cradek: do you know what size you need, or are going to fit them?
I have an educated guess only
calculated from the measured backlash and observed/guessed ball-to-screw contact angle
cradek: Well, if it's going to cost you more than a round-trip ticket to LAX (like you have a few ballscrews to hand fit), let me know.
surely not. If I can't get it in one or two tries ordering balls, I'd send it somewhere
what is in LA that would help?
do you know a source for balls?
cradek: I know a place in LA that will let you bring in your screws, hand fit the balls you need for free, then buy what you use.
that would be amazing
you have o hand fit yourself, but they have all the balls there to let you try and size up what you need.
cant you just get another nut and preload them against each other to take out lash?
what's their name?
cradek: I don't recall, I called and talked to the owner once a couple years ago. very very nice guy. They DO have sizing kits, but a lil prociey
fenn: there are two nuts a fixed distance apart. preload comes from carefully selected ball size.
cradek: I can look it up, but might take a while
uh, fixed distance? can you add a washer or shims?
cradek: if you need it let me know.
fenn: that would be ideal, but nope
why do they have two nuts if they arent preloaded?
fenn: they are preloaded by carefully selected ball size.
er, i mean, pushing in opposite directions
yes, they do
preloaded axially, not radially
one set of balls (nut) pushes right, one left
but it is in one assembly
i see. sounds... too stupid to be made up
JymmmEMC: I could really use the name or a lead (address to search for?)
fenn: as far as I know, it's very normal
yes, most things are like that
our big machines have shims between the nuts.
that would be nice.
better yet would be a fine thread with a locking mechanism
cradek: They make precision balls for NASA/JPL reference standards as well
cradek: Heh, I found it
it was Parker ballscrew or something right?
uff no that's a stepper drive
The overall accuracy of this system is one-million of an inch and is calibrated using standards traceable to NIST. the size of the balls to an accuracy of 10-millionths of an inch.
cradek: Bookmark it ***** http://precisionballs.com/
JymmmEMC: I need balls sized in at least thousandths
.0005 would be better
cradek: You can get .0000005 if you want they just have SO much info on their site, that it's hard to find it all
I was just trying to get you started in the right direction
to the sizing kits
I can't figure out how to order balls :-)
Jsut call em mondya moring, the owner is a good o boy
you know what grade you need?
plain old chrome steel would be fine I think
you can get flat spots on the balls
let me find the thompson grade again
cradek: max static load (in lbs) ?
there is no static load really
what's the diameter of the leadscrew?
[00:26:39] <Dmess> http://www.cbc.ca/listen/streams/r1_toronto_32.html
Thompsons shows a Max static load on 1" of 9000 lbs
it's a very small machine. the whole carriage can't be even 500 lb
ok, just wasn't sure the application.... pcb mill, or one where the angle iron is taller than all of us =)
it's a 2000 lb or so lathe
screw is maybe 16" long by 1" dia
sorry taking so long, thomspson is giving me info overload atm =)
loads in a machine tool can be a lot higher than you'd think by turning the handle, several hundred pounds sometimes
cradek: I seem to recall that a rule of thumb on preload was 5-10% of max load, maybe a ballscrew mfgr tech page would have that?
true, this machine accelerates fast
my only goal is to get most of the slop out of it
cradek: Refernce info, page 88 to the end... http://www.thomsonbsa.com/pdf/ThomsonBSA-Lead-and-Ball-Screw-Catalog.pdf
cradek: I'm still looking for the ball specs though
south bend 12"x20" supposedly rated for 2400lbs z-axis thrust: http://www.southbendlathe.com/lathes/1220specs.htm
just as a data point
cradek: I *THINK* I remember hearing "grade 3"....
Allowable Nominal Roughness†
Deviation from Lot Diameter Ball Gage Ball Diameter Marking in Microinches
Grade Size Range Spherical Form Variation Variation Tolerance Increments “Ra”
3 .006-1/2" .000003 ±.000003 ±.00003 – 0.00001 .5
at a very minimum
[00:38:23] <JymmmEMC> http://www.danahermotion.com/website/com/eng/products/precision_balls/precision_balls.php
cradek: Ah, here we go... look at all the PDF's on this page... http://www.danahermotion.com/website/com/eng/products/ball_screws_and_lead_screws/ball_screws/Installation%20&%20Maintenance_121082.php
they even tell you how to size up the replacement balls too
fenn: where did you find the southbend information?
is that dynamic or static or did it say
cradek: NP, I went thru the nightmare of having all that info when I thought the issues with my machine were related to the ballscrews.
dareposte: seeing how it's under the "motor controls" section, i'd guess static
cradek: luckily enough thomsonbsa is in San Jose =)
i dont even know how one would measure dynamic forces in a rigid system
cradek: I just drove down there and took them my ballscrews to inspect =)
aww this is so cute: http://www.southbendlathe.com/lathes/612main.htm
fenn: 2400 lbs is a LOT of static force for a lathe of that size is why I was asking. Not impossible but very high
dareposte: it seemed to be very common based on my very quick google survey
fenn: I found the page you are looking at and indeed it does appear to be static. Very interesting. Also the spindle motor is 15 hp
[00:48:57] <JymmmEMC> http://precisionballs.com/EvaluatingCNCMachinesDirectory.htm
looks like most commercial cnc lathes are around 1500-3500lbs
fenn: coincidentally about what a 1" ball screw is rated for...
our 6x13 cnc lathes are 6700 lbs
I believe if my 9x20 harbor freight lathe could generate 2400 lbs of thrust it would tear the casting apart
toastydeath: rated z-axis thrust, not weight
[00:51:47] <JymmmEMC> http://precisionballs.com/Introduction_Ladder_Logic.html
also something to consider is that rolling element bearings in spindles or whatever have to be way oversized in machine tools so the machine doesn't overcome the bearing preload during operation
not that the machine is going to actually hit those load numbers
cradek: Uh OH.... you better look at the MINIMUM ORDER on this page.... http://precisionballs.com/ordering_balls.htm
(i am trying to clue in on what you are talking about and I think it's the z axis load of the spindle)
KimK: Thanks for the comment! If I get something going, I'll post it on the users list!
toastydeath: i think they just look at the torque rating on the motor and see what that calculates out as, based on the ballscrew pitch
oh, the force of the ballscrew?
well, maybe not torque rating, maybe something to do with following error.. who knows
my 9x20 lathe with 5/8" ballscrew and 425in-oz stall motors generates a theoretical "850 lbs" of thrust
if you figure the required thrust for stalling the 1 hp motor though its more down around 200 lbf
right, that's because they have to be way oversized or you'll eventually damage them
this started out as a discussion of which ball bearings to pick for repairing a worn out z-axis screw.. so i assumed that's what the manufacturers were describing with "z-axis thrust"
mshaver: I'll look forward to your post. Those Mesa "anything I/O" daughterboards look pretty handy too.
z-axis thrust is the peak torque of motor * 2 * pi * ballscrew efficiency / pitch
Heh, they are drilling ball bearings using a EDM machine
and tapping too
drilling ball bearings for fun and profit!
we can make beads, necklaces, anything you want!
[00:57:42] <JymmmEMC> http://precisionballs.com/threaded_ball_2.html
Never heard of THREADING using a EDM machine before
also someone asked about the forces in the system
they make toolpost dynamometers for that
stuart s. showed me some carbide ball bearings with threads 'cut' in them
done with edm of course
toastydeath: for measuring Z-axis thrust it might be easy to use a thrust washer between the Z-axis bearing and the screw shoulder
a strain gauge instrumented washer
might be, but i don't know of anyone who has done it that way
a hydraulic cylinder (force tester) should work too
we measure chain tension that way quite a bit for machine PMS
where would you put a hydraulic cylinder?
the base/cylinder would be up against the spindle/chuck and the Z (using a dummy boring bar?) would push on the measuring face. Feed *very* slowly, watching weight/force gage, note highest reading before Z trips out or something breaks.
Well, you wanted to measure the *maximum* force, LOL
yeah i see
i was thinking of trying to measure the actual force encountered when cutting something
like taking a 1mm cut at x mm/min
my machining book has a formula for estimating the cutting force but it was probably made up in the 80's before they had widespread high efficiency linear motion
Oh yeah, a strain gauge would be the way to go there. That would make for a handy front panel display while drilling, very useful.
"just" put strain gauges on your toolpost
maybe modify emc2 to have "constant force" as an option ;)
you could probably do something with adaptive feed
Spindle load is useful on the front panel too, but Z force would be much more directly informative in drilling situations.
fenn: is adaptive feed already programmed into emc?
and it can be adaptive to whatever you want
what does it currently use as .. oh you answered my question pre-emptively
(it's a hal input)
KimK_: I wish there was an equivalent for the "rest your arm on the drill press handle" way to drill
seems that drills have a natural feed rate
depends how they're ground
I'm sure that's true
and the material you're drilling.. i.e. brass or plastic
constant force machining would be the equivalent of "rest your arm on the drill press handle" in this case
haptics is interesting, and i think EMC is probably the only thing around to do it with
so all that's needed is an instrumented drill chuck :)
why do want constant force machining
dareposte: and on to that... "then fart" so that you body has muscle reflexes and alters the pressure applied to the drill press.
cradek: You're right, they do. You can look them up in Machinery's Handbook or similar. Usually around .002"/rev to .004"/rev for common sizes
JymmmEMC: all the benefits, none of the foul odors
dareposte: depending on what cuttign fluid your using, if any at all
constant force seems more useful in rough milling or rough turning
so you take a constant area chip without too much wacky programming
but that does't happen with constant force machining
well, close enough
lol, it doesn't
it's been tried
it works in ultra-accuracy ID grinding
but that's about it
the tools break because the machine hits hard spots or has other force fluctuations
and it tries to do something retarded
sounds like not constant force then
what about doing a moving average of the input
or some other kind of filtering
if you want to experiment around with it you've certainly got the programming skills to do so
sense -> model -> act
yes but i dont have the motivation
the general consensus is that it doesn't impart a large enough benifit for the shortcomings
but then you might come onto a situation where it works, so whatev
just wait... in 2015 it will be all the buzz on new tooling :)
i will call it... roboslave
because it's just like having a real machinist!
in cylindrical grinding it works because it makes the deflection constant
ID cylindrical grinding, rather
is that constant force in the "x" or the "z' directions
oh my that's fancy
how's that work? take the vector sum, or slave each axis to each force sensor?
i have no idea
i guess it would be the same
just seen it in several books on high accuracy grinding principles
i would think it would have to be the vector sum to keep deflection constant
but really the only deflection thats important would be X... unless you are talking super high precision then the Z would deflect in an arc motion around the tool holder too
makes my head hurt
why on earth would you need something that PRECISE!!??
optics and electronics
yeah but when i press precision bearings onto my sloppy turned shaft they never work that well anyway
so use a precision shaft? =)
that would require precision equipment :)
"All tolerances +/- 0.002" unless noted"
if the bore is long enough the rotational deflection would be greater
i bet a home dude could make a high accuracy lathe/grinder with a little effort
i'd like to be able to turn something to a half thou first
abrasives scare me
half thou over how long?
that should not be hard!
you're telling me!
have you seen my lathe?
i carried it out of my trunk into my apartment and assembled it on the kitchen counter
its a 20" travel though
level it and get a'cuttin
well now its about halfway CNC'd and its more like a 16" travel
oh i leveled it all right... the test bar had 0.0005" runout over 12"
so why no cutting .0005"/1
the accuracy problems had a lot more to do with the feed mechanism than the bed and spindle
so now that its CNC with ball screws i am hopeful to reach my goal
and probably had a lot to do with my inexperience too
"advance 0.001... but it cut .002' wtf!!
i hate machines in radius
well it had metric feeds but inch dials
and they were less than accurate it seemed
or repeatable for that matter
i saw the spec for the south bend 20" bed lathe and it weighs over a ton
so i'm wondering if that has something to do with it too...
with a .010" DOC, you should be able to hold .0005"/1"
on just about any lathe
that was another problem
don't back out the cross feed when you stop, let it drag as you back it up in Z
to the novice it would seem that if you needed to reduce the diameter by 0.003", then you should be able to dial in a doc of 0.0015" and it would do so
then bump it in .010" straightaway
dareposte: while on the subject of radius/diameter, what does EMC2 (in lathe mode) show on the screen XYZ or XDZ? Or something else?
dareposte: so you know you need to not take really light cuts?
yeah now i learned that
i got a text book and read it a few times
KimK_: hang on i will load it and see
"technology of machine tools", published 1981, $4 shipped from abebooks.com
what the heck is the D axis
mine reads xdz
using the "Axis" display?
it also shows a bunch of velocity data and whatnot
OK, cool. X & D both at once, no waiting, LOL
it shows X, D, Z, Velocity
in that order from top to bottom
might be useful in MDI mode
so what is X
Velocity? As in SFM? Last programmed F? What?
why would you have both X and D?
mm/min for mine
rpm * feed per rev?
if i tell it G0 z100 it seems to show the acceleration up to rapid speed and then stays the same at rapid speed, then slows down again
toastydeath: just for convenience. If you wanted D, there'd be only X, and vice-versa. So it shows both. Nice.
i have a hard time multiplying by two sometimes
i prefer to leave lathes in diameter and just leave it that way
they probably put it in there for me :)
all our drawings are done in diameter
oh now dividing by two is horrible
what do you have to divide by two?
to find the x position
diameter is 10.3, and i need it to be 5.8
no dude, you can set most controls to read X as diameter
so there is no radius confusion
See, there's an advantage to "X only", multiplying by two is easier than dividing by two, LOL
just never use radius
except for g2/g3 obv
but then you can just i/j if you rly want
way over my head now
my text was published in 1981 remember, it talks about punch cards for cnc
i think they still had g2/g3 back then?
hey cool how did you get those extra buttons over on the right?
i don't see any vcp option or tab or anything
cradek: Nice, thanks. And where is the velocity display that we were talking about?
KimK_: it shows under the others - guess it's turned off right now on mine
KimK_: you could run sim/lathe without any hardware
cradek: OK, that's fine, just so i know where to look for it
that is very fancy
almost like a SCADA hmi
cradek: Re: running sim/lathe: Well, I'm having trouble with EMC right now, install? upgrade? removal? re-install? troubles that I couldn't fix, not sure what to do next, might backup /home and reinstall, but somebody smarter than me might be able to fix it.
whats going on with it?
did issy's gui make it anywhere? has it been added to cvs?
from the precisionball website: "An 800-pound sphere can present some serious shipping problems."
If I start sim/axis, "EMC2 Errors", EMC2 Terminated with an error... lots of info in a window follows.
oh i had that problem too i think
well at least that symptom
did you compile it or is it the preinstalled
It's preinstalled, but I have messed with it using Synaptic because of something (some accessory or improvement?) I was trying to install, but it didn't work out and I uninstalled, but it's not been the same since.
jmkasunich: Yeah, can't keep it from rolling off the truck =)
making the error messages available to those who might help you is a good first step. http://pastebin.ca
does it load the non-axis guis?
jmkasunich: But I do like that they didn't say it was impossible, nor that they couldn't make one =)
cradek: OK, thanks. Standby
just need to make a special crate
cradek: STAND CRAADEK STAND! Good dog
i found that after upgrading the gtk and or opengl modules it wouldn't load any graphical displays
it still worked with the curses based ones though
crashed with the same error
jmkasunich: Nah, I still like the idea of a no-contianer 800 ball and trying to see them put in on a truck and try to chain it down =)
800 lbs probably wouldn't actually be all that big
even with a crate - I wouldn't want to be the guy who needs to uncrate it
i mean yeah for a bearing its big but as far as a mass of steel it would probably be under a couple feet diameter
jmkasunich: LOL, no doubt... how does one actually STOP an 800lb ball
and gawd help ya if you're on a hill
estimated diameter of 800 lbs steel ball = 17"
oh, not as bad as I thought it would be
jymmmemc: slowly =)
imagine it rolling over your foot or worse... hand! OUCH
even a 8" ball would be tough to deal with
thats only 100 lbs, but hard to grab hold of
i figured at .238 lbs/cubic inch it would be around 18" diameter
i'm wondering how they even make a 17" steel ball
I want one, but I don't know why
i wonder if they could thread it for you
Here you go. http://pastebin.ca/1202248
I appreciate it if you can help. If not, I can reinstall after a backup.
cradek: an 800 lbs one?
Oh man... could you imagine if a 17" ball was to be used in a ballscrew ?!?!?!?!
could you imagine if a 17" steel ball was set loose in a shopping mall?!?!?!
Yeah sucker.... you get to preload the ballscrew this week!
if you think moving a 17" steel ball is hard imagine moving the ballscrew it would have to be used with
Now that you mention it, what's the diameter of a wrecking ball?
dareposte: those giant tonka trucks
heh, mcmaster has 4-1/2" ones
KimK_: google sez 3 feet
fenn: a 3' diam ballscrew?
fenn: OK, thanks. And we won't even subtract points for lack of roundness.
must be hollow?
4 1/2" ground ball
google also says 1500 lbs so that would mean they're hollow
i think my boss has some of those but they're brass
fenn: Really? Wrecking balls are hollow? Whaddya know.
probably something to do with shrinkage during casting
cast wrecking balls?
filled with water onsite maybe?
now you gonna make me look up how they make wrecking balls
or maybe they just don't need a solid ball to do the job
KimK_: that is definitely not the error i had
I just figured they were solid due to the abuse, never thought about making them lighter for transport.
seems like you'd want it to be as dense as possible
if it was too dense it might pull the crane over
no, if it were too heavy
Yeah, but these same cranes lift huge buckets of wet concrete, etc.
i bet they don't go swinging them around like that though
i propose that lifting concrete and swinging em are different
OK. I'm still surprised they're hollow.
a 3' diameter steel ball would weigh 25,000 lbs
you forgot to divide by 2 for the radius
no i forgot to multiply by the density of steel
or divided in wrong place or somethin
units '8g/cc*4/3*pi*(3ft/2)^3' lb = 7060.436
yeah thats what i meant
maybe a reason for the hollow is that they are cast - the outside cools first. once it hardens, as the inside cools and contracts, you tend to get voids in the middle
i did .2830 lb/cubic inch
but got 7000 lbs
are they really cast??
jmkasunich: actually wikipedia says they're forged
jesus christ, how do you forge a hollow ball
i'd think a cast hollow wrecking ball would be like beating someone with a christmas ornament
forged sounds more likely
"hollow" could mean 30" OD, 6" ID ;-)
true but still
still cast iron would be brittle even if it was solid
how do you forge a 30"x6" hollow ball
wrecking balls are steel
how do you forge a hollow, period?
that's crazy and i want to know how.
i guess you beat it around a mandrel
make the hollow cylindrical shape
they forge rifle barrels around a mandrel so why not wrecking balls
i dont think they're hollow
i think this is a case of bad data
dareposte: because rifle barrels are open on both ends
dareposte: Now that I look at the error list, I think it had something to do at the time with removing RTAI and installing (what, Xenomai? something else?) Anyway, it didn't work out and I went back. Only it didn't.
but the mandrel only comes out one end
sometimes they forge chamber and all in them
or battleship guns are made that way i believe
'They must have confused “three feet in diameter” with “three cubic feet.”'
over 9000 feet
3ft^3*8g/cc = 1498lb, problem solved
here's what happens when one gets loose: http://www.youtube.com/watch?v=PU8kT8Yv7VQ
also note 1500lb is not "one and a half ton"
maybe they come in different sizes?
naw that was from the same story (the video link)
can you order a "medium sized" wrecking ball if you only need to wreck smaller structures?
yeah journalistic prowess at it's best
yes, apparently up to 12000 lbs
is that the same as 12 tons
don't be silly
1 ton = 2000 lb
1 tonne = 1000kg
thats what i always thought too :)
Maybe this line from the error list ( http://pastebin.ca/1202248
) is pertinent: insmod: can't read '/usr/realtime-2.6.24-19-generic/modules/rtai_hal.ko': No such file or directory
KimK_: i think you should reinstall emc 'apt-get install --reinstall emc2'
KimK_: you are not running the realtime kernel. its name is something-rtai
fenn is incorrect
i am most likely incorrect
you need to reboot and pick the -rtai kernel at the boot menu
a short ton=2000lbs
I haven't checked, will it be present on the list? (based on the 'no such file' earlier)
depends how you installed emc2. you apparently did not start with the linuxcnc cd.
but if you installed the emc2 deb package, it DID also install the rtai kernel.
then all you have to do is boot it.
I don't recall what I did on this machine, but i'd guess that I did 8.04 and then one of the script tricks from the wiki?
And how do I pick a different kernel at boot time in Ubuntu 8.04? (Get to the grub menu, is that what I want?)
you can make rtai the default if you want to always use emc.
I think RTAI as default would be fine. Can I examine the grub menu in... what, /etc/grub?
you will need to change 'default'
there are some nice instructions in that file
It lists Ubuntu 8.04.1, kernel 2.6.24-16-rtai and recovery mode, plus 8-10 others (non-RTAI)
but not the -19 version in the error message. Hmm...
Can Synaptic get the -19 RTAI kernel for me?
you already have a -rtai kernel, right?
Yes, but the startup errors said couldn't find -19
it couldn't find the rtai modules for -19-non-rtai (because there are none)
boot the -rtai kernel and you'll be fine
Synaptic only lists the -16
* KimK_ is sorry to have buzz-killed a lively discussion about wrecking balls
cradek: I'm not seeing the instructions to edit 'default
google for "ubuntu change boot kernel" or similar ;)
hint: you edit /boot/grub/menu.lst
the other day our old OKK horizontal machining center
decided to try and turn the spindle on
while it was toolchanging
the locking pawls were engaged
it was a very eerie noise
and reset wouldn't kill it
a gearshift in mdi fixed it
but it was this weird low whine
cradek: I am 'finally' almost done with the post processor for the cinci. I started Thursday morning. I thought it would take about 2 hours max. 'This shouldn't be difficult' whew
stustev: are you using apt?
is there anything else?
not here : )
besides APT! apt rules
I have PostWorks from NCCS. It is a generalized 10 axis post processor/generator.
stustev: is it that different from all your other machines?
it runs EMC2 ;)
I meant the gcode
cradek: no - I had to reeducate myself on how to build a post with PostWorks. I had a problem with two things. Turned out to be the same problem in two areas. I didn't recognize it until just a few minutes ago.
the gcode is the same - I couldn't get it to put out a T (of all things) or a W
I had disabled the T because the cinci didn't use it.
I had never enable a W - nothing else uses it
I'd expect that to be the most commonly used of UVW, since it makes perfect sense for the knee of a knee mill
the enable for both are on the same page of the post generator
SWPadnos: I don't have a cnc knee mill with a controlled knee
me either ;)
It only took me two days to see both problems (the same problem twice)
that's good. it didn't take 4 days that way
just a day for each
If I was a golfer my front nine would be a 72
wow, velocity mode amps work. I can program f.0025 (inches) and see perfectly smooth motion on my tenths indicator (about a tenth every 3 seconds)
doing anything smooth with a tenth indicator is impressive
it really looks good
X or Z or both?
how big is an encoder count?
this is on Z
jmkasunich: 1/204800 inch
so a tenth is 20 counts
Z is rolling smooth then - you reassembled the ball nut well
and no stick-slip on the ways
what does that lathe have for ways? balls or sliding?
doubt I could see it if it was stepping 20 times per tenth
just sliding with tapered gibs
isn't it a sliding dovetail
hey, have any of you ever dealt with an Accu-rite MillPWR control before
it doesn't appear to have rollerpacks
for like, bridgeports
I should make a video to send to people who think step-servos are a good idea
i have one that's way, way overcorrecting for backlash - and there's no backlash comp option
send the video to the Mach3 forum
you're right Stuart
we talked to the new MACH developer at IMTS on Monday - he was friendly and nice
it's all the anti linux users that always get to me
not sure what they're afraid of
I get quiet when I'm afraid - some people get loud
heck I try for anti windowz
no one really understands religion/politics/operating system preference
how was IMTS?
this was the least fun/exciting one for me - EMC2 has really affected my perspective
there was nothing there that EMC cannot do
I was always seeing things I would like to do/have someday - I have been to all IMTS shows but one since 1990
with EMC I have all the capability and more
so do you
yes and CAM too with apt360
EMC and APT will rule the WORLD!
or at least my shop : )
stustev: do they make large machines like the cinci anymore? Did you see anything like it at IMTS?
they had some really, really BIG machines there
my largest machine is X 200", Y 100" and Z 40" - there was one bridge mill there that my machine would sit on its table
there was a vertical lathe with 20' (I think) swing
it was running at its lowest speed (1rpm)
which looked pretty fast at the edge of the table (about 60sfm)
stustev: I purchased a book titled "Numerical Control-Making a New Technology" by Reintjes- one of the guys who worked on APT. Interesting stuff for a machine nerd like me
how old is it?
got it off amazon for $3
does it talk about the development in the 40's and 50's or about today?
40's through 70's
all the gory details about the air force contracts
I just started reading it
look for it on amazon, there are more copies
I will look
I'm looking for evidence of public financing
just in case...
public financing of what
it seems that defense contractors don't count as "public financing"
which is too bad
the $3million dollars in development of APT coding
the air force put money into it too
didn't they release it?
at least in cursory skimming of the book
I haven't found anything official yet about releasing it
you could at one time purchase the source for $1200
on a reel to reel tape from ICAM in Texas
I thought that was the 360/370 APT code released in the late 70'
s or early 80'l
I don't know
I never got to look at the stuff from Icam
the guy wouldn't told me at the time that he didn't know if the tape was good or not
couldn't tell me...
anyway, get the book and have fun
SWPadnos: I thought you had succeeded this afternoon - I thought I didn't have internet access - :)
I didn't do it, I swear :)
It was just a little slow
I must get my beauty sleep - good night gents
good night Stuart
me too I guess :)
SWPadnos: PLEASE take extra =)
quick Q: 12V .39A stepper motor, and an L293, powered by 12V, do I, or do I not, need resistors on this thing, as the L293B that I have does not have the current limiting features
if you power the motors at 12V then my guess would be that you don't need any
you could measure the winding resistance to be sure though
well this makes sense,.. microchip uses red wires for GND, and black for Vdd
dareposte: will do
actually, the box says 75ohm, so,.. lol
you're golden then
if you can you might run them at 24v to get a better step speed
then i'd need a 24V PSU which i dont have atm ;)
besides, for a first time attempt, i'd say speed doesn't matter :p
if it works i'll be happy
yeah if you get it moving upgrading the psu wouldn't be hard
the 293B will handle up to 36v if you ever want to push it a bit
dareposte: yeah, but if i run it at a higher voltage, i'll just use the L298N, I just got these parts wrong, but seems to work out,.. I'm trading them in tuesday
they're not bad drivers for a 75 ohm stepper
this is just a stepper i have laying about
i'm trying to get it to WORK before i scale up and make the PCB's lol, currently doing some low-freq work on a breadboard
if you plan to use like a 3A stepper then yeah you'll definitely need a bigger driver
atleast one that has a heatsink, lol
this is a DIP-16 package, .. :p
i used the L293 as a reversible dc motor control once and it worked really well
I know that it's basically the same, just no current sensing, or multiwatt15 package
i think the 298 only gets you 2 amps though
it's not like i'll be making a bridgeport or anything :p
293 is only one amp
but you can piggyback 2 of them :)
did you get 293D ?
alex_joni: B :(
ok.. then you need protection diodes
otherwise it might blow up
ANYWAYS, done hooking the ICD2 up, time to hook the L293B up to the PIC.. (don't have a compatible computer to hook it to a serial port :()
so i'm doing USB->homemade ICD2->PIC->L293B->stepper lol
alex_joni: Hey, does the livecd have sim?
fragalot: I use those cheap usb->rs232 serial converters on my laptop and they work pretty well
it doesn't have any ports except for a usb port and video out
|dareposte|: I have a "not so cheap" Velleman one, i thought that that didn't work?
alex_joni: sim as in simulation, as One is too cheap/poor to have a REAL TIME system
people told me it didn't work but i never had any problem with it
dareposte: awesome o.0
JymmmEMC: nope, there's only one kernel on the live cd
and that one is a RT kernel
not much point in running sim on a RT kernel though..
here's the one I use: http://www.emtcompany.com/products/adapters/dxubdb9-usb-to-serial-db9-adapter-cable.htm?gclid=CNKigry125UCFQIfswodAicsDw
alex_joni: There are many that just want sim for testing purposes
yup, and it's _very_ easy to install
just apt-get install emc2
after adding the emc2-sim repo to sources.list
alex_joni: and that does sim AND RT or just sim?
if you hook something up, flick the switch, and it smells like somebody is soldering.. You're doing it wrong.
the infamous "magic smoke"
|dareposte|: no, just a very warm heatsink
on the 7805
I don't think there is an install script though
fragalot: ah yes the 7805. they are amazingly resilient though, i've thought for sure i'd smoked one and just flip it around and hook it back up and it still worked great
doing a LiveCD with sim should also be fairly easy..
hmm,.... I.. DID hook it up right.. can't see why it's heating up so quickly
so if you find someone eager to do it, I can nudge him in the right direction
alex_joni: theres enough room left?
emc2 + stuff is < 20MB
so it should fit on the standard one
|dareposte|: I just wish that they'd have an internal diode, 'cos if the 5V on the other end is somehow still powered, it feeds back to the 12V line
alex_joni: Well, if you are bored, I know some ppl would appreciate it.
can't say that I am atm
alex_joni: Hey, how much UTF-8 do you deal with on a regualr basis (non EMC related)
|dareposte|: there we go, it was powering the ICD2 aswell as the PIC aswell as the logic for the 293 :p
JymmmEMC: not much
alex_joni: you onlt read/write in engwish?
no, romanian too
and sometimes german
not UTF-8 though
alex_joni: ah, ok
yeah, sounds vaguely familiar
alex_joni: Damn, well your no help =)
what do you need?
alex_joni: might need some testing done
alex_joni: and I know you know all those funky characters too =)
well, not if it's kanji
alex_joni: nobody knows all 30,000+ Kanji characters =)
alex_joni: I might say screw it and just restrict to lower ascii chars (<127) do you see any drawbacks in that?
I can read my language just fine with US characters only
and most over here do
alex_joni: Since you know .DE, let me ask you.... does the lack of umlots (sp) change the definition ?
alex_joni: Ä -vs- A
usually I type AE instead of :A
btw, my IRC client doesn't do UTF-8, or my ssh client or something
alex_joni: Sorry, I dont know .DE, so I dont know if it changes the meaning
it does change things
alex_joni: how so?
you don't get valid words anymore
[14:03:27] <JymmmEMC> http://en.wikipedia.org/wiki/Umlaut_(diacritic)
alex_joni: could you give me an example?
use capital AE to distinguish from A
Angst vs. Aehnlich
one is with A one with Ae
Ahnlich is not a german word
I can't think of different meanings for A vs. Ae
but schon vs schoen are different
schon = already
schoen = pretty
ok, so it CAN change the definition
so what would schoen be written at correctly?
oe is the o with .. ontop
(not sure that works..)
What's with the xE ?
that's how you pronounce it
is that how you type it too?
make your mouth in the shape of O (like you would want to say o) but say E instead
sometimes it's used
heh, I sounds like a dying tape recorder =)
schoen - pronounced like that doesn't sound very pretty
[14:11:20] <alex_joni> http://en.wikipedia.org/wiki/Ä
sch is pronounced sh
good morning all. hey you guys ever seen this one: http://www.syilamerica.com/product_X4.asp
alex_joni: Ok, so if one saw "Ahnlich" would it be understood ?
alex_joni: Like if a word has different meaning if it has a different Diacritic?
no, Ahnlich is not really understood
or very hard
alex_joni: is it just .. or are there others?
- above .. above the char too?
I don't understand
alex_joni: right column, look at 'U' http://en.wikipedia.org/wiki/Umlaut_(diacritic)
the one with - is not part of german
neither any of the next
right, but I'm asking you is other other ones in the germna language besides ..
in german you have the letters from the US alphabet + ae, oe and ue
only 3 more
ah ok, and in romainin?
you have s and t with a , underneath
alex_joni: and the lack of the . undrerneith changes the definition?
[14:18:32] <alex_joni> http://en.wikipedia.org/wiki/S#Similar_letters_and_symbols
it can, but it's readable
there are also a, i with ^ above in romanian
and a with ( above
[14:19:46] <alex_joni> http://en.wikipedia.org/wiki/A-breve
alex_joni: See, I didn't know if it was just for properness like "Don't" but now I realize it's an added character to the languages alphabet
german has only 3, romanian has 5
other languages have lots more
Right, like Spanish
there was another extra letter in the german languages alphabet
but that one got dropped and replaced with a double s
[14:22:04] <alex_joni> http://en.wikipedia.org/wiki/Eszett
alex_joni: ah that was the other thing I considered it to be... masculine and feminie
and I thought the lack of them might mean gender neutral
I'm sure it's true for some cases
alex_joni: Yeah, and many asian dialitics, instead of gender, it's level of respect
respect for a boss, respect for an offical, respect for a teacher, etc.
strong booger, this is.
HAH, even halfstep works
[15:42:18] <fragalot> http://omploader.org/vcXYy/P1000950.JPG
most, simplistic driver,.. ever.
ve7it is now known as LawrenceG
fragalot: sounds good!
looks good too
very neat breadboarding
i've never had a breadboard look so neat
makes fault detection easier
where's all the ragged scraggly wires twisted together?
in a box i stuffed to the side, lol
rule #1 of taking pictures of your work to brag with: clean up first.
with clean up, I mean lay your arm down and shuv everything aside
i'm usually too lazy and or excited to do that
snap a picture, then go edit out anything inappropriate later
while you're at it you may as well program the PIC to accept serial commands with acceleration and speed
dareposte: nah, that's what EMC is for :p
I just used that PIC to generate the stepper signal xD
i've wondered if emc2 could be compiled for like a TI chip or one of the DSPs with good gcc support
which reminds me. I forgot to put the diodes in.. oops
dareposte: hmm, i have a few microchip DSP's
I dont think gcc supports microchip all that well
sadly one of the main reasons i use atmel's chips instead for most things that aren't trivial
i'm using the knudsen compilers atm to be fair
i've been reading through the source tree and it looks like it would take a lot of rewriting in the HAL, or just new code
mainly because i haven't attempted to run my ICD2 (as seen in the picture) in linux yet
i started out on microchips but moved on to TI's then now atmel, simply for the easy support with gcc and more open standards. I loved the microchip simplicity though it was veyr nice
and they have apparently come out with many faster and better chips since i quit using, i was using a 16F628 which was pretty good at the time
neither one would be enough to run some of this emc code though in my opinion
any one used HAL to control a VFD???
well, I still have a lot of PIC's laying about, just because we get classes on µC's and use a PIC in school... I just sampled boxes and boxes and boxes from microchip :p
maybe an ARM7 or arm9 might do it
dareposte: microchip has a few 32bit µC's, would those work?
Dmess: I have not successfully done it yet, but i am in the works of trying to implement a 0-10v signal for my hitachi vfd basic speed control
Dmess, lots of people use HAL to control a VFD, generally using a 0-5V (or other) PWM
i need 24 V....
yea.. A-B unit
24V should be to turn it on or off, not as the speed reference
as best ive read so far
what vfd is it?
Allen Bradley (A-B)
free for me
i went for it
I just realized I have the L297 and wonder why i spent time programming this PIC.
it says -10v to +10v analog input, or 4-20ma
oh that's the 160-C
wonder what the difference is?
no, not really
[16:14:14] <dareposte> http://www.ab.com/support/abdrives/160/datasheet.html
SWP.. i guess i want analog signal follower mode right???
is that good Dare??
does your drive manual say it supports "analog follower"
if it does then yes you definitely want to use that for controlling it
yes it does
you should be able to use a dual rail op-amp to generate -10 to +10v, or you could try out the 4-20ma signal, it says input impedance is 250 ohms so a regular bipolar transistor should be able to drive it
and where do i feed input?? thru the +/- 10v pot
yeah, the kitchen & garden transistor should work :p
bleh, that sucks when you translate it into english
s/kitchen/house/ but meh
but i wanted to use
kitchen and garden transistor? heh... is that the one in your toaster and also the weedwacker
HAL to drive it..
bui i wanted to
SWPadnos told me yesterday that you can use PDM outputs into an RC filter with an op-amp follower to drive it
which is what i'm working on right now
Dmess: AB apparently doesn't have the manuals online, so I can't get any wiring diagrams or anything :(
let me know how it turns out
i hav a 160 SSC manual.. was the closest i found yesterday
it just tells me that no results match my query
i'll get it from work we have the same drive in our HVOF booth
for any 160 series documentation
it was on a rockwell automation site
rockwell automation owns allen bradley
the hitachi drive i'm using had a +10v out supply, and an analog reference input right there next to each other
program a few parameters and it translates 0-10v into low speed to max speed. Mitsubish is the same way
i may have a + 10 v right there too
made to take a potentiometer
yeah that should work then
instead of a pot just hook up your op-amp in the way and it will never know the difference
and feed it back into itself
i'll need to reaserch the op-amp thing... im a machinist more than an electronict gur
yeah it gets a bit tricky if you're not familiar
or even if you are familiar... ;)
I thought I remember jepler making up a simple op-amp circuit to take pwm/pdm and outputting a voltage
the thing you should be wary of is that the signal needs to be referenced to the - terminal of the external speed pot inputs. that may present an isolation problem
dareposte: a "household transistor" would be what i meant to say ;) eg. common one,.... one of my teachers jokes, but it doesn't work out in real life or on IRC I guess
Dmess: if you don't want to mess with all the electronics here's a pre-fab one that looks plug-n-play
[16:34:06] <dareposte> http://www.kdntool.com/_sgg/m4m3_1.htm
would that be the pot wiper input at +/- 10V
i don't know what all he's got crammed into that board but it looks pretty fancy
output impedance he says is 500 ohms, so it should work great with the AB input
[16:36:19] <SWPadnos> http://homanndesigns.com/store/index.php?main_page=product_info&cPath=1&products_id=21
my controller input impedenc says 250 ohms
SWPadnos found one even cheaper!
mmph, my stepper just randomly stopped stepping.
input impedances are supposed to be high, which input does it say is 250 ohms?
switching it off for 1 minute, then back on appears to somehow fix it :/
dareposte: 's not even warm
is your PIC still running or did it get stuck
did you set the wdt?
scope tells me it's still running
oh, the watchdog, darnit
but that would halt the signal too
fragalot: i didn't see any hf bypass caps on the power rails in your picture, you might try throwing one on there
dareposte: already done that, aswell as diodes, but i just found a wire that snapped off
that causes problems sometimes
only in extreme cases
i usually find that one of the 6 scraggly wires with frayed insulation that i twisted together has shorted to something it shouldn't have
my breadboard is big enough not to have to twist wires together
hmm, for microstepping, does that only require another output sequence like halfstep, or something else too? 'cos i can't find the truth table for microstepping anywehre
microstepping requires some fancy pwm coding
half stepping is easy, microstepping not so easy
PWM you say.. meh, not worth the hassle ;)
it might be useful for 1/4 stepping but beyond that its a bit of a waste for cnc stuff
for the accuracy i'll be getting once i'm done (long term project >.<) even full-step will be too accurate :p
i think the way it's done is you vary the duty cycle of the PWM to each phase to attempt to hold the rotor in a position somewhere between half step and whole step
getting any sort of accuracy is questionable, and on top of that under load it will alter the shaft position
so if you want to accurately microstep you have to calibrate it under the intended load and use a lookup table
then if the load changes it won't be accurate any more
yeah, that's fairly useless for CNC work, lol
thats why most people recommend using only full steps to get your system accuracy, then microstepping to get better smoothness
how about halfstep
although 1/4 step is guaranteed to be somewhere between full step and half step, so it arguably does make it soemwhat more accurate
just to smoothen the ride, and be able to get a higher speed without "clipping" a step
half step shouldn't be used for accuracy either really, but people do
its sort of like reading between the graduations on your dial indicator
if i get accurate to within 0.25mm i'm happy, lol
if you need more accuracy they make 400 steps/rev steppers you can use instead of the usual 200 steps/rev
Yes they do, but this 200 one is more than good enough
i geared my x-axis on my lathe down 2:1, with a .200" ballscrew and it gives 0.0005" per step
well i just disabled the watchdog timer and it seems to be working just fine now
in theory half stepping would be .000025" but under load that may or may not be the case
you could add an encoder and then feed it back to the PWM duty algorithm and get it to be pretty darn accurate, but then you almost have a bldc motor and why not just go all out
dareposte: why not put laser positioning in while you're at it
ive seen many older stepper systems with tacho's as feedback to the cnc
no idear what them be
basicly a speedometer
i dunno what the old siemens 3m did with it but it seemed quite reliable..
very interesting, i wonder what they would do with such a device on a stepper
they were on the back of all 3 motors.. and neede to be cleaned regularily
they make sure the velocity requested is the velocity you got ( closed speed loop)
we had 3 machine
yeah i guess
as opposed to closed position loop with scales???
yes, lotsa kinds of loops to be closed :)
position force speed...
this was circa 1989-90
wouldn't it be better to close the position loop so you can integrate the speed and accel loops accurately
closing the speed loop only lets you get two of the three
all classic servo amps have it (speed loop) built in
position not need for lots of stuff, cnc is a level up
so true it becomes proxies and limit swtches
anybody ever tried using a cheap encoder like one of these for a MPG? http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=P12334-ND
quadrature output for $0.75
hey that pight work for my spindle encoder...
not tried, but its endless, so possible, on a spindle beware of max rate (its a tiny plastic bit ya know )
only rated for 50k rotations
heh, and life
oh actually 15,000 rotations
so that would take about 10 minutes to fry :)
or less om my machine
but it might be useful for a manual pulse generator
on the cheap
yup... but would you need an axis drive to feed it into??
i was thinking just feed it into EMC and let it figure out how to move the axes
hmm have we come SO far without me knowing??
don't ask me, i still haven't gotten my machine working
but it doesn't seem like it should be too hard...
my extent of using hal thus far has been stepconf
if HAL can take quadrature input from your spindle to figure out the speed, why can't it take quadrature input from the mpg
brb smoke break
fragalot: did you smoke anything yet?
for mpg BigJohnT's set one up . look at the wiki to see how an encoder can be hooked to hal and used to jog. Also last few days discussed a button jogger. check logs http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/
dareposte: two problems with those inexpensive encoders that both go back to the fact that they're mechanical rather than optical. First, contact noise at open/close gives a very low limit for max rotational speed. second, they are very coarse with as few as 12 detents per revolution, compared to 100 for a real mpg.
but it will work like an mpg using a very simple circuit with a few pull-up resistors and emc's software encoder counter if you can live with those limitations
dareposte: well, maybe i wrecked PORTB.0 and .1
steves_logging is now known as steve_stallings
That is the encoder that was used on the PMDX-112 parallel port training fixture that we supplied for Ray Henry to use at the CNC Workshop. OK as a light duty MPG, but far short of a spindle encoder.
thx for the good advice Steve
i just dropped the motor onto my icd2
were YOU at IMTS anywhere??
If you use it, be sure to see their recommended RC filter circuit for the outputs.
jepler: do they use like a wiper system or something?
Steve has not yet made it to a IMTS show. I make it to the hobby shows, been to the Plastics Mfg show a couple of time, used to get to the big mainframe computer shows.
WHAT a show... 1 walked 60 miles in 3 days
83 miles from the time i landed to when i left Chicago on my pedometer
5 buildings @ the Mcormich center.... my home town isnt that big
Yea, Plastic Mfg show is held there. Incredible equipment set up and operating, some of it 3 stories tall inside the building. They were running a plastic film extruder that made so much product that it kept two forklift busy getting it out of the building.
LOL... i have NEVER been so glad to land @ YYZ-Toronto.. to see a tree again
I would love see all the CNC machinery. Do they let you get close enough to touch anything?
we were at the hilton gargen inn right d/t
i touched anything i wanted to.. and made them try stuff i knew they couldnt do..LOL
Yea, that is good. I hate scripted demos that hide all the products flaws.
they kicked me out of the Toshiba booth bcz i made all the bells go off on their / my old VTL
i knew it couldnt go thu 0... the didnt i guess... ; )
its my old salesman and coworker who asked me to eithe leave or help him sort out the mess you made... ; ) i left stage right...
they probably paid big time money to get that booth up and running there :)
running big machinery in public is nerve racking. too many loose fingers and different opinions of what is 'funny'. I've run lots of IMTS booths. humor and machines just dont mix.
jepler: thanks for the explanation on the cheap encoders. I was wondering why they aren't more wide spread
tomp: I think the humor here is that Dmess had more intimate knowledge of the machine than the "experts" running the booth did, at least of some of the machine's problems
I would imagine that happens a lot, when you have someone running a machine daily for years they learn the quirks and undocumented "features"
i had the willies the night before i left.. and had to shake myself and remind myself i wasnt working IT
it is very easy to know more than the people running the show in the booth - they are not operators - they are maintenance men and salesmen
no i ALWAYS screw the APPLICATIONS guy... for i was 1
the salesman is too easy
except James Bonnell
he's quite sharp
but then agin hes sent me on hail mary installs from hell so we can joke and screw around pretty well
steve_stallings is now known as steves_logging
it was way too much to try and do in 3 days...
nite #2 i was lying FLAT at 7:00 PM
elsewise there would have been NO day 3
I have been advised that I need the latest verion of "encoder" to run a single pulse spindle sensor. I have installed the 2.3 pre trunk version but I am getting errors that the encoder.0.interpolated-position does not exist. How do I check which version of the encoder module I am running and how do I set it to the newer version?
scrub that question. sorry. I got it the wrong around it is position-interpolated.
I wrote interpolated-position in my email, I had it backwards
No problem.. I just did a cat on the .ko file and saw the text in the output.
there is an easier way
there normally is :o)
"halcmd show" will show you all the pins of all loaded modules
"halcmd show pin encoder.0" will show you the pins associated with encoder.0
where do I run that, if I run it in a terminal window I get RTAPI version mismatch
I am also getting an "unexpected realtime delay"
if you are running a compiled version of EMC, "run-in-place", you need to run the halcmd that matches it
by default, halcmd runs the installed one
to solve that, open a shell, cd to the directory where your run-in-place build is, and do: . scripts/emc-environment
that sets environment variables so that halcmd (and other programs) will be run from that tree, instead of from installed
I thought I had already done that, I'll try it again. Does it need to be re-run every boot?
the env vars apply only to that shell - if you open another shell for something else, do the same thing there
emc-environment sets vars for the current shell only
yikes, I'll never remember all this! Time to go and plug it into the mahine and see if it works!
Thanks for your help
mind the DOT >.< as in . scripts/emc-environment
"do it here"
I owould like to slow down g0 speed
but I do not remember where to put the proper value
OK, it is working on a single pulse from the spindle now... The reading in not in rpm though, I think it is in revs per second, the figures look about right.
Also I am now getting the "unexpected realtime delay". I had to load the rtapi-smi.ko in the release version, do I need to do something different to make the devel version work?
Gromit: when you say "reading", what do you mean? do you have a pyvcp widget to display spindle speed?
regarding realtime - if your motherboard needs -smi for released, it will need it for dev as well
Yes, I have a pyvcp widget bar showing the speed. I am feeding it "spindle-speed"
remember that spindle-speed is in revs/second - that is what encoder makes and that is what motion wants (for feed/rev)
so you need to scale it by 60 for RPM
you need a scale block in hal
ah ok, I already have one of those for the step control of the spindle, I'll add another and use that.
this is from my lathe config:
net spindle-spd-rps encoder.0.velocity motion.spindle-speed-in scale.0.in
net spindle-spd-rpm scale.0.out lowpass.0.in
setp scale.0.gain 60
net spindle-spd-rpm-filt lowpass.0.out
setp lowpass.0.gain 0.01
general Q: with the "net" command is the first word an alias?
I added a lowpass block between the scale output and the pyvcp widget input, so the value doesn't bounce around so much
net foo bar baz creates a signal called foo, and connects (pre-existing) pins bar and baz to the signal
if the signal already exists, it simply makes the connections
cool, that is what I had assumed.
you always need a signal name, and at least one pin, you can have as many pins as you want
"man halcmd" describes all these things
all the lines in a hal file are nothing more than commands to halcmd
ahhh, I see, thanks. You could write a "EMC for dummies" book. Those little snippets are the bits I needed to connect the dots! Thanks
So with the SMI. Am I correct in assuming that I need to copy the rtapi-smi.ko to somewhere within the emc2-trunk directory and is there a different rtapi.conf file to edit?
[20:34:59] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?FixingSMIIssues
I don't think you need to copy the .ko file
I think it gives directions for both rip and installed
you do need to add it to the rtapi.conf for your run-in-place version
That is the page I followed for the first install. I didn't understand the "run in place" bit at the time and it was working so I ignored it. Thanks
I will do.
Time to go back to the garage and reconfigure it all... Thanks
hey is there any easy (safe) way to make a degauising coil?
one time use sorta thing
I need to blank out about 100 tapes
Does this sound legit? http://www.staffordnet.net/degausscoil.htm
I'm sure it will work find for degausing CRTs as designed
[21:07:01] <JymmmEMC> http://www.eham.net/forums/BoatAnchors/3822
but for erasing tapes I bet its far too weak
Yeah, the one RS used to have is no more
only the manual is available =(
google "bulk eraser"
re: spindle rps2rpm ... scale.0.out = spindle-spd-rps * scale.0.gain = 2*60 = 120, ok, but what is lowpass.0.in * lowpass.0.gain for (120 * 0.01 = 1.2)?
re: bulk erase, find old garage, ask for the 'growler' ( a demagnetizer )
lowpass gain is constant at 1.0
the "gain" factor is a bit mis-named, it actually determines the filter time constant
"setp lowpass.0.gain 0.01"
oh, its time, ok
0.01 means that the output is 0.99 * the old output + 0.01 times the current input
"a bit" :)
so low "gain" values make the response slower
oh a weighted travelling average, how stock market-y
i make on screen meters jitter less with that
Hello, I'm having a problem installing EMC. I just upgraded my working box from 6.06 to Ubuntu 8.04 and emc is gone :-( The script to install Emc2 tells me that my python 2.5.2 is too advanced. What should I do?
Hugomatic: make sure you get the right emc2 install script
[21:56:09] <alex_joni> http://www.linuxcnc.org/hardy/emc2-install.sh
run that in a terminal, and post any errors that come up
if it's more than a couple of lines use pastebin.ca
alex_joni: same problem... http://pastebin.ca/1202842.
Maybe its a Ubuntu 8.04.1 issue.
Hugomatic: you have the wrong script
or had it initially
you have http://www.linuxcnc.org
and you need hardy
try this: sudo gedit /etc/apt/sources.list
then look for the last couple of lines, there should be :
make sure they read:
hardy base emc2.2
hardy base emc2.2
alex_joni: My new error is: FAILED to fetch http://www.linuxcnc.org/emc2/hardy/dists/hardy/base/binary-i386/Packages.gz
alex_joni: I didn't fill the repositories fields correctly
you still have an extra emc2 in there
alex_joni: In the dialog box, there is a URI, Distribution and component. Where do I put the 'base'?
sorry, wrong answer
URI is hardy
distribution is hardy
component = emc2.2 base
but it's easiest if you just edit the file directly as I said above..
alex_joni: sorry, I didn't know
alex_joni: thanks again... its installing something now...
I don't get xkcd
I think I know who the two characters are supposed to represent
I like the new pid saturated outputs
I used CL to check for saturation for N periods, and feed that to axis.N.amplifier-fault
heh, I recall when you were a ladder skeptic
do you remember when I was a HAL skeptic too?
for certain things, it's much easier to write ladder than logic in hal
I used emc1 longer than most, because I thought hal was scary
HAL blocks like AND and OR should only be used when you tneed a tiny bit of logic
more than one or two rungs, and CL is the way to go
I did use hal and blocks for my home indicators (AND the two home switches)
but that's the extent of my hal logic I think
man, the UPS is going nuts
we have a lot of wind here - I haven't noticed light flickering, bu the UPS has kicked in a half dozen times
but anyway, pid-saturated is cool. it tripped when I tried to get an impossible velocity
I'm pretty sure it would also trip if I lost feedback
if you were moving at the time anyway
if stationary, maybe not
maybe not right away, but definitely before any runaway happens
if cmd and fb are both stationary with very small error, you won't get runaway
as soon as you try to move, you will
you'll get crawl-away, at least until the integrator ramps up
crawl-away isn't quite so destructive...
I saw a full runaway once on the old control (broke the coupler to the feedback)
it moves pretty darn fast
and hits pretty darn hard I imagine
believe it or not, I stopped it
it had full travel to go, and I stopped it halfway to the end
damn, tempus fugits when you aren't looking - I should start dinner
I stared some soup at noon - smells pretty good now
come on over and have some.
I'm planning on making soup tonight too
chicken corn chowder
but that will take too long so I gotta make something else for dinner
yum, corn in soup is good
corn, redskin potatoes, roasted poblano chilis....
I'm off to bed, before you make me get a midnight snack
good night all
I have not found any more spindle problems yet :-)
oh, before I got off cooking, an EMC thing
in fact, trunk is working really well for me
the pyvcp entry widget - I've been thinking about providing an "input" pin for it
when the pin value changes, the widget is loaded with the new valul of the pin
cradek: that's good news
I plan to remove quite a few more things, so if it still runs ok for you it'll be great
jmkasunich: so it's an I/O pin?
jmkasunich: I think using an entry widget for a readout is a gui blooper
I dunno if it should be a bidirectional pin, or if the widget should have two pins, one in and one out
it could lead to fighting with the user who's trying to edit the thing
there might be a better approach, let me explain the intended usage:
I want the user to be able to measure something and put the actual value into the widget, then use that (via M66) to modify the program
rough, M0, measure, M66, calculate, finish
if the initial value of the widget is something like zero, the preview will be all messed up
but if I can preload the widget with the nominal value, then the user's value will only be a minor change
M65 to set it?
so, rough, write nominal to widget, M0, measure, enter, M66. calculate, finish
well, M65 doesn't do analog right now
who knows what axis's preview does with M6x
I was gonna cheat and use M101 P value -> setp value
probably just returns zero
oh, the axis preview doesn't interact with the rest of EMC, does it?
getting preview right is impossible, so I am tempted to recommend not trying
damn, I didn't realize that
right, it doesn't have access to motion's IO or anything of the sort
think of it as a special case of sai
that makes the preview useless for this class of program
well, useless is a strong word
which sucks, because I really rely on it when I'm writing non-trivlal g-code
but same for probing
is this adjustment like tool len offset or radius offest or z offset ( hole too small or pocket too shallow)
if I rough something to 0.760, then ask the user to measure, then finish to 0.750, the preview will think it is massively undersize, and who knows what will happen
tomp: diameter check
rough to say 0.760 diameter, then measure, and if it winds up being 0.7593 or 0.7602, modify the finish passes accordingly to get 0.7500
I wonder if there can be some #variable that says "this is a preview run"?
alternatively, M66 during preview should not modify #5399 at all, if it can't do the actual read
How would the interp know that it is a preview run?
lerman_ no clue, brainstorming
AXIS interprets some magic comments, I'm not sure of the list
actually, the better solution is the one I started to dscribe
during preview, the canon command for M66 would simply leave #5399 unmodified
that way I can preset it with the nominal, during preview it will assume I measured exactly the nominal value, and make no correction
during real run, it will get my actual measurement and use that
stuffing some arbitrary value like 0 into #5399 during preview is no more "correct" than leaving it alone....
the interp will always call GET_EXTERNAL_ANALOG_INPUT() and set the var to the returned value
I don't think there is a way to write "if previewing leave the value alone"
during preview, the GET_EXTERNAL_ANALOG_INPUT() that is called isn't the same as the one called during actual run, is it?
doesn't preview basically replace all the canon functions?
its accomodating something. (asking 760 and getting 7602) what is the something? else you cant say what to adjust. Some flex on hvy cut wont happen on last .010. some err on screw may not be consistant...
cradek: dunno what those special comments do
I'm suggesting you try it (tell AXIS to not run the M66 line, thus leaving #xxxx alone)
tomp: so, you take a dummy finish cut. Then measure, then take an adjusted rough cut, then take the real finish cut.
tomp: you gotta be smart, for example, last pass before stopping to measure takes off 0.005, then after measuring you do two passes, one that is tweaked to leave 0.005 for the final pass
so the pass before the measure and the final pass are the same DOC
cradek: so (AXIS, hide) causes axis to ignore the subsequent g-code until it sees (AXIS, show)?
I think so
where is that documented?
dunno, I looked in the source
Maybe we should add a parameter set comment. (Parameter #<this_0ne> = 5.234)
tried it in MDI, didn't work there
but in hindsight that was stupid
lerman why a comment
why not just write #5399=5.245
I don't see the magic AXIS comments in the docs.
Yeah, that's dumb. (Me, not you).
I write a program with the axis magic comment approach after dinner
7pm, time for food
If we have other needs to know about preview mode, we should be able to know that pretty easily.
I also wish we had an analog output similar to the M66 analog input (M65 is digital only, at least according to docs)
if the user has to do extra work to generate a decent preview, it seems like we've failed somewhere along the line
when the program branches or calculates things based on runtime variables, the preview is going to be borked
if the user can say "during preview, use this number", he can make it do something more sane
that also applies to probing
for probing you have a sane representation - just use the entire move
it may end up shorter when run
jmkasunich: I'll add the analog output in the next few days
how is M66 so far?
for example, tool length probing - if the preview "probe" always returns zero, it will probably get messy, but if the user can say "use this value during preview, and the measured value during run" you get something reasonable
bbl, dinner here too
how about measuring the tool, updating tool table and reload the preview?
M66 seems OK, all I've done so far is "setp motion.analog-in-00 <something>" then M66 E0, (PRINT, #5399)
it appears to work
alex_joni: I thought you were going to sleep?
and I thought I was going to eat.....
yeah.. same situation :)
man, I think we're getting Ike's leftovers
hope the trees in the back yard hold up
I have two trees where "distance from tree to house" is significantly less than "height of tree"
if wind direction is house to tree.. then it should be ok :)
ohno, the house blew over and hit the tree :)
its starting to hiwl here too
alex_joni: keep us posted. i'm interested wether John can wire the analog output to modify the path ( we used to have thumbwheel cutter comp on wedm's for offsets )