I found my keys this evening, in the glass recycling crate in the back yard. I have no idea....
was my fluke meter in there too?
and the log book for my bike?
I think the log book for your bike is in my van...
andypugh: best place ever to find keys was inside the bass reflex tube of my old speakers, we would have never looked if our daughter would not have reached in there again while we looked for the keys
* JT-Hardinge wanders inside to make green eggs and ham
Tonok2_ is now known as Tonok2
"EMS moves the smoothest of all three, the cuts are consistent and the
code seems very robust. The backplot screen is great on Axis. The
other screens are pretty good, but I'm loving the simplicity of axis."
interesting comment i think, found it while googling EMC2 backlash
realized i never asked a single question about backlash when we took it out of X on the Fanuc controller this morning
cradek: so your email said you are the ladder logic wizzard to call upon in times of need? :P
if you have a laser, you can do better and actually map the screw errors for emc to compensate
backlash comp is simple to configure but sure assumes a lot
cradek: converting a knee mill in a few days is promising, the tech who services our Johnford mills thinks 4 weeks down time with rewiring stuff
somewhere in between is where you'll always end up I bet
which we thought sounded high
we guessed 2 to tops 3 weeks on first retrofit, less on second
whether that's reasonable depends hugely on the machine, what schematics you have available, and how well you plan.
cradek: any docs on this laser mapping of screw errors? googling bring up a cnczone emc thread , but no info on how, just that it's simple
all i ever seen involved dial test indicators
you put the errors in a text file and emc reads it in
yeah you can measure however you think you can get good results :-)
cradek: i am wondering about the laser part of it, just intrigued by another method
oh there are systems where you set a target on the axis and move it, and the laser head tracks it and gives you the precise position of the target.
cradek: unfortunately taking out backlash this morning made the second part worse, and it being the last part, no time to find out if part moved, or whatever the root cause was for the cut in X+ being the one cutting too much, not the X-Y+ or X-Y- cuts
if you were discussing this earlier, I didn't see it
nah, just sharing a bit of my work day, aside, Fanuc backlash comp is not realy #emc stuff
ah I see
just glad EMC has it, since i had not asked about it before, google made me happy though ;)
yeah screw comp is a nice emc feature
what we don't have is periodic corrections (measure, curve fit, give to emc, it corrects that)
someday that might be nice.
relatively simple, but nobody has cared to do it
How does EMC handle backlash comp? Just add (+/-) distance to the axis move?
as opposed to a unidirectional approach
it would have to consider direction change of the ballscrew only
sort of. you have to add it in gracefully - not quite abruptly
but yeah that's the general idea
you're always (ALWAYS) better off fixing the machine though
still, milling is not an easy thing to do with backlass present, lathe work....fine
cradek: realistically expect a ballscrew to go a year longer than it should
software backlash comp, no matter how well it's implemented, doesn't work very well to remove the effects of a sloppy machine
* morfic is all psyched none of the lathes turn flat spots on the apex of balls right now
My manual mill has a little lever on the long feed table with a climb mill pictogram. Tighten up on that nut!
^^ backlash comp works well to keep flats off balls
whose comp or whose balls, i now wonder
* morfic sees .cc files in /emc2-dev that's good, i unfortunately took a turn towards motion.c first, .cc is better for my brain, much better
all realtime parts of emc are pure c
i don't see myself touching realtime stuff
clock is cruel, screaming at me saying it's late, so thanks all, 'night
I am trying to compile the latest download of 2.4-dev. when I run sudo apt-get build-dep emc2 I receive not found errors:
checking for rtl... not found
checking for mbuff... not found
checking for rtl_time... not found
checking for rtl_sched... not found
checking for rtl_posixio... not found
checking for rtl_fifo... not found
What have i missed?
0.12.0 HeeksCAD is nice.
a few new features
like what? :)
zig-zag roughing and finishing?
[04:50:20] <elmo40> https://sites.google.com/site/heekscad/news
thought this was cool, if anybody is trying to make a delta robot: http://www.societyofrobots.com/robotforum/index.php?topic=10906.0
SWPadnos_ is now known as SWPadnos
open source dwg viewer ?
you need it, you write it.
I need reading dwg file . for my own software.
so I need open source code dwg viewer lib.
I use python but I can use .dll. Are there any dwg viewer lib.?for python or any language
MarkusBec_away is now known as MarkusBec
i don't know anything about it
i can't believe youtube doesn't support ogg theora
I received some cards from Mesa Electronics today and they included some star shaped orange plastic bits. Each "arm" on the stars looks like it is supposed to break away, but I don't know what these little arms are for. Anybody have a clue?
[11:44:36] <alex_joni> http://www.youtube.com/watch?v=AENJxqR0g48
alex_joni, I was sure this would be some mad 9-axis machining or something... but still cool
findux: can't you use qCAD for a dwg viewer? Or is it only dxf?
alex_joni, ok after watching the whole thing, WOW
the first part is ok-ish.. but the augmented stuff is really nice
not hard to do either
elmo40 actually I need dwg library for use my own software. I developing a soft ware.but software not cad software. I need to read dwg file but I have some problem dwg file library source.
it's like having the laser on the circular saw just much better
anyone around? :)
[12:13:45] <frallzor> http://pagesperso-orange.fr/fabrun/project%201/indexproject1.html
what is this type of model called in english?
cant figure it out
sheets of wood you assemble
frallzor: Hey Mr #57!!
good work.... so now you are going to make little cars in 3D?
would be cool to make one and have on a shelf
will be fun to see how my dust shoe performs today
will be assembling it later =)
simple design made for the Mechmate =)
frallzor: I designed the shoe so it moves with the router... what was the reason you make it staic?
simple, allows more movement atm, but i will mak a moving one later
but for sheet work this is good
since the fan blows air down
if it lifts then some dust might escape around
frallzor: ic.... I have a router, and I need to make sure that I move my air back up, else the router blows more air then the fac..
i see it as a good thing
if you have brushes or similar and then suction
flow makes the dust never settler
and is easy to suck
findux: I timed out, did you receive my message?
qcad yes receive
frallzor: I thought so to, for my fac it was just to much :) The dust blew all over the plce
no, freecad :P
when i did have some suction using old vac, it was allmost ok, the new one will make it work nicely i assume
frallzor: good.. I must admit, my vac wasn't to powerfull... I like the idea of your design though, I am going to consider that aswell..
but im just using what I got, the spindle is aircooled, so its blows alot of air =)
frallzor: how did the item come out?
[13:12:49] <frallzor> http://pici.se/p/ppOKrFhVV/
a couple lines at the right.
tool length move or something?
the fact that I dont have proper fastening yet =)
after some material was gone it started lifting
need to figure out a nice way to fix that
no vacuum at all =)
there is your solution ;)
will be making some kind of T-slot solution
not bad, made the machine in 5 months according to the log
t-slots will work. vacuum is easier
I disagree =P
but your machine is for large flat parts
vacuum is simple to setup.
t-slots need bolts and such
and you have to avoid clamps when machining
well how to fix easy vacuum then?
you have a vacuum cleaner? it will work, a little... but a compressor will be better.
then the hose to the back. and a table with holes in it.
a vacuum pump is the way to go
we use piab
split the hose to each hole. close off the ones that don't fit the material (no leaks)
that seems like a messy work if you need to split for every single mount
just make channels and block off unused ones
channels on the top that is? and block with rubber?
[13:34:20] <JT-Work> http://www.piab.com/en-gb/Products/Vacuum-pumpsgenerators/
yea or a bunch of holes and the bottom as a small box
the smaller the size of the box the faster the vacuum will work
[13:38:25] <elmo401> http://www.klitgaard-design.dk/?p=314
very simple design ;)
but for plastic forming... same concept, though
a 'pro' version: http://www.techno-isel.com/cnc_routers/technical_section/Vacuum_Hold_Down.htm
one could mill those tables pretty faste
ah got a nice idea now
some piping required :P
how is mdf for vacuum?
first styrocut test with emc ->http://www.pictureupload.de/originals/pictures/210410155936_werbung_sim.png
generated by heekscad
i will try it on a VF2
frallzor: mdf should be good for that
a few holes, milled zones in the mdf and some gaskets should do the trick
the table is 2750x1630 aroundish, divide into maybe 8 zones of 2750x200mm
and just limit to the holes for smaller areas
would a back valve be of some use?
or what ever its called in english?
so suction kan be kept if the vacuum is lost in the motor
it's called a check valve and it would not work unless you had a perfect seal
well I assume a gasket will make sure that happens
The_Ball: >> I received some cards from Mesa Electronics today and they included some star shaped orange plastic bits.
I believe those are "keys" used to key the plug on connectors to a certain socket. I haven't installed them yet and may not use them..
I don't recall seeing anything like that
Dave911: yes they are keys
JT-Work what do you use as gaskets?
our vacuum applications are not for wood cutting
I assume it must be quite flexible to allow it to sink so the material lies flat on the table? yet not too soft so it can be vacuumed
JT .... so do you insert those in the socket slot and then remove the little tab on the plug to create a key situation?
I had a couple fall out after...
Dave911, aha! now it all makes sense
cradek, I didn't receive these "keys" for my older Mesa card, but the older Mesa card has a black style connector, these new ones are green
cradek, screws are on the top instead of on the sides
Thanks JT.... I think I will skip the keys and take my chances .. yep I know... I'm living on the edge ;-)
oh ok, replacing those black connectors with something else is good :-)
The ones I received about 6 weeks ago also have green connectors..
The green connectors work fine..
I put in a couple of these "keys" and they fit perfectly, but they don't stop you from inserting the terminal block at all
I have some of those, also green.
they only work if you have the "blocks" in the terminal strip
(not mesa though)
Dave I used a fine tip marker to mark mine as well
SWPadnos, I have the "fins" that will block against the "keys", and it does make a difference, with no key the terminal block will drop in by its own weight, with one key I need to apply 50-100 grams (guestimate) of force to enter the terminal. So I guess it works
Yesterdays advice to set P,I, and D to zero and FF1 to 1 worked great!
There was still an accumulating error over long rapids, but that was corrected by slightly reducing the max velocity.
A little P and I should fix that error
(you need a little because the stepgens xtal derived rate is not an exact match to the PCs clock)
Now the error is a small amount, about 0.0009, right from the start of a move.
Falls to zero or so after the move stops.
Without at least some P and I you are running open loop so you will get accumulated errors in velocity mode
FF1 alone works well because the stepgens generated velocity is so close to commanded but you still need feedback
Got distracted, had a visitor. back now.
Tried some P & I, it just seemed to make it jittery.
About P = 0.75 it starts to jitter, adding I did not seem to help.
what is your encoder resolution?
4000 count per turn
ok, so it matches the step resolution (?)
8 turns per inch = 32000 counts per inch.
turn ~= inch :)
turn = .125 inch.
You only need a little but if you have none, (only FF1) the position will creep away...
you need deadband of around one step, or 0.00025
One step would be 0.00003125 in theory.
no, that's one encoder count
OK, I was using 0.0005, so that should have been OK?
yep, should be
Now that I have the 5i23 card I can actually step the signal generator faster then the motors can take. So...
I plan to redo the signal generator firmware to only make half as many tiny steps per commanded step.
Thus changing it to 1000 steps per turn.
If it jitters at P-=.75 , use 1/2 or 1/4 that you still need feedback...
This should maybe reduce the deadband to half?
make the signal generator do 1:1, there's no need to have any multiplier in there
I'm not sure if the 5i23 card can step that fast?
(in other words you are not actually using your encoder unless you have some P or I)
the 5i23 can do millions of steps per second
the encoder will still cause a following error if it stalls - I think that's still the only possible advantage of this setup (and you don't need velocity mode for that)
as for peter's statement, FF doesn't use feedback, it's command only
P,I, and D are the terms that incorporate feedback into the equation
so you're effectively open loop when you only have FF0, FF1, and FF2
So open loop with crash detection?
more or less
and stall detection added for free!
closing the loop will also work better if you decrease your ustep size
OK, I'll redo the signal generators firmware for 1:1 micro steps, that will make it 25000 steps per turn.
I suspect your encoder resolution is better than your step size this will lead to jitter
huh. I thought it was doing 8 microsteps per step ...
Oh, no, 50.
oh. that's a lot :)
so that'll be 200ksteps/inch. how fast does this thing move?
I was thinking of going to 25, for 1000 steps per turn.
you may have to look at the inputs (if they're optoisolated) - optos may not keep up
0.55 in. / sec.
the Mesa card can step pretty fast, it's a question of the optos
ok, 110 kHz probably shouldn't be an issue
No optos, direct.
then 110 lHz is really fine :)
optos on most other stuff though.
But I'm not sure if the dsPIC30 can read that fast, 16MHz internal clock.
is the step input on an interrupt line?
it is polled and debounced.
remember, you can get rid of all the timing stuff in there - there's no need to count the time between pulses or anything
just increment/decrement the sine wave index whenever a step is detected
no, it is slow enough that none is needed.
even 16 cycles is pretty close to enough to poll and update the DAC outputs
(which would give you 1 MHz step inputs)
160 cycles is plenty
I don't know, I was kind of disappointed in the dsPIC30 as far as it seems very slow.
hmmm. do the dsPIC's have the clock divided by 4 like the regular PIC?
* frallzor gotz screws
I has a choice of 3 clock dividers, I have it on the fasted, had to add a heat sink.
I'm talking about the machine cycle being different from the clock cycle
DSPics have a PLL the ones I'm used to have 30 or 40 MHz CPU clock capability (1 clock.inst)
* SWPadnos hasn't used a PIC since the 16F84
Maybe it could have a faster XTAL, but I ordered it as a demo board, so ...
frallzor: You finally called 1-900-Happy-Hookers ?
All I had to build was the DAC board that plugs right onto it.
frallzor: oh man, you made my day =)
Well Thanks all. I'm going to call it a day and get ready for work. (night shit^Hft)
jon e's spindle encoder is clever
it is neat
having index for tapping is neat because all your threads end up oriented the same way - but it's not necessary if you don't care about that - I just wired one of the AB phases to Z
we just got a RFQ from one of our customers and their spec called out our jet orientation device so we are going to get part of it for sure
I guess I can't "peck tap" either, but who cares, I use proper taps instead
have you used a thread mill yet?
no, they're too expensive for me
I have one but have not used yet LOL
"Borowed" ? http://sfbay.craigslist.org/sby/wan/1702922099.html
err "Borrowed" ?
wonder if he can spell lock now
hay you if your reading this add
He can spell Lock Out
that's how we spell it down here LOL
Either he didn't lock it up, or it got towed.
been waiting for my replacement tank for the way lube for 2 weeks :/
cradek: what was that coolant you used for the hardinge?
"soyeasy nucut something"
the "something" is different weights or something - if you call they'll help you match it to your work
Jymmm: smells like french fries
ok, I got it now
cradek: LOL, oh that would kill me. You're trying to finish a job, hungry, and the place smells of french fries!!!
JT-Work: the manufacturer: http://www.elmusa.com/
I ordered mine from them directly - not sure if they still do that.
thanks, I'm an oem equipment manufacturer so maybe they will sell to me
a bit crazy http://www.crabfu.com/steamtoys/rc_steam_turbine_tank/
[16:48:07] <Jymmm> http://v5.tinypic.com/player.swf?file=apa9ti
now that is funny
lol, how so? The hunter/hunted part?
no, more that there are far fewer ducks as pets than almost any other animal ;)
in 'apt-get install build-dep emc2' is not 2 files? 'build-dep' is a switch ?
build-dep and install are two commands
I think you can only use one or the other at a time
as a kid my duck was 'herkimer'
SWPadnos: Ah. I guess ppl dont realize how smart birds can be.
the command was from the wiki page
the command line you typed should have tried to install a package called "build-dep" and one called "emc2"
do "apt-get build-dep emc2" and "apt-get install emc2" separately
works 'apt-get install build-dep emc2' does not work 'apt-get install build-dep'
could you rearrange that in the form of a sentence? :)
This phrase works 'apt-get install build-dep emc2' . This phrase does not work 'apt-get install build-dep'. ( no idea of colon placement for correct grammar :)
that is a strange result though
"apt-get build-dep emc2" should work
as should "apt-get install emc2"
oh, well duh
"apt-get install build-dep" tries to install a package called "build-dep"
which probably doesn't exist
tom3p, were you looking at the "Installing EMC2" wiki page?
stupid computer, do what i mean (R J DALEY )
somebody commingled "build-dep" and "build-essential", I think :)
( client keeps dropping out, sorry wlan0down... wlan0up... rinse lather repeat )
I'd suggest lathering before the rinse, but that's just me :)
oh well, it got something
yesterday i was looking for how to find dependencies, found that you can get a dot (vizgraph) style plot of relations from dpkg using the 'dotty' option
tom3p: OR, you could just connect the machine to the net and be done with it.
yah, can you help me carry the eight towers into the library ;)
tom3p: No, but I can give you the color code so you can run a cat5 cable to them =)
isnt there some offcolor mnemonic device to remember that color order?
I am trying to configure EMC2 do control my spindle motor and am having problems understanding what I need in the HAL file. I am using a Mesa 7i43 card driving a 7i47 daughter card. This in turn is plugged into a Homann DC-06 board that turns a step/dir command into a variying DC voltage (0-10v) that is connected to a motor control board.
I started with a base config that was generated by pncconf (an alpha utility in dev EMC).
I pasted the spindle portion of the .hal file to http://pastebin.com/JxXE7anq
possibly you want to use mesa's stepgen in velocity mode to generate your pwm signal
Yes, I do want stepgen in velocity mode.
ok, cool. what is your specific question then?
well, I have a scope on the output of my mesa card and am not getting any pulses
you may need to set the stepgen pins to output mode (?)
ok, that narrows it down. while you are expecting but not getting output, could you do a show pin hm2_7i43.0.stepgen.03 and pastebin the result?
grommit, could you post the whole hal file on pastebin>
yes, on both counts. Hold on a min....
LOL "...and the police reloaded it for her" love it!!! http://www.cnn.com/video/#/video/crime/2010/04/21/dnt.elderly.woman.vs.intruder.kcci?hpt=T2
that's about as funny as the duck :)
Ok, entire config is here: http://pastebin.com/ByvhgUc4
will try show pin now....
ok, I don't see anything that sets the stepgen pins to output mode, but I confess that I don't know if it's necessary (I'd expect the driver to default to output mode for pins that are assigned to output functions like a stepgen)
how do I do a show pin with AXIS running?
grommit: my enlish not good enoph to anser this quest
grommit: open console and type 'halcmd show pin show pin hm2_7i43.0.stepgen.03
err sorry, open console and type 'halcmd show pin hm2_7i43.0.stepgen.03'
ok, show pin is at: http://pastebin.com/aVTJhswe
how about the results of halcmd show pin | grep "*is_output*"
what hardware pins are you scoping?
grommit: if your running Axis you can select Show Hal Configuration from the machine window to view pins
what emc2 version and firmware package version are you running?
and also important, what does dmesg say about which pins are assigned to what functions?
I see that for at least some firmwares, stepgen 3's pinout in this firmware was bogus
halcmd show pin | grep "*is_output*" does nothing, just returns the unix prompt
right, no stars needed
haldmd show all | grep is_output
(they're params, not pins also - oops :) )
non-gpio pins do not have an is_output parameter
that solves that problem :)
ouput is here: http://pastebin.com/dmKCb1RW
grommit: lots more questions have been asked - we'll quit bombarding you with more while you answer them :-)
Lets see... I am running the latest EMC2 (2.3.5).
Firmware for 7i43 is the lateset from Peter Wallace for my 7i47 card
The steppers on my x,y,and z seem to be working
I have commented out the encoder stuff for the time being (but will work on that next).
cradek: what firmware was stepgen 3 bogus and where did you see that?
Have I caught up on questions?
except for what physical pins are you scoping
How do I determine that?
you tell us where you hooked up the oscilloscope
oh, duh. I am currently on the output of the 7i47 card for stepgen 03. hang on and i will tell you which physical pin
I can't find the PIN file that matches your firmware - it is probably on your system somewhere. would you pastebin it too?
yep, hang on....
well I bet SWPadnos is right and we should check the output in dmesg too
PIN file is: http://pastebin.com/Pw3kVUmH
My step signal is on 7i47 card header P2 Pin 19, which is logical I/O 18
ok, I think that's an error :)
I/O 18 is pin 39
yeah seems like you should be on 39
37 for step, 39 for dir
no, that is 37,39 of physical pins on 7i43 card (header P4).
onto that header, I have my 7i47 plugged in
you're using a physical oscilloscope, aren't you?
I'm so confused
look at pins 37 and 39 on the P4 header on the 7i43 first
I was confused as hell. PWC gave me the mapping and it took me hours to get it straight
if you see step pulses there, *then* look at the pins where you expect the outputs from the 7i47
It is very difficult to get a scope on there because it is a ribbon cable that ties the two together.
unplug the ribbon cable
just to be sure, you're using the 7i47 "12 channel motion oriented RS-422 interface" card, not the 7i37 "Isolated I/O card"?
[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"