MarkusBec_away is now known as MarkusBec
do you have to addf a constant?
OK it finall all seems to work
MarkusBec is now known as MarkusBec_away
Thanks for the help chaps
do I need to change anything with my emc if I enable parport_pc (using the latest master)
emc config that is...
have to reboot, brb
mozmck: i don't think you need change anything (i didn't)
dgarr: thanks! looks like it works
tomp is now known as tom3p
Valen1 is now known as Valen
* oPless wonders what spec of stepper motors people use on their cncs
depends. the right sise for the job plus a bit
you dont want steppers to lose step
and too beefy, their acceleration is too slow
that's exactly what I'm trying to determine :)
I saw a very nice cnc mill being sold on farnell for £200ish
but I think it's a little small for the size of things I want to do
what do you want to make
well the volume is roughly 42x42x25 of the largest piece I've prototyped
there is a spreadsheet on the wiki about picking steppers
travels being too small is my bugbear
jthornton, remember the name of the page off the top of your head?
no, but I can find it
[12:19:10] <jthornton> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Mechanical_Spreadsheet
[12:19:25] <jthornton> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Stepper_Formulas
a simple search for 'stepper' doesn't reveal anything other than volt... ooh *clicks*
hmph open office?
oPless: there is quite a bit of reading material on the wiki
you on a windows machine?
they have open office for mac
I'll see if google docs has Oo import :)
ahh it does, lovely :)
IPM = inch per minute?
* oPless curses the lack of SI units :)
* oPless assumes TPI is also threads per inch too
this is awesome, thanks jthornton
MarkusBec_away is now known as MarkusBec
Valen1 is now known as Valen
hmm, so ... I guess the next question I have is ... how do you typically set the XY axis to 0,0 and the z-axis to the top of the material (and I guess you manually enter the depth of the material)
bring the machine to x and touch off etc
I set X0 as the left side of the material and Y0 as the back usually
and Z0 as the top of the material
the depth doesn't matter
* BJT_Shop goes back to cleaning up the shop
so you typically manually home the devices. right
homing is a different activity to touch off
homing established the machine limits of movement
ok, perhaps there's a terminology issue in my head that needs some attention :)
archivist, did you get my pm ?
did it make sense? :)
you may want 5 axis or you need to hand polish
yes, 5 axis worries me a little on how to implement
my mill is 5
there's umpteen gantry designs I can use but many of the 5 axis ones (on youtube) seem to be industrial :)
oh really? is it self built?
mine is built from scrap and cheap ish rotary tables
cool, have you any pics or a website about it?
[14:11:54] <archivist> http://www.collection.archivist.info/archive/djcpd/PD/2009/2009_08_15/102CANON/IMG_0246.JPG
crikey, that's a monster :)
some is scrap industrial
its not that big at all
not in terms of scale :)
Very impressive in a frankenstein sense :)
column is an old lathe bed
xy is a tool setting jig
rotaries are Vertex and Soba
the milling motor ... does that have to be anything special? I'm guessing the larger (fsvo large) the motor is contributes some significance to how the mill operates?
mine is a standard 3 phase with a variable frequency drive to it
I'm considering some kind of dremel, or AC drill
well, maybe a DC drill, thinking about it
the amount of information to absorb is giving me one hell of a headache ... haven't felt like this since going to uni :)
its a good learning tool, you also get a bonus of a machine to keep and use
retrofitting an old Denford mill or Boxford is another way
yes, I was looking at a "PROXXON - MF 70" on farnell since it was reasonably cheap
but it's not really suitable for the volume I want
someone suggested building a 5/6 DOF robot arm, with a turntable
which is a possibility I guess, but then the servos will probably have to be beefier
archivist: that sure looks like a nice setup for gear cutting
except for visibility to the work :((
it does a magic helical knurl
cradek, I tuned the steppers on the lathe a lot better today just trying the screw cutting again
cradek, bug/feature in the toolchange/offsets area
offset during M6 ie before you change the tool depends if incoming is longer or shorter
example screw.ngc did not retract enough to allow me to put the next tool on
I mean g76.ngc
docs use g43 h- and code in example has no h-
tomp is now known as tom3p
archivist: set a tool change location at the top of travel
no h means current tool
no h need adding to docs
g43 h is in the user manual
g43 without is not
how does g43 differ from g43 h?
uses the tool you just selected with T=
jthornton, you can see the g76.ngc for an example without, I read that in conjunction with the docs and noticed
ok I'll look at it to see
got it thanks archivist
I got a much nicer thread today
to be more specific it uses the tool that is LOADED which is possibly different from the prepped one (the one selected with Tn)
you mean with the M6?
I was only saying that archivist's summary (the tool you just selected with T) is not precise enough and is not strictly right
I didn't read your doc fix :-)
this is how I get my commit count up :)
pretty sure "the loaded tool" is the right set of words to use
"selected/prepped tool" is wrong words
LOL I used "loaded tool"
I habitually write "Tn M6 G43" all on one line
the difference between prepped and loaded is small or none for that usage
I almost bought a cnc lathe last week but the price went too high :(
I passed on the other one we talked about
that 16c hardinge?
I'll keep looking
you'll find one soon
now that the conduit is in the ground I'll have 200 amp service to my shop instead of a 8-4 extension cord on the ground
going to add on the north end another 25 feet or so soon and move the metal working equipment in there
nap time here :)
if I were to buy a 20mm end mill holder would I probably find that the hole in it is in the range 20.02 - 20.10?
the kind with one setscrew on the side
cradek, they tend to be made to H6 tollerence as that is what cutter should be thats put in there also what you looking for?
what is H6?
I know the hole is bigger than the cutter, but I don't know if 20mm end mill is <20 or the holder is >20
here is what I want to make: http://www.renishaw.com/media/pdf/en/bece1e25657c4ba99dd4d8b03c9fb6ab.pdf
just looking for shortcuts
starting with the center hole nicely centered and the right size would be nice
i have one of them for my probe :) so if u need any unknowns just showt
[17:54:26] <jthornton> http://www.wisetool.com/fit.htm
robh_: I have a CAT40 but need BT40
renishaw has them for the low low price of $400
im not sure id bet cutters are made down on nominal, while holders aimed to be up, so hydrolic holders shrinkfit would work but with out checking it up in cutter and holder catalogues
mines BT40 Cat40 not all too diff but means dont fit in pockets with out change
it works but can't go through the tool changer
u can buy BT40 blanks
yeah - I've started with a cheap shell mill holder before too
all sorts of ways easier than starting from scratch
have u looked on nikken site, they may tell u holder info
well i could measure one for ya lo
yeah I did not find real specs on their site
3 phase power......
Do many of you guys have 3 phase power available in your home shops? If so what additional charges do you pay per month. I'd have to have another service run to the shop but I am seriously thinking about investigating that as I have a pole 50 feet from the shop already and 3 phase in front of my property. I thinking of bringing my bringing my big lathe "home" soon and it has a 30 KW...
...spindle motor, 4 kw servos and a 5 hp hydraulic pump. So I think I will need more juice. ;-)
Dave911: we're residential/rural here and were quoted $20/month added fee for 3 phase. installation cost was $75,000.
what's the justification for the extra monthly fee?
atypical = fee?
that's what you called it
who knows what it would be for anyone else - just an interesting data point
I would happily pay the $20 but not the $75,000
well, the $75k killed the deal I suppose
I just don't see any justification for it
$20 I can live with..... the $75K would be an issue... I'm hoping my install would be cheap since they would only have to hang 3 xfrmers and run about 200 feet of wire. The poles are already there.
couldn't hurt to call and ask
That is what I am thinking. Did the 75K include running 3 phase on poles for a mile to get to your place or ??
yeah it wasn't available close
I bet I could buy all of the materials, wire and transformers to get the power to my shop for $1000, perhaps less. But they seem to have a problem with guys tapping into the 5 KV lines on their own.. ;-)
I don't have any problems with 460/500 volts but that 5KV and up stuff can reach out and get you.. :-( Actually I think they have 13.8 KV out there now as they upgraded the service in the street last year. The overhead wires were raised and the cables must be 5/8" in diameter or so now. So there is plenty of power at the street. I just need a little more of it..
this is cool http://www.linuxfordevices.com/c/a/News/NorhTec-Gecko-Info-Pad/
cradek at work at mo, states H5 for endmill holder so -0 +9 microns
Hi there flks
Has anyone used G76 for coarse threads?
I am wondering if there is some PID tuning needed or something, the carriage velocity is varying rather wildly, and thus not cutting a good thread.
[19:03:02] <andypugh> http://www.youtube.com/watch?v=NDA48nLdbiI
andypugh: thanks, so it would be much less than 20.02
andypugh: what emc version?
also what is your spindle speed, encoder resolution, and servo cycle period?
Sorry, I was searching the wiki
Spindle speed is 200 rpm, encoder resolution is 20ppr and servo thread is running 500,000nS and base thread 25,000nS
(Hmm) I am running encoder.update-counters in the base thread to debounce it, but encoder.capture-position is in the servo thread.
that is correct
what emc version?
ok we had exactly this discussion recently - wonder if we can find it in the archives
the problem is your encoder feedback is way too coarse
you're getting 66 pulses/second and your servo cycle is 2000 times/second
I did that on purpose to suit the p-port...
so for 30 servo cycles it thinks the spindle has not moved. then for one cycle it thinks it did move a lot
that is hard to track obviously
the solution for someone else was to use the interpolated output of the encoder component
However, it seems to "pulse" once per rev, not 20 times per rev
(or about once per rev)
what seems to?
The carriage velocity
that period is because it's obeying the acceleration constraint you've configured
I thought it looked like a controller tuning sort of problem
do you know that your spindle feedback is correct? does motion.spindle-revs increase by 1.0 for each turn of the spindle?
Yes, that part is right
Though I have not checked that it doesn't jump from (say) 0.9 to 1 instantly or something smilar
hm, I don't see the log
well you know you will only be able to get steps of 0.05 because of your 20 ppr
that's why you want to use the interpolated output, which is smooth
so try that
ie wire motion.spindle-revs to encoder.0.position-interpolated?
I think I tried that, but I can't be absolutely sure I saved, reloaded etc.
Is it possible to reload the hal config files without quitting EMC?
yes and no. you should quit and restart emc.
And re-home, and re-load the G-code. Ah well
Let me give it a go
andypugh: in this case you could stop the program, unhook the signal, and connect the new one to position-interpolated
but that easy to screw up, and it won't get saved
so a restart is 'recommended' ;)
Well, I tried it and it made no difference. motion.spindle-revs is perfectly smooth in halscope with either encoder output wired into it
that's not possible since there are only 20 ppr at 200 rpm. you are not looking at the right zoom level, maybe
There is certainly no sign of any steps in the position input corresponding to the carriage speed variations.
Is there a way to log to file? Both outputs have umpteen decimal places in hal-meter, but it is hard to tell if the non-interpoated ones are quantised.
However, I thought that threading was possible with 1ppr in the newer versions?
yes using inteprolated position
So is the answer to disconnect my 20ppr input?
Yeah, it seemed counter-intuitive
andypugh: use halscope
grab a trace of both the output from encoder-interpolated and the axis.2.motor-pos-cmd
then we'll see further
would motion.spindle-revs be better than encoder-interpolated?
Give me 5 mins.
Now to put the screengrab somewhere useful
[20:28:20] <andypugh> http://www.atp.pwp.blueyonder.co.uk/G76.png
cradek: YOu were right about the encoder output being quantised, and -interpolated is much smoother. However they are both very smooth compared to the commanded position
hmm.. something is fishy there
Sorry, my browser crashed.
I didn't see anything that happened after I posted that url
23:31 < alex_joni> hmm.. something is fishy there
can you pastebin your hal file?
err.. better the output from 'halcmd show thread'
I suspect the problem is the encoder function beeing run after the motion controller has run
e.g.: "addf encoder.capture-position servo-thread" instead of
"addf encoder.capture-position servo-thread -1"
[20:36:46] <andypugh> http://www.pastebin.ca/1580708
Though a few things have changed since then
hmm.. that looks correct
but make sure that encoder.capture-position is before motion-controller
not that you changed it in the mean time
In the HAL file?
no, in the thread execution order
the order of the addf's in the HAL file doesn't matter
only the params to the addf
So how do you change the execution order?
you can add a function at the beginning or end of a thread
best is to 'halcmd show thread' and look
line 478 in your pastebin
OK, bear with me. The lathe is in a sperate building, but has a horrid keyboard.
take your time
evening all, has anyone here ever used / interfaced a sony mangescale DRO type linear scale
OK, the actual current hal show is in pastebin
what kind of pastebin? .ca .com .org?
what kind of pastebin? .ca .com .org?
Sorry, got in a bit of a mess opening pastebin in the same window, rather than a new one.
[20:54:19] <andypugh> http://www.pastebin.ca/1581821
yeah, got it
the only thing I see is that debounce should probably run before encoder.update-counters
Considering that the index pulse is a step-input to axis velocity, and it has to catch up in order to cut the thread, I am thinking I have a tuning problem.
Although changing the spindle speed doesn't make a great deal of difference.
Oddly, I do get the occasional smoother move, about twice the frequency..
but you're running steppers..
I don't see what tuning should be involved there
Axis accel is as high as it can be. Lower accell makes the problem go away, but only because the carriage never catches up with where it should be.
So I got a constant-pitch thread, of the wrong pitch..
What I am wondering is how the axis catches up with where it needs to be? It sees the index pulse, and is immediately behind where it needs to be due to axis accel limits. It needs to then catch up, but not overshoot
maybe try to plot the index on the halscope too
see if it's related, it might not be
as it looks to me, the accel/devel happen exactly once every msec
I was meaning the initial "start the cut" index pulse.
that's once every servo-thread
No, it is far slower than that
andypugh: not judging by that picture
200msec / div
so 5 times / second
Indeed, it's actually bizarrely slow in this world of microseconds and nanoseconds
can you try and plot the index pulse too?
I wonder if the dimples are related
at 200 rpm you should get 3.33 indexes/second
Yeah, though eating would be good, it's 10pm :-)
it's midnight here ;)
Once every index, but the index is not visible in the spindle position trace
hmm.. can't say I can think of anything else
I will link everything to a message on the mailing list. Perhaps somebody there has an idea?
how bad is the index without debounce?
Thanks for trying. Hopefully someday I will have answers instead of questions
I'm sure you'll get more useful replies from the mailing list, with all this data already gathered
Bad enough to really mess things up, though I have an RFI filter from eBay for the VFD which might make it redundant
I notice your index is pretty long
usually the indexes I've seen are the same length as the A/B channels
I'm not sure if something isn't happening on the hi/low transition of the index pulse
It's possible. I thought that 50% duty cycle on both tracks was sensible.
the pos-cmd is more or less steady between index pulses
the accel/decel happen during the pulses
maybe cradek drops back in and has some insight
It certainly looks like it is related to the index somehow.
afaik during the threading .. only position needs to be used
and index to sync passes
The odd thing is that I can't even see any aliasing in the position trace, it looks perfectly linear, and that should be all that the trajectory thingy sees
Unless G76 is integrating velocity.... But that would be stupid.
In my day job I would be looking hard at the value of a D-term (I tune car ECUs)
Anyway, time to turn off the lathe and lock the garage, I think.
andypugh: at 22:30 sounds about right
just pastebin the ini for reference too
maybe someone on the list will want it
(for your accel/vel settings)
Too late now, the EMC box is shut down.
But if they ask, I can get them
good night ;)
Silly client, often fails to show what you have typed until it is too late.
alex_joni, are you the person that built a robot-arm based cnc?
jst_home_ is now known as jst_home
is it safe to test hall effect switches with a magnet?
I'm working on connecting servodrives but i have had no way of checking if limits are in place. There is a simple way, rite?
I think proxes are triggered by plain old metal - hold a wrench in front of them or something
huh thanks... i figured theyd be flipped all the time that way.
cradek: you know what I'm working on... I powered X and got a runaway servo. Most likely that means tach wires reversed, rite? f/b through resolver encoder should *not* be able to produce this result...
position loop (resolver) backward will also run away, but emc will stop it right away with a following error
do you have docs? correct phasing for the tach is usually explained somewhere (like "positive axis motion should make a negative tach voltage")
you got full speed runaway, not just creep?
if full speed I bet you're right and it's the tach polarity
zoom. quik lil devil
this machine rocks. good thing im quick too.
what happens when you reverse it?
haven't tried. i wanted to reconnoiter first. good to see I'm starting to understand.
The only other thing I thought it could be was a grounding issue, but i'd expect creep then i think.
what worries me is that this was at power on, not a command. tach shd read 0 forward or reverse. If it is running off based on noise or something I can't predict which way it will turn.
"Phasing the velocity transducer"
apply a small signal of either polarity to the velocity command terminal.
the signal from the tachometer that will eventually be connected to TB1-J should be of opposite polarity.
If it is not, reverse the transducer wires.
amp enable must be miswired for it to run w/o command, rite?
it says amp enable is active low - hook through an opto to the amp's own signal ground to enable
open = disable
you have docs for these servo amps
didn't realize i was missing those until I got into this project. Anything in there I can't work without?
Say, what kind of current draw is seen on the +-12v ins? I went with the supply in the fridge but it's fair to say that mite be overkill. Plus it buzzes.