okay, running the ubuntu live it seems to work a little, at leasr
I'll play with this at the shop tomorrow
good luck, and tell us what happens
you know I will
long as I'm here...
I've one of the older (by now) BDI versions. when I went to make it current the other day, it couldn't find the host, so I'm guessing it's since been moved?
can i make my install current, or do I need to do a fresh one?
hi all, cradek still here?
we've eaten him... just his bones are here
just wanted to keep him up with my progress
I think he's AFK
Waive a $50, and look needed...
AFK: unknown acronym
Away From Keyboard
if you see him, let him know that the onboard video was messing with me
okay... you could use /msg memoserv as well. :)
okay... "/msg memoserv help"
doesnt seem to work, I'll try him tomorrow
I'm off to play, thanks boys
After 25 pages of the manual I need to go make chips for a while :-) bbl.
hmmmmmmmmmmmm.... just realized that I'm a little screwed on power on the stepper motors. oops.
DEFINATLY WORTH WATCHING --> http://www.metacafe.com/watch/252065/video/
their FLASH APPLET runs in my flash player to tell me that my flash player isn't new enough
they only want windows users to see their videos
120VAC socket has a live/dead/earth....
should the case be connected to earth?
of a power supply?
A-L-P-H-4 is now known as A-L-P-H-A
should the transformer also be touching earth? [it's housing]
What should be isolated?
how'd my lappy get killed?
A-L-P-H-A: I don't know what you are building, so I'm hesitant to make any recommendations
rebuilding my cnc mill power supply
it was working, but I actually don't think it's right.
ac line, to xfmr, to diodes, to cap, to drives?
Usually, the outsides of things should be grounded, yes.
ac line, xfmr, bridge rectifier, cap, fuses, inline cap (Mariss' design), drives.
just wondering from the ac line, should the housing be grounded? As well as the transformer's housing, and say the heatsink for the geckos.
all exposed metal should be grounded
if you can touch it, ground it
Imagine the alternative. You could have the outside floating,
This way if there's arcing in the machine, or a wire comes loose and falls against the housing, it shorts to ground.
sbailard_, I don't understand... but I'm taking a guess meaning, if you can touch it while it's on, and it's metal, ground it. If I can't/shouldn't touch it while it's live, don't ground that... [fuses for example.]
Rather than bringing the housing up to 120ACV.
Do you have a multimeter?
I have a multimeter.
ground the machine frame, the box that contains the electronics, any exposed surface
hmm... I should get another piece of sheet metal, to form a base to mount stuff from.
Then when you're done, check the AC voltage between the case and ground, between the 120 live line and ground,
I'm I little scared to do that...
LIVE voltage testing it?
That sounds like fine, only make sure you don't have traces or components on your board touching the sheet metal.
yeah, I'll isolate the data signal stuff...
but see... I have kinda like two devices in the same box...
I have one power supply for the geckos, and then I want to mount a second device (power supply for breakout board powering pullup resistors/limit switches, etc....)
Well, I'm not really competent to troubleshoot something I can't see, but that should be find.
Are they off the shelf, or did you build them?
The power supplies.
Here's my list of things, that 'can' be grounded... the two transformers, geckos, and the case itself.
no, these power supplies are based off of Gecko Drive's design.
That sounds correct.
The transformers, what are they like? Do they have cases of their own, with grounding tabs?
I'll draw it by hand, and then scan it... easiest way for me.
let me find the schematics... sec.
A-L-P-H-A, I don't know that much about power supplies. A 'yes' or a 'no' from me in this context doesn't mean that much.
it has two INs, three out leads model 167L50 http://www.datasheets.org.uk/datasheet.php?article=64037
Opening it now.
... I'm not sure either.
what's the question about the transformer?
do I ground the transformer to earth?
sure, it's chassis mount, it'll probably ground itself
well... right now, it's kinda not... as I have everything mounted to particle board... [don't ask me why, cause I'mkinda stupid with electronics.]
I know enough to cause trouble
And the chassis is separated from the primary and secondary wires, so it won't hurt anything.
As long as you put it in a grounded metal box.
okay, drew everything now.
And ground your mill, and optoisolate your computer, you probably should be safe.
(optoisolators are probably built in to the design of your controller)
jesus this image is huge.
[03:18:13] <A-L-P-H-A> http://220.127.116.11/emc/powersupply.png
not sure I think the fuse should go there
if your cap is very big, you might blow it with inrush
which means you might need slow-blow fuses, which is bad
but that's the only problem I see
cradek, it was Mariss' design, it's a 5amp (can't remember slow/fast)
I don't see any obvious errors. (I didn't spot that bit with the fuse.)
5amp fast blow
A-L-P-H-A, is Mariss' design up online somewhere?
what's the cap directly on the gecko?
it's off of Marriss' design schematics...
sbailard_ yeah... sec
cradek, power smoother... it's a 470uF 100V one.
A-L-P-H-A: don't care whose it is, you wanted opinions...
cradek, just saying... not being defensive or anything
right I understand
at least not trying to sound like it... [I don't know better]
I think the fuse should go after the last cap
[03:21:46] <A-L-P-H-A> http://geckodrive.com/photos/Step_motor_basics.pdf
ask jmk :-)
the cap right after the bridge rectifier (filder capacitor) is based off of C = (80,000 * I) / V, which turned out to be something like 12000uF, or something like that.
what's the xformer voltage?
in 120VAC, out, I forget... I think 50VAC.
so you want about 70v on the drives?
[03:36:25] <A-L-P-H-A> http://18.104.22.168/emc/powersupply2.png
oops... that 70VAC should be 50.
[03:37:18] <A-L-P-H-A> http://22.214.171.124/emc/powersupply3.png
breakout board has 5v regulation on it?
yeah... there's a 7805
13v is probably too hot for a 7805
you might want a 6.3v transformer, not 9v
6.3v are common as dirt
okay... I know the 7805 can handle it... if it has a heat sink.
true... I think I have one or two right now in my toolchest
maybe, but there's no need to overdrive it so much. heat is bad.
don't want it blowing up in the middle of a run
A-L-P-H-A, good luck. Sorry I wasn't hugely helpful.
(i might be wrong about this, but) dont ground the parport cable connector to the chassis of the motor drive box
it makes a big ground loop between the box and the computer
yeah, that was isolated.
I was just grounding the transformers.
not the breakout board...
[04:48:53] <A-L-P-H-A> http://126.96.36.199/emc/powersupply4.png
gee you might as well fire up eagle at this point :P
actually, eagle sucks. use kicad
eagle wouldn't fit this.
oh the schematic would.
but definitely not any design... 6x4 is pretty small..
now I'm wondering if the bridge rectifier should be grounded...
why would you do that?
cause on it's base, it's metal... the top (where it has four prongs) is where the connections go.
the base should be screwed to a heatsink
depending on how hard you are gonna push it it might not need _much_ of a sink, but it should have something
it's rated for like 25amps, I'm like using 2-3amps??
then just screw it to the enclosure wall
which of course means it will be grounded, and that is fine
and gawd help ya if you got some funky exptoc bridge where the case isn't islotaed =)
ackhmm... it isn't grounded right now. :/
I'm uber that way
jmkasunich I dont know why I didn't know about heat-sinking a bridge, but glad you said something. I would have never thought to ask.
I picked up a 40A 400V one a couple weeks ago
I'm trying to think of everything to ask.
probably for likr <$10... I'm getting $5.
getting = guessing
I paid $4
I was so close
but I wanted a HUGE bridge (overkill)
that reduces the need for heatsinking
bridge is the least of my worries... I think my transformer is under amped.
if your gonna use the bridge right at its rating, you need to heatsink it wekk
I anticapate changeing the ratings of the PS as needed.
if your using 10% of the rating, just fastening it to a flat sheet of aluminum is probably enough
jmkasunich: Nah, I justed wanted overkill
if I need 40a I'm in BIG trouble =)
moving cars with those motors? :)
hot wire =)
but that doens't need a bridge
electric car, I meant.
Nah, that's jmkasunich job
he's the one that plays with "Da big shit" =)
hello.. anyone still around?
A-L-P-H-A: can you check a thing for me?
A-L-P-H-A: check if you reach robcon.ro or ns1.nextice.com please :)
neither are working
ok.. thx :(
just started to get some answers from there
fenn: probing does work quite nicely in emc2
hey... alex_joni how's your electrical knowledge?
fenn: there is even a file called gridprobe.ngc which does .. a grid probing ;)
alex_joni, there's two methods of calculate the amount of amperage required, one is rating, and the other is via work.
right now... I've got this.
The motors are wired for speed... I have two motors that are labelled 4.2V3.5A, and one 3A and 1oHm.
man... I gotta find that email.
so around 3A @ 4V
well, my transformer says thisn... 100VA 2A 60Hz.
3A @ 4V is 12W
if you multiply that by 3 you are way under 100VA which the transformer can handle
I don't know the formulas... so I'm way safe at the moment?
power = current * voltage
P = I * V.
lets make some simplifications just that you understand things
is there another Power calculation?
say you have 100V mains
V = IR
alex_joni, k... 100V
a 100VA transformer can handle 1A on the 100V mains
on the output you have less voltage
how much current can there be on the output?
* Jymmm raises hand
fenn (No, I menat I knew the answer)
this is highly oversimplified.. but just so that you get a feeling of rough values
A-L-P-H-A: so if your steppers take 3.5A @ 4V I think it's safe :)
If you REALLY want to calculate the transformer needed ---> http://www.mcitransformer.com/i_notes.html
what I'm confused about is... aren't I running these motors at a higher V then the sticker?
scan the WHOLE page
A-L-P-H-A: you are, but for only a tiny amount of time
[06:12:08] <A-L-P-H-A> http://188.8.131.52/emc/powersupply4.png
A-L-P-H-A: Stepper motors are typically ran at 20 to 25 times their rated voltage.
alex_joni, oh... so it's just burts, it'll be okay?
Jymmm: the stepper whitepaper says 3 to 20 or so :)
4 to 20
alex_joni which paper?
url and page# ?
Jymmm, all the gecko stepper drivers white papers say so.
[06:14:35] <alex_joni> http://www.geckodrive.com/photos/Step_motor_basics.pdf
and page #?
and I bet you remember how to search
alex_joni, okay... so if anything... for the future... lets have pretend numbers... 5AMP 2V steppers... 3 axis... so that's about 30VA... So a transformer I have, should be rated for at lesat 30VA (lets give it more, 50VA to be safer)...
I'd say 100 to be safer
okay... so just get my desired secondary voltage... but rated for 100VA.
you arent even going to be using the steppers all the time
fenn: a 100VA transformer is tiny
so you can use less than 30VA, no?
you can .. but it might get hot :)
A-L-P-H-A: if you can (and there's no reason you couldn't) get a toroidal
Mine are running at 30V@ 3A = 90VA * 3 axis = 270VA + PS Loss
very small size, no problems
Jymmm: I doubt your motors are rated 3A @ 30V
the transformer though, says it's rated for 2amps... that's where it's confusing the hell outta me... the gecko white papers, say the sum of the sticker plate (2/3 or 1/3 dep. on wiring) of the amp rating.
alex_joni: and you would be wrong... that's exactly what I'm running them at.
A-L-P-H-A: the transformers are rated on the primary side
alex_joni, availability... don't konw where to get them locally.
Jymmm: I know what you are running them at
Jymmm: I said about rating
alex_joni and I said running.
stepper motors have a small resistance
so if you put 3A through them they only take something like 4V or so
you do drive them with more to get the current flowing faster
and the stepper moving faster
but that's not the amount of power the steppers take
they still take about 12VA not 90VA
but were tlaking the ratings of his power supply, not his motors
Jymmm: bet your transformer isn't rated 300VA
alex_joni Wanna bet $20 USD via PayPal?
just after you convince them to accept .ro for paypal
I'm using a switching PS rated at 10A
Jymmm: 30V 10A ?
that's about 2 or 3 times you need
but it can't hurt.. except it's more expensive
alex_joni Since when do you use a PS for the EXACT rating? You never put that much stress on it.
if you want it to last that is.
I usually overengineer at 1.5
Switching PS are cheap
also far away from perfect
hell you get 5V@30a from a PC PS
at least for driving another switching drive
anyhoo... what secondary voltage does he need (AC) ?
I'm doing 50VAC out... atm.
A-L-P-H-A: thought you don't want to change your PS?
Well two of these would get you 48VAC... http://www.mpja.com/productview.asp?product=7845+TR
alex_joni, don't... rebuilding it, to clean it up.
[06:28:34] <A-L-P-H-A> http://184.108.40.206/emc/powersupply4.png
<-- check that out... that's the current setup/design.
Ok, two of these will get you 48VAC @ 10A
[06:28:44] <Jymmm> http://www.mpja.com/productview.asp?product=7845+TR
wouldn't that be 8A?
Jymmm: I don't see 10A
[06:29:25] <Jymmm> http://www.mpja.com/productview.asp?product=7846+TR
wrong link =)
okay... I see can 48VAC @ 10A
Jymmm: that's actually a 5A transformer ;)
it's not 10A
A-L-P-H-A: No it would be 4A, because you are double the voltage not the current
Jymmm: that's one of the BS of transformer manufacturers
alex_joni it say 10A
Jymmm, I was referring to your first link
they rate them at 12-0-12 10A
A-L-P-H-A so was I
but that means 12-0 can handle 5A, and 0-12 can handle 5A
it's either double the amps, or double the volts... can't be both, unless you do 4.
if you want to use 12-0-12 (24V) you still can only get 5A
A-L-P-H-A you wire the transformers in series to get doubel the voltage and the same current
I know it's a bitch.. but got bitten by it too
alex_joni Yeah, I gotcha ya =)
Jymmm: that's why it's so small
pics are deceiving
check the wire sizes
already closed the apge
alex_joni okay... so, if I'm running 70VDC... but the motors are rated as above (3.5A x 4.2V = 14.7VA) x 3 = 44.1VA, I'm okay... cause the transformer is 100VA?
alex_joni, so why's mariss suggesting 1/3 or 2/3 of the sum of the amp ratings stated on the motors? [just wondering]
lies! all lies! =)
Stepper motors can generate electricity
and do when deaccelerating
so is this why I'm doubling the VA for safety?
alex_joni, thanks a lot for the information and clearification.
alex_joni, much appreciated
Good Afternoon alex_joni
danex: I tested the M110 and M120 you suggested, and it does work as expected
from MDI and from an program in AUTO
In the Head Version?
Then the problem is in my install of the head version?
danex: but you must have the full path : /home/user/emc2/bin/halcmd sets signal true
otherwise it will use the halcmd from the installed version, which can't set the signal
danex: I suggest you update
I have the head version Run in Place is this correct?
danex: but that's from CVS, and you can get a newer version
go to emc2/ and issue "cvs up -dP"
it will get the latest changes automatically
you need to rebuild then
Let me make sure I am correct in the install sequence, First Apt-get the Build dep for Emc2 and Emc2 axis, (this will install EMC2 if not already installed) then cvs the head version
and run in place the head version
same as me :/
I hope this day ends quickly so I can go home and work on the mill
If I can get internet at the plant I will let you know how it goes , they are quite picky about access,
otherwise it will be late today or early tomorrow your time.
have a good afternoon
anonimasu, good luck with your day and your mill project
alex_joni: it should not be necessary to give the full path to halcmd. scripts/emc makes $EMC2_BIN_DIR the first element of the PATH, see line 90.
echo "M101 P$P Q$Q: put your code here"
export | grep " PATH="
type -path halcmd
M101 P-1.000000 Q-1.000000: put your code here
jepler: right.. forgot about that
anyways.. it did work OK for me..
Marc van Doornik's setup is fascinating, he's very clever to set that up
sub 5um precision is really something
actually it proves only one thing :) given enough skills & time you can do basicly anything with emc2
people who say you can't probably lack one or the other or both :)
wow. That is pretty cool. I could see emc replacing the laser controller we have (fanuc with some specialty boards to do duty cycle)
cradek: guess the TP is working great if it can keep it up with sub 5um :)
again - warm fuzzies.
I get an excided euphoria feeling every time I think of what emc is capible of (warm fuzzies)
emc2 - the anti-drug.
so how much of the lead screw tracking could lesw impliment so he would not have to do that funky cam
actually what they Marc implemented as kins existed in emc1 as axis compensations
but it's dormant in emc2
and I'm afraid I don't understand parts of it..
It was over my head.
alter to EMCMOT_COMP struct, for dynamic altering
of compensation points by external processes, e.g., thermal comp
oh - thermal comp also. cool
skunkworks: buzzwords :)
it's basicly a way to add an offset
now that I read it :)
I had asked about that the other day - our big machine used the spindle temp to adjust for expantion of the spindle in the z axcisa
right.. this might be done
I finally figured out why I couldn't run emc2 when I ssh'd into my machine from remote. I knew for some time that it was because the "max locked memory" was set very low (32k) but I finally figured out it is a bug in pam_ssh, the authentication module that allows X and console logins with the SSH passphrase.
jepler: the one you wrote?
[13:54:08] <Jymmm> http://seattlepi.nwsource.com/business/1310AP_Japan_Battery_Recall.html
alex_joni: no, someone else wrote it
[13:58:25] <jepler> http://pam-ssh.sourceforge.net/
oh.. I think I remembered something I read on your blog about pam_ssh
and I assumed it was yours
no, just something I installed and like
and which not many people know about, it seems
heh.. btw blog
my aether doesn't show right in IE :D
not that I'm very concearned about that
alex_joni: there are some old posts on cradek's blog or on mine about that
jepler: right.. I kinda remember, using tables instead of css
but I'd rather not, and screw the IE users
40% of my blog visitors are MSIE
jepler why are you using PAM?
Jymmm: I don't understand what you mean. All modern linux distributions use PAM by default.
That's news to me
Jymmm: bet you use pam_login
thought it was all passwd and shadow passwords now
jepler: yay.. made it show in IE ;)
jepler: had to change the order things get written to the html
jepler: [html <div id=left>][_insert_html _left][html </div>] before the id=main
jepler: in index.cgi around line 657
so there is a guy around that runs emc on his laith with a servo for a spindle? and he switches between it being an axis and a spindle within axis?
skunkworks: what does he do with it?
rotary work with mill spindle?
I don't know - I just heard it in passing.
yes but I forget his name
he was the guy who makes some kind of bead-making accessory I think
I think davidf
is threading in the live cd?
skunkworks: parts of it
skunkworks: I have some numbers about estop speed
Ok - shoot
* skunkworks is just a user though - not a programmer.
Michel1: go on
If I estop brutally from 450mm/mn, there is about 0.02 mm error; if I stop smoothly, it takes 0.082 mm
what kind of machine is this?
(doing a english conversion because I need a perspective 450mm/m=17ipm)
Those numbers are consistent on X & Y. They are for my sherline with shims and backlash tight (ie. lots of dry braking...)
and little inertia.
I just tested the Mcodes again with the path set as described still failing in the head version. I did a complete reinstall of ubuntu. and the emc2-head
danex: I can't imagine how that can happen
Could it be related to the Motenc Drivers?
danex: try routing it to parport
or maybe even use halmeter to look at the signals value
I'll be back in a few , someone just crashed another machine
I suspect though there might be something else overwriting it
So the price to pay for not losing steps on estop is at most .06 mm at full speed (ie. 0.008 seconds :-) )
Michel1: how far would it go if you where running 5000mm/min?
how about on a system with phisical brakes on the motors?
This is the max speed of my axes
Sherline 5410 with 8760 stepper unit (vanilla sherline)
the robots I work with estop pretty damn hard
* anonimasu nods
depending on speed and inertia estop is hard..
Michel1: imagine a big (and when I say big I mean big) portal
something like 600kg+ moving at > 1m/sec
deceling that can take 2 seconds
hitting an estop stops it in under .2 seconds
needless to say it's not a very nice stop
Michel1: I understand that for smallish machines it makes sense to do it your way
Back in the 300/2400 baud days, what where the names of some of the HW/SW compression algos used then?
but I guess you won't be able to convince people to integrate it in emc2 by default
for stops like that I'd use pause
isnt there some more requirements about estop killing off other stuff(industrial wise)
anonimasu: lots of them
to have your machines certified..
but probably area dependent
I have the papers at work I'd read it otherwise..
e.g. EU vs. US
* anonimasu is doing (CE) stuff at work for another machine
canada is pretty crazy too from what I heard
skunkworks: you can test it by yourself: it's a 3 line patch;
well, sauna time :)
I can just compare the abort to the estop.
or do the math but I don't remember what my accelleration is set to.
[17:35:36] <skunkworks> http://www.electronicsam.com/images/KandT/DSCgantry.JPG
skunkworks: toy size :D
no offense ;)
my machine can run over 2000mm/min
skunkworks: :P like I said ;)
skunkworks: in stepgen.c 1) comment the full enable test on line 540. 2) comment addval=0 on line 742
alex_joni: lets see your little toy ;)
let me find a pic :)
not of your welding bots :)
3) add at line 556 if((*stepgen->enable == 0)&&(stepgen->addval))rtapi_print_msg(RTAPI_MSG_ERR, "make_pulses() joint:%d pos_fb:%d deltalim:%d addval:%d\n", n, (int)(*(stepgen->pos_fb)*1000), stepgen->deltalim, stepgen->addval); // MG
alex: For such a big portal, open loop command (ie. just steppers) seems strange!!!
actually I have had it up to 7600mm/sec but that is just scary.
[17:40:15] <alex_joni> http://www.robcon.ro/emc/IM000398s.JPG
I mean 7600mm/min sorry
alex_joni: thay are cool
Michel1: AC servos not steppers
skunkworks: to test, you just have to press manually a limit stop -> not dangerous
skunkworks: it seems small in a picture ,)
Very usefull for PCB engraving ;)
Michel1: adding encoders to your axises solves the problem.
.1mm accuracy :)
Michel1: or just keep your change locally
skunkworks: I understand better your concern now :-)
and because you have servos (closed loop), you are not concerned by lost steps, which is *not* my case
Michel1: I'm more concerned with death.
skunkworks: do you use stepgen or freqgen?
Michel1: I have steppers on that machine.
skunkworks: is that the really large k&t?
anonimasu: that's thesmall one
skunkworks: I know, but I have no need; steppers and openloop works OK for small mills
is the work on the big one comming along?
or is it stalled/sold?
I was busy this summer - I have started accumulating parts though :)
I can't wait to see that doing full 3 axis moves :)
(it was 2.5 originally)
how fast will it go?=
Michel1: the ecoders would not be used for closed loop in the servo sense - they would be just used to keep track of posision
on a stepper machine.
I cant find the real pic of it
anonimasu: i would be happy with 200ipm
[17:56:53] <alex_joni> http://www.electronicsam.com/images/KandT/DSCCurrent.JPG
skunkworks: that's really cool!
thanks (the gantry?)
skunkworks: I think both points of view makes sense. As my patch would change the old behaviour, I suggest adding a variable in the .ini , in the axis section, something like "SMOOTH_ESTOP=1" with a default value of 0
michell: I can see smaller machines might want to use this and making it configurable in the ini would be cool.
maybe bring it up on the devel mailing list and see what people think.
alex_joni I can watch the signal in show. In the head version it momentary in the release it is maitained
skunkworks: OK I'll do that and submit the patch.
skunkworks: Sorry, that's what I was meaning.
if you want a controlled stop I still think using estop is incorrect, since it turns off the machine
can't you get the result you want by not hooking up the .enable inputs of stepgen?
even if you decelerate first, you cannot maintain position with the amps/choppers turned off
or using a component that lengthens the enable pulse enough to permit you to stop
This nickname is owned by someone else
If this is your nickname, type /msg NickServ IDENTIFY <password>
cradek: what did we end up doing about ABC and inverse time feed mode with g38.2?
ABC is allowed
alex_jon1 is now known as alex_joni
not sure about inverse time
If the probe does not trip even after overshooting the programmed point slightly, an error is signalled.
it doesn't do this either, does it?
I think it simply stops?
I think it makes sense to allow inverse time probing, after matt T explained the
point of IT
probing: allow rotary probe, remove minimum length restriction
the gcode aborts? I think it's supposed to
jepler: I guess I do too
jepler: might be another departure from the spec, but that doesn't bother me much
those last things I said are from the CVS log...
oh did I change it to allow already?
cradek: I think you did
so I'm agreeing with myself?
I'm not sure if that's reassuring or not
cradek agrees with cradek_ too
I also wonder if it's an error to start a probe move when the probe is already tripped (it's supposed to be)
seems one of us needs to get a probe
it stops right away
* alex_joni plans to build one :)
[19:20:32] <alex_joni> http://www.indoor.flyer.co.uk/probe.htm
cradek: I thought you had one
I think I do ... somewhere
I don't recall how it mounts; my spindle is all different now
Maybe I'll write a sim-probe
SMOM (machining) maybe
03jepler 07HEAD * 10emc2/docs/src/gcode/main.lyx: document g38.2 probing, including (PROBEOPEN) (PROBECLOSE)
the probe looks like an interesting project
alex_joni: we *did* test (PROBEOPEN) didn't we
[19:25:04] <jepler> http://www.linuxcnc.org/docs/html/gcode/main/#sub:G38.2:-Straight-Probe
jepler: been meaning to ask you..
what's the difference between http://www.linuxcnc.org/docs/html/
alex_joni: in the future, docs/html and docs/*.pdf will refer to the released version, and docs/devel will refer to the development (CVS) version
but presently there is no html documentation corresponding to the released version, so docs/html is a link to devel/html
what is the difference between lpr and cpr in encoders - is there a difference?
I bet they're different by a factor of 4
would lpr be "edges"
I bet l is "lines"
03jepler 07HEAD * 10emc2/docs/src/gcode/main.lyx: inverse time feed is allowed with g38.2
and cpr would be the holes?
c is "counts"
hm still confused. 250 line would be 1000 counts?
or the other way around
I think that's right but I bet each manufacturer specifies these differently
I can picture a probe algorithm kind of like homing - move quickly until the probe triggers, then reverse very slowly until it untriggers and call that the probe point
otherwise you should feed at only one step or encoder count per realtime period to get full resolution
I wonder if probes "untrigger" as accurately as they trigger
if not, you could do two reversals, the last one being forward slowly
Your not talking to your self - I just have nothing to contribute.
cradek: it gets problematic
imagine scanning a few hundred thousands of points
I think the reverse-until-untrigger is no worse
two reversals is definitely worse
yeah.. that's what I'm thinking too
too bad g-code doesn't allow relative moves
sure it does
hmm.. then why not just do this in an O-loop ?
btw.. what's the relative move?
you definitely could probe a grid using a loop
you don't even need relative to do that
cradek: relative to clear the tripped
and go to the next point
the gridprobe is already in cvs
I see, you want relative to the probe trip point, I'm not sure that's what you'd get
[20:07:22] <alex_joni> http://cvs.linuxcnc.org/cvs/emc2/nc_files/gridprobe.ngc
isn't it relative to the current position?
I'm a little fuzzy on how probing works
cradek: how so?
I guess so
hmm.. so programming G91 (incremental mode), and retracting Z a bit might be enough :)
I still wonder why this should be in motion not in the g-code
what happens if you retract z a "bit" then move over and before you reach the posision the probe trips. Would you stop and move up until the probe un-trips then move up the same amount
you don't know how far to retract with your g91 to clear
cradek: probably retracting to the position where it barely doesn't touch is not good for any subsequent moves
you would still need to go up a bit
sure but you can use g91 safely then
* alex_joni goes to a quiet place to think about it :)
neat idea using reletive movement.
still would need a contingency when the reletive movement isn't enough to clear the part for the next move.
that could be done on o-words also I would think
you definitely have to write the probing program with knowledge of what you're probing
psuedo reletive could be done also I would think (I just like the word psuedo)
cradek: to a point - a generic probing routine would work - but if there is edges that are higher than your probe then yes - you have got issues.
* skunkworks thinks alex's "quiet place" is his hot tub
skunkworks YOUR hot tub?
I will have to admit that was a poorly worded sentence.
Nah, just pulling your chain a bit is all =)
I'm used to it :)
that'll be fixed in english 2.0
* sbailard_ is still running the pre-beta release.
Multi-lingual? Hell I can barely speak one.
I noticed a discussion of "Marc van Doornik's setup" above. Is it written up anywhere?
[20:41:15] <skunkworks> http://sourceforge.net/mail/?group_id=6744
Thanks. I'll look it up.
* sbailard_ returns to lurking.
Jymmm, jepler, cradek, skunkworks, sbailard_, alex_joni.