03jmkasunich 07TRUNK * 10emc2/src/emc/motion/ (control.c motion.h): convert homing from edge to level triggered
" (control.c motion.h): convert homing from edge to level triggered" - any reason why?
to detect errors on machines that have a particular problem - where the home switch is not at the end of travel, so you can get on the wrong side of it
Kinda like that over ride switch?
jmkasunich: i think i once asked the estimated refresh rate for a servo system under emc.
I'm sure it's a function of cpu speed and thread speed, but iirc, it was a guestimated 5ms ? ( 200 hz ?) Sound reasonable?
(I'm looking at how fast i need to sample the edm gap to develop a new velocity , so nyquist strikes again ;)
servo update rate
even with an ISA servo controller emc can do 1kHz servo cycle
2-8? kHz is possible with various hardware
the default is 1KHz
guesses on the mesa?
thanks, so i need 2x+ sampling rate, more if i average
not sure I follow
oh, you want to sample the gap volts at 2x EMC's servo rate
i think i need to sample the process faster than i calculate the new velocity
03jmkasunich 07TRUNK * 10emc2/src/emc/motion/ (command.c control.c motion.c motion.h usrmotintf.cc): fix override-limits so it only overrides the limits that have been hit, and prevents further movement in the wrong direction
03jmkasunich 07TRUNK * 10emc2/src/emc/task/taskintf.cc: fix override-limits so it only overrides the limits that have been hit, and prevents further movement in the wrong direction
03cradek 07TRUNK * 10emc2/src/emc/task/taskintf.cc: fix override limits readout on the guis
a-l-p-h-a2 is now known as a-l-p-h-a
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-06-15.txt
alex_joni_ is now known as alex_joni
level triggering. finally.
Hi there. My machine has a Y-LOCK actuator, which should be engaged when the Y position has been attained. A good case for synchronised digital output?
It does this because the Y axis is a big heavy drill assembly which runs on a sloping track.
hcseb, do you know the locking/unlocking times?
No, I don't
hcseb: I suspect you need to make sure emc waits till it is unlocked
before it commands motion again
hcseb: I would set up a simple ladder
which sets motion.feedhold
then emc2 can't move..
Ah ha, that's the information I'm after.
use y-velocity as an input, if it starts to increase, set the feedhold
command the unlock
and wait a certain time before letting motion go
(at least the theory sounds easy/simple)
hcseb: familiar with HAL?
motion.feedhold - is that an output from the motion controller? And I'm fairly familiar with the HAL now
input to the motion controller
motion.in-position (or something like that) is the output
Ok, So I set motion.feedhold when I have engaged the lock.
motion.in-position commands the lock
but you need to make sure emc2 doesn't run off while the lock disengages
when it needs to move motion.in-position will become false
and the lock will start to disengage
but if in the mean time the Y axis starts to move, it might result in bad things
so .. you need to prevent emc2 from moving until the lock is disengaged
* alex_joni loves his new monitor
Disengage lock, wait for any motion in the axis, then as soon as some is detected, engage emc motion control?
lets say your axis is locked
you want to move now (in the program)
first detect the movement (YVel != 0)
then set motion.feedhold true (prevents emc from any further moves)
command disengage lock
when disengage lock is finished, program motion.feedhold false
then emc2 will start to move
the only problem is the tiny movement when it detects that Y wants to move
I feel like I'm going to need to sit down to work this one out
just before it blocks with motion.feedhold
and sleep on it a couple nights :P
<- is standing next to the monitor which is awkwardly just to low to view while standing straight up...
I got me one of these: http://www.trustedreviews.com/displays/review/2006/08/03/Samsung-SyncMaster-205BW-Widescreen-Monitor/p1
Of course, before I get the Y Lock stuff sorted out, I have to work out why it is that I am getting no input signal for the Y axis..
from the encoder?
lerman_ is now known as lerman
Was hoping this was the channel of the IT company EMC :)
Hello! Can EMC do an automatic measurement of the tool after a tool change? Would like that after every tool change it automatically moves to a sensor and measures the tool before proceeding with work.
x3rox: with g38.2 (probe) you can measure the length of a tool
x3rox: in emc 2.1 (the latest released version) you can't directly use this length with G43 tool length offsets. You can use G10 L2 ... or G93 ... to change the coordinate system offsets instead of using tool length offsets
Newbee is asking... ;-D
x3rox: in emc 2.2 (the version under development) you will be able to simply give the tool length to G43.1 (transient tool length offset)
So I initially set the working zero point once and from there all moves are relative, and then I can change tools as often as required without again setting the zero point?
Sorry for asking possibly stupid, but I habe never used EMC.
x3rox: here's how it works in the development version: you put a switch that can be pressed from above by each tool in the corner of the table
This is what I want to do.
x3rox: you load the first tool into the spindle and "touch off" to the top of material or otherwise set the Z=0 of the workpiece
x3rox: then you use a sequence of moves to go above the switch and then down until the switch trips. This move until the switch closes is a G38.2 "probe" gcode
x3rox: later you put the second tool in the spindle -- it may be a different length
you do the "probe" sequence again, and the difference of those two numbers is the difference in the length of the two tools
Ok. is then Z=0 again in exactly the same height to me?
you give that number to G43.1 (transient tool length offset) and now when you command Z=0 the tool tip goes to the same place the old tool tip did b efore the change
All this can be done by a "macro" inside tool-change?
x3rox: in emc 2.1 the G43.1 code doesn't exist yet, so you have to hack around it by modifying a coordinate system offset.
Well, I will get my mill not before 1-2 months. Is then v2.2 out?
x3rox: you have to explicitly call this code separate from the tool change. You command T1M6 to load tool 1, then the code to find the reference tool length. Later after you load the second tool with T2M6 you include the code to find the difference between the reference tool length and the new tool length
this is done best in an "O-word subroutine", and I think that emc 2.2 will include some feature that makes it fairly simple -- just something like "O1001 call" for the first tool and "O1002 call" for the subsequent tools
x3rox: We haven't chosen a date for emc 2.2, but I hope it will be released this summer. In any case, the upgrade from emc 2.1 to emc 2.2 will be fairly easy for most users.
Would be great to make a tool change with one command. ;-D
I agree, but realistically emc 2.2 is not going to have such a feature
Well, I surely will need some time to get the mill working again (must build new electronics,...), then I can play around with it in spolystyrole foam with version 2.1, so I assume I will not be ready before August.
in emc 2.3 I hope someone will choose to work on advanced tool changing stuff -- automatic length detection and tool-pockets-on-table are the two items that seem to be coming up most frequently
right now both can be done through gcode "Onnnn call" but not through a simple "Tn M6"
jepler: as long as "O1001" is the only thing I have to enter, it is one command to me.
Am I wrong?
* archivist can think of a really evil tool measuring problem
x3rox: depends if your CAM or whatever is generating gcode will be able to write the required Onnn call line
does anyone here know the status of the "O- call from file" feature in trunk? I hope I'm not mistaken about it
Is there a way to add a REVERSE switch to common household fans by chance?
I'm thinking about getting this breakout board, can anyone see any pros/cons with it? http://www.cnc4pc.com/Store/osc/product_info.php?cPath=33&products_id=48
jmkasunich: are you doing live testing on the mazak machine, then coding to suit?
JymmmmEMC: I notice the silkscreen of the board says "pins 2, 4, 6, and 8 are for high-speed frequency signals"; if your step signal is on odd pins it sounds like they've engineered the board not to work
jepler: Sorry, had a visitor... If all I have to do is "search & replace" normal tool-select-calls by "O100x", this is easy...
jepler: Sorry, I dont quite understand what you mean by "sounds like they've engineered the board not to work" ?
started writing up what I think needs probing for a clock gear miller
JymmmmEMC: if you want a high frequency signal on pins other than 2, 4, 6, or 8 it sounds like it may not work, based on that text
(if these are for high frequency, the others must not be)
Has somebody a really easy tutorial on G-Codes? Would like to lean that when I have time.... PDF preferred.
jepler: I know that STEP woulg be hi-freq, but DIRECTION be considered that as well?
x3rox: our gcode manual is http://linuxcnc.org/docs/2.1/html/gcode/main/index.html
with a quick reference http://linuxcnc.org/docs/2.1/html/gcode.html
but neither is really a very good introduction
jepler: Thanks, I'll email the guy after I've had a chance to RTFM some more.
JymmmmEMC: I don't see this mentioned in the pdf manual, all I have to go on is the silkscreen
[17:43:18] <jepler> http://www.cnc4pc.com/Store/osc/images/c11r3_1.jpg
near the lower right
jepler: You make a good point, I did find a couple things lacking in the docs myself. Thus my inquiry. (Yeah, got it open right now =) I already asked him if the SSR can be changed from 10A to 15A too, my spindle takes 12A.
it's also on the board picture in the last page of the pdf manual
I do like the fact that almost all IC's are socketed though.
jepler: Has nobody thought about making a pretty PDF book instead of this very thin appearing HTML "manual" about gcodes?
jepler: Silly Question... do YOU see the SCR in that pic (is it the heatsinked left of the fuse?)
x3rox: contributions are always welcome =)
Is the whole "manual" of g-codes this one HTML file?
jepler: On this page http://cnc4pc.com/Multifunction_CNC_Board.htm
at the bottom it says "Check out the scope images at 93.9khz on pins 2-9." http://cnc4pc.com/C1_Performance.htm
I guess s/he doens't love us anymore.
Hmmm.... From the manual... Output pins 1,2,3,4,5,6,7,8,9,14,16,17.
• Input pins 10,11,12,13,15.
03cradek 07TRUNK * 10emc2/src/emc/kinematics/tp.c: second half of fix for rigid tap being position mode
^^ we think rigid tapping works, and we're looking for the right tools to actually try it on the mazak
I witnessed it. :)
oo cant see it on the web cam
I am the guy on the right :) (your right)
hey there. got a link to the webcam?
[20:08:14] <archivist> http://linuxcnc.org/compile_farm/festcam.shtml
thanks, got it.
hey, that mazak? a buddy of mine has 2 of those, running on original controls.
whoo hoo -- Rigid tapping works :)
show us out in the world
they are going to take pictures
Hey guys, if you have a PayPal or Ebay account, you REALLY should order one of these tokens. They say $5, but when we ordered them, it was free. https://www.paypal.com/securitykey
3/8-16, tapped on the mazak, and the bolt fits!
My gf got her ebay account hijacked last month, this would have prevented it.
[20:40:06] <JymmmmEMC> http://www.theregister.co.uk/2007/03/21/ebay_hijack_plea/
They're in the testing phase, so might limited on the wty they issue out.
jmkasunich: cool beans!!!
Now, if I only knew what RIGID tapping meant =)
head rotating and feeding at the tap screw pitch tap tight in the spindle
add a comma after pitch
just did one in 1/2" thick aluminum plate (the first few were in plastic)
machine taps (spiral)?
emc has done rigid tapping on the mazak -- code is in cvs.
rigid tapping photos: http://axis.unpy.net/index.cgi/01181942178
just seem em nice, try with a spiral tap in steel next
better at swarf clearing at speed
Ok, i'm ignorant here... rigid tapping -vs- what tapping?
loose eg tap in a sliding sleeve
JymmmmEMC: at the bottom the spindle reverses and the tap comes out backwrads
(beats me; I'm no machinist)
[22:00:55] <jepler> http://www.mmsonline.com/articles/0703rt1.html
jepler: does g96/97 show now?
jepler, in rigid tapping the controller knows the tap is say, 1.2mm pitch. as the spindle turns, it moves Z down by 1.2mm for every rev, it has to keep it sync'd as it slows at the bottom and then revereses etc
alex_joni: I'll check
jepler, basically the Z axis becomes slaved to the spindle
I only looked for it to show in the gui
didn't check if it really works
robin_sz: yes sounds about right
alex_joni: g96/g97 work mostly
you should know :P
we tested it on chris's lathe
next add power sensing on the spindle and then tap wear can be noted
and sister tooling would be nice too
archivist: there is an input called adaptive feed override
basicly you feed an analog input (0..1) to select real time FO
alex_joni, no, thats not what he means
so you can check for wear, compare with a preset value and use that
not control, sensing
its the error signal from the servo drive
robin_sz: you can probably check speed vs. programmed speed?
a worn tap needs more power
its the spindle load that matters
and too much power breaks it
alex_joni, most modern CNCs read the actual power (watts) in the spindle ... you set a limit on the tool, and if the spindle exceeds that in a cut, then the tool is declared worn out, it then either:
or b) swaps to a sister tool (if available) .. you can declare certain tool stations to be sisters of other tools, with maybe a differetn tool length offset
alex_joni: yeah it looks like the display is right
have you guys ever had issues with dvi monitors?
sister tooling is fairly common on machines these days
alex_joni, only the price
* alex_joni just got a new one, and it turns off from time to time
might be the graphics adapter though
it just goes blank, and restarts in ~ 1 second
or a mobo that can;t really supply enough amps?
robin_sz: wouldn't that lock up the machine?
it is resolution-dependent
lower res doesn't do it
the AGP and PCIX slots often have a separte on board regulator on the mobo
this is agp 8x
older dual DVI card
anyways.. off to bed now
good night all
jepler: I didn't manage to make it say G96..
alex_joni: really? the startup should be g96 (spindle in RPM mode)
the alternative is g97 s- d- (spindle in CSS mode, S FPM, maximum D RPM)
fwiw if monitor makes 'clunk' sound, it can be the +12V is overloaded ( similar to de-gauss sound )
hmm.. let me check
Rugludallur: I've been studying the NewPlasmaDesign.odg, there's a few connections missing. would your .hal files fill in the missing bits ( eg Position Sel B's select line, AdjustHeight's in0 ...) ?
tomp: it's a tft.. no sound from it
alex_joni: My tft monitor goes blank if it is too cold in the shop.
It then stays blank until it warms up.
(It's a Xerox LCD display.)
lerman: it might be too warm now.. definately not too cold
[22:23:21] <jepler> http://icanhascheezburger.com/2007/06/09/halp-8/
did you see the rigid tapping photos?
alex_joni: er, I guess I had it the wrong way around
yup, very nice
g96 s- d- is CSS mode -- e.g., g96 s50 d1200
well, if you say it's working .. I'll take your word
from the code I looked at it should work :)
more lathe progress?
good night all
ds2: not since earlier this week -- we added CSS and FPR
if you're watching the camera, everyone's clustered around it because we're trying to make a tool holder for the mazak -- there aren't any holders for smallish imperial tools
ooh CSS :D
jepler: when are the stock removal canned cycles going in? ;)
ds2: not this fest
ds2: the camera is http://linuxcnc.org/compile_farm/festcam.shtml
where is the fest? looks like someone's shop
[23:41:44] <jepler> http://cnc-workshop.com/
it's a place called Cardinal Engineering in Galesburg, IL, USA
ahh... is the galesburg thing going on today?
forgot all about it
Ziegler: we've been here all week (some of us since last saturday) but yeah today and tomorrow are the biggest days
im only about 20 minutes away
wish the cam put names on everyone ;)
Ziegler: come on down
ds2: you probably saw me, jmkasunich and cradek in the last frame
I might manage to come down tomarrow
I actually just finished up my machine last night
well... 98% finn
Ziegler: what machine is ever 100% finished?
[23:43:56] <Ziegler> http://images.myonlinesite.com/cnc/construct/20070614
Ziegler: just look for the big blue maxnc, and you'll find the emc folks nearby
if you go elsewhere you'll be invited to do crazy things like buy mach3 :-P
jepler: the one that is in the crusher or remelt furnance is 100% finished!
ds2: yep that's true
03jepler 07TRUNK * 10emc2/docs/html/gcode.html: document new gcodes added at fest: rigid tapping, constant surface speed, feed per revolution
need higher rez so I can read your screen }:-)