you have the servo motors yet?
yep. had those for years
I don't think these brackets are for NEMA42 motors, but I can make an adapter plate if necessary
(those are flat, and are easy to machine with a manual mill ;) )
SWPadnos: Who is this Man-U-ell you speak of?
my Mexical friend
eh - whatever
SWPadnos: Ah, Si Senior
I picked up a pair of treadmill motors from surplus city - I was really surprised how responsive they were with a 1.5V AA battery considering they are rated for 90V cont 130V intermitant and about 6000rpm peak
it's the current that matters
OK - so I cheated 1.5V 2000Ma
the motor coil is probably 1-2 ohms, so a 1.5V battery would try to put 0.75 to 1 A through it
super batts for digital cameras
(not that an AA battery can supply that amount of current)
they can - but not for too long
I didn't think you could even get the magnetic flux going that low =)
though it only needs the current for a little while, to accelerate the motor
SWPadnos: Actually, they can.
not for long, without getting hot
kinda like the batts the RC car people use
AA are rated at 2850 mAh
D are rated at 20500 mAh
newer ones maybe. I'm speaking of alkaline batteries
so am I
alkaline AA are ~1800-2000 mAh
[00:08:50] <JymmmEMC> http://data.energizer.com/
then select Energizer Alkaline
(they used to be anyway - could have been advances since I last bothered looking)
for product group
these are high output lithium Ion
sinc-manganese dioxide, according to the datasheet
they won't hold a chrage more than 3 weeks without recharging
2850 mAh at 25mA discharge rate
Huge differnce between D and C
at 500mA, it's under 1500
SWPadnos: Hey, you weren't suppose to read the WHOLE datasheet ;)
yes I am - that's my job ;)
SWPadnos: Eh, you're fired!
SWPadnos: Go become the fry cook you've always wanted to be!
yer not the boss o me - nyah nyah
SWPadnos: I'll tell your wife on you!
SWPadnos: You know... that rug that somehow mysteriously got a stain on it
yeah but zinc-manganese dioxide - thats OLD school - I was taking those apart to get zinc strip to make some free H2 (from HCl)
back in the early 70's
remember all those ads for energizer and duracell alkaline batteries? how they're so much better than X, Y or Z
all alkaline batteries are within 3% of each other
best buy the cheapest that has a fresh mfg date
Except, Energizer has a lifetime guarntee that ANY of theri battereis won't leak or they'll repair/replace the device
I think consumer reports did a test, and found that some generic-label battery actually had marginally better performance than the big brands
JymmmEMC, well, that's a good thing (but not part of battery performance :) )
SWPadnos: $1200 Camera and you begin to care
oh sure - liability and warranty issues are important
but have nothing to do with performance
true - funny thing was each battery performed differently based on discharge load vrs time
ther are related to overall value though
oh sure - you have to match the battery technology to the type of load
for best performance anyway
SWPadnos: You get a bill for $75 from Nikon and see how much you care about Performance for the $0.99 pk of battereis
so technically one could be best for flash - another better for walkmans...
I charge Nikon, buddy
I recharge my 2500MAh AA batteries =)
Tha'ts Million, not milli =)
all you need is a billion AAs in a case
they are nice batteries though... the 2500mAh Energizer rechargables.
back to motors - those 90V treadmill motors turned faster with the AA batt than my 37V Ametek servos - but the 90v had some noticeable cogging where the ametek was perfectly smooth.
it's probably different forward vs reverse as well
may not be, but probably is
not that I could tell
ok, that's good
you can tell cogging by just shorting the leads and trying to turn the motor quickly
it's a 0V battery :)
the Ametek has provision to be a 4 brush motor - although the one I was testing had only 2 brushes & holders installed @ 90 degrees
the treadmill motors are 2 brushes @ 180 deg
not sure if it means anything but the ameteks also had 4 PM where as the treadmill motor had just the 2
what makes the ameteks so smooth is the 60 deg skew in the armature
"smoothness" is what differentiates a servo motor from a normal DC brush motor (for the most part)
if you gear them down.. The cogging will be less.. lawenceg has had good luck with cheap motors for servos. using windshield wiper motors.
I know he names, but not the details, of the methods they use to make good servo motors (ie, I won't know what I'm talking about if I keep talking much longer :) )
Dan Mauch @ Camtronics has been selling the Ameteks for years for BP conversions
the little Ametek I got from surplus center would have a very hard time running a BP, unless it was turning the spindle speed knob
but I think he is selling a 4 brush version
I may be wrong but I believe the 4 brush can take 2x the Current.
* Skullworks-PGAB remembers when 100IPM on a BridgePort CNC was "Rapid". ( It also loaded programs from paper tapes....)
apparently the new 6000 ipm machines don't even have windows and feature advanced door interlocks
just because they move so fast
8 feet per second is a bit fast to get away from :)
a wee bit difficult, yes.
that 3/4 second average reaction time would let the machine smash through me from head to toe ...
apparently with a 40k spindle in aluminum
a 1" carbide tool
exhibits almost zero tool wear
if you feed it properly, that is
SWPadnos: Not to worry, with your hollow legs you got a good 10 mintues before it'll be over
I really like the datron Dynamocs ones (with videos!)
.003 to .005 per tooth
when we were at NIST 3 Fests ago, they were doing some experiments on tool wear with a 30 or 50 HP lathe
they found that when the tool starts to screech, turn it up a little more, instead of turning it down :)
they were machining bullet shapes out of cylindrical stock in <1 minute, for like 4 or 5 inch diameter stock
and I'm pretty sure it was steel they were turning
sounds like a roll lathe
roll lathes are for roughing out the basic forms for rolling mills
no, this was turning final shapes
5-6" depth of cut
.250-.5 feed per rev
it was a standard big lathe, with some extra stuff
like a 6-axis dynamometer
basic extra features, amirite
(I think. could have been something else - it's been a while)
and they were using very good measuring tools to check insert wear (standard Kenanmetal inserts)
[01:23:19] <LawrenceG> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=140152947574&ssPageName=STRK:MEWA:IT&ih=004
anyone know about er32 collets?
I am wondering about this for myshoptask as I only have a couple of MT3 collets
looks like a nice set, even has 1/8
yea... that is key for most of the stuff I cut..... I am currently using a 1/8" to 1/4" adapter.... not the best solution
a pain to change bits
I think I break more pcb drills prying them out of the adapter than I do drilling :}
your spindle is MT3?
I have a few drawbar mount collets, but the er32's look easier to change
easy to change tools of one size, not so easy to change sizes
Ziegler: awesome, I just watched and listened to your video .. I did the "O words" to set everything up and play chords; cradek did the arrangement of Daisy.
that's still so funny to me
skunkworks: you didn't run gopt on that pcb code, did you?
hmmm... flash... not workinhg here
LawrenceG: too bad
I should make a video - it's nicer with all 4 parts
cradek: your blog entry is more likely to be useful to someone if you say what "the right diameter" is
jmk said that too
"the right diameter" is the diameter of all your tool holders that actually fit
what if I don't have an existing tool holder?
I wish I knew what the right number is
it must not be too critical
I'd say the right size is "large enough to catch any retainers, and to provide a datum plane. small enough that it doesn't interfere with anything" :)
very nice jepler and cradek... my kids get a kick out of it... (the guys at work to... but the dont openly admit it)
Ive got the 4th axis... just didnt have it hooked up
cool, you should try it with
well, I guess steppers do have an advantage over servos then
working on it!
wget worked to retrieve the video... playing, but no sound... bummer
LawrenceG: what are you trying to play it in?
totem, xzine, mplayer
you need mp3 playback capabilites
mplayer plays it fine for me... but then I encoded it :-P
ok... works in mplayer now..... guess the crashing browser tied up the sound device
hmmm. Willie Nelson must be quieter than Blue Oyster Cult
ya would think this was windows!!!
most bantu probably don't have internet access
[02:06:32] <Ziegler> http://www.linuxquestions.org/questions/showthread.php?t=449244
that was for you LawrenceG
Ziegler: thanks... all is well now
gives 46.0mm ~= 1.811in
very cute.... Jymmm probably wont hear too well as his mid band resonances are damped out :}
bahhh!!! my forth axis config is gone!?
just pop it off the stack
jepler: thanks, I added that information
LawrenceG: Eh sonny, did ya say something?
hi Jymmm ... ears burning?? have you played daisy yet?
LawrenceG: You did say you got a &%) and not a ^%)
750x is on the way
it's %($*&#(% drive
hopefully pull the smarts out
LawrenceG: Talk to skunkworks about the X,
the manual shows it as a plugin.....
LawrenceG: It is, but I don't remember what jumpers you have to replace when you pull out the daughter card
that may take some digging or a call to customer support
LawrenceG: It SHOULD be in the manual
I wish the schematics were as well
LawrenceG: Also, if you dont want the daughter card, let me know what you want for it, I just wanted to play with one for shits and giggles.
how do you have yours mounted?
LawrenceG: Tha'ts a little personal don't ya think?
LawrenceG: Per the specs listed in the manual on a 1/4" aluminum plate (at the moment)
I think the right answer was 'often'
It aint quantity, it's quality
hmm... when your married you cant be choosey
the shipping on the drive was more than the drive
sad, aint it
[02:20:45] <LawrenceG> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=330158919233&ssPageName=STRK:MEWN:IT&ih=014
LawrenceG: oh fsck me.... the seller is just 30 minutes south of me
input scale for angular...
I have it 1/4 steppting
direct direct drive
steps per degree of rotation (assuming you're using ANGULAR_UNITS=degrees)
1.8 deg / step
the stepper is literally direct coupled to the turntable?
200 steps / rotation
in this case I just have a stepper laying on the table
Jymmm: that would be a fun outfit to visit
so input scale is 200 x 4 ?
then pick some easy value
like - copy the number from the Z axis :)
if you were actually using it as a rotary table, you'd put 4*200/360 = 2.2222
hmmm. that may not work as expected. got to try it to find out
right... I realize for the reason it doesnt matter, but I am trying to get the ini file back to semi normal
if the goal is to play daisy the only important thing should be to put the same scale in daisy.ngc
what table would you normally have it attached to?
havent built a table yet
the little ones are usually 90 turns/rev or thereabouts
I just have the 4th axis with all the plans in the world to do something with it
I think if you enter 2.2222 the stepper will make 1 revolution when you command move from A0 to A360
is Z not setup as "A"
program doesnt seem to move A
no movement on A
from the program
Ziegler: try this version instead: http://cvs.linuxcnc.org/cvs/emc2/nc_files/daisy.ngc?rev=1.2;content-type=text%2Fplain
I had to modify that one to run on my 3-axis machine
M20 eh ;-P
hmmm that sounded ugly... somthing is off
Ziegler: check near here:
(set XYZA axis scales)
O100 call    
where is the bounds at... like y -1" 1"
this older version always stays right near 0,0,0,0
LawrenceG: for what it's worth, the seller for that collet set is the same one cradek got his drill chuck and arbor from
ah... other one moved much less for me
aahh... had z scale typo
[02:54:16] <cradek> http://www.practicalmachinist.com/cgi-bin/ubbcgi/ultimatebb.cgi?ubb=get_topic;f=1;t=015636;p=2
odd set - they omit 13/32, 17/32, 21/32, but have many other 32nds
beware the guy I bought my chuck from - maybe I just got lucky
A playing a low tone?
cradek, I've had a problem with him, as has a friend of mine. slow or no delivery, discourteous response to inquiries, difficulty remedying problems
jmkasunich: thanks... I was checking that vendor out as they seem to have some useful stuff
SWPadnos: sounds like your experience is typical
sure wish I had another chuck for 0-1/2" or so though
and another arbor ;-)
any particular reason you bought one so big?
not paying much attention to the small size
HAHA!!! cool.. 4 axis does sound good!
Ziegler: time to make a new video for us?
my biggest chuck is a 5/8, for my drill press, and I have a 3/8 to go with it
for the lathe and mill, 1/2 is plenty
I have a few big drill bits, handy for the first pass of a large bored hole on the lathe, but they are all morse taper shanks anyway
wonder if 1/2" goes down near 0 (he doesn't say)
FWIW, my albrecht is 1/32 to 1/2
dunno if the clones have the same low end
I think mine does, but it's an "expensive clone", not totally cheap crap
(it's an R-8 chuck though)
almost done uploading
[03:22:08] <Ziegler> http://images.myonlinesite.com/cnc/use/20070902/videos.html
that play sound for you all??
though it does sound a little like a warped record
sounds better in person
around 17 seconds it gets disharmonious
but it is cool :)
I think it might be because the stepper is not attached to the table
could be - it may be stepping and bouncing
probably not the worst musical instrument ever, but close
like the bagpipe, it has no volume control
bagpipes are really great ...
on someone else's mountain
but up close, they'll rattle your teeth out
I like bagpipes
(in moderation of course)
moderation like being a few miles away
they're very nice in the right circumstances - mountain meadow, misuc wafting in on the breeze, etc
its nice to have a 4 axis ini file working again
seem this new file you posted move back and forth in one place alot... the old file actually made a path
the new song ngc
I was thinking that it would be neat to actually trace out a path that made something
but I didn't want to think about it enough to actually do it :)
I like the path also
was HOSTMT12 intended for m5i20 with 3pcs 7i33?
toast is now known as toastydeath
toast becomes toastydeath at bedtime?
toastydeath is what i have registered
and is opped on other channels
I'm running a run-in-place version of emc2, I've tried to use a hal configuration file that I usually use with my installed version
this file uses new threads
but with the run in place version I get an error
insmod: error inserting '/home/xemet/emc2-trunk/rtlib/threads.ko': -1 Operation not permitted
HAL:2: ERROR: systemv failed, returned 1
HAL:2: ERROR: insmod failed, returned -1
should I use sudo?
xemet: when you get an "error inserting" a module, look in dmesg for the the rest of the error.
xemet: maybe... run the emc-environment script this way notice the space between '.' and text " . ~/your-emc-directory/scripts/emc-environment'
xemet: you should never use "sudo" with emc2, except for "sudo make install" or "sudo make setuid"
xemet: this wiki page lists most of the .hal and .ini differences from 2.1 to TRUNK: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions
tried with emc-environment
but the error remains
what does dmesg say?
so look at the output of "dmesg"
I see this:
ERROR: new thread period 500000 is less than existing thread period 988960
the fastest thread must be created first
threads creates thread1 first, then thread2, then thread3
fix your hal file
but but servo thread is created in another file
so I move those lines in the core_stepper
paste your ini and hal files at pastebin.ca, and paste the url here
oh, you mean the threads created by 'loadrt motion'? You'll have to reorder things, then, so that the faster threads are loaded first (i.e., loadrt threads before loadrt motion)
why are you explicitly creating threads anyway?
motion gives you three, and that should be enough for most setups
I thought it created 2. it creates 3?
duh, right - 2
I was thinking of traj, but thats not a thread
I was using them for the pid control of the spindle
it works on the 2.1.7
you're right that it does not give an error on 2.1.7, however it does not have the intended effect either
motion creates two threads, the base at 50000 ns
and servo at 1000000
has the base floating point?
why not run your PID in the servo thread?
I was using one at 50000 with floating point to calc the spindle speed and one of 500000 for the pid
I remember that I tried and the pid would not work as well as with the 500000 thread
now I try to run all PID and the spindle-velocity cal on the servo
"encoder" includes a velocity output; why aren't you just using it?
in the official version I had problem with velocity of encoder
because I've only two pulse per rotation
so if the thread was too fast I get 0 velocity
so I had to write a component to calc the time between two pulses
and calc the speed using that as basis
what is your speed range?
since you're trying to use the development version now, you should see whether the improved velocity estimate is good enough
yes...but mine is a old-development version
3500 = 116 pulses per second
the one I used to modify for the nurbs
I bet the velocity signal in the encoder module will work well at that speed
you mean the encoder module in the official version?
(it measures the time between pulses AND the number of pulses)
or the one in the develpment version?
in the development version
I don't know if the new velocity is in the released version
was it there in march?
jmkasunich: I'm pretty sure it's not
I've installed this version in march
and never upgraded it
xemet: I don't know; if you're using the development version of emc, you should update your emc using "cvs up"
the very worst version of emc to use is "cvs from weeks or months ago"
jepler is right
yes, but I've used it for the nurbs modify
either use a released version, or keep it updated
so if I update it
I've to check every time if my patch continues to work
if not I've to search why
what exactly is your nurbs patch?
just a try
based on what jepler done time ago with spline
added a new g-code
if you think it might be generally usefull, you should consider submitting it
then it will become part of EMC and you won't have to worry about if it breaks when you update
well...I'm not a programmer
so I think my code is not so good as the original
it apporximate a nurbs with biarcs
I really don't know what to suggest. I do know that if you stick with an old cvs snapshot, you will gradually fall behind. we cannot support random CVS checkouts, not when the most likely answer to most bugs is "we already fixed that, please update"
xemet, that's the life of a patch maintainer. you get to see if it breaks whenever there's an update to the project :)
yes, I understand
but you can't expect others to take the time and effort needed to fix problems in your setup that have already been fixed in the latest code
[13:15:41] <jmkasunich> http://cvs.linuxcnc.org/cvs/emc2/src/hal/components/encoder.c?rev=1.26
sorry I was asking only beacuse this was working in the 2.1.7 just to know why
the improved velocity output was added to encoder.c in January
now I have pid hooked to the servo-thread and everything works
I'd love to see your nurbs stuff in the EMC codebase (though I don't know if there might need to be some discussion or changes to the actual G codes used)
I've just to tune the PID better
sorry if I seem a little harsh there - haven't had any coffee yet :)
don't worry, I know it is a problem of mine configuration
let's pass to simple questions
what if I want to use halcmd from the run-in-place version?
just to know
if one would like to add a thread in the development version, what is the right procedure?
if I want for example basethread 50000, one with fp at 500000 and the servo at 1000000
is there a way?
do you need FP in the base thread?
let's say no...
let's say I want only a thread intermediate between the base and the servo
and 50000 is an ok BASE_PERIOD for EMC?
xemet: you can't easily add a thread between base and servo in either version
so what's happens in the official version when I type loadrt threads and give a thread the 500000 value for period?
swpadnos: yes, in my configuration for now it is ok
xemet: in the released version, you probably get a broken thread
the CVS version checks for that case and gives an error message instead
what's a broken thread?
the thread works because it has functions hooked to it
xemet: are your sure your thread is really running at 500000nS? it might be doing 1000000
I'm not sure of nothing. I can say some things:
I've added two threads with floating point in the 2.1.7
one at 50000
(the same value of base, but with fp)
and other at 500000
tuned the pid
and the pid works really well
what do you get from "show threads" after that?
if I hook the functions I use to control the spindle velocity to the servo-thread
the pid doesn't work as before
so I think the thread is not running at 1000000 like the servo
this is the result of show thread http://www.pastebin.ca/679907
what does "spindle.0" do?
clacs the spindle velocity
is it reading counts from counter? or are the spindle encoder signals directly going to "spindle"
it has a counter that is incremented every period if the signal if false, when it turns to true it calcs the velocity
no, counter is there for nothing
sorry, I've to remove it
btw, you should move the parport writes to after pwmgen.makepulses and counter.update-counters in the base thread
I tried to use it to get the velocity
how? just I write them after?
yes, or you can use the optional argument to addf
plain addf: "addf function-name thread-name"
optional: "addf function-name thread-name number"
if number is 1, the function will be the first one in the thread
if it is -1, it will be the last one (same as default)
-3 means third from the end, etc
and...just for curiosity, why should I move them?
you are calculating the step and pwm signals in functions 6 and 7, but you don't send them to the real world until the next time the thread runs
ah ok, of course
(its only 50uS later, so it doesn't make a huge difference)
any idea about the threads?
are they broken?
well, if it works, they must not be too broken
I'd have to study the code carefully to see exactly what happens
about 98% of users have only the base and servo threads
and another 1.99% have only threads that were created with "threads", in descending order of speed
very few people create threads in any other order, and very few have two threads at the same speed
I think they are really stupid :) :)
yes, I've to fix that
RTAPI (the RT code that lies under HAL) creates each thread with a priority one step lower than the previous threads
the goal is that fast threads have high priority so they can interrupt slower ones
I've to undesrand why my pid doesn't work with the servo thread
when I talked about 98% of people I didn't mean that you were doing a bad thing
with the thread I've seen I can set Pgain to 12 and the system is responsive
I meant that you were doing a poorly tested thing
if I hook it to the servo I can't set Pgain over 2
I think the problem is in the component that calcs the spindle velocity
it is not stable
the output from that component is only updated when you get a pulse from the spindle
thats only a few hundred times per second
I should have more ppr
but I can't...
you should download encoder.c from CVS head and try it
can I hook to the encoder only one phase?
because I don't have phase-A and B
I believe so - in CVS head, counter is deprecated, and there is a "counter-like" mode for encoder
(I think - I need to check the man page for encoder)
I'm compiling the new trunk version
so I will try on that
yeah, there is a parametere "counter-mode" (a bit) set it true if you have phase only
crap - I hope the man page is wrong about counter mode
"When true, the counter counts each edge of the phase-A input"
it should NOT count both edges - unless the signal has 50:50 duty cycle that will give bad velocity results (and bad position)
trying to use my standard configuration files with the new trunk...I get this new error:
HAL:103: setp requires 2 arguments, 3 given
the file is the core_stepper
and the line is:
setp stepgen.0.position-scale [AXIS_0]INPUT_SCALE
in your ini file, is INPUT_SCALE a single number?
some older ini files had two numbers, a scale and an offset
now I check
EMC1 used both
EMC2 only uses scale, older version would ignore the 2nd number, newer version complains
ok you were right
now, I new misterious (for me) problem
emc/task/emctask.cc 312: interp_error: Parametro richiesto mancante
Parametro richiesto mancante
emc/task/emctaskmain.cc 2643: can't initialize interpreter
I'm afraid I don't know much about task
this is why I suggested that you only copy the CVS version of encoder.c, instead of the whole thing
emc/task/emctask.cc 312: interp_error: Coordinate system index parameter 5220 out of range
Coordinate system index parameter 5220 out of range
the latest EMC2 supports xyzabcuvw coordinates
yes, I know but since you said me to ugrade my cvs version I thought it was a good chance to do that
I think that param error is related to the new uvw axes
updating is good, but your nurbs patch probably doesn't work with uvw
there is no nurbs pathc at the moment
I'm trying to run the normal version
I think all the old configuration files doesn't works...
I tried the sim/axis also
are you running it with one of your configs?
I think it has something to do with the var file
old var file doesn't have variables that are needed for uvw
there is a simple way to fix that
unfortunately I don't remember it
yes, they were my configuration files
delete your var file maybe?
you need to add some vars to the var file. I'm llooking it up right now
nope - deleting it won't do it
ok, the sim/axis of the trunk works
I can easily use the example files
my configuration is little different from the standard
one sec - I think you only need to add one line to the var file - I'm just looking up what it is
wow, axis shows the limits
heh, lots of neat new features
I always thought "why doesn't it show the limits?"
when do you think will be released the 2.2?
couple months maybe
we are getting close to feature freeze I think, then some testing before release
the more people we can get to use CVS for the next couple months, the more bugs we will find before we release ;-)
well I'm here
[14:12:38] <SWPadnos> http://cvs.linuxcnc.org/cvs/emc2/configs/sim/sim.var?annotate=1.2
the lines with ! at the start are the added ones
unless you've made changes to your var file, you should be able to copy one from a new config over your old one
if you do that, you'll lose any offset coordinate systems you've set up
never modified the var file
then you should be OK copying one from the CVS configs over yours
now I want to try the new encoder
do you recall how many files had to change to add the NURBS suppot to the interp?
were the mains
I will try to launch diff to see how many changes had occurred between my old version and this trunk
well, there's another way to think about it
and I will see how those are related to changes
you can look at the changes in CVS between the version you based your changes on and TRUNK
then try to apply those to your changed files
well, you can CVS to give you a diff between any two versions
I don't remember the exact command line, but it's something like this:
cvs diff -u -r first_revision_name/number/date -r TRUNK
something like that will give you the changes since <first revision name>
I will try
you can also the changes you made as a diff
returning to the encoder
you need the original source, then just diff -u original_file modified_file
so should I use it in counter-mode?
I think it's easiest to just run cvs up, and then fix any clashes afterward
cvs up gets a lot of things right automatically, any other method is just more work
that's probably true
and, the more often you do this, the easier it is, because there are fewer/simpler clashes each time
though you can just keep (or mail around) a diff if you get one somewhere along the line
if I use the counter-mode, shoul I hook only phase-A right?
I just tested counter mode, it counts only on the rising edge, which is good
I'm correcting the man page now
the output velocity is in?
rotation per second?
units per second
units as set by the scale factor
for a spindle, you probably want to set scale to the ppr, so the position is in revs and the velocity is revs/second
so, the scale means the ppr?
I remember the old encoder has a ppr parameter
the scale is the number of counts per unit of measure
no ppr in the old one
scale is same as before
example: for a linear axis, if you have a 2mm screw pitch, and 1000 counts per rev, setting scale to 500 would make encoder's outputs be in mm
(since the machine is 500 counts per mm with a 1000 count encoder and a 2mm screw)
if I want rotation for minute and I have 2 ppr I have to set it to 120
lemme think a minute
I think you want 2/60, not 2*60
2 pulse per second = 60 rpm, so 2/60 of a pulse per RPM
so scale should be 2/60
don't apologise - it is confusing for that
the scale is the number of counts per unit of measure
so for every rotation
I've 120 pulses
no, for every rotation you have 2 pulses
I want to say
for a revolution per minute, you have 2 pulses per minute
2 pulses per minute = 1/30 pulses per second
I want the velocity in rotation per minute
not per second
so I set it to 2
if you set scale to 2, the velocity output will be in rotations per second
if you set scale to 0.033333333, the velocity output will be in rotations per minute
velocity = counts_per_sec / scale
loadrt encoder will load three encoder
if I want one should I type counts=3
loadrt encoder num_chan=1
most newer components use "count" and default to 1, encoder is old and unfortunately we haven't change them all to work the same, because it would break existing configs
(loading 3 by default was to make things simple for people with 3 axis machines, but I think it was a mistake)
03jmkasunich 07TRUNK * 10emc2/docs/man/man9/encoder.9: fix man page - encoder 'counter mode' counts only on rising edge, not both edges. (Both edges would give inaccurate results if the signal isn't a 50/50 duty cycle.)
it is WORKING
all on the servo thread
PID and encoder
no need for other stupid threads!
and very stable
i love this PID, it was the first time I used one when I configured the spindle control
thank you very much jmkasunich
good morning all
has anyone use emc with a pci express video card yet?
yes, but not with a real machine
what kinda card did you use
Nvidia 7300LE, 7800GT, and Quadro FX 3500
also onboard Intel 965(?)
did it use these cards native,or did you have to do anything to make them work
I think it worked OK with vesa, nv, or nvidia. of course, graphics performance and latencies varied somewhat ;)
I'd look at my notes and tell you how things worked, but I don't have any notes
sounds like my notes
Morning all. has anyone seen these: http://www.amtencoder.com/
Roguish, nope - but they look cool
i can get a 7300le on newegg for 30 bucks
check the price: http://www.digikey.com/scripts/dksearch/dksus.dll?Detail?name=102-1307-ND
BOSS, yep, that's why I had one :)
plus, they're low power and fanless
geez - 30krpm at up to 512 PPR settings
not that we're going 30k rpm or anything.
gotta build a new box for the trade school setup,agp is gone
are you buying a computer?
no,build all my computers
well, that's what I mean - are you getting a new PC to run the machine?
ok. there's an embedded PC I've experimented with lately, which seems to have phenomenal RT performance
this is with no X, and lots of stuff disabled
(like printing and that sort of thing)
it has a PCI slot and a PCIE slot, though the onboard graphics seem to work fine
with a celeron 1.8GHz, I get latencies of ~1-2 microseconds
with a core 2 duo 1.8 GHz (with a busy do-nothing running in userspace), I get ~200 ns most of the time
was gonna use an amd athlon64 socket am2 processer
there's a blip every 5 seconds or so which I believe to be kjournald, so I'll be setting up a hard disk with no journaling to try that out
on a micro atx board
everything i have is amd
uhm SWPadnos, I used cvs diff
yeah, me too except fora few specialized computers where there wasn't really a choice
in the src directory of my old modified trunk that is from june I think...not from march
cool another amd guy in da house
it seems that nothing is changed...it gives only the changes about my modify
hmmm. I don't know the CVS command line well enough to give you specific help :(
but there's the cvs book online
[15:19:05] <SWPadnos> http://cvsbook.red-bean.com/
why cant emc run the newer onboard video chipsets
like the nv6100
I've run EMC on a 6150 MCP system
it sucked worse than anything I've seen
not, the command is right. Jepler or alexjoni gave it to me
I think we're expecting to update the liveCD release when Ubuntu 8.04 comes out, since that will be an LTS release like 6.06 was
BOSS, yeah - it's an nvidia integrated graphics chipset
though I may have installed feisty and compiled EMC+RT - don't remember
thought it looked familiar
I'm booting up that machine to check
it'll be a minute
xemet: did you run the command in the top directory of your checkout? if you run it from a subdirectory I think it will only show the changes in that directory and its subdirectories. (I'm not a CVS expert, but I'm learning.)
brb - gotta check on the cat.
hi all, is Ubuntu 6.06 LTS still the preferred platform for EMC2 ? I'm installing on a new computer and wondering if something new is just around the corner...
yes, but I need the changes only in that directory and subdirectory
BOSS, it's just reported as a C51 and MCP51 - and it's feisty now so I can't tell you anything concrete about the RT performance
but it sucked big time
anyone here knows about the CHKS macro?
was it a replacemente of the CHKF?
sounds like i need to get the 7300le from newegg
Roguish, I'll stick one of those encoders on my next digikey order (which should be this week)
BOSS, note that the accelerated driver doesn't play nicely with RT
on the 7300?
what does it do
I haven't looked at physical output on a scope, so I don't know how bad it is
Roguish, I'll stick one of those encoders on my next digikey order (which should be this week)
I don't recall if there are crash issues also, though I haven't had any
I know the RT latencies can be great for a while, then you'll get something terrible (milliseconds instead of microseconds)
ok. i really like the price. looks like a good alternative to the HP.
my agp nvidia seems to work ok
with the nvidia accelerated driver (and RT)?
Roguish - looking at the specs, it seems they may need an adapter ring to fit on midsize or larger servos
didnt load any driver,just emc disc,i assume its real time,stepper mill
I'm pretty sure the encoder mount on my servos has a 1.5" diameter depression
BOSS, ok - then you're using either nv or vesa, not nvidia
you have to manually install restricted modules to get the nvidia proprietary driver
hmmm. I don't have anything I can put those on. my motors use shaft encoders, not hollow
it says nvidia
as the driver, or the vendor?
for the card,you saying emc has me running vesa?
I don't know what it would do by default - it could be nv or vesa
k,but im not running the nvidia driver,correct/
you tell me. I can't tell you without a login to your machine
lemmee reboot into ubuntu
no rush. I'm pretty busy today (again) :)
i here ya,lol
swpadnos:im in ubuntu now
where can i see what driver i have
look at /usr/X11/xorg.conf
though there's probably an easier way
like grep -i nv /var/log/Xorg.0.log
but I don't recall exactly what you see there :)
im lost now for sure,lol
look at /etc/X11/xorg.conf
use less or something
then look for the word driver (case insensitive, type "-i" inside less to toggle that)
there will be several for things like wacom tablets and mice and things
but it'll be pretty obvious when you get to the one for the video card, which should say nvidia or nv or vesa
ok, that's the open source, not-too-well-accelerated, RT compatible driver :)
it seems to work
that's the one that works
would that driver work with the 7300le
I think so
I'm not sure though. check the google
i have a agp nvidia ti 4800 that gives me an unexpected realtime error when i try to use it
well, there are at least two aspects to "will it work"
one is "can the driver even be used with that video card"
the other is "will it screw up realtime"
(another is "will the performance be adequate")
shouldnt any new pci express card be adequate if it has a driver that works in emc?
not necessarily when you think about realtime
SWPadnos: is that a common problem?
you guys using video cards as drivers?
boss is now very confused
err.. .to comunicate with the driver?
Ziegler: i think driver here means kernel-module for handling gfx
oh... anyone tried using a vga card to send step / direction to a driver?
not that i know of, interesting idea thought :)
A video frame with a binary pattern, and repeat the same motion 60 times a second...hm
I think that was talked about on either cadcamdroedm (whatever) or the gecko yahoo user list
yeah... I think I have read about it some place... I didnt come up with the idea
for higher speeds you could repeat the patterns
xemet: sorry for the delay. not sure why you aren't seeing more changes - unless there really weren't any others.
heh - that circuit board was milled with accelleration set to 50in/s/s
hey guys, can I use an floating point number in the scale lines on my stepper_inch.ini file?
swpadnos, thank you, my nurbs pathc was succesfully applied to the new trunk
just had to add in some points the u,v and w new coordinates
I just ran trhe figures on a pulley set that will fit, and the z's 16000 turns into 34285.71
great, one less thing to sweat, thanks
next Q: McMaster Carr's acme bolts and bronze nuts, what lub will stop the squeal, or do I have to wear the black off before it shuts up?
can anyone help me out with this simple program and emc?
[21:51:50] <DanielFalck> http://pastebin.ca/680446
I am getting the 'Radius to end of arc differs from radius to start' error
I know I can make it work with an 'R' value, but I want to understand why this doesn't work
the development version of emc has improved this error
on my system, the error reads:
Radius to end of arc differs from radius to start: start=(Z0.7500,X1.5000) center=(Z0.7500,X1.7500) end=(Z0.5000,X2.0000) r1=0.2500 r2=0.3536
center is calculated by adding I and K to start
but it's supposed to end at X2. Z0.5
I see you have that
X moves 0.5, Z moves 0.25 - that's not an even multiple of quarter circles
I'm visualizing in quadrants
I'm doing a lathe post for apt360
I suppose (a) graph paper or (b) AXIS preview would be helpful here :)
I've been thinking lathe diameter stuff all day : )
emc is always probed with X as a radius
so I have everything backwards and all X's *2
need more coffee...
i wonder why qt4-dev depends on mysql
that lathe diameter stuff is evil
[22:07:28] <DanielFalck> http://pastebin.ca/680462
was that created in apt?
yep. you're moving 0.25 in each dimension, with a 0.25 radius = 1 quadrant
the last one looked like you were trying to go two quadrants in X, but only one in Z
but I had to alter the X1.5 to be X 1.75 by hand. Now to work it out in the postprocessor
SWPadnos: yes it did
what's the intended move?
In lathe code, you always think in diameters to keep it simple for the operators (and me :)
oh, so that's the problem. that should be easy (tm)
just dont do anything dumb like showing the wrong X coordinate
I had multiplied all X values by 2 in the python code for the post
emc only does radius - shows diameter on the screen though (in axis)
I've been playing with apt this weekend, trying to work out some simple roughing for the lathe
[22:13:37] <DanielFalck> http://pastebin.ca/680467
that's the apt code for anyone curious
a lot of code for a little bit of output
geez does it at least do multiple passes to depth?
with the subroutines it will
Using G41 | G42 in G18 can sometimes cause strange movements or errors
on most CNC's
Skullworks-PGAB: in emc, G41/G42 cutter radius compensation is supposed to work in the G18 plane specifically for lathes .. if you run into any bugs, please report them.
fenn: I'm guessing qt4-dev depends on mysql because qt includes a database interface API
[22:30:56] <jepler> http://doc.trolltech.com/4.0/qt4-sql.html
jepler: did you see this? http://www.youtube.com/watch?v=Og6V-nS2dig
skunkworks: yeah, I criticized that it didn't look like you'd run gopt on it
it keeps rapiding from one side of the board to the other
sorry - I had actually used pcb-gcode.. Because I was running it on windows..
huh -- another emc video I hadn't seen before. http://www.youtube.com/watch?v=I9OAmw7-A2c&mode=related&search=
starts off with my favorite program shown on the CRT :-P
[22:40:13] <skunkworks> http://groups.yahoo.com/group/pcb-gcode/
Guest365 is now known as skunkworks
if configure wants Python.h it is probably set up wrong right?