
so the next BDI will be 2.6 based?
may well have both 2.6 and 2.4 kernels
* alex_joni adds to his todo list playing around with 2.6
probably I'll do this with the next BDI
will there be a unpatched kernel for the BDI ? (e.g. no realtime?)
or is it enough to do a apt-get kernel-2.4.2x ?
why would you want an unpatched kernel ?
I'm thinking of changing the distro
even for other machines I install
if BDI works nice... I see no need to install it for EMC only ;)
there would be nothing to stop you using any of the Debian kernel images....
Back in a bit...
* paul_c notices a Scot joined....
* alex_joni wonders what clan :-)
hi paul and alex
I must test that abort bug with the jogging
I have not seen it
and I have ofen done an abort on running code thousands of lines long
might be related to a repeated command only
I get it in my programs all the time
les: did you try emc2 lately?
cradeck: nest time I fire up the machine I will try your code
you can even see it with sim.run
alex: have not tried emc2...am in large production and cannot experiment much
I see.. ok no problem
I am still searching for people who did use emc2
hmmm...well sounds like a bad thing so I should check it out
and can try the new autoconf built ./configure
not such a bad thing
if you haven't used emc2 before it's not that concludent
(although I'll be happy with every test done)
I am running bdi 2.xx with a june 2004 emc build
on bdi it has been tested... should work
I slightly modified emcmot.c on it
BDI 2.18 is on the compile_farm
so there it works
The only thing I do not have is per axis accel I think
But in my ap accels generally need to be matched
well ... better get back to autocad...must draw up CUSTOM router bit profiles
expensive to have made but saves big machine time
* paul_c greps the generic.run script and thinks it could do with a major cleanup.
alex_joni: Did the vote go your way?
* paul_c sneaks off to get a coffee - Then to write up on the 2.6 changes for emc-dev...
* paul__c announces EMC running on a 2.6.9 kernel.
EMC ver. 1
I'll have to try it on my fedora 2 box
Just way too much to do to get emc2 to work with 2.6.x kernels.
of course, without actual hardware I'll have no way to verify that it works
jepler: you have an o'scope
jepler: that's kind of like a milling machine
So in the traditional stepper pinout, it's direction on pin 2, step on pin 3, etc?
[20:29:36] <cradek> http://cvs.sourceforge.net/viewcvs.py/emc/emc2/configs/standard_pinout.hal?view=markup
cradek: I was checking code during lunch for that abort bug
I was just starting to dig into it too
you reproduced it, right?
not in emcmot anyway
have not fired up the machine (office day) but if you say it reproduces I believe it
if you can run sim.run you can easily see it there
question was if it has always been that way or something got recently broken
Issuing EMC_TASK_ABORT -- (+503,+12, +28,)
emcTaskPlanClose() called at emctaskmain.cc:1893
emcTaskPlanRead() returned 0
emcTaskPlanLine() returned 1
emcTaskPlanCommand(g0x0) called. (line_number=1)
look at this
after the abort, line number 1 is executed
Because I routinely abort stuff running in the middle of a large 20,000+ line program
ah...that's it...my line 1 is never a motion code
is your emc old?
a few months old
so....something other thana motion code in line 1 would be a work around?
Issuing EMC_TASK_ABORT -- (+503,+12, +33,)
emcTaskPlanClose() called at emctaskmain.cc:1893
emcTaskPlanRead() returned 0
emcTaskPlanLine() returned 2
emcTaskPlanCommand(g0x0) called. (line_number=2)
I put % at the beginning and end, and it runs line 2 after abort instead
my cam programs usually put in %...
but still the first real line is something other than motion codes
but that doesn't help, darn
usually stuff like G64 etc
working on one now...let me see
emcTaskPlanCommand(g64) called. (line_number=2)
ok, that works around it for this program
but I don't think it's a workaround for every program
no not every one
my pcb files will do it and they have a long preamble
cradek: elm320.ngc does start with a g0
but as I check the first 6 lines of code for me are spindle, units, etc
jepler: but it doesn't execute that line, because it doesn't go straight up
the dangerous part was going to 0,0,0 without z going up first
change that line to g0x0y0z0 and see if it still goes somewhere random
the z is always the z value in that first line
imagine that emc thinks it's at (1,2,3) and it sees g0z.04 -- it'll go to (1,2,.04)
(just a guess, testing it now)
* paul__c has blown the primary parport...
yep, looks that way
Gosh I have 5 tool changes in the one I am doing now...and an array of at least 20
I changed the top line of elm320 to G00 Z0.040000 X1.234 Y1.390
paul__c: One place could do the copy but wanted $45 for it. I'll keep looking.
and sure enough that's where it when when I hit escape
rayh: So it's not you downloading at the moment then ?
I don't think so. Unless they went ahead in anticipation.
Is it to a location near here? Or can you tell.
Started just before 17:00GMT
Jepler: now stick a MO6T1 or G80 on line one and 2 and try it
So they've been at it three hours.
(not near a linux box that is up)
well les, let's get with the program;)
les: whatever's on line 2 doesn't make a difference (G80 anyway)
still goes to 0,0,.04?
getting with it ray
You need me to come down there and set up a box for you?
Well, with the G0XYZ I mentioned above it goes to 1.234,1.390,0.04
doing a 20-50 array 5 tool change program now...
emcTaskPlanCommand(G01 X-1.469906 Y1.570106) called. (line_number=510)
emcTaskPlanCommand(G00 Z0.040000 Y1) called. (line_number=1)
I changed the program again. These are the latest emcTaskPlanCommands in the terminal. The machine ended up at -1.47, 1, 0.04
ray: have one emc box in the shop and one in the office (it's off though)
guys, please put whatever you figure out in the bug tracker
Oh, and I don't have a % on line 1, so when you say "insert G80 on line 2" it goes below that g0 motion
Oh. Okay Les, you're forgiven for using those other oses.
I have never ever seen that kind of stuff happen in two years of running
My preambles are usually 6-8 lines of non-motion commands
I am on xp now and I hate it
paul__c: you mean me? :o
but that is what the software runs on so...
Timbo just stirred ?
joined the channel from work, just remembered now i'm home :)
paul__c: The guy I was talking with about the download is out for the day.
Won't know till morning if it's him.
bluh I have just about had it for today...I hate office days...today I am billing dept.
les: Know that feeling.
I was out in the weather (20f) today so happy to be back in.
But on cold days I have to ramp up the shop heat very slowly...if dew point > than the cast iron...
well I don't know what good it does to know this, but here's the stack in minimilltask when it re-executes line 1:
#0 emcTaskPlanCommand(char*) (cmd=0x80c1130 "G00 Z0.040000") at rs274ngc.hh:442
#1 0x0804e457 in emcTaskPlan () at emctaskmain.cc:1040
#2 0x08050b84 in main (argc=0, argv=0x0) at emctaskmain.cc:2887
jepler: latest emc source?
les: emc1 cvs from a week or two ago
updated on November 15
so it's a bit old
I need a copy of the exact file you are running
you mean the ngc file?
yes...or just a description of it
use my simple example from the bug report
ok...it is exactly as in the report?
I simplified it for a reason
I'm using a different file than in the bug report
let me try the file in the bug report and make sure it happens to me
I will really need to check this...I have emc pushing around half a ton at very high speeds
my pcb drills cost $8 each and I don't like breaking them
(you win, but still)
I use them a lot as well
down to .02 diameter
carbide drills amaze me after trying to drill fiberglass with HSS
ha...yeah for sure
yes, it happens to me with the file described in #1074914 (though I repeated the line only 1000 times)
I got that traceback by attaching gdb to minimilltask, then breaking on emcTaskPlanCommand when readLine == 1:
(gdb) break emcTaskPlanCommand
(gdb) condition 1 emcStatus->task.readLine == 1
cradek: the latest maxnc.ulp puts a comment and then G64 on the first 2 generated lines, if I can trust my reading of the source. So new files from eagle should be OK.
so it really ought to work with the first several lines non-motion preambles
jepler: I'm not at convinced that we understand the pattern
cradek: I think I understand it
jepler: heh, ok, I trust you
jepler: so it runs the first command with the rest of the current "state" (non-specified coordinates) taken from the read-ahead spot?
yeah, I think that's right
In the many many aborts during running it just stops cold for me...sometimes a delay though
but I don't know enough to propose a fix, other than to say that emcTaskPlanCommand or emcTaskPlanExecute shouldn't be called here
but it stays on the tool path during the delay
paul__c: Is there a package missing from rc46 that prevents a good compile of today's sf?
speaking of rc46, is there a way to get apt-get to work without doing a hard-drive install? I have a USB flash drive if that helps. apt-get complains that there's not enough space, no matter what package I try to install.
rayh: Not that I know of.
All the tickle stuff is there?
emc1 -- I just did a CVS update. Do I need to clean before I rebuild?
I think it's supposed to work right without a clean, but I doubt anyone knows for sure
well, cleaning anyway
(there are auto-generated depends anyway)
I think no important files changed, but I just want to reassure les this happens to me with emc1 HEAD
jepler: I extract the source using a script in emc/generic named packupsrc.
That makes a 2.5 Meg tarball that I move into a pristine directory.
When that is extracted you will see a couple of compile scripts.
I'm comfortable building out of my CVS tree
I'm not going to commit anything from it, since it's an anonymous checkout
I would try it today except for the several hour shop warmup time I have to do
just hate condensation on $100.ft rails
I'm doing all this on sim.run, no real machine involved
The file plat/linux_2_4_20/bin/minimilltask does not exist or is not executable.
I use a single developer checkout so I try not to corrupt it with compile files and such.
iosh.cc:43:23: tclExtend.h: No such file or directory
is linux_2_4_20 the version you are using.
did something change to require tclX?
well I could as well.... but best for me to fire it up and see it as long as the computer is in the cold shop
jepler: nothing has changed
I believe that tclx has been required for quite a while.
required to build minimilltask?
I've never been able to build mini or tkemc, but that's fine
no, for tkemc I'm sure
try various -k incantations
I think that paul put a -k in the make files a while back.
tickle stuff should not be required for the c stuff.
jepler: Are you working from a bdi?
rayh: no, redhat9
But you have been able to compile from this before?
maybe I manually made in emctask last time. That's what I did just now, and I'm running
yep, still happens with a new HEAD compile in the simulator
I compiled the sf this morning under 46 without problems that I could see right off.
Probably the problems have to do with lack of tclx
but I don't know why make -k wasn't continuing on to do all the direcotires
erm, re-spell that last word please
cool new word...
I must run to the bank (still doing billing dept today)...be back in a while
my eyes! my eyes!
* jepler got mini to build
also strange: PLAN_SYNC isn't called until I hit ESC the second time
jepler: now be nice
sorry, I'm just a jerk
jepler: please put your new discoveries in the bug report
what news from your end? any notable developments??
Not to much muddling through mostly.
oh hey robin
finally (after many months of messing about) finished my so**ding plasma design .. 2 going out next wednesday
hey les ...
how long to fabricate?
1 to 1.5 day :)
for a machine?
* robin_sz nods
all prefab parts I guess
* robin_sz nods
* robin_sz points to the CNC laser, the CNC pressbrake ...
I had to do two custom router bits on the cad today
a little pricey
yes C2...it's farmed out...tempted to make em in the shop
but would have to buy a lot of diamond wheels
well, if you had a cnc grinder .. :)
right....that is what the vendor uses
I suspect 1 single diamond 'cup' would do 90% of jobs
only about $160 setup...I can't touch that in house
sometimes its best just to shop it out
it was just hard to find a facility that was not in China
I shopped out my cnc turning ...
bearing pockets sized at +5 thou
I am looking to aquire a cnc lathe and another cnc mill
+5 is big
theres some cheap cnc lathes on the uk ebay at the moment ..
would make nice emc donor machines
..and I gave away two of those (olod) in Chicago
yeah, I remember
just got my fill of trucking the stuff
theres nothing wrong with the Hardinge stuff though .. well made machines, with dinosaur controls
trying to do 20-50 up parts here on the router
have to use variables in the gcode I guess
* robin_sz forgets whether emc has subroutines
sadly it does not
cant be hard to add though ...
Art Fenerty managed it ...
I run out of g5x fixture offsets
so must pop new parameters in emc.var I guess (with gcode)
isnt there a 'start this program again from the beginning' gcode ...
emc does variables at least
prob...but with 5 tool changes it gets a bit complicated
still on a manual change too I guess
but even 5 manual changes is better than a lot of sanding
how about a nice pneumatic collet ...
there is a buggered perske with automatic changer on ebay cheap
a little risky
if bad bearings (likely) it is a good deal
bad windings...bad deal
it is at about $350
yeah, and a rewind costs how much again?
40krpm after 10 min warmup...so prob the bearings
rewind=throw it away
yup rated for 700 hz I think...
so likely the bearings are toast
why are they so much more to rewind than 3 phase motors of normal spindles?
just no economy of scale I guess
also big big g forces
I doubt the economy of scale thing is relevant, rewinds are all one offs, no?
perhaps...but then a 3" armature at 40krpm is...
I guess it depnds how toasted it is ..
if it still turns, buy it!
77kG...must have made an error...
* robin_sz fails to find it in search
* robin_sz nods
still doing plaques?
yes...even though the magazine ads are shut off
damn those pesky cutomers huh?
but getting a lot of web hits
oh...and the numbers on the spindle were ok....therefore the armature must be much smaller diasmeter
6" diameter @ 40,000 rpm= mach1
should use one of those wren jets you have over there...160krpm turbine
you can getr compresssed air powered engraving spindles that do 120K rpm
and some electronic sub-spindles for milss that do simialr speeds
I need high rpm for the signs but this big production turkey thing just needs plain old horsepower
how many hp that perske?
only 3. I need 5 or 7 I think. More =better until cutting forces rupture or burn up the part
yeah, guess so.
still, a 3hp spindle with pneumatic collet ...
gwaan .. you know you want it ...
no, wait ..
you NEED it :)
yeah when you suddenly have thousands and thousands of the things to make...
sets of 50 must be enough ...
but to not kid myself...75% 0f production cost is still sanding/finishing
have to do 500/week
emc is getting a real commercial workout here
thats the idea ...
yup...it is really an ideal project for the machine
yeah, you built it to do just that sort of work I guess
(well, except it needs a toolchanger ;)
I am actually most worried about getting some workers quick
you know, the sanding thing we talked about
hardly skilled work though
whats that, 10 bucks an hour?
just some dc motors, special chucks, and handheld sandpaper
yeah 10 or so if they are reasonably reliable
we do it differently over here
now spray painting operation...I am out of luck there
phillipinos or east european immigrants on half that rate ...
ah yes, lets see. painting. let me guess, somehow, winter has unexpectedly arrived again?
and they go for about 7 but a little more if they are good
* robin_sz nods
yeah...and still no cross flow heat exchanger
that is a $.40/hr problem...I should not dwell on it I think
I only am dumping 4kW...at $.10/kWh
If I remember ...
so that's down in the noise right?
there were extended periods you couldnt paint at all last year
even with heat
was using only 1.5 kW
4kw will help, I suspect it will need more though :)
bought some more of those oil filled finned electric heaters...only $40 each
I calculated the temp rise/flow power thing
humidity was my concern there
4kw will get me about a 20c rise typically
wow, that must be low flow then
yes very low
about 250 scfm
right, I assumed you had several cu ft a minute
haha if I did the calcs right
seems to be in the ball park from experiment
normally 10 times that much flow is used
you have to wear breathing apparatus with a flow that low
1 j / gramme / c
well, 1.006 at 1013 mbar
for dry air?
* robin_sz nods
ach forgot the density@ 0c
pilots should have that memorized haha
soo ... 1 m^3 needs 1300J to raise it one C
so 1.3 kJ /m/c
m^3 I mean
and 1 m^3 is ~32 cu ft right??
1 m^3/s=1.3 kW
I am running only 250/32/min
so .13 m^3/s
.17 kW/degree c
yep your right ...
going to be nice and warm :)
3.4 kW at my flow then
just going to stack the heaters in the inlet I guess
yes I made room in the blower box...just stacked em up
3 1.8 kw units
that should do it
I have the opposite problem
and $.40/hr....no worries
gets to damn hot!
laser sucks in 45kw
so 700w ends up as light ...
but even that ends up as heat eventually
I might have to push the chiller outside ...
yeah bad in the summer
I thought it would be great insode in winter, but its too much
forget about climate control...just blast huge amounts of makeup air in I guess
this is england
we dont do climate control
Oh yeah haha I forgot...from when I worked a while at Lucas
Id say less than 1% of homes have aircon
prob 95% here
but we have nastier temp extremes
in chicago I remember...
highest temp= 107f
134f range in one year
whats that in C ??
70c or so!!!
I guess we do 30 odd max in the summer
-10 in winter
you know whats real cool with this laser,
little TV camera wayches the cut
nice to sit in the office and see it still going
laser and plasma would be good for me to have really
laser, for wood/plastic?
when I grow up and get rich here in a little while we will have to talk about that
nice little 100W co2 tube from china then :)
maybe even 50w
less than 1000 bucks ...
actually...our Miller and Hobart divisions might be interested in your low cost units
put in a word when I'm there?
sure .. go right ahead!
need brochure or web site...
could be arranged, I pulled the site for a revamp
well it will be a little while I hope..but they are bugging me to get up there
(have that turkey stuff to do)
basically I have interet from Cleveland Motion Controls, basically I'm fitting their CNC control to it at the moment.
its very industry standard I guess ..
probably used on 70% of cnc tables in the plasma industry ...
ITW likes that sort of thing
need a marketing hook or two
which is good I guess, because it makes it easy to sell and service
they want me to come back to corporate but I will not do it
can't take the stress
yeah ... 8hrs in an office is a killer
but I just consult for them...max 50% of the time
* robin_sz nods
the money is no fun if the stress makes you sick
worked myself right into a hospital starting the Chronotherm division
along with some of my people
life too short
without doubt, not worth putting yourself into hospital just to feed the corporate machine
just can't take that politics and jetting all over
* robin_sz nods
I dont do politics
I just do what needs to be done, and let others get upset about it
when you work at headquarters of a $12 billion corp your job IS politics
don't like it
then dont do it : .. oh, you arent. fair enough :)