just a repitition that made me think something was approaching, and a green sky over DeKalb
jmkasunich: maybe the 'forearm forklift' is enough for smaller moves
[00:02:33] <tomp> http://foreaadmin.web.aplus.net/igallery/viewer.php?albid=3&stage=1&imgid=51
I prefer to rig moves such that I can stop at any time
you can, it's just tiring
look at how happy those guys are
for a while I had no xvid at all, and the monitor said 'out of range'... this is a useful tool that got me back into range http://xtiming.sourceforge.net/cgi-bin/xtiming.pl
the fridge must be filled with helium or womething
fridge fulla beer?
maybe it was
thx all, gotta go
Yale has a nice library - Manuals/OBSOLETE_PRODUCTS/tons-o-stuff
I asked a question last night that no one knew the answer to... mind if I pick your brain about it?
well, you can try :)
What is the correct step length, step time, etc... values for the keling 4030 driver?
I have no idea whatsoever
Its not written anywhere.
and I keep getting the realtime delay error on emc2
oh. well that problem needs to be fixed before you worry about the step timing
I can make it run, but I think it's probably missing steps, so I'm loth to actually cut any material
I thought that the incorrect step timing was causing the error?
nope. realtime delay means your computer isn't doing so well running realtime tasks (like path generation and step generation)
[00:20:32] <SWPLinux> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting#Unexpected_realtime_delay_check_dmesg_for_details
actually, that wiki page may have older test procedures
have you run the latency test on your computer
yeah, I read that
and I ran the latency test
what kind of numbers did you get
but part of that tutorial requires you to know the step time, length, etc of your drivers
something like 12606
for overall max
the true but usually the cpu is the biggest issue
what is your base period in your hal file
sorry, trying to do this from two computers
where do I look to see that?
in your config directory the *.ini file you are using if I remember correctly
AndyTx: when you ran the latency test.. Did you surf and run glxgears.. (load the computer?) and how long did you run the latency test?
I ran the latency test for about 5 minutes
no I didn't run glxgears
[00:26:01] <SWPLinux> http://www.cnczone.com/forums/showthread.php?p=373160#post373160
post #9 has the timings
I just installed linux on this new system yesterday, I'm a bigtime linux newbie
if thats the thread I'm thinking of, I think thats not for the 4030
it says it is, but I don't know for sure
those are insanely slow timings
oh yeah, I saw that thread
hm, I can't remember if I tried those numbers or not
AndyTx: You need to run the latency test longer.. a good 20min or more while doing other things. Running a few glxgears from a terminal is good also.
theres another cnc zone thread for a different keling driver. I tried those and it didn't work any better
if you have realtime problems, you have real problems ;)
which need to be fixed before you bother tweaking timing
ok... how do I run glxgears?
that realtime error is telling you that the PC can't guarantee that it will keep to the timings you set
open a terminal (Applications -> Accessories -> Terminal) and type `glxgears`
(or type glxg and hit tab twice - it will auto-complete for you)
you can also hit alt-F2 to get an application launcher, which has prettier completion
ok, so run that a few times while the latency test is running and plug in the overall max in the stepconf wizard?
yes, I think so. let me look at what stepconf asks for
but if you get a number that's very high, like >25000-30000, then that computer pobaly can't be used
unless y ou get lucky and it's something that's listed on the wiki troubleshooting page I linked
right, I remember reading about some hardware tweaks that would lower latency
I think the main one is the smi fix, which only works on certain chipsets
or on-board video issues
*shared video memory
oh true - putting an old AGP video card in can be a big help
hey skunkworks, different problem - what is a joint 0 error ?
Ok, stepconf asks for 1: Step time 2:step space 3: Direction hold 4: Direction setup 5: parallel pin port (standard) and 6: latency test results
agh, afk for a few
ok, numbers 1-4 come from that cnczone thread, or from the manufacturers docs :)
on stepping machine setups - it usually mean you don't have enough acceleration headroom in the step gen.
6 gets the higher of the two max jitter numbers in the latency test
yeah, I don't think those numbers are accurate, but I'll try them with this new latency test
I seem to periodically get one when I have two axis doing G0
and the manufacturer's docs are useless
the realtime error is unrelated to the stepper timing though
yes, though it does look like you may be able to call them up and ask
Thanks very much for the help
also, if you open the drive and see what chip is used, that will give you some idea
lol, you ever call keling?
sure, good luck
no, not me
I honestly cannot understand them
though I saw the guy at CNC workshop a couple of years ago
the guy speaks with a very thick chinese accent
brb, thanks for the help. I'll be back to pick your brain some more later
pminmo: raise your STEPGEN_MAXACCEL a few more % and see if that helps
for each axis
not in traj ?
or reduce TRAJ / AXIS_# max accel/vel
if the axes are all the same, then changing TRAJ may be easier (unless the axes override the TRAJ values)
if there are MAX_ACCEL and/or MAX_VELOCITY settings in the AXIS_# sections, then those are the ones that need to change
have number in each acis
what is confusing me those is I see velocities higher than my axis numbers
if the STEPGEN* numbers are "real", then you should reduce the maxaccel and maxvel numbers instead of raising the stepgen limits
on axis dro
for overall vel, that's normal. the displayed vel is the vector sum of the 3 axes
so if each is at its max, the overall may be sqrt(3)*max
and that is when I hit an error, when x and y G0
yep, G0 X+ to G0X- would do it - full decel from full speed forward, but if the stepgen has no headroom, it can never catch up with the command position
so increasing the max_accel in each axis is the thing to test
the stepgen isn't quite a rate generator - it's got a pre-tuned pseudo-PID in it, and it needs to be able to output a little faster rate than the TP will ever ask for
no, decrease max_accel or increase stepgen_maxaccel
similar with the maxvel settings - make stepgen faster relative to the axis limit
check dmesg also, if the stepgens can't go as fast as you want them to, then changing stepgen_* ini settings won't do anything
I don't think there is vel settings anymore for stepgen. (defaults)
(stepgen will print a message if you configure the max step rate to be higher than stepgen sees it can actually generate)
raise stepgen_maxaccel correct?
I haven't asked how old this install is :)
latest live cd
try raising the stepgen_* numbers by a few percent
well looking at it there isn't any stepgen numbers
ok. skunkworks is right about that then :)
ok then how do I satisfy the headroom need?
wait - there is no STEPGEN_MAXACCEL in each axis section?
err - I'm not sure how that's set up now. what is the config you're working with here?
(stepconf-generated, stock, some old config you're updating ...)
It's a modifed config out of stepconf
I guess I can fire it up and look at the hal info for stepgen
be back in a second
stepconf doesn't output a STEPGEN_MAXVEL, but it does output STEPGEN_MAXACCEL
and it's a pretty huge margin - 25% over the axis max accel
there is a maxvel and maxaccel parameter in stepgen hal
you need to look in the INI
the hal file should have something like `setp stepgen.0.maxaccel [AXIS_0]STEPGEN_MAXACCEL`
that reads the value from the ini, where stepconf puts the actual value
apparently the margin in 2.2.5 is ~5%, nt 25
I see it, I was looking in the ini file
instead of the hal file
ok, now I'm confused
so it does look at the ini file to fill in the hal parameter?
the hal parameter is set with the value from the ini file
ok, I'm blind the stepgen value is the next line down in each axis of the ini file....
all caps - hard to read ;)
and I've only got about 5%
make it 10-20
back to tweaking....... thanks
next question, how can I change the machine actual position numbers?
try the G92 g-code in the mdi
G92 x0 y0 z0 will zero the machine for example.
thanks, no gui (axis) equivalent
machine->touch-off->(whichever coordinate system you want to change) I think
touchoff changes relative numbers
there's a shortcut also - check the quick ref
ok, what is it you want to do? home, make this spot "somewhere" ... ?
Define a spot. No "home" on this machine at this point
and I keep getting out of range errors
because of the "actual machine" position
if you have no home switches, then hitting the home key sets the current spot to home (as defined in the ini file) without moving the machine
with no switches, you should probably set the soft limits to about 2x the actual machine limits, so you can home anywhere and still move within the entire envelope
SWPLinux, weren't you the person that told me about wallyplexing i/o pins?
maybe it was DIlbert
I forget the name of the technique
damfino- that's the first time I've seen the term :)
that I can recall
with no home switches, it still makes sense to "home" the machine at a consistent location
put a mark on the carriage and a matching mark on the bed, line them up by jogging, then hit home
I spun a brushless servomotor with my own electronics today, first time
what did you use? uC? FPGA? analog?
I've done it by hand, but that isn't as impressive :)
it was a dSpace controller, about a $5000 pci board
I built the driver portion
heh, I know of dspace - they use it at work
technically, it uses a TI dsp
my opinion is that motor circuits are about all dSpace is good for, others disagree
your control algorithms running on the dspace CPU
what about the output part? how do you go from dspace's logic level signals to power?
right now I'm using IR IRAMS
it's an integrated gate driver/igbt module
I looked at the fairchild devices that Pete uses, but chose IR for no good reason
do you have current feedback?
working on that :)
so this initial test was open loop?
* jmkasunich is very interested
I have several orphan servo motors, several VFD power stages, a 5i20 FPGA board, and a desire to make them spin
mostly I wanted to assure myself that I could get ac out of my driver
I know, I'm tired of spending days looking for drives on ebay
"your driver" = the IRAM module
it is opto-isolated, etc
the IRAMS is isolated, but at $5k, I wasn't going to trust my ability to maintain that isolation
neither would I
it was tempting though
I had problems with some of the junk box optoisolators I had available
which IRAM are you using?
it puts out 10A I think
* jmkasunich is looking at datasheets
the one I'm seeing isn't isolated
the logic is tied to the low side of the DC bus
you may wish to look at the fairchild devices, they are cheaper
the gate driver chips advertise isolation
for the high side maybe
I don't consider it isolated unless both sides are isolated
(I work with VFDs where the DC bus is the rectified AC line, so you can't ground the negative side of the bus)
isolation is a good idea anyway
I'd like to keep in touch with you as you progress on this - I know a bit about induction motor control but permanent magnet control is new to me
a few photos of the X and Y home switches recently added to my machine: http://axis.unpy.net/index.cgi/01217986161
what do you plan to use for current feedback?
I'm using the lem modules
beats the heck out of resistors - isolation again
the dspace already has analog inputs doesn't it?
that's the cheap one, the expensive one has lots more
I need to make a board with serial A/Ds, and do some VHDL to talk to them from the 5i20 FPGA
there is no "cheap" dspace stuff
I know, I spent over $20k on 2 boards
you are either independently wealthy, or doing this for work ;-)
I would never spend my own money on a dSpace, it goes against my instincts
but the $5k board was worth every penny
I want to do the actual drive control in HAL, with a mix of generic blocks and probably a custom module or two, plus the 5i20 FPGA for PWM generation, A/D interfacing, and some hardwired fault logic (overcurrent trip, etc)
yeah, as long as you have the fpga doing the pwm it should work
the dSpace is using some obsolete PowerPC chip
hooked up to an obsolete TI dsp that was made for motor control
probably 1/10 the processing power of a modern PC
the insane thing is that the expensive version of the board is ISA
thats not insane, that is bu$ine$$
the only people buying/using the ISA version are the ones who are already locked in and have no choide
I felt insane buying ti
so they rape 'em hard
but it's hooked up to the pc over a fiber optic link
_that's_ how you isolate!
except for the power supply
yeah, but you can use different supplies for the PC and the power - no noise coupling, etc, and no risk to the PC if something breaks down
the pc turns out to be the least of my worries
I assume since you are doing this for work you won't be open-sourcing your code or algorithms, eh?
its all done in Simulink, would need porting to something else
I can give it to you
thats rather remarkable
I'm doing health monitoring, not drive building
ah, the drive is just a means to an end, not a product
actually, the rc plane guys have released a lot of open loop control software
if you can spend $K on simulink/dspace, etc, why not just buy off-the-shelf servodrives?
because you can't make them screw things up on demand
oh, now I remember talking to you about this before
the health you are monitoring isn't people
it woulda been easier, no doubt
actually, the next iteration drive is going to ruin it's own health
gonna teach it to smoke cigarettes?
there will be smoke involved
biright lights too I bet
hoping for less of that
well, often they go together
also sudden loud noises
well, we intend to find out if there are any indications prior
to those exciting events
there are, for some value of "prior"
the problem is that usually prior is measured in nS or at best uS
don't really care about those
this is cool - mainstream computer store selling an ubuntu system:
[02:11:19] <jmkasunich> http://www.microcenter.com/single_product_results.phtml?product_id=0291873
well I did another latency test...
max jitter was 19340520
3 instances of glxgears and tetris running. 20 minute test
were you doing the test that scrolls numbers in a terminal window, or the fancy GUI one that only shows the maximum numbers?
I just opened the command window and typed latency-test
its still running
was that followed by rows of numbers marching up the screen?
no, I don't know how to end the test w/o closing the window
I really don't know what I'm doing with linux
look at section 2 of this page:
[02:15:52] <jmkasunich> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
but the max jitter hasn't gone up any in 10 minutes or so, so I figure thats probably the max jitter.
cool, I'll scan that
that number has about 3 digits too many
it sounds like the test you are running is the one that tries to be all user friendly by only showing the worst case numbers
ok. I tried ending the test by typing ^C before, but it didn't do anything
the one on that page I linked shows a new set of numbers every second, which includes both the worst, and the numbers for the previous second
ok, I'll try running that one instead
lol, let me get my pen and paper
gotta walk into the next room where the cnc computer is and type it all in the window there
the CNC computer isn't on the network?
gotta fix that ;-)
anyway, I was going someplace with all my rambling
the "lat max" column is the maximum latency in the last 1 second period
the "ovl max" column is the maximum since you started the test - that is what the simple test prints out
I bet you'll find that the "lat max" column is usually a 4 or 5 digit number
but every once in a while you get something huge, 6 or 7 digits
if that happens every 64 seconds, it is a strong indicator that you have an SMI issue - there are workarounds for that, read farther down on that page
if the big numbers happen when you do something specific, it points to whatever you were doing
for example, on one PC with onboard video, dragging a window around on the screen would cause big latency
the solution was a new video card (cheap old ones work)
cool. I hope its an easy fix
I didn't even look inside the case, it might be an onboard video card
brb, gonna run a new test
I gotta run some errands - goodnight and good luck
hm, well the max latency seems to be somewhere between 8000 and 14000
in 20 minutes, there were 3 instances where the maxlat was a little over 19,000,000
and it had "overruns"
this may seem like a dumb question... but how does one stop the latency test? ^C does not seem to do it for me
yay, got my mesa stuff
AndyTx: if you had even one instance of 19,000,000, then the max latency is not 8,000 to 14,000, it is 19,000,000
right but I'm talking about the maxlat of the last second
19000000 seems a tad too high
a ton too high
yeah, as in totally unusable
unless you can identify the cause and fix it
yes, I'm guessing that it had to do with the smi interrupt
did the three instances coincide with anything you were doing?
did they happen at any specific interval, or random?
yes and no
it seemed to run fine for a few minutes
and then every few minutes i'd get 3 spikes
like maybe 40 or 50 seconds apart
then it would run fine again for awhile
you might want to look at the things mentioned on that wiki page, including the SMI one, and try any that don't cost money
did you figure out how to stop the latency test?
you need to be in that window when you hit ctrl-C
yeah, I hit ctrl c
I'm trying to understand how to disable smi interrupt, but I really don't understand half of what I'm reading
ok, you said earlier that it didn't work, just checking
yeah, ^C must be shorthand or something for ctrl c
I was trying to type it
yeah, it is
hmm, looking at that SMI instructions - it is a bit intense for a newbie
and also a bit outdated, since I bet you aren't running 6.06 - if you just installed recently you probably have version 8.04, and those instructions would need modified
I think it might be easier now, but I'm not sure - I'm still running the older version
yeah, its the latest version of ubuntu
er, well the latest that came with emc2
ok, I'd suggest trying in here again tomorrow, or sending a message to the emc-users mailing list describing the problem
I can think of a couple people more likely than me to be able to tell you how to solve SMI on the latest version, but at least one if them is in a very different time zone
do you just have another machine to try?
I have not dealt with this SMI thing yet - I think all my hardware is too old
yeah, I've got my current computer I'm running windows on
heh, ubuntu would be a great improvement ;-)
oh, you don't have to compile it, I bet - just get it loaded
It ran the trial version of mach3 just fine
(although I understand if you want to keep windows on at least one computer)
hm, can I run the latency test from the livecd?
I think so
sorry, if I have realtime trouble with some machine, I toss it on a pile, grab another one from another pile, and continue :-)
yes you can
cradek: normal people don't have piles of computers laying around
on this ubuntu 8 install do you have a rtai_smi module? maybe you just have to load it.
jmkasunich: yeah yeah
where on the cd would it be?
k, let me check
you could just try to load it
in a terminal window, type "sudo insmod /usr/realtime<hit-the-tab-key>/modules/rtai_smi.ko
there doesn't seem to be a usr directory
k, I'll try just loading it in the terminal
where I wrote <hit-the-tab-key>, hit it and it will fill in the full name of the realtime directory, which has some version numbers, etc tacked on the end
yes there is a /usr directory
oh you're looking at the cd? you need to look on the booted system
I think he might be
(he has two computers - the windows one he is talking to us from, and the ubuntu one in the other room with no net access
he's been testing the other one, but asked about latency from live-cd to test the doze one
sorry, I came into the conversation late
theres an rtai_smi.ko
you can load it with "sudo insmod" then
so that should disable smi interrupt?
then run the latency test again and see if you get good results
wow, thanks so much fellas
thank us if it works ;-)
I'll go run another test and be back in 20 minutes to beg more help probably
(tomorrow - bedtime here)
sure, thanks again
whenever I execute m06 (tool change) command the spindle stops and I have to restart the spindle every time in my g-code program. Is there any way to avoid spindle stop?
cnc_engineer: that's pretty standard behavior in cnc controls.
there will be times you don't want to start the spindle: an air stop, or maybe a probe. what if my indicator was in the next pocket:)
why would you ever want to do that?
how do you change a tool without stopping the spindle?
on a commercial control I'd just change the tool macro to look at the spindle direction before it stops, do the swap, and then restart it in the correct direction
dunno how emc handles m-code macros with the crazy O word thing
toastyde1th: trying to change tool with the spindle on is cute.
-- whats your opinion on this lil' thingy for engraving purposes ?
iirc there are some free plans about as well
I've been looking, but i can't find those
atleast, no decent ones
fragalot, basicly I cant see the detail I want, to say I like/dont like
me neither, thats why i asked :p
just trying to find something that 'l get me a running machine for under 400 euro
proves to be a tough challenge
things that I look at are bearings and flimsyness of shafting
well I built from scraps so no need to spend too much
the most expensive part i'm worried about is the steppers + driver
yup thats where most I my money went
I'm just trying to figure out if it's cheaper to make a driver myself, or just to buy one.. :/
atm the hobbyCNC drivers look nice
£30 down the scrap yard, £10 machine from an auction, free motor and vfd, lathe handed down from dad scrapped for headstock and col
so does that router,.. it's fairly simplistic and looks like it can be easilly resized
i'd spend more money on fuel trying to locate a scrapyard that has anything usefull, wouldn't even know where to START looking for auctions that don't involve antique slippers, ..
but drivers and steppers nearly all new £, 30 V PSU was a free one as well
look for factory closure auctions
have, nothing usefull sofar
mainly just pneumatic and hydraulics that went out the door on those. bit over my price range for most things too
£40 for slides for the headstock
y'know, i'll just postpone this untill after my exams, then take it easy, step back and see what i need, and how much it's going to cost me
by then, prices might have dropped on a lot of things (or increased >.>)
about £250-280 for the steppers and drivers iirc
[11:25:43] <fragalot> http://www.hobbycnc.com/products/hobbycnc-ez-driver-board-packages/
this looks promising, it's just shipping that's going to double the price, as per usual
last time i tried to import something... I got a Sun Fire server for free from a buddy of mine.. I ended up paying 600 euro's to customs for it >;>
what country are you in
ah ok, just over the water
it's not far away, but it's a world of difference
fragalot, found another http://cgi.ebay.co.uk/CNC-Router-Plans-for-Computerized-milling-or-cutting_W0QQitemZ170245923941QQcmdZViewItem?hash=item170245923941&_trkparms=39%3A1|66%3A2|65%3A7&_trksid=p3286.c0.m14.l1318
or look at a few, design your own
tried that,.. I ended up with: http://omploader.org/vbWhu http://omploader.org/vbWhv
(cellphone camera, don't shoot me :p) which is just pretty much utter fail
learning the various forces and bending actions would help
when i made that i just said "sod it" to physics
but physics will win
thats why i decided to get plans i know work
only available cheat is inertia if the tool moves fast enough, the material and machine dont get time to move out the way
archivist_ub: I doubt that a home-made device will achieve that speed.
I'll be happy if it does 10"/min
well thats what dremel systems are doing to some extent
archivist_ub: and i'll be even happier if nobody else bids on those plans.. :)
Im not going to
ah, you mean "just rev up the tool more"
I thought you mean if you crash or do sudden stops
anyways, its more fun to do something properly the first time, than it is to make scrap first, and use that scrap to make the 2nd, and waste 'precious' money in the process - even tho it's great for learning purposes
cradek_ is now known as cradek
jepler_ is now known as jepler
depends on what the part is
i work on parts all the time where it's not advantageous to scrap the first part
if it's some cheap lathe part that takes two minutes to make, it's gonna be scrap
but if it's a 800 dollar block of stainless, that's a bit different
you 'ease' into the first part.. ;)
in the HNC manual it says to set the tool offsets .030 short so every cut is oversize
they warn you a few times to NOT move drills and reamers the .030 though
yeah that would be lol
I don't understand the point of the magnet trick, but I think it's interesting that he uses a spring and gets good results on PCBs.
"When you magnetize the tool it becomes polarized to the magnets so that they prevent it from slipping when the tool is spinning."
I suppose it is hard to drive it while having it free to move up and down.
won't work with carbide cutters..
I can't see how you get a uniform cut depth -- why wouldn't vary with the feedrate?
I think he says it does
I think it would be too much futsing
i'm confused to it's purpose
the further you nudge it in, the greater the force of the spring will be etc..
fragalot: when doing trace isolation milling with engraving bits, the depth of the cut is very important because the width of the material cut away is proportional to the depth
I can imagine this working tho for not-that-accurate things, with bulgy surfaces
jepler: and theres always a chance you go too deep & just engrave the PCB into 2 PCB's
fragalot: so everyone would like to find a scheme to cut to the perfect depth while riding over whatever curvature or warping in present in the board's fiberglass layer
lasers measuring distance the whole time. *nod*
I've a question for the experts: what is happening when G53G0Z0 works but G54G0Z0 gets a limits violation. Soft limits are set at +0.2 and -7.5?
G54 is 0.0, 0,0, 0.0
tool length offset?
or G92 offset
set for my min tool length
to my knowledge I don't have a G92 set.
try g49 g92.1 g54 g0 z0
I think it's gotta be one of those two things
OK ... will get back later.... got to go do things outside now...it is going to 100 today.
make sure you check g54 settings again too... g10 l2 p1 z0
does EMC have a distance to go menu
if it does, dave_1, look at the dist to go and see which axis wants to move outside the limit
that will give you a starting point
he is only moving Z so it must be Z
can't go wrong with cradek's suggestions though
my suggestions were easy - there are only so many offsets - so it must be one of them!
offsets are funny things
had a co-worker who is a little slower than average set a job up
then when he went to let it run
sat there and watched a roughing endmill blow .750" deep worth of material, 1" over from where it should have been
through three of four parts on the machine
he had 1" in the work shift =/
hm, maybe stopping it would have been a good idea
the other question was why didn't he spot it before it even began to cut
1" is definately visible by eye?
but that's one of the mysteries of the shop, why he does the things he does
no one knows, they make no sense.
dangerous to have near a machine
he's scrapped runs of 50 parts before
promote to manager
and his machine has the brand new, 3000 dollar vises
two of them
i fear for their safety
i've had a machine go sideways on me, still don't know how it did it
you set the machine up, you make ~300 pieces with it,.. then when it starts with another piece, it does what it's supposed to do.. Then instead of getting the 6mm drillbit it should take, it just switches over to a dept gauge i had put in ages ago for aligning some things, and tried to drill with that
obviously you hit the stop button, it stops,.. but before that it tries to change tools again, and halts halfway the process
long story short, the toolchanger doesn't work anymore,... lol
yeah,.. manual toolchanging ftl lol
you have power drawbars and still you whine about changing tools!?
(wish I had a 'tool release' button)
I wish tools went back to that same position every time
i complain all the time about not having quick change tooling and tool turrents
yeah pretty soon we'll have R8 people here making fun of me
Im using morse 2 collets
or worse, morse collet guys
cradek: if the toolchanger works aswell as it did on that VMC-1060 trust me, you're better off with a manual one.. (altho the power drawbars are a bliss)
if it worked, it kept crashing into things
the motor ended up frying
it would be cool to work on a machine with a tool hive
like a kid in a candy store and all the candy is free
although more recently i have started whining about umbrella changers
i always hit my head on a tool in those things
plus they take a year to change tools
i shouldn't be laughing tho.. I've gashed my head a few times doing that too
i think it's a rite of passage, in some way
or at least i tell myself that in order to avoid feeling like a total moron
it was pointing sideways,.. but i was cleaning the machine out,. so i opened the window on the side, leaned in, scooped some goop out, went to go back to an upright position,.. and thats when it went wrong
and hitting your head on a tool doesn't hurt until you pull away
it's just a solid "thud"
and you think
blood everywhere and a trip to the bathroom later
boss does that here, at least he makes the coffee before I have to take him to hospital
he had a module .6 finger
a what? :p
a gear cutter size
reckon 400 euro would be a reasonable price for a DIY engraver ?
already have the dremel & range of tools for it, so can save on that. can get the rods for the guides cheaper, aswell as the angles & bearings,..
holy crap i only just realized how many parts i can get at like 20% of their normal price
huh being a python lover, this sure looks neat: http://myhdl.jandecaluwe.com/doku.php/start
jepler: indeed it does
why does linux maintain two copypasta buffers
ease of operation
i like it
I DO NOT
i highlight stuff and frantically hit ctrl+v and nothing happens
and have to remember to r+l click
or worse some stuff from like three days ago gets pasted nine times
i hilight, and use middle-mouse button to paste
i don't have middle mouse
ctrl-C also works, but not in terminals, which is the main reason for the "select copies" way of doing things
i like select copy!
ctrl-C & ctrl-V work for me
i am pro-select copy
i just wish select copy and ctrl+c/v were the same bloody thing
[16:20:11] <jepler> http://www.jwz.org/doc/x-cut-and-paste.html
i'm a programmer, I like multiple copy buffers, just incase you overwrite one by accident. & need what you copied. :p
I've used X for so long and other systems so little that I'm completely comfortable with X's way of doing it
i just don't have a middle mouse button
(though I notice that as time goes by, gnome and kde apps get shittier and shittier when it comes to actions on the PRIMARY selection)
[16:26:28] <fragalot> http://www.cncroutersource.com/images/xframec1words.jpg
<-- why does his just shout out "FAIL" (if this was a "place on top & engrave" type, eg. w/o bed)
fragalot: I am sure it's intended to have a bed attached at the front and back like this router:
[16:29:15] <jepler> http://axis.unpy.net/files/01188441458/img_7045-medium.jpg
jepler: the same page had one that didn't have a bed, that's why i typed FAIL, then read a bit further that a bed needed to be added, and why that design was bad,..
jepler: that's the whatshisname isn't it
but if that part were shown you couldn't see the linear bearings
an older model; seems he's changed a lot of stuff in the one currently for sale
shame that he only sells via ebay tho
the price is actually pretty damn good.. $550
shame that that doesn't include the drivers tho
yeah--at a minimum, you have to add stepper driver and spindle
actually on a few, he sells the dremel along with it,.. or it's somebody else, re-selling his work
[16:33:58] <fragalot> http://www.clearchoiceprep.com/zenbot/images/collage.jpg
things he makes with it are stunning tho
I wonder what the SCALE is on the current zenbot.. must be much more than the 1/20 inch of the old style with simple 1/4-20 threaded rod and 70 in/min rapids (SCALE=8000 for half-stepping or 32000 for 1/8-stepping)
er, much less since higher SCALE means fewer steps per inch
Rapids: 800 inches per minute with xylotex driver and 24v power supply. We recommend setting the rapids to 200 inches per minute or lower until you are familiar with the machine. (other bipolar drivers might work equally as well or better, but the machine has been tested with xylotex)
argh saying it wrong still
the new one uses timing belts instead of threaded rods
shipping to the UK is $140,.. I can't even imagine what shipping to belgium would cost
and how much customs would ask for it after
ah, 140 too :p
* fragalot scratches his hair, wondering wether or not to get it
mmph, stepper motors are so much more expensive here, and i won't be able to reproduce that quality build here at home.. *dilemma* :p
I guess i'll just wait for his next one & save up meanwhile
surely $700 is nothing in european money these days
at my job its a lot!!!!
about as much as a loaf of bread or something
jepler: 1.53 conversion rate last time i checked
and as apart from my summer job i have no income, it is a lot.
ah, currently the rate is 1.54191
hm the SCALE must be pretty coarse on the belt-driven zenbot -- a 12-groove pulley for MXL belts is the smallest size I found right away. That gives 0.96 inch / revolution or only 208.333 full steps per inch. Bigger pulleys would give even coarser SCALEs.
what about using smaller microsteps
(not that i know what i'm talking about or anything..)
usually microsteps don't have a lot of positional accuracy
for small apps I keep thinking about the panasonic 40 W ac servo motors with controller Steve had at fest.
hey crakek ... no luck ... also tried leaving just 5220 (at 1.0) and a z offset in the var and that did not help.. really grasping at straws. ;-(
going back to shop to try other things. :-)
a) go with linear bearing system b) go with L angle + rollerskate bearings
* fragalot ponders
there is a homebrew rail system that uses round bar and normal ball races arranged in a triangle made with milled out plywood
mm, normally they use an L-profile for that
but that's the b) option
just figuring out which will be the easiest to use / most cost effective
huh both videos on the zenbot auction are now saying "this video is no longer available" when you play them
they worked when we were talking about it..
weird -- works after reloading the page
lol that is strange..
hah that blue phone video is not what I'd call the most optimized motion
the wood thing is also kinda,.. well.. sub optimal (then again, optimal way of doing it (eg. tracing the curves in a circular motion) would be too hard to program :p
archivist_ub: y'know that ebay page you linked me to?
archivist_ub: that's one of the FREE designs.. JGRO design
some cheap bar steward making a $
[19:04:44] <skunkworks> http://au.youtube.com/watch?v=TWbMTk1rQ0g
I hate ebay's way of saying "once you bid, you go trough or die"
I think you can retract with good reason
yeah, doing so now
as the JGRO is for personal use only, and can not be re-sold to anyone,.. I should be good
yup, maybe a mail to epay as well about invalid auction
would i just send that or "finetune" the message
meh sent that off
archivist_ub: I'll wait reporting this, depends on wether or not he accepts my withdrawal
it appears that ebay's "file a report" isn't designed to let you report violations of a third party's copyright
i have yet another question.
jepler: indeed it doesn't.
mr_jimjames: ask away! :)
i get the error joint 0 on limit switch
i don't have the switches hooked up
i would like to disable that
did you use stepconf?
mr_jimjames: invert the singal
or don't hook it up in hal
mr_jimjames: more like, if it's now a 0, try setting it to 1 instead
on the stg hal file right?
I haven't had the chance to properly play arround with emc yet, so i don't know, .. It's just my programming brain that kicked in, ;)
i don't think there is a 0/1 value
mr_jimjames: in the stg_io.hal file you could comment out the lines like "linksp Xhome => axis.0.home-sw-in".
linksp Xminlim <= stg.in-02
linksp Xminlim => axis.0.neg-lim-sw-in
i don't have the file in front of me
yeah, what jepler says
and so on for the Y and Z axes
sounds like a plan
er, yes, you want the 'lim-sw-in' lines, not the 'home-sw-in' line
* cradek shivers while thinking about a servo machine without limit switches
put "#" at the beginning of those lines
the machine has switches, just not to emc
silly people using inches
that will remove the software connection between the I/O point stg.in-02 and the limit switch logic
well, as they say
the inches, they are real
later when you add a limit switch you'll remove the "#" to uncomment the line
archivist_ub: metric ftw.
if god wanted us to use the metric system, there would have been 10 diciples. ;-)
fragalot: most machines 'over here', especially older ones, are natively inch machines
there will be a total of 6 lines for the X, Y, and Z axes and for the negative and positive switches
(but I know you were trolling)
ok, then, i'll get to it.
if you work with old gear and service/make parts for old systems you need all measuring systems
cradek, lignes :)
archivist_ub: yeah, but all new machines should be metric ;)
* fragalot nods
the only thing still in inches here is plumbing gear
BSP rulz except in america where its NPT
* fragalot wishes he had magic levis
y'know, the type where everytime you reach into your pocket, you find a 20 euro bill
interesting in Britain people think BA is an imperial thread, except its metric based on the swiss thury thread
awesome how all schematics for CNC mills don't include a parts list
I still can't believe archivist_ub made me bid on a free CNC router design
* fragalot thinks archivist_ub is the one that's selling it
thats right,.. I found your secret identity!
now.. reply to my email!
* fragalot grabs a calculator and starts converting measurements
wow, cnczone's registration email looks nasty
the heat is getting to me
i jus dropped a carriage on my finger
i just printed something out, looked at it, and went "holy f*ck it's upside down" .. untill i realized i had to turn the piece of paper arround to "fix" that
hahahaha nice man
right, so you go downstairs to grab a dial caliper, and some bearings with some rods
you come up with the caliper.. realize you didn't bring the rest, go down again, put a basket of clothes in the washing machine as it's idling there anyways, go back up, realize you forgot to take the bearings again,...
this just isn't my day
old age gets to you
or insanity :)
insanity is bringing the bearings and calipers with you, and measuring the clothesbasket
and then washing the bearings
no, I did that last month
wrecked the washing machine
I had some metal shavings & other stuff in my pockets.. didn't empty them... you can guess what it did to the machine
our machine is immune to that
lucky, mine isn't
no, that pours
sigh why do the local hardware stores suck so hard
i'm off to bed before i break something again
we had a machine mystery just a few minutes ago
i was running one of our HMCs, and the parts were good
the next guy gets on it, and a pocket that was supposed to be .709 deep is suddenly .883 deep
i have no idea how this occured
i don't think the tool slipped, because I made like 20 parts with that tool and no problems
a co-worker and I managed to work together and break its predacessor, but i don't think a problem with a new tool usually shows up 20 parts later
* anonimasu nods
* anonimasu has the same issue
it's a new tool
we broke the old one and had to replace it
I cut some parts that were like 0.01mm off.. a week ago..
it was a true "none of us are as dumb as all of us" moment
.01mm, isn't that like .002"
and then I were going to cut some pockets for a shaft..
more like o0,0003937007874015748031496062992126
anyway.. the hole I was like 0.25 too small..
err I cut..
I still have NFC why.
worn tool, forgot some compensation, machine spring
pocket cycle.. tool measured at 10.00mm..
metal was iron and sounded nice while cutting
bump cutter comp up .125mm?
and I took a finish pass to eleminate springiness
like 0.025mm.. last pass
toast: I did bump it up alot, but why it didnt match is something I dont get
why are you taking tenths passes!
.025 is like .001 according to google
that's awfully hard on the tool
yeah.. for a finish cut 10mm doc.. exterior wall
the other cuts was 5mm doc and 10mm woc
the rough cut..
* toast shrug
0.025 millimeters = 0.000984251969 inch
0.025 inch Yes, that's a rough cut.
toast: the chipload I use is 0.06mm/tooth
if that makes more sense
if it was just one piece of cast iron, did it have a chill (hard spot)
no, not cast iron
some kind of hot rolled construction iron.. not cast..
ah that horrid stuff
I think it's like got a strain limit about 300Mpa
it cuts pretty well
I think hardness varies in it, dunno if its specified
it's specified for the stuff I buy
* anonimasu is beat
it's like "wtf off?!, (measure tool), measure workpiece, measure workpiece, check caliper, take another pass, measure)
one of my many hand wounds is red and puffy
go get it checked out
i'll wait until it's obviously in bad shape
Good evening, can anybody help with setting EMCs machine position, not offsets or coodiante systems, machine position
pminmo: do you have home switches or anything like that?
pminmo: if you dont, just hit the home button in axis, or on your keyboard
not at this point and thats what I'm trying to work around
that sets the macine coordinate system to zer0
for me home key just sets the relative position
well, the coordinate system you are working in(touch off) are relative to the home position of your machine
but if the machine position is out of the soft limits range it becomes a non movement problem
so with a set of "home" switches I'm looking for the equivalent to tell the machine where it is
Yes, that's right
excuse me without home wsitches
I dont know how to do that sorry
I dont have any either
It's driving me nuts
it's on my top priority for when I get my machine togther
when you have no home switches, "homing" an axis means making the current motor position equal to the [AXIS]HOME position, typically 0. on my machine before I added home switches, I made X=0 be very nearly the end of travel, and set my soft limits to -.05 to 6.0. My startup procedure is to jog the machine to near that end of travel, then press the home key.
pminmo: there's a switch you can enable to make emc save machine position on exit
If I lose position during a session, I may not be able to jog back to that position because of soft limits. in that case I just exit and restart emc.
(in emc 2.3 we are going to add an "unhome" feature which will lift the soft limits so that you can jog back there)
isnt there a ignore limits button already?
anonimasu: that only overrides limit switches, not soft limits.
it would be easier to just have a "zero" command rather than unhome
jepler: that's kind of odd.
how is it odd
overriding hard limits..
the operators is my knowledgable about thel location of the machine than emc is when it starts up
is more knowledgable
so setting any coordinates is up to the operator
you could probably hack a program togther in python that lets you set it..
in 5 lines or so
soft limits have no value if the operator has no method to tell emc where the machine is absolutely in liu of home switches
pminmo: emc does have this procedure. it is as follows: jog machine to reference location, press home key
that is relative, not absolule
unless I have something setup wrong
home isnt relative it's absolute in terms of machine coordinates
the other coordinate systems end up relative to home
thats would I would think but not how it's working
jepler: if you jog from home and re-home your position the soft limits move too right?
what are you seeing that is different from what you expect?
for example if I jog to what I visually consider home, and hit the home key it zero's the relative coordinates but if that position is outside of the soft limits any gcode won't execute
if i look at the machine coordinates they are something entirely different
i'm using axis btw
i could have something setup wrong
but I can't figure out what or why
pminmo: so your reference location is outside of the soft limits?
not in absolute terms
emc does what's expected of the soft limits..
but the soft limits have to know where the machine is in absolute coordinates
pminmo: when you home you move the soft limits to as they are relative the the machine coordinate system
and that's what's to be expected of them
does the home key pull the .ini home variables?
pminmo: though I'd like that feature to disable soft limits :)
* anonimasu dosent know
if the home key pulled the home location from the .ini file I'd be a happy camper
set the soft limits 2.5x the machine envelope?
toast: yeah, that's another solution to it
then they are of no value
this is a 4' x 2' x 6" wood router
pminmo: why not enable the switch that lets emc remember the position between restarts?
hmmm, what switch
* anonimasu goes to read the docs
it's a ini file setting..
anonimasu is talking about [TRAJ]POSITION_FILE -- see http://linuxcnc.org/docs/html/config_ini_config.html#sub:[TRAJ]-Section
can I define home in the [traj] section?
Ignore [TRAJ]HOME: "This value is only used for machines with nontrivial kinematics. On machines with trivial kinematics this value is ignored."
consider this inifile fragment: http://pastebin.ca/1094277
here we have an axis without a switch, because SEARCH_VEL is unspecified (0)
pressing the HOME key with this axis active will define the current motor position be 0.0 in machine coordinates, because HOME_OFFSET is 0.0.
after that, you will only be able to move to -0.01 and 8.00 because of the soft limits defined by MIN_LIMIT and MAX_LIMIT
I need to go double check, be back in a second
hm, actually if you want to call the reference location something other than 0.0 you have to have both HOME and HOME_OFFSET -- stepconf doesn't write the right thing in this case
the home key starts the axtive axis moving to find a home switch, shift home does nothing, ctl home does nothing the quick ref says one of them does a g52 the other
does al axis home
all axis home
I type baddddddly
OK, if your inifile specifies an [AXIS_#]HOME_SEARCH_VEL then it is set up to home to a switch
remove, comment or set HOME_SEARCH_VEL to 0 to configure it to remove the home switch
I had it setup for when I do have switches
off to try that
see you anonimasu
"home all axes" only does something if you have a valid [AXIS_#]HOME_SEQUENCE defined for all axes