wow. streetview is incredible.
03jmkasunich 07v2_1_branch * 10emc2/configs/m5i20/m5i20_motion.hal: update sample 5i20 config to support index pulses
03jmkasunich 07TRUNK * 10emc2/configs/m5i20/m5i20_motion.hal: update sample 5i20 config to support index pulses
notranc is now known as kc6lbj
* alex_joni wants to go back to bed
Well, your a newlywed... of course you do
what are you up to?
At the moment, watching deadwood. I kicked all the customers out a lil while ago =)
Yep, the bldg is empty (checked all the cameras )
When I got to work today, I was PLANNING on running thru the eagle tutorial, but that never happened!
I got to work a few minutes ago
and I've just started coding
I started a new project with making a ui for emc yesterday
that looks/acts like the commercial controls do
You know these darn customers... I can't get to my servers, my OC-48 is throttling, etc, etc ,etc nag nag whine whine LOL
anonimasu: is it hard coded, or customizable?
hard coded at first..
then I'm going to write a GUI editor..
[07:57:12] <anonimasu> http://imagebin.org/9157
I wouldn't mind one specifially tailored for touch screen
that's the idea
touchscreen + panel
or either one
anonimasu: I hate to criticize so early on, but is that the button you intended to have in the final product (basically)?
or just a placeholder?
it's the button..
make it work in 320x200 resolution
320x200 would allow people to use very cheap LCDs
Jymmm: where was the criticizm?
ds2: are you Nucking Futs?!
ds2: mainly it'll be tailored to my touchscreen for a start..
no i am not nuts
ds2: are those vgc compatible?
320x200 is MGA iirc
kind of... the right VGA cards will drive them (need C sync)
well, for a start I dont care :p
or CGA it's been a while. 640x480 at the MINIMUM
800x600 or 650x480..
that's what im thinking..
atleast until I have it good enough to run..
anonimasu: You know what I've REALLY wanted... FrameBuffer =)
Jymmm: what's wrong with that button?
anonimasu: is it going to change based on the state of the control?
or better yet, fix the text mode UI ;)
the gray box is the icon box.. where you can stuff graphics..
Jymmm: you can stuff porn inside of it..
on/off, hi/lo, disbaled, unavailable, etc
well unavaiable dosent draw stuff there..
anonimasu: only if it delivers the real thing in 20 minutes or less guarnteed =)
anonimasu: like greyed out
Jymmm: no.. thoose are the tab's at the side of the screen(softkeys)
or alarm condition blinking, color change, etc
that's in the status bar on top of the screen..
anonimasu: This is your baby, but I hate tabs personally.
Jymmm: well give me a better way to get panelmount buttons to work..
they are just tabs because they attach next to the screen border(next to the softkeys)
anonimasu: yeah, I hear ya.
[08:09:58] <anonimasu> http://www.hhrobertsmachinery.com/CNC_Machines/CNC_Controls/Heidenhain_CNC_Controls/Heidenhain_530i/itnc530b.jpg
anonimasu: Ok, back to the button for a moment, instead of having a seperat earea for current condition/status, why not incorporate that into the control itself?
because that's not how it's done..
Jymmm: well, some things work like that like single step and such..
they are toggleable..
it doens't mean it has to be like everyone elses, and anything more intuitive is usually better anway.
no, that's right..
did you read the last line?=
they work like that though things like machine status are shown in top of the screen
if estop is triggered..
"Spindle error".. stuff like that
See, I think that current status should be in/near it's control. People will begin to remember where certain things are as they use them, and once they the know the spindle control is top-left corner, then can see it's on/off/fwd/rev/speed is there too, they don't have to look all over to find what they need.
there's a status area for that too..
but why seperate them?
where plc related stuff goes..(some modal codes) coolant and spindle on..
anonimasu: do you use windows?
yes I do..
how is that related?
anonimasu: ok, on the taskbar, you can have the netwpork adapter display an icon when it's connect. you familure with that?
Jymmm: if you like windows axis is what you want..
are you familure with the icon I'm talking about?
ok, that icon changes based upon traffic, half of it blinks on inbound traffic, the other hald blinks on outbound traffic. You can click on the icon to make changes (control) the settings or get further details. The control and the staus are "intrgrated" togher, that's kinda what I'm talking about. here.
Jymmm: are you familiar with any commercial cnc controller?
anonimasu: No, but I am of POS systems, where the users are clueless and the controls HAVE to be as intuitive as possible.
And that's Point Of Sale, not Piece of shit =)
though, sometimes it's hard to tell them apart =)
well, have look at how they older heidenhains work..(that's where im going from)
hang on a sec, let me find something...
the thing is once you are used to them you dont have to think much while doing stuff..
look at tnc-310.. in the documentation area..
[08:26:34] <Jymmm> http://www.artlebedev.com/everything/optimus/demo/
Mind you, this is a kybd, but different
well, have a look in that documentation and we'll continue :)
hit the CAPS LOCK in the demo
and numlock too
or click PHOTOSHOP aqt the top
anonimasu: do you see how intuitive it is and gives status feedback, that's kind what I'm talking about
Jymmm: Well, that's how the buttons do work.
they toggle and change status, based on what options you have avaiable..
anonimasu: but didn't you say the stauts is somewhere else and not part of the control itself?
where the status (estop and stuff shows) never moves..
it's there in auto mdi or program edit mode..
Look at this, pretty plain jane to me http://www.centroidcnc.com/t400.htm
[08:36:31] <Jymmm> http://www.centroidcnc.com/images/m400console1000.jpg
Hell, TCNC almost looks better than that =)
anonimasu: Anyhow, let me know how your coming along =)
centroid looks f-ugly..
though it's useability that matters..
and the logic how you do stuff.
like the fact that you cant do coordinated motion with a machine that isnt homed yet.
I wish I had a touchscreen to play with.
Jymmm: it's thoose little things
well, it's the behind the scenes stuf fthat would take care of all that for you.
Heh $1600 for that kybd
well, emc dosent right now
as I said have a read in thoose manuals..
shouldn't there be lots of emc descusions going on?
yeah im waiting for that too
this plc does positioning
like runnin g-code
with interpolator/planner and stuff..
sounds usefull, where to steal one from....
[13:54:13] <anonimasu> http://www.3s-software.com
though that's just the runtime
though you need their cnc library :(
sounds a bit $$$$ ££££
what axes xyzabc xyzw xz xyuv xyzrpw
haha just 2½
what systems of axes are useful to have? any that aren't on that list ^^
* archivist hugs emc for being in the correct price range
I just noted it..
that's a cheap plc system..
though the hardware is expensive..
200eur for the runtime(interpreter and *..)
archivist: it's probably for cutting paper or something
I just thought it was cool :)
or you can implement your own motion planner and stuff;)
jepler: I think pretty much any combination. for instance xyzaw might be a mill with knee and rotary table. a 5 axis mill might be xyzbcw
a lathe with tailstock might be xzw
it's tempting to say we should just add another pseudocartesian triple uvw that works like abc
(at least I assume that's what you're working on)
cradek: yeah something like that
I wouldn't bother with rpw, it's a can of worms because those letters are already used
but uvw are not used now, it should be straightforward (but not necessarily simple)
what I'm working on would let you have xyzuvw but not xyzabcuvw
i.e., the total number of axes remains fixed at 6
and I think that mixing linear and rotary within a triple (e.g., bcw) would also lead to tears
many machines with quills have z+w, it would mean you couldn't use any rotary on those machines
because the cartesian-like distance in 'bcw' makes little sense
even less than treating 'abc' as cartesian-like
I understand a 5 axis machine is generally programmed in inverse time mode
that would probably work fine
it would probably also work reasonbly well if you never commanded bc and w at the same time
is this the configuration helper?
sounds like interp work first :P
how did the science project go?
skunkworks: no, I haven't actually done any work on that yet
jepler: your new pluto firmware works as is. Have not tried any of the fancy stuff yet.. :)
skunkworks: OK, that's a promising sign
skunkworks: didn't you see the great tuning I did?
[14:21:57] <alex_joni> http://18.104.22.168/~juve/iozsi/jog-step1.png
that doesn't look very good to me
skunkworks: joke aside, thanks to jmk it started moving a bit, which is more than I was expecting
jepler: indeed it's not, but it's a great achievment as it is :D
too weak motors, too little reduction, too little encoder resolution /mm, etc
alex_joni: cool. - now all I ask for is a video ;)
it stayed within a few mm... could be worse
skunkworks: there are some more pics in that dir
cradek: 3.56 counts / mm
about 37g wire force to combat friction
and motors have about 42g rated force
as jmk said.. it's a couple magnitudes away from useable
motor starts moving at +/- 24% pwm, between those it just sits..
I like that - may have to use that quote in my day to day converstaions
feel free :P
I ended up with something like P=20, I=200, D=0
you need to use a transfer function to get rid of the deadband .. http://emergent.unpy.net/files/sandbox/linear8.comp
something like: n=4 x0=-100 y0=-100 x1=0 y1=-25 x2=0 y2=25 x3=100 y1=100
maybe with another term right around 0 so you can actually get a zero output
or just put x1=-1 x2=1
pid output => linear8 => pwmgen input
(note: I haven't tried this myself)
jepler: yeah, I was thinking about something like that
alex_joni: is it going to work for what he needs it for?
skunkworks: yup, I bet that would improve things..
but in the mean time it's a tad bit late :P
I ment the machine in general
it's more than enough
even showing that encoders work, pid and pwm and all
I hope this was for collage and not high school... I feel I am falling behind as it is.
:) see what I mean?
more transformer load testing: http://www.anderswallin.net/2007/06/1800-w-80-v-psu-for-servos/
still not sure it works right...
a resistor rated at 2000W? that must be quite the resistor.
yep, it's about 20cm long and 2x4cm in cross section with heatsink-like fins
the transformer regulation could be a bit suspect
testing only the AC side shows a smooth drop in voltage from just under 60 to 55V at 6.5A load
I would be happy if the DC load tests would be straight lines like this with the same(ish) slope
bigger caps but the pulse current is way above the transformers happy state
what do you mean by that? pulse when charging the caps for the first time?
no each cycle
awallin: even for simple PWM servos without current feedback, you can get good tuning even from a sagging DC supply -- cradek had a similar experience to yours
jepler: that is comforting to hear! the guy who commented on my page thought that the variable (decreasing) gain with load would be a problem
yes but what percentage of time are all motors pulling full load
caps: Mariss' guide on geckodrive.com says C= 80000*I/V and with 20A and 80V the formula gives 20000uF. I have double that.
archivist: that percentage is very small
then dont worry
let the caps handle it
awallin: you will have to select maximum speed & accel based on the worst case supply voltage..
yes. but a G0 move on one axis while another is close to stalled is also pretty uncommon
* skunkworks has a 6kva 3phase transformer to use for the servo drive
I am sure to make a lot of smoke with that..
getting it wrong at that size will be entertaining
I rule: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250135380771
skunkworks: do you have a metal box you hide inside when testing?
jepler: I may actaully put the electronics in a box.. Maybe. A bit odd for me though
explosions contained in a metal box can be bigger
JonE said when he was working on inverters - he would crawl under his desk - turn it on.. if it didn't go boom - he would slowly get out from under the desk
most fun I had was telling a student to screwdriver the terminals of an audio power amp.... it went bang
it was supposed to short circuit proof. it nneded testing
we used to stick small electrolytic caps across the switched 110v outlets for the next class to find.
has anyone waded through the entire gpl3 yet?
I will ask - gp13? google isn't being much help.
noo but seen a few comments on groklaw
it said somewhere that you can't mix gpl2 code with gpl3 code in the same project. I stopped reading about then...
skunkworks: you need a better font
awallin: that's why they've been promoting "gpl 2 or later" licensing for a long time
oops - that makes more sense.. :)
Dallur: did you recieve your new THC?
hey Martin_Lundstrom: I got the new THC, Installed it in the case but I'm still waiting for answers from the designer about some options and HF things
bill2or3: You just paid $82 for the paper wrapping the THK bearings come in, not the actual bearings themselves! ;)
bill2or3: There's no actual bearing pictured in that auction =)
yeah, I just collect the wrappers. :-)
bill2or3: I guess you missed the fine 2pt print... "What you see is what you get"! WooHoo, is that like a rubber band collection?
skunkworks: 6KVA huh? I was wiring up a panel once, and one tiny single straind fron 18ga control wire was touching chassis and I blew up a 2KVA xfmr. The shop stunk for 2 weeks after that! All the resin smell.
Case and point... twist those wires securely before inserting into a blind terminal strip =)
Dallur: cool, I tried the config some more
Dallur: did you experience a slight dip when loading a gcode file?
should I code or should I sleep
code in your sleep
So anyway. This is what we know. The steady state error at 100ipm is aprox .0008. This coincidentally is also the distance the axis will move in .5ms - my servo period. Now I have heard that ff1 should take care of this. but as you see from the pictures - ff1 as ff2 seems to only affect accel/decel. FF0 set at around -15 is what seems to effect the 100ipm steady state error. From what I understand - that is wrong. http://
And it is my guess - (looking at the pid only tune) that the error is actually ahead of commanded positon.
(and also that I need a negative ff0 to correct it.
skunkworks: how does your drives work?
? it is just an pwm driven h-bridge
err do they work well..
so far.. :)
have not burned one up - but only running at the most 10A so far.
they don't get warm at that. (which is good)
PID error = command - feedback. so if the movement is in a positive direction, a positive error means that the position lags behind the command.
Martin_Lundstrom: yes I did, heading home now, will log on at home
jepler: ok. (I need to scope cmd and fb - which might help me more.)
(I am pretty sure the movement was in the positive direction now that I think about it)
(up arrow and the numbers get more positive ;))
sonic youth - catholic block
hum if I write "G0 X3/4" using the non-ascii "3/4" character, axis gives an error on the terminal
UnicodeEncodeError: 'ascii' codec can't encode character u'\xbe' in position 0: ordinal not in range(128)
that's terrible, it should be supported
how exactly did you decide to try that?
if you use a "wrong" letter in gcode it tells you "bad character used"; I was changing the message to show *what* character it was
lerman was once talking about line and column numbers for errors - I think that would be smart
hm, line numbers is nice
then the gui could jazz it up however it wanted
that gives you a good pointer
anonimasu: we have line numbers already
I never write bad gcode so I wouldnt know ;)
sure, me too
I've only heard that
emc/task/emctask.cc 309: interp_error: Bad character 'a' used
^^^ error given by the interpreter when running an XYZ-only machine
on valid gcode like G0A0?
I need a shave.
yeah I typed G0A90 or somethin
emc/task/emctask.cc 309: interp_error: Bad character 'y' used
you still have to have the unused "axis 1" in the inifile, because making the first two axes be 'xz' changes the meaning of g17
jepler: you are doing this on a live machine to make music, right? ;)
not sure what this has to do with music
jepler: I don't follow about g17
cradek: maybe I'm going about this in the wrong way
I'm basically changing the interpreter so that you can choose what letter invokes the routine that reads the value known as "X" all through the rest of the interpreter and emc, and so on for Y, Z, A, B, and C
I also considered that for a simple way to accept UVW (call read_a, read_b, read_c instead)
so if I make the axes be named "X", "Z", and the rest without names, then the gcode "G3 X0Z0 Rwhatever" is an arc in the plane known as G17
maybe you "just" have to fix g17, g18, g19 too? they should sometimes be disallowed anyway
and ... and ... ijk?
AXES = 3
COORDINATES = X - Z
by the time you have write "AXES=5" "COORDINATES=XY-UV" you start to feel like maybe you're going about it the wrong way..
is AXES= needed?
you mean, could I get rid of it and use the value I get from parsing COORDINATES? Yeah, probably..
maybe accept any of COORDINATES = [X-][Y-][Z-][AU-][BV-][CW-]
(always require 6?)
I think I want 'XYZW' to be legal, not force 'XYZ--W'
but maybe I'm wrong to want that
random G codes can turn a machine into a nice percussion instrument ;)
cradek: are you implying A/U B/V and C/W are mutually exclusive?
ds2: read back
ds2: in the context of the hack I'm pursuing to enable UVW axes, yes
with my hack, the number of axes stays fixed at 6 but there would be flexibility in naming them
it's not an ultimate or perfect solution
in fact I may end up so ashamed of it that it never sees the light of day
(cradek has been kind to point out problems with my approach in the other layers of emc2)
sorry / you're welcome
does anyone have a link to some 4-axis foam cutter gcode? I have googled unsuccessfully, finding mostly screenshots of foam cutter cam software
I wonder specifically whether inverse time feed mode is used, and whether arcs are possible on the second linear axes (UV)
jepler: someone (dave-e?) posted some on the user list way back when
cradek: I'll look
I would think it would also be similar to wire edm code.
>> N012 G03 X-011934 Y+008356 I-011934 J-004344 Q+000000G41
I'm looking for advice on fitting a VFD to my turret mill spindle
dave's code uses X, Y, I, J, Q, R
... and no decimal points
so it's XY +- IJ and UV +- QR? or are UV not specified for some reason?
there's no UV
Guest168: what do you have for hardware now? Drives/interface.
regarding motor control, what is scalar, vector, dtc?
oh maybe QR are just the offsets (UV in our language)
Guest168: I bet you should come back tonight when jmkasunich is here
(tonight = 4 hours or so)
cradek: thanks, but I hope I'll be in bed by then;-)
* cradek looks around for another motor control expert
I don't want to get too tech, just basic advice for now
cradek: the other interesting thing about this code is that it keeps issuging G41/G42 on nearby lines .. I wonder if those mean something other than radius comp in this dialect
I remember it was a bit otherworldly
have a url?
[20:18:34] <jepler> http://firstname.lastname@example.org/msg01575.html
skunkworks: my m5i20 is running my servos and I've still got variable belt
(hm steve stallings sure seems to think foam cutting people will want independent arcs on both sets of axes)
I can see (from the pic) that dave-e's part doesn't require that
I can't fathom the reason for the G41/G42 stuff in this code. I bet you're right and it's unrelated to the normal 41/42
page 14 starts your journey
page 19 defines the differnt drive types.
skunkworks: many thanks.
Guest168: how many hp is this? it's a 3 phase ac motor I assume?
cradek: yes, I guess 3hp
what power do you have to run it on?
oh ok, that's good, that's pretty big to run on 1ph
cradek: I'm a bit worried about torque loss and wondering if I should go to 5hp?
I've been pointed to these but haven't tried one yet: http://web6.automationdirect.com/adc/Shopping/Catalog/AC_Drives_-z-_Motors/GS2_(115_-z-_230_-z-_460_-z-_575_VAC_V-z-Hz_Control)
I'm going to get the GS2-22P0 for a 2hp bridgeport mill spindle (I have 1ph power)
cradek: pricing looks good, I'll have to pay that in £sterling here!
yeah they definitely seem cheap, an expert friend says they're good, he uses them a lot
cradek: can you tell me how an oriented stop is done?
I don't think you can do that very easily with a vfd and ac motor
cradek: so what about toolchangers and orienting boring bars on m/c centres?
are they using dc drives?
I know there are ac drives that will do position - they have used some here. but I don't know enough about them.
vfd - not ac servo.
any idea what the lowest speed a vfd will run?
there is lowest speed and then there is how much torque
yes, this is my question really
for the occasions I run the back gear for speeds below 500rpm
cradek: how will emc communicate with that VFD?
there is torque mutiplication, hence, if I use a vfd should I go one size up.
jepler: I think they all take an analog voltage, so you can generate it any old way
jepler: but, I haven't looked in detail
with regard to orientation, I kind of thought if you can run the speed low enough,
then hit the air brake, you might get get it within 5deg?
the mazak at cnc workshop uses an encoder on the spindle and a pid loop going to the spindle drive in order to orient
1 user-configurable, 0 to 10VDC (input impedance 10k ) or 0 to 20mA / 4 to 20mA (input impedance 250), 10 bit
it is a little finicky but it does eventually orient to +- 1 degree
like you say, you could also go slow and try to stop at the right place.
what spindle drive does the mazak use?
it's just an AC motor but I don't know details about the drive except it takes +-10v for rpm
I'm guessing the pid loop is just for z axis following?
no, position pid on the spindle specifically for orient
but, that same encoder is used for tapping too
with emc, you can pretty much do whatever scheme you want to get it oriented
but, that's pretty advanced configuration
you can look at the config of the mazak: it's demo_mazak in our cvs
it uses classicladder (EMC's internal PLC) for a lot of the advanced stuff.
ah rigid tapping, maybe one day...
rigid tapping is very new in EMC, we demonstrated it this year (last month) at workshop
you can see the spindle orient for the tool changes too.
THis is a closeup http://youtube.com/watch?v=JCEwlfJj__A
I wish we had done the videos of the aluminum instead of plastic
I'm impressed ...dreaming of mazak with emc...
yeah it's a nice machine.
that spindle is being inched around for alignment, rather than stopped to a point
yes, that's the pid loop hunting
that's all under the vfd control
that's for me then!
it has no mechanical method of orienting
(I'm not sure it even has a brake)
the backlash in the gearbox makes it a little uncertain, otherwise it would be pretty fast I think
but you could try both methods
because your drive will always have a bit of offset, it might be hard to reliably go "very slow"
pid (with a lot of I term) automatically compensates for that drift
okay, I'm going to do some reading
come back any time
thank you very much to you all for your help
good night all
Dallur: are you still awake?
quick simple question
touch off doesn't seem to have any effect at all
all coordinate axes are unchanged
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-07-02.txt
not many here eh
ler_hydra: what is your coordiate system st to? g54?
ler_hydra: are you running a sample config?
skunkworks, ah, none at the moment
the axis spash screen ;)
from /etc/emc2/sample_configs, or from your home dir?
(if the var file is not writable, it won't work)
wait, network hosted
that would explain some things
thats a bit cryptic imo
shouldnt it complain rather than fail silently?
but now it is faaar too late for me
1 AM, at school, work tomorrow at 8 :/
oh for the days of college! :D
guys i figured out why rigid tapping is "better" than floating tapping
though nobody probably cars
rigid tapping tells you quicker if you taps are dull by snapping cleanly inside the hole with no room for any extraction? ;)
super-high preformance taps
are not self-starting
coated form taps with weird geometry
and if you try using them in a floating tap holder, you will break them and screw the hole up
prevents the next guy from reusing the tap and messing up his work ;)
well you can kind of do it in a floating tap but it is hard
also, these taps can run however fast your machine will tap
like, you can tap with these at 10k rpm
are these forming taps or cutting taps?
if your machine axes can synch
nice but at 10K, isn't friction going to be a little bit of a problem?
not with the lubricant these taps like
and they're coated
just like endmills
but endmills don't move metal around like a form tap does
the specifics of it aside, that's what they're using rigid tapping for in industry
non self starting, high performance form taps
i don't know the friction issues or heat generated
as i've never done it
what exactly does it mean by non self starting?
in a floating tap holder, it will compress the holder
and not go in
it will not start in the hole
then, once the holder bottoms out
it will start
and break if it needs any more rear room
so it doesn't have the 1-2 turn taper?
toastydeath: since you actually work in the industry, how often are forming taps used?
they're pretty much all that's used in a lot of higher end shops
like, mom and pop shops still use cutting taps
but anything in major production or in aerospace, serious manufacturing
is using form taps
they're sensitive and not at all ideal for hand tapping
I would guess if you stop - you have a broken tap.
Hmmm I find hand tapping form taps to be just fine
plus it must be a pain in the a$$ to start
which is why I wonder why everyone isn't using form taps
* skunkworks hasn't tried
form taps are the best thing since sliced bread... no chips to clean out and don't have to reverse
ds2: there are form taps, and then there are these taps
which are non self starting, remember
it's very hard to get them going by hand
i'm just talking about regular ones
now if they would just make AFFORDABLE form dies...
use a die head!
i dunno man, a die head is kind of like an investment
because you only need one set of either form teeth or cutting teeth
resharpenable, and fits any size
like a house, forklift, backhoe, etc? ;)
nah, die heads are what, 300-500 used?
and unlike normal dies, they will fit a whole range of sizes and threads
but add in the roller things and it is around $1K+
uh, roller things?
the thing that actually forms the threads
if you buy used, they usually come with the whole range
usually 3 or 6 of those per thread size
i don't think many people in here are too opposed to buying good gear on ebay
i refuse to do business via ebay so :P
well, then i suggest you purchase yourself a good die set!
i am reading up on suturing
i am thinking about practicing one handed
in case i ever hurt myself at work and want to close a wound
this is probably a bad idea