jmk_dinner is now known as jmkasunich
Igor! It's a LATHE!
Wow.. The compound and crossfeed on this thing moves like glass.. Not bad for a machine built in '47. :)
And the ways are just as smooth. :)
it's called quality iron
And good care/storage for 99% of it's life.
It only got wet/weathered in the last 6 months when they moved it to storage in a leaky garage..
* Phydbleep goes back to scraping dust, grease and dead spiders out of it.
Everything works better when you scrape out the fossilized cosmoline. :)
03jmkasunich 07kbuild-0-1 * 10emc2/src/emc/ (8 files in 4 dirs): merged a fix from the trunk
had to look up cosmoline.
WW-II surplus gun/metal preservative.
I need to figure out how much material my mill can remove...
feed rate, doc, rpm, and stuff.
Hm...here's something annoying
I want to use Axis for a friend of mine
The job is doing inlays in wood
Max width = 0.500"
height? or width??
Working with a 0.313" dia cutter
So, things are small
Now, the machine needs to run at 18 IPM
Now, when I choose to run Axis with these programs, the update interval makes a mess of the live plot
I know my box may be a bit slow - 600 MHz Duron with a GeForce2 MX card
Grrrr...Am I going to have to get a > 1 GHz CPU to make Axis run better?
I've tried changing the .ini file parameters for polling intervals....no dice.
it can probably be tweaked...
does axis require 3d acceleration?
I don't know...I thought it did
I can't remember offhand how to check if it is
did you configure 3d acceleration?
does the nvidia binary driver work with adeos kernels?
Yes, it does...I get the Nvidia splash screen when X starts
asdfqwega: how about using mini?
does have a backplot .. :)
I know it's not exactly Axis..
all this is beyond my knowledge right now.
damn, hungry... maybe I'll make some sushi.
Time: 03:05:38 -0500 GMT, Windows XP Professional, Service Pack 2 (5.1 - 2600), AMD XP 2800+ running at (1-AMD , 2123MHz, 512KB (100% Load)), DDR400 RAM Usage: 821/1024MB (80.18%), C: 43.67gb of 75.42gb free, D: 0gb of 0gb free, N: 27.91gb of 372.62gb free, Current Uptime: 2days 4hrs 47mins 54secs, Record Uptime: 3wks 12hrs 48mins 50secs, 3 Samsung 19" flat monitors (1 Trinitron, 2 CRT).
sushi is great
My main goal is work on a setup that is easily visible, and shows a preview of the program
sushi, the food?
or is that a command line tool?
Time: 10:05:51 +0200 GMT, Windows XP Home Edition, Service Pack 2 (5.1 - 2600), Intel Centrino 1400 running at (1-Intel , 599MHz, 0KB (0% Load)), DDR400 RAM Usage: 339/503MB (67.40%), C: 3.1gb of 37.25gb free, Current Uptime: 1wk 6days 21hrs 11mins 25secs, Record Uptime: 3wks 9hrs 30mins 56secs, 15'' TFT.
emc2 compile time
on 2.4 ?
should work like a charm
* anonimasu hopes it does
hm emc2 ferrors on rapids
hmm.. dunno ;)
but at least it builds :D
try reducing your max vel
I think pulse output doesn't keep up at those speeds
works now I just increased min ferror
I long for a vital.. so I can have insane speeds ;)
or well insane for my machine
I like emc2..
glad you do
it seems more tidy then emc1
that was one of the aims.
what nevyn said
* anonimasu nods
maybe I should really take my time and play with NML
alex_joni: you've gotten further than me if it builds.
nevyn: it builds now on 2.6
but it's not fully workable yet
so.. when I was at lca
paul mckenny was talking about his recent work on PREMPT_RT
asdfqwega: You around?
which is a way to do hard realtime in the kernel.
hmm.. that sounds interesting
but depends on the definition of HRT
deterministic scheduling and latency
in my opinion hard realtime is only if everything else is switched off
including interrupts & such
* anonimasu nods
basically this allows userspace code to run like it's an interupt handler.
alex_joni: agreed, like on a microcontroller or such..
* Phydbleep has finished scraping cosmoline and de-gropping a lathe. :)
what's done under linux is .. kinda hard realtime (but not fully hard)
so interupts preempt etc are all off..
more between soft & hard
I guess you should just run your RT tasks, on a dedicated machine
and your userspace tasks on another one..
posixrt is useless..
how about communication?
an0n: only polling
anonimasu: That's why I use twm on my bdi.
that would be a pita to synch 2 machines
kind of like what we do with X today..
alex_joni: this is new...
nevyn: tell me more
but well, without X on the box running the motion part
alex_joni: basically it's a way to have a userspace application running with the same privilages as an irq handler.
so you'd have a realtime blob that provided bits of rtapi functionality that was just a app running as root.
sounds like the way RTAI is doing it
it sortof is.. but sortof isn't
* alex_joni has to read more about it
instead of sushi. I made 2x(english muffin + egg + 2 strips of bacon + slice of cheese).
I feel full.
hm, peltiers are cool
anonimasu, cooling your computer?
nope, just looked at some papers regarding them
A-L-P-H-A: You forgot the spam. :)
I was looking after piezoelectric sensors..
spam = evil... be it in can or electronic form
I think I have some back bacon in the freezer.
but that'd involve effort.
delta max at 73c
now that's what I call cool..
anonimasu, what about from like zero... what's the delta then!
stack a couple of peltiers to a X kg of alu/cu heatsink
I wanna make liquid nitrogen.
A-L-P-H-A: but I dont think that'd work in real life..
A-L-P-H-A: that's probably absolute max ratings..
anonimasu, I know... I was gonna say.
I can see like something like 200oC, to like 127oC.
watercooling + peltier heat exchanger..
or whatever it's optimal heat removal is.
anonimasu, radiator... more economical in electrical costs.
I enjoy passive cooling, as I enjoy quiet.
max cooling effect is 57.8w of thoose peltiers I was looking at
what's the energy costs? 57.8w?
no idea, but that's the max
that's like half the energy for a light bulb...
and I don't call that energy efficient.
as long as it's silent
Oy vey, thats gonna be a lot of waste heat.
but the heat's gotta go somewhere, it's just sucking heat from one side and passing it to the other side.
but ADDs even more heat on the other side in the process.
ah found the sensors I was looking for
Thermal differentials with the stacks will be an inverse log curve.
can that work? like stack peltier one on top of each other to increase the cooling?
Yeah, But it hits critical mass and melts down.
Phydbleep: depends on what you exchange heat to..
exchange it for $$
peltier > pipe > house
The cells can only move so much heat and the thermal waste will cook a 12 pack unles you're heating LN2 to start with.
12 stack that is.
hm I should tweak the accel on emc2 in a bit..
the parts take 40 minutes instead of 4
* Phydbleep would settle for a complex part taking 3-4 hours.
Load the program, Set the safeties and lock the door.. Wake me when it's done. :)
what were the last lines after my last message?
Phydbleep: that's scary
Phydbleep: this is a simple plastic part
* Phydbleep is talking 0.050 cuts in Al.
Phydbleep, I have this... for production runs, I use my relays.
Phydbleep, what's the HP on the motor?
I take 4mm cuts :)
1/4" endmill and a part that looks like a 1" deep binary tree.
what's the RPM on that endmill?
see... how do you guys calculate the RPM and feedrate of stuff?
it's like confusing me.
I'm not experienced with milling, but am with lathe work.
A-L-P-HA2 is now known as A-L-P-H-A_laptop
what unit is the accel in the config ?
anonimasu, how do you calculate rpm/feedrate/doc?
I usually calc it in visualmill..
because it's easy..
when I mill this plastic I dont care..
plastic is like cutting air.
I can take 10mm cuts with a 4mm endmill..
I'm talking harder materials, like al, or steel.
hold on a bit..
I'm talking like plunge rates, finishing, roughing (amount of overlap), etc.
slotting, surfacing, etc.
I usually use the calc in visualmill.. and I cant remember how you calc it by hand
hmm, I'll have to play with visualmill. I'm currently using mastercam.
and I can get the toolpaths done, but I really don't think the speeds are correct.
visualmill is affordable ;)
$1000 USD = visualmill.
I have _access_ to both.
[08:23:44] <anonimasu> http://www.unionbutterfield.com/tech/sc/milling.asp
* alex_joni goes to do some work
bye guys ;)
is there a mastercam post that works with emc?
A-L-P-H-A: you have the calcs there
what's the name of it?
Ft x T x RPM
Ft = feed per tooth
T = tooths
10k rpm on the spindle
0.05 * 3 * 3000 = 450mm/min
* Phydbleep is looking for a pick of the part.
0.05 * 3 * 10000 = 1500mm/min
[08:26:57] <Phydbleep> http://www.tfcbooks.com/images/patents/valve1a.gif
I can't go that fast.
A-L-P-H-A: slow down your speed then..
until you get the chipload you want..
what kinda of chipload do I need/want?
otherwise it's dust making
I usually go for 0.07 in alu
The thing at the top is what I want to make.
0.07mm? or inches?
anonimasu, what's the HP on that?
I'm only I think 1hp.
somwhere around 3000rpm
but I think I have a bit more..
I'm like 2560rpm. or something slow like that.
since I got the VFD cranked up to 200% torque..
damn it, I'd really like to crack up the rpm + power of my motor...
Phydbleep, what's the HP/rpm of those threadmill motors?
A-L-P-H-A: vfd's are good/cheap
I am just waiting for my motor to die so I can get one with more hp..
[08:30:53] <anonimasu> http://www.maintenanceresources.com/referencelibrary/ezine/nurbs.htm
but well I might build a small highspeed spindle for milling alu
now that's cool.
I've got 1.5hp @ 4.5k rpm
Phydbleep. 1:1 gearing?
* anonimasu wishes he had that
That's at the shaft.
No gearing.. I was going to go 1:2.2 got get 10k
my mill sucks ass for that... it's a bench top mill, not a highspeed machining centre.
and I should be happy with what I got. though I'd like something better.
chiploads seems to be about 0.15 in commercial apps..
* Phydbleep is building the whole SOB anyway....
anonimasu, that's PER tooth?
anonimasu, what about how deep I go?
A-L-P-H-A: depth = dia
like how much material could I remove with a 1hp motor?
anonimasu, so like, a 12mm endmill, going 12mill deep of material?
but I usually go for 50% since I dont like the plunging
anonimasu, where do I set that up in visual mill, the depth of cut? And Ican't seem to get visualmill to spit out working gcode.
I was going to feed the Z at a constant 0.050" per iteration.
like, everyhthing is only on 1 Z level depth.
hm, add finish passes
and roughing passes when you create the toolpath..
anonimasu, that was roughing...
I never made a finishing pass.
I can help you out with it later tonight
need to work a bit
cause when I tried to test the output, everything was on one level... like it never cut past the first one.
anonimasu, that'd be cool, thanks.
i THINK you can do DEPTH=DIA but well, dont take my word for it take it easy ;)
my plunges was at 250mm/min in alu..
without helical entry
the spindle moved.. since I had some screws that was loose :)
Helical entry == 1 big chip?
nope.. ramped x y z
in a spiral..
cutting your way down
used in high speed machining, to go down in a part to pocket and such
very smooth way to enter a piece..
Ah, The it dug a nice screwthread syndrome.
I love doing holes in the plastic that way
it takes a couple of secs per hole :)
hahahahah... someone on ebay is selling EMC BDI cds, for like $7.
hm.. maybe I should increase MAX ACCEL globally
the helical boring stuff dmess had was scary
what'd dmess do?
he gave me some papers on helical boring
you have a tool and a spindle with a encoder..
you calulate the angle of the edge..
and turn parts on the mill..
requires insane machine speeds
very very cool
I've got a lathe... boring is done with that... or on the mill, my boring head.
Phydbleep, how did you go and ask those excercise places?
"hello, do you happen to have any dead treadmills that you guys are going to toss?"
Pretty much.. You want to talk to the service dept more than likely...
And If they come up with one take them a 12 pack of beer. :)
are you talking manufacturers? or excercise equipement sale places?
well I guess it's time to try my mill for real..
www.lloydleung.com/gallery is some of my crap.
Dealers... Anywhere they sell them to health clubs.. Call a couple of the health clubs and see who sold them theirs.
I'll be cranking out another prototype tonight....
I can take pics while I mill it :)
Phydbleep, good idea.
I want a motor!
anonimasu: Did you see the pic of what I want to mill?
like your motor would be awesome for my mill... I'd just make a new bracket, and a new pulley
it's large :)
no, Phydbleep's 1.5hp 4.5Krpm motor
Hehehe.. .Mine is 130V and screams.
I wish for a better mill someday
ROFL.. I think hte neighbors would bitch if I tried that.. I'd need a
BIG diesel gen-set.
need to try this out
2.0hp motor, @ 130VDC. hmm............ hmm.......................
What's it look like? is it pmdc? Most important.. Is it SEALED?
chips will eat a dc drive unit quickly.
it's on the back of the mill... chips are not likely to get up around there.
[08:56:56] <A-L-P-H-A_laptop> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=15280&item=7137479689&rd=1&ssPageName=WD1V
That doesn't look sealed.
how come treadmill motors are so much smaller looking than my mill motor?
That also looks like a 3/4hp continous duty motor.
[09:02:17] <A-L-P-H-A_laptop> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=15280&item=7154837627&rd=1&ssPageName=WDVW
that one is open
'Treadmill' duty and pmdc.
[09:07:11] <A-L-P-H-A_laptop> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=26226&item=7513744319&rd=1
? wth.... 10 minute duty cycle?
* anonimasu is happy
it worked just got the Z accel and velocity left to set up..
runs happily at 1000mm/min without trouble..
although that's still slow ;)
mine goes at like 2000mm/min
ballscrews are good
[09:11:27] <A-L-P-H-A_laptop> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=26226&item=7514332205&rd=1
That Leeson is a jumper-pump motor
They use those in the hopping lowriders.
10 minutes is about the length of time they compete..
<-- Canuck... knows nothing about lowriders
Think '67 Chevy Impala with blue metal-flake paint, green fur interior and a chain steering wheel bouncing up and down ~ 6 feet. :)
I rather not.
I think they're dumb.
I prefer class... over ugly looking cars.
* Phydbleep prefers performance and reliability and doesn't give a flying fornication what it looks like..
Hehehe... I'm the type who gives something a paint job so hideous that nobody with eyes would want/steal it.
I was talking to my buddy over a coffee.
if I won the lottery... what would go into my car garage.
I think it'd be all american iron in there... no jap, or nazi stuff.
Hehehe..I've had the same mountain bike for like 4 years.. Paid $15 for it at the end of the day at the flea market.. Hot Pink..
Pontiac Solstice would probably be definitely one of them... with a roll cage, and swapped engine.
Hehehe... Once it hits the ground here and some maniac tweaks it, It's American. :)
it it's made in North American (Canada, US), I'll think about it.
No crackhead self respecting or otherwise would be caught dead with something that color. :)
I'm a GM kinda guy... I don't know exactly why
* Phydbleep has an '82 Grand Cherokee in the driveway at the moment.. American?
I've owned a 1987 Le Baron GTS... GrandAM SE '98. GrandAM GT 2002. Cavalier 2004. Buick Century 88.
Then why all the replacement part stamped "Made in China"?
I can accept Quality Chinese Grade A parts.
I can accept Grade A parts made anywhere.
And once they're re-assembled here they're American. :)
bearings made in China, are sooo cheap... and I can't figure out why
Phydbleep, that's made in America, I can accept that.
I'm really not a jap car kinda person.
All the parts are imports... Screwed if you do, Fscked if you don't. :)
I'll take BMW 645i. but that's only cause it's $120K USD car.
wow... 645i is $99,000CDN.
ROFL... I'd average 240 cars for that. :)
A-L-P-H-A: What is make that lathe? It looks like a Logan..
Ah.. Kind of hard to see it under all the accessories. :)
lots of stuff.
lots of junk
[09:47:38] <anonimasu> http://www.masak.se/beg/Abene-150.html
most likely that mill will go to work..
it's been used in a prototype shop.. never for any production runs
seems like it's in nice shape
* anonimasu is a jap car person
anonimasu, gonna get that?
my next car will either be one like the one I have now or a audo s4 avant..
I like s4 265 hp from factory
yeah, the S4 was talked about
dunno what I'
d like for a 3rd car.
well I am to get a dog so my small jap car wont do
and i cant live without AWD
APD == All Paw Drive
there's like, subaru
A-L-P-H-A: what post did you use in mastercam for emc?
I'm tinkering with tcnc.
<-- registered user of Turbocnc
if you want a emc post for visualmill I can give you one :)
I'll make a emc one someday for mastercam
but I am short on time
I would like that.
can you accept dcc's ?
A-L-P-H-A can autoaccept it.
dosent seem to work
let me check
A-L-P-H-A is blocking messages from unidentified users. You must register and identify with NickServ before talking to this user.
/mode A-L-P-H-A -R
still the same..
when "ripping" a piece of aluminum flatbar on a bandsaw, what speed and blade should I use to get the best finish cut?
you won't get a good finish...
just mill the ends smooth
no mill (yet)
and it's rip, not ends
Jymmm, also depends on # of teeth
that's why I asked what blade to get/use.
what the heck is ripping?
36" long cut lengthwise
you want to cut flat bar into 36" lengths?
why not just buy it in the proper size to begin with?
you've got it now
I'll be off nwo
nevyn_ is now known as nevyn
* alex_joni has been away
how's the lathe?
How you doing today?
Good we have tool changer running in manual. Still a bit of work
nice... going for some automatic toolchanger?
Paul is supposed to get here sometime today.
We'll work on the big software issues then.
nice.. then you'll have some qualified helpers around ;)
* alex_joni counts rayh, matts, paul_c
We could have a party.
How is the weather there?
now it's ok.. stopped raining
Rain is okay for a while but sunshine is better -- at least where I am.
there were some heavy floods around here
some dams broke, too much rain, et
Weather has been acting up of late.
Floods that destroy old stuff are painful.
rayh: how much is servo_period now in emc?
[14:18:15] <alex_joni> http://www.cjtimis.ro/inundatii/UIVAR_OTELEC/UIVAR_OTELEC.htm
(click on the links)
set from ini. default is 0.001 sec.
and base is 0.000050 ?
that's 50 usecs ?
Those pictures show some real damage. What a shame.
yup.. well that's life ;)
Period is used primarily by the stepping generators and can be set from >50 to about 10
I hope my house still stands
Are you at that level of flood plane?
I have a house in the same village
my own ;)
You would be welcome to move in with me if it's not?
unfortunately not assured :)
I don't live there
but I bought it as an investment (was very cheap, so I don't mind very much)
about 1k EUR ;)
Yea but any loss is very real.
Matt, the real estate agent, laughs at the 1k EUR.
well.. if he wants I can get him some very cheap now
especially bc they are under water :D
Okay. Buy it all. <g>
He'll advertise it as waterfront vacation property.
you might have a point
with indoor pool
Matt says to send him some listings and he'll see how much he can sell.
think I'll go there this weekend to see what happened
Matt says he was in a row house in Baltimore the other day that had 18 inches 1/2 meter of water in the basement.
There are some pretty run down places in that neighborhood.
rayh: what have you guys got going on with the lathe so far?
We've got the most of the hardware conversion done.
Got pics up on linuxcnc.org/dropbox
They start with Hxx.jpg
Our plan was to get everything inside the two end boxes.
Good so far.
yesterday, you mentioned something about slaving encoders with HAL being nice
Having a bit of trouble with the 12 volt logic for the tool indexer and some limit switches
but a uln2803 handles the 5 up to 12 and 1k resistors and zeners are doing the other
Oh yes. To handle single point threading this would be great.
FredP and a few others worked out how to ramp an axis to spindle speed and then jump into geared
mode for the actual cut.
I don't know where the code guys are with that yet.
Once we get all of HAL and BDI together in one place, EMC2, we will be able to move forward again.
alex_joni you guys got quite a way with HAL and 2.6 yesterday.
rayh: yeah.. but I'm afraid smthg's stll fishy in there
I managed to run tkemc, but any attempt to move would crach my pc
hal stuff isn't fully done (a lot of places need twisting to allow module extension .ko)
but I think it won't be long
I thought the hard part would be the build, but that seems to have been pretty easy
But it looked like the job of pulling it together is underway.
yup it is
When I left codeFest, the committment had been made but little coding.
I'm greatly encouraged by this work.
well.. I put myself as responsible for the 2.6 work on SF
so I had a reason for doing it :D
I'm glad that you did. I really like your work.
[14:46:35] <alex_joni> http://sourceforge.net/tracker/?atid=356744&group_id=6744&func=browse
rayh: did you take any pics at CodeFest?
steve: how's it going?
drooling at the prospects of all the progress you guys are making on EMC2 updates
No. Fred and Paul took a group pic at the end.
I hope to see some stuff progressing from Paul's & Fred's work on the TP
Don't know where it went. Will ask Paul when he gets here.
Martin took some pics
rayh: just /msg'ed you the link
I seem to remember taking a couple. Will see what I got.
steve: maybe we should set up a page about fest, and outcome (or maybe add those to the wiki)
I can put one together. Send mail to email@example.com with links to photos and other source material.
Whoa. Megapixels. That stalled out this box.
Someday I've got to learn how to do this stuff.
what stuff ray?
Somedays I think I was born in the wrong age.
same here :)
sometimes I wish I was born in the PDC2 age
These images of martin's were taken mostly during the NIST tour.
rayh: you're far more advanced on IRC & www & such than most people I know :D
I'm more comfortable at home cause I've got many of the commands written on the basement wall with a perminent ink pen.
that's nice :D
reminds me of what I do with robot-newbies
I make them write on the factory floor ;)
to draw axes (X- / X+, Y, Z ;)
till those wear out, they got them learned
I did that when servicing a German made wood veneer splicing machine.
Each limit and solenoid was written in english on the floor.
anyways.. I'm going home now
Company bosses thought I was crazy until I got it running.
got my massage hours :)
Catch you after a bit.
yeah.. hopefully later
Let me know about your investment home when you get a look at it.
* anonimasu yawns
* Imperator_ is desingning on his machine
have to produce some parts before i change my job, because at my old one are staying arroung good machines
Imperator_: Makes me wish you worked at a place that builds anti-grav units.... :)
alex_joni: Laugh if you want, but that SouthBend was a pita to move out of the shop and onto the truck. :)
/me wonders how to figure out the module extension used :)
* alex_joni wonders how to figure out the module extension used :)
alex_joni: 'ls /lib/modules/2.???/dirvers/kernel' ?
from inside a user-space .c file
it would be nasty to do that
where do you need that ?
I could use a define (-DLinux_2_6)
Not really, Just test for a .o and a .ko file..
but I don't like that either
test for the kernel version
from halcmd ?
darn.. now this is stupid
it's already there
if .o fails it tries .ko
why do they have changed this ???
for 2.6 (dunno who did it.. but I'm glad he did )
I got movement ;)
alex_joni: Hopefully you wanted movement. :)
emc2 does move on 2.6
but... it hangs
no ideea why
not sure how I can debug that :(
don't know strace
what does it do?
I htink it will give you a running view of program execution/stack/memory..
running is not ok
strace - trace system calls and signals
I need debug on what happened before I needed to hit the reset switch
The ptrace system call provides a means by which a parent process may observe and control the execution of another
03alex_joni 07kbuild-0-1 * 10emc2/src/hal/hal_lib.c: added a few more DEFINE_SYMBOL's in order to allow scope_rt to function
process, and examine and change its core image and registers. It is primarily used to implement breakpoint debug-
ging and system call tracing.
not sure it works on RT stuff
it should let you log to a file for forensics.
I had to use it years ago when I was re-writing a driver for a Happauge TV tuner card on slackware.
* alex_joni wonders if Phydbleep could do it again :)
Probably.. I had 75% of the code (commented in German) so it was a dual translation job.. German > English, then EU spec >US spec.. I had all the docs for the Phillips chips on the board so it wasn't too bad.
I never did get a decent translation of the term "Einstellung" though.
That was the concensus, but hte context was wrong.
I was using the find/replace translation method..
So the context was pretty mangled to begin with.
And I only know enough German to get shot by a jealous husband or over-protective father. :)
03alex_joni 07kbuild-0-1 * 10emc2/src/rtapi/Makefile: added MODULE_EXT to rtapi.conf so that scripts/realtime can figure out what extention to use on the modules to be inmoded
03alex_joni 07kbuild-0-1 * 10emc2/scripts/realtime: added MODULE_EXT to the modules to be inserted, so that it works on 2.6 too
I have about the same grasp of Russian/French/Latin/Farsi..
I found a bug in emc2
* Phydbleep can scream "I didn't know she was your sister!" in 9 languages. :)
it locks in a mode (MDI/MANUAL/AUTO)
and refuses to move the mill..
stepgen, siggen, _posemath, sincos,
alex_joni: When I switched from MDI to manual..
sorry bout that.. was using IRC as a scratchboard
the mill wouldnt move or change modes..
I had to restart emc to get it moving again
hmmm.. that's a new one
I was going to do a mdi move to mill off some stuff in a box for mounting the plc's in..
brb... Child + Cat + Tape == ERROR
OK.. Back. :)
care to dig into emc2 on 2.6 ? :)
I just got a clean compile (no warnings ,)
03alex_joni 07kbuild-0-1 * 10emc2/src/rtapi/rtapi_math.h: added rtapi_math.h to act as a wrapper for math.h, it should get included by RT modules which need math.h
03alex_joni 07kbuild-0-1 * 10emc2/src/rtapi/Makefile: added rtapi_math.h to the list of HEADERS
03alex_joni 07kbuild-0-1 * 10emc2/src/ (4 files in 2 dirs): replaced include <math.h> with rtapi_math.h, to clear some compile warnings
an0n: if you're sure about the bug, file a bugreport pls
I've already gone inside..
it'd be nice if I could get some more data when it appears..
I ordered components for spindle current feedback today
alex_joni: Yeah, I'll take a whack at running the demo.. No servos/steppers here at the moment.
same here ;)
running dry at the moment :D
do you have a 2.6 patched?
I CAN dummy up a logger on another box to get a log of the output though. :)
so I hope to be able to get the stuff togther for it next week maybe
Easy.. PLIP cable and monitor the port on the second box.
I could also just prog a pic to watch the port and stream back the data on a serial line.
* Jymmm pokes MrMath!
he knows who he is!
in respect to water -vs oil quenching (like with acme rod), does that effect the finich as well?
will the oil leave a 'shiny' finish?
Viscosity of the oil changes things too.
I'd quench in water.. if possible..
ok, will water leave that black finish?
no oil will
more then water..
I am not sure if there's any diff but I've always thought it's a bit scary quenching in oil..
you need loads of oil to do it right
I know I'm goin cheap, but I was just thinking that if I could buy acme rod for a couple bucks more to get a better finish on it, that be a good thing
oh, I dont think that's possible..
you will most likely distory the rod by heating it up
Oh, no... I mean buying oil insted of water (as example)
err drillrod (sihg(
* Jymmm <---- Drain Bamage
You have to heat it evenly and then flash it with the coolant or it will warp...
Worked on a design for a few days to find out at 4am it won't work.
Phydbleep I'm talking buying pre-done.
Jymmm: Are you talking about re-heat-treating it yourself?
Phydbleep oh gawd no
I think you need a oven to do it good
Ah... OK.. I was referring to diy. :)
Phydbleep I figured you were. =)
anonimasu: That's why I'm building a 'pizza oven' :)
man that's some pizza oven..
Phydbleep you just want pizza, forget the metal!
0,3 sec baking time..
900F for the hot end of the dial. :)
most pizza oven run acg 750 F
* Phydbleep found a way to cast furnace/oven parts from refractory concrete. :)
how many c is that
I dont know F
or rather I dont care for F so I dont know it
way too cool :)
anonimasu google: 750f to c
Hehehe... The materials are all rated for 3000F, So it really just depend on how much air/fuel I can feed it. :)
melting point for iron is at 1535c
[20:35:37] <anonimasu> http://www.threeplanes.net/toolsteel.html
have a look
900F = 482.22222222222222C
3000 F = 1648 C
I really start to hate this...
* Phydbleep wanted the pizza 'well done'
ontinued heating to the hardening temperature (1450 - 1500� F) assures complete conversion to austenite. At this point the steel is no longer magnetic, and its color is cherry-red.
an0n: from there it depends on how you bring the steel back
cool it fast you get marthensite afaik
* anonimasu nods
I hate emc2 crashing on me
* anonimasu does too
I hope it doesn't crash on 2.4
oh, I consider locking up the UI as bad ;)
anonimasu: Waht wm?
that page was pretty good with info on hardening
deepfreezing my PC is
Phydbleep: that has nothing to do with it
Phydbleep: the ui stopped caring for mode changes..
or well emc
* anonimasu whines
OK.. Beacause I use twm for embedded stuff... Don't need all hte fluff and bloat.
have you guys played with hex bar at all?
I need to put some debugging in so I can check it myself..
you should ;)
Hex bar is OK, but curse bricks are better. :)
Is it strong and less prone to warping/flexing than rod rod?
Depends on the alloy.
* Phydbleep has seen tubing that is stronger than barstock.
03alex_joni 07kbuild-0-1 * 10emc2/configs/emc.ini: changed motmod.o to motmod.ko, perhaps we should figure out extension during runtime and only have motmod here
yeah, that's kinda why I was asking. It would SEEM that it would be stronger, but nfc in real life.
Same weight/length, just the diff in the shape.. The ridges/flats SHOULD make it marginally stronger.
I was thiking that if you hit the metal store you're less likely to pick up a piece thats bent/warpped than you would round stock
even angle you can found warrped (over 20')
Jymmm: You'd be better off with hex tubing.
Hmmm, haven't seen that yet, will have to look aorund.
Jymmm: How big is the bar?
3/4".. So.. ~1.4" flat-flat with a 1/4" wall for the same weight.
03alex_joni 07kbuild-0-1 * 10emc2/src/ (6 files in 6 dirs): changed some Makefiles to copy modules to rtlib/ when 'make module_install' is run
can't even find hex tubing
Jymmm: Even round tubing in the larger sizes is more resistant to bending than the rod.. And a lot easier to check quickly.. Pocet laser and sight down the center. :)
one of those pointers
but how would you check straightness?
mount and look?
If you're really paranoid machine a pair of plugs to align the laser @ one end and an acylic target at the other.
very cool though
* Phydbleep is full of strange ways to test stock. :)
I wish you could find drawings on how to build a interfereometer
the lasers are cheap nowdays
Hehehe.. I had a laser on a lathe to set the centering 5 years ago. :)
although I better start saving to buy one
* Phydbleep spends <$3 for semiconductor lasers.
* anonimasu nods
well having a laser dosent really help..
still need the optics and stuff
Even a raw laser is really usefull at times. :)
tell me more about it..
I was thinking about a interferometer ;)
they arent exactly cheap..
and I dont really need one..
hm, my source of electroncs dosent have lasers
for a diode
I have one of the first "retail" laser pointers, very nice and runs off of 2 AAA batteries =)
[21:01:33] <Phydbleep> http://repairfaq.ece.drexel.edu/sam/laserdio.htm#dioint
Interferometers Using Inexpensive Laser Diodes. :)
[21:02:55] <Phydbleep> http://repairfaq.ece.drexel.edu/sam/laserlia.htm#liaint4
well.. I think I have all I did commited
running emc2 again
I'll look at in a while
dollar store has lasers now too.
with batteries to boot!
you could slvage diodes off there
$1.15CDN out the store.
stupid 15% tax.
well guys.. think I'll call it a day
actually 2 days.. as it's tuesday already
do you happen to know how to calculate the amount of material removal from a 1hp motor?
trying to figure out the proper feeds and stuff for my mill.
I'm not a machining guy
I do HW (and a bit of SW)
ALPHA: what are you cutting?
A-L-P-H-A: Assuming a sharp tool, 1hp available power, and cutting plain carbon steel - and a stiff machine - you can theoretically remove from 1.5 to 0.8 cubic inches per minute
Your Mileage May Vary
asdfqwega, sorry. back... needed to help with dinner.
sharp, HSS 3/8" - 1/4" - 1/2" 3-4 flute cutter. on a 1hp motor [but let me find that plate rating].
1" per minute sounds about good to me.
I'm just gonna put some alu 6061 in there, and see what I can get.
watch me break some endmills.
There's other factors, yet - is it belt drive?
asdfqwega: I got the Logan here and cleaned up. :)
Yeah, I'll have toolpost measurements for you later.
And I found a Celeron 333 to throw in the DTK board.
A-L-P-H-A: You said 1hp motor, but how it's coupled with the spindle reduces the available hp
A-L-P-H-A: Gearup loses torque, geardown adds torque.
Not torque - efficiency
* Phydbleep always used the rule that 'rpm/X = torque * 0.95
rpm/X = torque * (X * .95)
* Phydbleep ducks the flying flash-bangs.
asdfqwega: ~ 2"(W) x 2.25"(L) x 2"(H) for the main block..
* robin_sz bounces
today, I tried lasering some aluminium
the spare parts for the laser shoudl arrive tomorow :(
it went THAT well...
if it doesnt pierce, that energy has to go somewhere
Reflected back into the emitter usually. :)
"turning in the widening gyre, the falcon cannot hear the falconer, things explode, mere anarchy is loosed upon the world"
robin_sz: Got a nice pinhole in your upper mirror?
in the laser
OW.. Don't tell me you slagged the yag. :\
nah, no damage to the laser itself
what did you break?
just the ceramic on the nozzel holder
Maybe he melted the WARNING sign?
exploded it did
compounded alignment issues?
nah, just tried to go too quick
robin_sz: What the power rating on that beast?
output, 700W of yag
well, 1K, de-rated
Hehehe.. Lot bigger than the yag rod I have. :)
so it can keep the lamplife up
24kw into the lamps
* Phydbleep has one of the cigarette size yag rods somewhere for something to play with later.
robin_sz: me? Nope.. Optical pump in the uv is what i was thinking of trying.
or pump it with laser diodes
I can get a terrawatt (peak) pulse from a 4 foot nitrogen laser. :)
does it make a popping sound on the target?
If I wasn't worried about being shot as a terrorist I could go up on the mountain and give Mexico (400km south ) a light show..
what color is it?
uv.. You have to pump an intermediate media to get a visible output from it..
dye laser huh
I saw one of those at ANU
And all kinds of things lase when you pump them with enough uv. :)
by the way, when are they changing you national anthem?
they were trying to store a qbit in a crystal with it.
to remove that bit about "land of the free"
<Phydbleep> If I wasn't worried about being shot as a terrorist
we've lost havent we?
* Phydbleep is getting ready and learning all the words to "Dumbass, Dumbass, Uber alles"..
coming again to save the....etc
robin_sz: We havent lost till the concrete sets. :)
Phydbleep: I picked up a coherent innova 90 in hard rubbish ;)
* Phydbleep is just waiting for all the jackasses to all be in the bunkers at the same time...
it's only 6watts but I got galvo's ;)
which reminds me..
Who wants to grab a shovel and have a pothole patching party? We'd be doing community service. :)
I just have the 700W yag, a 450W CO2 and a 1700w CO2, oh and a hene tube :)
ROFL @ robin_sz... # lasers and a flashlight with an attitude. :)
I still yearn for a 632nm green laser
now, theres a good use for your YAG rod.
I was looking at a multiple discharge path for the N2 laser.. Rotating the arcs around the yag and flushing the whole tube as a cooling measure...
Paul is off to matt's place
and to see fred some more
les: did you get segmot cured?
Nope. Heres what we found:
Sadly we have pronounced SQ dead. No one can get the damned thing to work...not in 6 years
We have all but talked fred into doing or at least helping with a more conventional cubic or quintic planner
which could happen fast
* robin_sz nods
les: and you know what killed segmot? I do.
So what is segmots major malfunction.
specifically...a cubic spliner is already in there...it's just in the wrong place
PHY it was just too ambitious and complicated. The major malfunction is:
les: what killed segmot is neither the "too ambitious" or "too complicated"
what killed segmot is the crazy world of code it tried to live in
in a nice ,clean, well thought out desing, a sement based planner would be doable and understandlable
the math was right...I know that.
but in emc as it is now? nightmare.
emc2 is better but still bad
we were running emc2 here
but, everytime I say it, certian people just accuse me of being a troll, so fsck it. its dead as far as im concerend Im afraid
today, after playing for a few days with it on a old Brodgeport series 1
we ordered a Baldor ESB ...
it is bad. Our tests indicated that it could not even do smooth machining at Bridgeport rates
and bridgeport rates are very slow
* robin_sz nods
I thnk jmk has the best chance of getting this thing beaten into shape
Well without a solution We do not have a certain "product"....
may order some iron anyway
it's on the pacer site now
stick a baldor or G200x on it and be happy :)
So what you're saying is segmot might live if somebody went through and rebuilt it except for the math routines.
I think it will never live inthe current framework
and the cubic/quintic planner will have to be bodged on top of the current mess,
No PHY....emc will only live with the emc2 tree cleanup and a new TP
proabbly further discouraging anyone from sorting out the current mess
* robin_sz hi fives les
* Phydbleep is really only interested in emc2.
It has been decided that all work will be done on that tree
emc2 caught my attention for a short while
the old one is such a mess...
I was just getting interested, when someone dumped the entirety of the old tree into it
it never recovered imho
It's better...but not quite ready for prime time
we ran it ok here
its still got 700 line switch statements
Bleah!.. The old tree is for reference only.. Maybe steal a math routine or 4..
it should have been a clean sheet. the rebuild of the messaging layer was what was needed
bu the old messaging layer was installed and used as it was
Dammit...I'm good at math, and I'd love to help, but my mind bogs down when I try reading code
I lot all interest at that point
still, apparently Im just trolling.
les: I still haven't seen a test program that kills SQ
We have also determined kind of a speed limit when using general purpose CPU for real time
I've killed SQ
les: I think paul pronouncing it dead because he can't figure out a bug in a couple days is a little premature.
* Phydbleep steals robin_sz's trolling motor.
I guess test and real world differ some
les: go on, send him emcfine.tap
We have a new test program
I did the math and paul coded it up
makes a 3-d spiral with increasingly close points as it goes in
will release it but paul wants to gpl it or something.
Take the .dxf output from autotrace, scale it to my work area = 100000 moves of 0.001" or so
yeah, I wrote that too a while back. I didn't get a crash but I got pauses.
Chris: We are trying to get fred and anyone else to help with a simpler but more viable cubic or quintic tp
I tried that a couple times. I got pauses a few times, and other times it would just stop
Fred says this is pretty easy
les: You're limiting it to 5 axis?
les: I've seen you guys talking about that for a year but nobody has written a single line of code.
* robin_sz suggests digging out the foundations before piling any more towers onto the castle might be a good plan
emc has a cubic routine in it already
just in the wrong place!
les: also nobody has been able to say the title of the book that describes the algorithm, even though several people have asked for it on the list.
[monty-ish] And they said you couldn't build a castle is a swap!
hmmm ok hang on
so I built one.
it fell down
Does this mean my BDI is going to burn down, fall over and sink into the backyard?
It sank into the swamp.
So I built another.
"Introduction to Robotics" j. craig
asdfqwega: from wood. that burnt down.
Puls sat around outside in the nice spring weather reading it cover to cover
Phydbleep: Maybe ;)
les: thats a good start, I get the impression that some of the coders on emc must have read "robbie the robot goes fishing" by mistake :)
Well Fred and I both had this text in school
you were in school in 1989?
[22:41:06] <Phydbleep> http://robotics.uta.edu/me5337/
I bought a more recent version later!
les: can't find the message to answer
sigh, why wont my doze box log into this samba server
* robin_sz tickles it.
there are lots of old IEEE transactions on the subject too
of "robbie the robot goes fishing"?
They cost money if you get them on the web, but not much
the onely things that seem to cost money are pr0n and IEEE transactions
Cubic and quintic planner algos were research paper fodder twenty years ago
nowdays it is time optimal stuff etc
But anyway yes Chris there has been talk for years about this
talk and no action
got fred down here...
got rogier down here from holland to work on SQ...
I did what I could
I set aside a week of shop time for this recent testing bit
[22:47:56] <Phydbleep> http://www.rvg.ua.es/publications/gallardo98b-IEAAIE.pdf
Try that for TP math.
just a plain old four point look ahead cubic would be about 1000% better than TP/TC
Maybe I'm just stupid, but i thought that all the major tp work should be in the g-code perverter..
Well it's a level down
Think of it as just another form of g-code
it is causal
so you could calculate a plan at any time
even store it in a file
IE.. Retract the z before you traverse to mill the next pocket should be there.. Having the software second guess the operator is a bad thing as far as I'm concerned.
Well the tp we are talking about is just to allow proper smooth motion.
The current TP will choke on .1 mm spaced point and a requested 1.5 cm/sec feed rate
What's the difference between the following: end milling, face milling, slit milling? Is end milling, just using to mill material away? face milling, like with a fly cutter, to remove tiny amount of material? slit milling like milling slots?
trying to determine feedrates for my mill. :/
milling slots is the hardest because one side is climb milling...the other not
les, yeah... I'm gonna rough it out, and then go with a finishing pass.
single flute sometimes works better in that case
I need a 1/4" slot, so I'll probably use a 3/16" endmill to rough the slot, and then pass the sides to get a nice finish
Yeah, That's wht I was wanting to mill 0.050" per pass.. reduce the strain on the spindle.
ALPHA: that will be safe. Climb mill the two sides for a smooth finish.
climb mill? [I'm a newb]
k. I think I know.
climb mill.. You're cutting UP the hill.
cutter always compresses the material
my spindle is CW. So if I'm milling a hole/pocket, to finish, i'd go clockwise?
actually the opposite
let me guess.
with climb milling the cutting forces will help to move the feed along
it is unstable though
It feels kinda like power sterring if done manually
to finish mill something, lets say I want to mill the perimeter of a box... I'd go what direction if my spindle goes CW [normal]
it'd be like CCW around the perimeter right?
and CW for a poket/hole?
example: you are cutting the LEFT flank of a slot
and feed away from you
les, uh, that's still some confusion. Lets say we have a donut. I want to mill the hole. What direction, say we started at 12oclock high
most manual machines like wood table saws, jointers, etc do not climb cur
rip a plank by pushing it in backwards from the back
(don't do it)
No thanks, I like my walls with the number of holes they have now. :)
but with power feed machines it makes a much nicer finish
climb milling works OK on very rigid machines
well, I do'nt think mine's very rigid.
on a flexible machine, you can get runaway
screw it, I'll just figure it out by trial and error.
I run climb milling on the saws and moulders
but they have power feed
and almost all cnc finishing cuts are climb milled
the edge of the rotating tool is moving opposite of the feed direction...that's all it is
A-L-P-H-A: You might be able to get away with 0.001 cuts on it.
* Phydbleep has used that trick on sloppy machines.. Hog it out rough and then run 0.001" slow-feed finishing passes to get a good finish.
The less the cut...the less difference between climb and conventional milling
conventional pulls the chip up and forward...
climb down and back
easier to show than say I guess
asdfqwega: You still here?
les, I understand now.
les, so climb milling a pocket would be would push the chips down the hole, while normal milling would shoot the chips up.
conventional milling tends to recut chips
But if the machine is not rigid you must do it that way
climb milling wants to "suck the work in"
I really don't mind going over the part to get a clean edge. I'm doing it via CNC anyways... so it's okay fo rme.
I usually do a .01" climb pass for a finishing cut on most stuff
I'm just going to undersize the slot, and mill it out to proper size. 1/4" slot is the target.
used a 3/16 endmill, and then just clean the edges.
a .25 cutter will make bigger than a .25 slot dure to instability
yeah... so I'll be good then.
the rest of it, I'll just use a 1/2" endmill to rough, and finish the outside and pocket. And then a 3/16" endmill for the two slots.
I think is the model.
that will work.
Oh Chris if you are still there...we had a faster box so I got my first look at axis
A-L-P-H-A: Wrong url
I REALLY liked the appearance
I think is the model.
looks very familiar and un intemidating
The requested URL /current/bracket.zip was not found on this server.
I think is the model.
All that effort and it's a #%&#%&@ dwg file. :)
dead link, but I have seen it
I'll repost it as what?
A-L-P-H-A: Don't worry about it.
for the future then?
I'll see a jpg when it's done. :)
Who's set up to mill steel in here?
* Phydbleep needs a dovetail cut.
60 i think..
start first roughing a rectangular slot.
Yeah. Then 30/45/60
well at least a small diameter cutter so you don't have to try to hog the whole thing at once
* Phydbleep needs new crossfeed blocks.
MrMath is back!!!!!!!!!
I want 1 for a milling adapter and 1 for a spare. :)
les : Hey, what's the significance of 1.732 ?
it's not e...
is it PI/2?
I don't know
nfc, that's why I was asking oyu
let me check
hi dmess and robin_sz
no, it is root 3
les : I was messing with hexagon, and came across 1.732
robin_sz root cubed?
confirmed it is root 3
* Jymmm with deer in headlight look... what is root cubed?
who said that?
the number mutiplied by itself that equals 3
ok what is root 3 ?
sqaure root of 3
sqrt of 3
like root 2 is sqaure root of 2
I like sqaure better
that looks neat
robin_sz I get it now, never heard it phrased that way before is all
its standard maths speak
someone has to invent them!
robin_sz Come on, I found that (1.732) the backwards way =)
Jymmm: and what is the value of e?
I just think of the formal definition
robin_sz Hey, I was out getting stoned during those years of school =)
ok, easy one then ...
e is the root where e^x is the derivative of itself
that is why it comes up so much in nature
isnt it i before e except after c
the original "natural" number
dmess : Yes, yes it is =)
* Jymmm is gonna order some of these http://www.xylotex.com/StepperMotor.htm
les: did you read that piece recently about pi not being as irrational as first thought
when done in hex
oh now thats interesting...
it turns out ...
3.14 in hexidecimal?
you can calculate the nth digit of pi, in hex quite easily, without knowing any of the rest
well in some number bases it would be rational I guess
like base PI heh
but we do this ten finger thing
should have been 21?
google for bailey plouffe
les how you do 3/16 with the ten finger thing?
[23:55:38] <robin_sz> http://www.math.hmc.edu/funfacts/ffiles/20010.5.shtml
the 3 and the 16 are ten finger type numbers.
remember I said 21
for guys anyway
22 for the gals
well, a few of the gals anyway
you aint been looking hard enough
right, and on that note ...
Where's that leave me? I can count to 32 on 1 hand..
robin_sz : if a "gal" can do 23, sHE's all yours!
And if i use both hands I can get to 1024. :)
without the extra one he was thinking of...they'd be not much fun.
lol @ les
I must go to bed :)
I must go on that note too
night guys ...