sorry les_w was AFK
aluminium is a git ...
jmk_sleep is now known as jmkasunich
c02 is fine for it, you just need lots of it .. YAG is good too, its easy to etch
but you do need "significant" power
well, I will be up in chicago helping a freind start a business. He just bought a 50 watt yag.
doing that among other things
so you like leds huh bill?
Me too. I developed the first aircraft led annunciator, in the boeing 777.
did you see that led-as-input-device link that from earlier in the week?
on the latest invention I ran the ir led on the short wavelrngth side of the fermi energy. Much better tempco
no I did not see it
[00:23:25] <bill2or3> http://www.hackaday.com/entry/1234000873073550/
there's a video linked from there, that's worth watching.
ever tried using an led fas a photodiode?
a bit, I was into beam robots for a few months
works pretty well
les_w what material are you wanting to laser?
my noisy co-workers.
they need etching.
If the latest invention goes...I will need to do an automated impedance trim of a fourth order acoustic system
and zirconium doped lead titanate
les_w so you want to laser aluminum?
yag is the baby for that .. q switched
very very quicky
* bill2or3 runs along home
les_w: From what I've learned CO2 wont do anything to metal, yag is probably what you want. Ironically CO2 can etch stone - very nicely I might add =)
well I want to trim transducers to frequency
I ran a yag for years
Jymmm: co2 wont do anything to metal?
les_w: shh ;)
it was q switched optically pumped
co2 is not so good on al
I am trying to help a freind that just bought a 50w doide pumped yag
well, co2 is indeed worse on ally than yag, in terms of absorbtion coefficient ..
I designed 2000w co2 laser cnc a long time ago.
but with ally, its all about melt power
so, usually, co2 wins
getting decent beam quality for cutting above about 1200w is tricky with yag
mainly due to neam delivery issues
I just want to trim an assembly consisting of a pzt wafer and an aluminum resonator to frequency
in a heatbeat
compensated for cutting temp rise and all that
Jymmm: for info, 99% of all laser cut metal is done with co2
I did hit the r.
heatbeat was a good typeo
cellphone crystal trimming is prob a good model to follow
I want the thing to sound like a machine gun
how many of these thing you need to trim?
brrrt... box full.
it would be big
actually that is little
remember 4th order
so to things to trim to match
input is an impedance curve
id talk to the laser marker boys ..
that 50w thing is a laser marker
but remember, "from what ive learnt, co2 wont touch metal" ;)
I used copper, brass, and aluminum as stencils
cut steel ok
co2 on copper is like a mirror
the mirrors in my 1700 watter are copper
emissivity of non ferrous base metals is about .1 at 10.6 micron
the yag cuts copper though.
yeah makes ok mirrors
gold is better ;)
well just thinking ahead
everytime the nozzle gets moved by a bit of sticky up metal, the YAG makes a little cut in it .. and goodbye nozzle
because if it goes (and it looks good) the factory is mine
I like it
so I have an invention with an already established market
If it works, it is sold.
I like that plan
perhapsa a seneca then?
well, a nice twin anyway, rtaher than a single
plane I think
THAT seneca :)
an archer would suit me fine
I guess I can buy one.
get a pair .. one and a spare
full airways, of course
don't want to get a doctor killer...
lots of money...buy a plane beyond your skills and experience
bonanza is the classic doctor killer
just practice a lot. problem solved :)
gotta go get a new barrel on the rifle tomorow
4130 or something?
took it for testing ... previous 13mm group had opened up to 18mm +
I still have not finished my flintlock...too busy
since 21mm is the 9 ring, thats not alot of leeway
so .. nice new stainless barrel
Lothar Walther match grade
I have a friend in the trade :)
we are all doing well . the economy is good.
Prob live off of porrige next year
today, Lexus. tommorow, tin cup.
Alms...Alms for the poor...
today, chevy and the rest in the bank
tomorrow, still chevy
aw john you are to smart and practical
my parents lived thru the depression, and their principles stuck to me
quit that job and start a business. High on the hog today...starve tommorow
My parents did too
the principals did not stick to me i guess
principals just put me in detention.
you know john, If I had real financial security.....I'd go crazy.
thats what drives you?
well...to be honest....
my kids are grown. I am single. I don't owe a cent. I won't either.
I can do wild ideas and stuff.
As long as I DON"T BORROW MONEY!!!
it's a great feeling
agreed about borrowing
I really feel like a freed slave.
again, to be honest, inheriting this old farm helped.
no mortgage ;-)
I pay the electric, gas and water.
and car insurance
all the time (porridge = whatever you happen to be eating)
I am throwing a lot of money into expanding the shop.
why not? Sales are growing exponetially
where are you expanding? near the existing shop?
But soon I may have to commute.
yeah adding to the back
separate rooms for an electonics lab and metalworking
keep the sawdust out
also, I am having my employees do the wood stuff.
I don't need to be there.
I need to be fiddling with impedance analyzers I guess
But soon I will have to commute.
I will only live here on the weekends I guess
I cannot get skilled labor here for what I need to do
so it's atlanta or greenville, sc I guess
i'll find out more when I get to chicago and see what they want me to do.
forthe latest invention I am to be the manufacturer they said.
If it flies
and it's looking good in the lab
I think we have a weiner here.
if not, porridge.
Hi LawrenceG you watching this
hi tfmacz and lawrence
yup... supper time here
rain all weekend here I think
no yard work
the dafodils have bloomed though
tfmacz is a new emc user.... he built a small router from scratch (about 10"x10"x3"... started with 3 steppers from junked printers and it grew from there
he really likes the new ubuntu emc2 combo...
nema 17 power!
He is already cutting 700k gcode files.... actually I think they are close to nema 23 size... running 1/4 step drivers
I have not made any changes other thanthe new spindle...a little busy.
I had to rewire the shop
have you made it spin yet?
wire is more like stiff copper rods...
yea... some of that stuff, you need a conduit bender
no, I haven't. I have to install a bit more #6 and 8
6 and 8? thats skinny stuff
I had to stop working on it a bit. too busy.
I made a mistake underwiring the place
you can never have too much power
I have been helping a buddy put a new 200amp service in his house.... the feed from the meter is some Aluminum direct burial tec about the size of ones arm
I am learning that
got myself educated in high poer current limitong fuses
class t and semiconductor
Its pretty awesum cranking up a stick welder and watching the neighbors lights flash
what are you doing that need those fuses?
morse code practice
The vfd ap engineer reccomended them for the vfd
ok, that makes sense
for plain ole wiring they seemed very strange
less melted metal I guess
faster clearing time, less let-thru
still not fast enough to protect the drive
might make the difference between a glob of molten metal and just replacing an igfet I guess
if they blow the drive is probably toast, but they might limit the explosion
how big of a drive? (HP)
well you would know
little by yor standards...4 kW
modern drives under 10HP (heck, sometimes under 50HP) aren't meant to be repaired if it IGBT blows
so I could have just as well used class H fuses?
the IGBT module (6 IGBTs, 6 freewheel diodes, and often the 6 input rectifier diodes in one block) is soldered to a PC board
I just did what the booksaid.
nothing wrong with that
like you said, they might mean the diff between molten metal/flames, and a simple "bang" and a dead drive
changed all the control wiring for the stg from ttl to 24v too
with any luck at all, you won't have either (bang or flames)
Fire hazard is a big issue
with all the sawdust
remember, I work on them in development, so I see all the bad ones - I don
don't see the 99.5% that run just fine for years
is the drive enclosed?
well, we have real bad power quality here
I put it in a nema enclosure for safety
cut vents for the fan air
nema1 just means you can't get your mitts in
nema 12 is drip-proof
nema 4 is hoseproof
5 I think...I followed the book for size and airflow
5 is outside good I think
before you cut the vents ;-)
anyway popping 'lytics won't blow out of it
thats the idea
assume the drive will blow up, and keep the fire inside
(keep the flammable dust outside too)
got a new dust collector...but the fines blow throughthe filter a bit
not good for the lings
or the electronics bench
hence the new room
in mild weather, might just want the dust collector bags outside under a lean-to or something
let the fines drift away
dunno if the loss of warm/cool air during winter/summer would be a problem
I am trying to be careful here....4 kW in a half inch end mill can do some damage
I take it you now have your high speed spindle?
yeah. A lot of money. Class 7 angular contact bearings. Big. 24,000 rpm.
3 phase 240delta or 380 y.
amazingly, no load noise is 74 dba at one meter at 24,000
more when cutting of course
I expect with care to get 5,000 hrs out of the bearings...I hope so. replacements are $800.
Manufacturer said always do a 9000 rpm warmup for 10 minutes before cutting
forced air cooled.
google is going down the tubes
search for an IC part number or an instrument, and you don't get datasheets or info about the item
you get 100 dealers selling them
I just get serches of serch engines
it is getting a bit cluttered with crap
looking for a manual for an HP 1653B logic analyzer
a coworker said he found a pdf
I'm coming back from chicago with all of corporate advanced technology's tek scopes
out of production June 1994
they are replacing them
what era are they?
some of the newer ones suck
oh about early 90's I guess
prob 8 bit
no idea what models?
They told me onthe phone, but I forgot
things like TDSsomenumber, or just numbers, like 2440?
They don't do math like fft and stuff
we have a lot of TDS scopes at work, they're nice at times, but can be annoying
first or second generation digital tek I guess
response to control inputs is slow
that could be a good thing
I have a 2440, early tek digital
I like it
some of the TDS ones, you turn the time/div knob 4 clicks, and it takes several seconds to get its wits about it
want to here a laugh? I am having to do impedance runs on an old 2 channel analog tek
voltage and current
you need an impedance analyzer
I need something better.
last year we had an auction of unused/rarely used test equipment
the HP impedance analyzer went for about $20
I need to see a graph in real time of real and imaginary
I have to think ahead of a rt computer based system that will move poles in real time to tune resonators
what frequency range? Khz, Mhz, Ghz?
I have to move two poles and two zeros to a specific place
data aq card and some software that runs a laser
hmm, looks like this logic analyzer is also a scope
price was right too
I'm getting more into the thing where labor is everything and tools are cheap I guess
well, at work that is true
at home tho, you are still working, and I'm playing
when playing, cash outlays are bad, time isn't so bad if you are enjoying it
But I turned all my hobbies into work
that's why I took up golf...no hobbies..except maybe flying airplanes
and that's often to a business meeting
Anyway, private pilots fly in the same airspace and are held to the same standards as the bus drivers
not much of a hobby
you are just a robot listening to ATC instructions
wow, agilent is actively trying to make manuals for obsolete stuff available
I am teaching a wood finishing class to a hobby club soon...I hope that will be fun.
I need some agilent stuff.
don't they make impedance analyzers?
well I'm writing phase angles on a legal pad now...
we had a very nice HP (agilent) impedance analyzer, went for pennies about a year ago
it wasn't bleeding edge, but would easily do 100KHz
fine for my stuff
have you checked ebay?
heck, I spent the day making tiny foil hydroformed diapragms, mashing them with tweezers, and watching phase angles
looked at ebay a little
piezo resonators canbe pretty complicated..very non linear. Lots of empirical stuff because it would take 10,000 of nastran fem runs to figure it out
cheper to just use tweezers and see what hapens
I can converge pretty fast with a little physics and a lot of tweaking
it's kind of like tuning a cavity magnetron in your bedroom...
scrape a littlre metal off...watch the poles move.
anyway I can bill for even using ancient equipment.
But if I had more modern stuff I could get done quicker, bill anyway, and goof off!
jmkasunich is now known as jmk-sleep
Jymmm is now known as Jymmmm
Jymmmm is now known as Jymmm
[11:46:03] <giacus> http://vanessexpressions.com/images/table_background_default.gif
hi alex_joni :)
well, started playing around with coolcnc a hour ago
right now am checking out emc2 testing and the "stable" tree
there is no stable yet ;)
thats why ""
but it should be stable enough
my howto says :)
cvs -d:pserver:email@example.com:/cvsroot/emc login
# !! "stable" !!
# cvs -z3 -d:pserver:firstname.lastname@example.org:/cvsroot/emc co emc2
# !! testing !!
cvs -z3 -d:pserver:email@example.com:/cvsroot/emc co -r TESTING -d emc2testing emc2
need to update, the online version is not correct anymore
err.. not really
the HEAD can be broken once in a while
new development is there
is it ok now ?
-r TESTING is what users should use
right now yes, but it might not be tonight
ok, so its right the other way ?
and because CVS is laggin behind, there's no way of telling what right now means
you SHOULD use TESTING
i remove the other
unless there is a STABLE (also a CVS tag, but that's not there yet)
but on the emc wiki and webpages there allways is written " cvs -d:pserver:firstname.lastname@example.org:/cvsroot/emc login"
yes, that needs to change
but I would like to replace the wiki engine, and I'm not very sure if the data can be kept
what wiki do you want to use ?
I thought about mediawiki
ld: no input files
make: *** [../rtlib/hal_vti.o] Error 1
the testing doesnt compile on my system
the other does
that's 2.4 ?
what 2.4 ?
ok, was wondering about the build system
can you look at your emc2/src/Makefile ?
check for this line near the end:
../rtlib/hal_vti.o: $(addprefix objects/rt,$(hal_vti-objs))
right after ../rtlib/hal_stg.o: ...
obj-$(CONFIG_HAL_VTI) += hal_vti.o
hal_vti-objs := hal/drivers/hal_vti.o
no more hal_vti.o
ok, find the ../rtlib/hal_stg.o:
and duplicate it to read hal_vti.o
../rtlib/hal_stg.o: $(addprefix objects/rt,$(hal_stg-objs))
seems TESTING is missing that, HEAD has it (it was added a bit later)
ok, i will not put this in th howto as it will be corrected ?
it already is..
just TESTING doesn't have it..
but I thought 2.4.27 and newer use kbuild
I'll fix TESTING in a minute
hmm, ok, and its ment to be make installed
scripts/emc: line 130: /usr/local/share/emc/tcl/bin/pickconfig.tcl: No such file or directory
ok, ln -s /usr/src/emc2testing /usr/local/share/emc should do
Error in startup script: invalid command name ".main.f1.f2.f3.vscroll"
"$res.vscroll configure -relief flat"
(procedure "SW" line 3)
hmm, ok got some issues :)
i check again
ok, if it's meant to be installed you cannot run it locally
you need to install it in order for it to work
the is no make install target
alternative is to use : ./configure --enable-run-in-place
and then it works locally
there should be
a ok, wrong directory
EMC2 - TESTING-2006-02-19
Error in startup script: invalid command name ".main.f1.f2.f3.vscroll"
ok, still this issue
or what config?
i just run emc
ok, that might be in pickconfig
try emc /home/cncuser/emc2/configs/sim/sim.ini
ok, so pickconfig is the problem
* alex_joni checks out testing
can you put emc in a pastebin?
the top 30 lines or so?
whats that ?
[12:28:56] <alex_joni> http://pastebin.com/
[12:30:58] <cncuser> http://pastebin.com/571744
i could pack together a tarball of my cvs checkout if sf cvs isnt working for you right now
no, it is..
but I can't check in the change to Makefile, because TESTING isn't a branch
I would have to move the tag, and that's not trivial
well.. cvs foo is trivial
but it involves other things aswell
I'll try replicating it.. to see what's wrong
you didn't use any special places for the install directories (e.g. --with-sysconfdir=/etc/ )
ok, but you could .. just that you know;)
yes i could, maybe i will change it someday. for now i go with the defaults
it runs ok over here..
try emc -v -d
and paste the output to pastebin
I only did (./configure && make && sudo make install )
[12:38:36] <cncuser> http://pastebin.com/571753
as im with puppy and uid:0 all the time i did the same without sudo
ok, I think I know
do you have bwidget ?
try running /usr/local/share/emc/tcl/bin/halconfig.tcl
it works with emc2 1 month old cvs checkout
that will test for it
i click around :)
bugger.. my knowledge of tcl is not good enough :(
the thing is
try commenting out line 188 in /usr/local/share/emc/tcl/bin/pickconfig.tcl
i cant grep for in the emc source
its not there
it must come from somewhere else
no, it's in tcl/bin/pickconfig.tcl
but not as .main.f1.f2.f3.vscroll
no, with variables ;)
set s1 [ SW $f2.f3 -scrollbar vertical]
but commenting out the whole line doesn't work, s1 gets used afterthat
try removing the -scrollbar vertical
Error in startup script: invalid command name ".main.f1.f2.f3.vscroll"
dont get it
has no effect
am i editing the wrong version
strange.. did you edit /usr/local/share/emc/tcl/bin/pickconfig.tcl ?
or the one in emc2/tcl/bin/pickconfig.tcl ?
"/usr/local/share/emc/tcl/bin/pickconfig.tcl" line 188 of 293 --64%--
let me see what it says
#set s1 [ SW $f2.f3 -scrollbar vertical]
set s1 [ SW $f2.f3]
try a space before the last ]
set s1 [ SW $f2.f3 ]
I'm really sorry, I have no clue how to fix this tcl code :(
++ /usr/local/share/emc/tcl/bin/pickconfig.tcl '~/emc2/configs:/usr/local/etc/emc2/configs:/usr/local/etc/emc2/sample-configs'
Error in startup script: invalid command name ".main.f1.f2.f3.vscroll"
"$res.vscroll configure -relief flat"
(procedure "SW" line 3)
invoked from within
"SW $f2.f3 "
invoked from within
"set s1 [ SW $f2.f3 ]"
(file "/usr/local/share/emc/tcl/bin/pickconfig.tcl" line 189)
nevermind : thanks for the help
what version of bwidget do you have?
not that I know how you would check that.. but I suspect it might be important
I have /usr/share/bwidget1.7.0
whats the bwidgte thingy again ?
the stuff that makes trees in tcl
mus search it
i just have the axis bwidget
did you place that in /usr/lib ?
there is /usr/lib/bwidget
hmm, i see no version
is bwidget a part of tcl or tk ?
I have nfc ;)
no fucking clue
# This file is part of Unifix BWidget Toolkit
# $Id: tree.tcl,v 1.32 2001/06/22 01:56:50 ericm Exp $
I have 1.48
2001 seems a bit 'old'
ill get a new one
check the one in AXIS
AXIS package contains (contained?) a bwidget dir
i think its working if i install axis, but i want it correct from bottom. so i do a replacement bwidget
yes i know
no, that one doesn't get installed, it's only distributed with it I think
it does get installed
oh, ok :)
rainy day. after chores are done I guess I'll prepare materials for that wood finishing class.
I'll use walnut cnc sign blanks
after students are done i'll just run em through the sander again
axis installs its own copy of bwidget unless you set the environment variable USE_SYSTEM_BWIDGET
jepler: seems pickconfig barfs on older bwidgets
I have nfc why..
on debian systems, we make the emc2-axis package depend on the bwidget package of the proper version, so that's why the option is there
[14:36:36] <alex_joni> http://pastebin.com/571753
maybe old bwidgets don't have ScrolledWindow?
I think so..
alex_joni: then it should be changed to say 'package require BWidget 1.7' instead of just 'package require BWidget'
this wasn't for a deb system
then at least you get an error that indicates it's a version mismatch
can you do that in tcl?
* alex_joni is tcl-clueless
% package require BWidget
% package require BWidget 1.8
version conflict for package "BWidget": have 1.7, need 1.8
03cradek * 10emc2/tcl/bin/pickconfig.tcl: we require bwidget 1.7 for ScrolledWindow
argh.. hate you
was just about to commit
* alex_joni goes back to his sushi ;)
I'll have a bowl of cereal instead
Sausage or bratwurst works for me.
definite carnivore here.
BTW did you get a formula figured out for base_period
base_period=5+min_base_period (where min_base_period = base_period where PC locks up)
So to test we start a realtime process then speed it up until it locks the computer?
why should base_period be the lowest you can achieve on the hardware? Surely there is a point of diminishing returns, and if you use a higher base period you have more performance left over for userspace
that's why I was looking at calculating base_period from maximum step rate
Ah. Right. In the old days, with the binning routine (steppermod)
The fastest you could go was period/2
The next fastest was half that speed.
This made big changes in speed near the rapid.
So what we tended to do was make period as small as possible
and run rapid at only a small percent of possible.
* rayh is not being very clear with this explanation.
It's snowing again 20 inches and counting.
ray - was the issue yestery day with feed overide?
Feed override did make the blending problems obvious.
I was running a 300% feed override (scale) and could get really interesting overshoot.
rayh: I want to move TESTING today so alex and I can work on TP in HEAD. Are there changes you want to get into this week's TESTING?
Lots I need to do but nothing ready.
I've got a follow up image for the overshoot problem. It's not overshoot it's undershoot.
I'll put this image, the ini I ran, and the gcode in the dropbox in a few.
rayh: is this with feed override active again?
It is an always thing. It just behaves a bit different with different feedrates.
[17:06:15] <rayh> http://www.linuxcnc.org/Dropbox/tp-test.tgz
This is fascinating stuff. I wrote on the screenshot so we can compare thoughts.
rayh: I have almost finished the changes alex and I discussed yesterday - let's try your same program when that's done
rayh: well actually I should probably move TESTING first, argh
IMO, the first blending move is in the wrong direction. It's simply an issue of the sign of it.
Right. Whenever you are ready I can up and build a new.
What I was running to create the image was a couple hours old.
rayh: I think this is the expected result after the bugfix, but I'm not sure it's so great
notice I changed the backmost loop to come to a point - I blocked blending at that kind of turn to avoid the double-accel problem
It's blending all right.
yeah. for better or worse.
Is the location of the switch from green to red at 3"
doubt it. the motion_type switch isn't at the midpoint of the blend
That was the thing that had me fooled. I added a rapid line at that point so I could see where it really did it's thing.
(look at the backmost corner
I mean look at the one where it turns 90 degrees to go up
Can you re-run using the gcode in my tarball. That would tell you where.
I think what you've got there is pretty darn good. Should make nice surface finishes at normal cutting speeds.
you must have turned the feed way down for this program
I'm seeing excellent path following
It's settable using the #1000=1 line. Make it what you wish.
[17:28:00] <cradek> http://timeguy.com/cradek-files/emc/overshoot-fix2.png
this is your program as-sent
As you can see the intersecting line just before each switch from g0 to g1 shows that you have done something very different from what was done in the older blender.
yeah there's no reversal or any other evilness
I'm anxious to watch it work here.
I hate to check it in before moving TESTING, but I have to run for a bit
when I get back, I'll move TESTING and then check it in for you to play with.
oops, I have to do the arc-arc case first, I've only done the arc-line and line-line
How bout just an email with the changed file.
I can wait, if I have to...
catch you later.
sorry, it's at least 3 files, let's just wait a bit
I'd also like to finish the arcs first (another dozen lines of code or so)
ok, be back in a while
Sure. I'd like to see it become a part of the new testing.
oh really? heck I could check it in now then
I was assuming it was too experimental for that
We can always revert.
I'll check it in if you do damage control for me :-)
ok here goes.
* rayh goes into damage control mode.
03cradek * 10emc2/src/emc/kinematics/ (tc.c tc.h tp.c): experimental tp fix for ray, incomplete
Now that's finger pointing...
there's an unfixed case - think of two arcs coming together at a point, kind of like the number 3, that will still violate accel constraints
you should be able to turn your STEPGEN_MAXACCEL back down now (you had it twice MAXACCEL)
np. the gcode code tests both.
right. see you.
giacus whats the url to your gallery?
rearanging the office
I see why I was running out of bookshelf space
too many old digikey and mouser catalogs!
which is too bad
let me double check though
ok electronics shelf
then mechanical engineering
machiavelli's the prince and prophecies of nostrodamus...where the heck to those go?
looks like things are coming along with tp - cool. got side tracked - sorry
had to run and get a key chain drive for $29 1G
Wah! The latest TP stuff is a world of difference.
I read back but can't really see tangential anomalies in a backplot
There were a bunch of em that really varied widely with feedrate.
So we havw pauls #ifdef out, No misblend, no overshoot onreversal?
The circle in my test program showed at some speeds and overrides it would bounce 0.1 between.
Sure wish you had a emc2 setup on your machine les_w .
I'll need to change out the computer
I'd like to see you report on what you hear.
This last stuff from alex and chris made a huge difference.
is this the modified old code or simple_tp?
btw I tried to capture some of the arc-arc bounce but just couldn't get it.
I don't have any idea.
graphing position log is where you can see it better
recall all those graphs last year
looks like old, at least at the top.
Sure and halscope would handle that nicely.
arc-arc and line-line were virtually exact stop at most speeds.
but paths were followed
My testing was a rapid of 180 ipm.
The blending looks pretty clean.
More blending than I expected or have seen before.
Don't know what they used for constraints.
Well, in general softer blends are calculated if slower accel is specified
spose I could test by changing accel.
yup. see if accel=0.5 and x1, y1, x0, y0 makes a circle.
kick it off today and see if its up to speed by nightfall.
at 120 that's a 16 inch circle minimum circle size
accel is a bit slow you think?
les_w: it's the old tp, with paul's changes in it, but fixed a bit more
actually it's a good test
cradek and I figured out there were problems on blending reversals
if it can make circles smaller than 16 in diameter, something's wrong
so he put in code to prevent blending on reversals
ok understand alex
because blending that will surely prevent you reaching the commanded pos.
It seems to work well in the tests here. Wish I had a bigger machine to try it with.
what do I need to run axis/emc2?
as far as a box
les_w: a 300+
but 133 might work too
alex - to run 500+ipm?
skunkworks: les is running servo, so yes
What I need to do is put a 1 gig+ box in there with the motenc
so it would be better for us to get servo drives and a card
the stg is not up to the task
for running higher speeds
skunk, for commercial stuff servo is a must.
we are not that commercial yet ;)
for hobby the issue is usually cost I guess.
servo is just plain going to be a couple grand in tronics.
about 1k / axis
unless it's a tiny system
yeah ik/axis...I'll but that
probably more for a serious machine..
we have some servos we would like to put on our big mill - peak 59v 82amps 80ft-lbs
I mean :)
that would replace the hydrolic servos and have almost the same torque
old moog hydra point?
ach "ACCOUNTING PRINCIPLES"
that goes on the boring shelf.
"business Web Page Design"...copyright 1995
moog hydra point?
Oh too cool...bound copy of year 1943 "product engineering" magazine
that's a keeper
neat machine tool ads
scan it and toss it
article for forming plexiglas for the b-17 flying fortress
father has a bunch of bound "machinery" mags from the 1900's
seems ray has nice results on the latest TP
my change is not done yet
good work :)
can one of you plot accel on reversals and make sure I got it right?
cradek: if you can wait ~1 hour
* alex_joni is a bit busy right now
I'll be back later
I am probably not going to be near the machine all weekend.
ray sounds like he is testing all day
I'll dig into arcs and maybe I'll get it right on the first try...
cradek: I got good plots here.
rayh: did you test arc-arc like the number 3?
It looked like what you did somehow affected arcs as well.
I might have seen some near stops there but no reversals or bouncing.
only arc->line with reversal
crap this is hard without drawing a picture
Um... how long have you folks been fighting TP for?
cradek: You're looking for some blends that are not in the overshoot program?
arc into a tangent line?
arc into a tangent line going the opposite direction
okay. let me see what I can cook up.
also line into a tangent arc going the opposite direction
do you see what I mean? sort of like the letter u
Silly me, but wouldn't it be a good it to breka it down, comment it like crazy, then have it reviewed by Fred for clarification ?
also two arcs that meet at a point but the direction reverses, like lowercase greek omega
feed or rapid or mixed.
I've got the mental image.
I'm looking for accel constraint violation like we had recently at the beginning of Chips when reversing on those rapids
those are the cases I'm fixing now
while you are thinking about this - is this where there are issues with backlash comp?
skunkworks: no, that's in another place
skunkworks: I think backlash compensation should be an added segment before the TP but whenever I say that, people smarter than me howl, so I must not know what I'm talking about
looks like it decel/reverse/accel just like it should.
rayh: did you plot accel? I'm pretty sure some cases are wrong...?
no I didn't.
rayh: yesterday alex was seeing some accels of 20 when your max was 10
rayh: those are what I'm working on eliminating
My head is to stuffed up to get around making halscope do it right.
In my humble opinion, backlash compensation rarely works or does any good....in a cnc machine.
That is the same opinion I got from the head of US sales for mitts.
les_w: I agree that's true, but not sure it's relevant unless we are willing to discuss just taking it out.
les_w: before I fixed backlash in my machine, I struggled with finding the "right" backlash setting to get my parts the right sizes. Of course there is no "right" setting.
next shelf....Feynman's "the pleasure of finding things out"....then "Kama Sutra."
well I have no problem with it...as long as it has an "off"
* rayh dislikes gnome!
sorry - should not side track you guys.
rayh: back to your testing, I think line->line, arc->line are fixed, but no other cases yet (arc->arc, line->arc)
rayh: it would be nice if you could test accel on each case so we know if I get them all right
working on the basement. bbl - good luck guys.
Could I get a few hints on displaying accel with halscope?
just been down to see my gunsmith :) ...
rifle meet mr hammer and mr brass-drift
tap the retaining pins out and then hammer seven shaded of crap out of the barrel until it leaves the receiver
Yuk. That seems like a mess.
well, it got the barrel out, which was the general idea
alex_joni: you about?
alex_joni: cloos "robomat" series? any good?
jmk-sleep is now known as jmkasunich
robin_sz: around ?
I am now
robin_sz: still there?
raeady fo carneval
k4ts comes here for few days
we have to shot some nice photos
thinking about camera settings and bad weather we have..
someone sayd to play round ISO settings
without use the flash
03cradek * 10emc2/src/emc/kinematics/ (tc.c tp.c): experimental tp changes for ray, part 2
good score today, found some THK rails at a local lost-freight warehouse...
bill20r3 how much?
Just mail those to "Dumb as a post" Crystal Falls MI 49920
Jymm, $2 each
they were in with the drawer-slides
bill20r3 killer deal!
yeah, big score.
now if I could find trucks for the same price...
this is just the rails.
are the rail profiles a standard? (ie: can I use non-THK trucks on thk rails?)
rayh is now known as rayh-away
anyone that knows a bit about networks?
back from reality
cncuser: any idea what SYN_SENT means?
alex_joni: what do you want to know ?
er.. not what it means, what causes it
hmm, a syn is followed by an ack...
I have a connection to a webserver that gets halted in SYN_SENT
you get nothing baack
yeah, I know the theory.. but I wonder how I can make it timout faster
on the clientside ?
not even sure ;)
launch ethereal and hang strace on the client
this happens once in a while.. when the connection is poor (I think the ACK is lost)
but then nothing will work for the next couple of minutes (10, 15 I think)
except ICMP or other protocols
what do you use for http get ?
I tried a telnet foo.srv 80 last time
but it shouldn't matter