Les_away: Is paul_c still there?
* Phydbleep has checked the parts stack and found 1 P-II board (DTK PC100), I Dual P-II board (Gateway? PC100) and a Compaq PW6000 (P-II 300, Dual box, 1 cpu/riser, 128 meg ECC) that needs a drive.
The Compaq is a beast and has a heavy case that should survive a cleanish shop enviroment.
Oh.. And the RAMBUS P-III whatever board that I dont have ram to test.
seen alex_joni lately?
I wish.. Blew out my back trying not to drop sack-crete on the dog. :\
He was here a few hours ago.
aren't dogs fun... always running around like crazy people, getting underfoot, etc
~4.5 to be more precise.
late in his neck of the woods
I'll have to resort to email I guess
Frikken, Fur-covered, Four-Legged, Roller-Skate wanna-be's.. :)
(I guess I should look at this screen more often :) )
Les_away: Is paul_c still there?
Did he see the hardware list?
what h/w list ?
posted at time 20:28 by Phydbleep
is Paul there as in
Are you (or someone) planning on committing the Pico-systems drivers?
"can I ask him something"?
1 P-II board (DTK PC100), 1 Dual P-II board (Gateway? PC100) and a Compaq PW6000 (P-II 300, Dual box, 1 cpu/riser, 128 meg ECC) that needs a drive and a P-III/rambus board I cant test due to a lack of ram.
Les_away is now known as paul_c
starting to look at the lathefork merge
got the cvs book out...
did you happen to tag the point immediately before you started the lathe fork
(doesn't look that way...)
the branch point _is_ the tag
you have cvsbook handy?
I do - are you looking for a second opinion? :)
theres a section in there where they recommend making a tag somethign like "root-of-something", and then the actual branch is another tag called "something-branch" or similar conventions
If I were doing a "merge", it would be just in the interp directory.
there are also changes in the calling code (emctask mostly)
Yes... There would be some minor changes to reflect the C++ class change
another compilication is that Fred added custom M codes to the head during fest, and I don't want to lose that during the merge
(or changes to a C++ class)
The custom M codes *should* be in the Lathe_fork interp
so I'm reading up and trying to get a grip on everythinb before I do anything
some preliminary grepping doesn't look that way, but I could be wrong
cant grep much in the interp directiry because of the breakup into multiple files
but there are custom Mcode changes in head that are _not_ in lathefork
If the code isn't there, it would be a trivial exercise to add it.
OK... Two choices....
trivial once I extract it
a) leave the merge until I get home, or
c) merge, and add the M code changes in at a later date.
I don't like a - I've only been at it an hour or so, I'm not gonna give up so quickly
And 'b' would be?? Have a beer and decide in a bit?
b) would be do some diffs between pre-fest & post-fest, generate a patch, and apply to lathe_fork.
wow... I was thinking along the same lines
Then merge it back in?
that's why I was wondering if the root of the branch was tagged
fred's changes aren't the only ones made to head since the branch
as for Jon's ppmc driver, I'm going to have to hook up a floppy drive
I think there are two drivers at this point - the BDI one, and the HAL one
and mail the altered files to someone with www/cvs access
I'm planning to dive back into that once this merge is done
my plan in a nutshell:
1) merge the lathefork stuff
2) make a branch for the 2.6 build system work
3) work on drivers on a 2.4 system
alex has indicated that he'll be able to work on the 2.6 stuff
I've already got the 2.6 compile farm slot running, and all slots are attempting to compile the bdi-4 branch (only the 4.20 slot succeeds of course)
once we have a branch for the 2.6 work, and all slots are attempting to compile it, we have an environment that supports the build system work
Is there a reason that the USC HAL driver can't be merged earlier?
not really I guess... but it isn't finished yet
I'm hoping the lathefork merge takes only a few days
creating the 2.6 branch and adding it to the compile farm is a matter of an hour or so
then I can focus on drivers
OK - I was hoping to play with it a little, but that'll have to wait for 2.6 anyway
damit... I just wish I could get one of my computers hooked up to Les' dsl modem
When will emc2 compile/run on a 2.6 system?
just take out the fixed IP stuff ;)
Phydbleep: when we fix the build system to do that
Phydbleep: "When It's Done"
may be several weeks
what's the doze equiv to ifconfig ?
paul_c: Yeah, Set it for DHCP nad it should grab an IP from the router/modem.
depends on the doze
or ipconfig for Win2k
if, ip, what's one letter
for XP it's in the net connect properties.
or are you looking for configuration?
(as opposed to info)
He needs to swap to dhcp from a static ip.
nope - he was running Linux at fest, so this is a different question
(but maybe the same answer)
getting info from the win box to setup the linux one maybe?
Should work the same for 90% of the US cable/adsl links.
depends on whether it's PPPoE or just DHCP
my dsl need(ed) pppoe
jmkasunich: Ick! :P
(needed because I'm running a little black box router that does the pppoe now, so I just use dhcp)
yeah - a little router is often the easiest way, plus you get sharing
the isp provides a doze pppoe client that handles authentication and such... easier for their windoze users
* Phydbleep has a Athlon-900 for a router/dhcp server. :)
hoe are your electric bills? :)
* jmkasunich now has 5 200MHz pentinums running 24/7, plus the dsl modem and router
Astronomical with the other 5 boxes that run all the time. :)
that doesn't count my actial development boxes which are only on when in use
heh - I'm bad enough off with only 2 machines on all the time - I'll really be scerwed when my server gets switched on
well.... If I need to set up pppoe in order to use the dsl modem.....
no - just doenload ... oh :)
* Phydbleep has been using hardware for heaters all winter.
does les know what he has there?
Looks like Comcast.. That should be dhcp.
les needs more computers anyway... just run down to the corner computer store and get a router... you know, the corner computer store in Tiger?
It's just a modem with an XP box to connect to.
jmkasunich: Why not go to the Tiger branch of CompUSA ?
that's the one I was referring to
at the corner of Main Street and Fat Chance Ave.
* paul_c looks out of the window to see if downtown Tiger is still open....
All the lights are out in Tiger's shopping Mall...
sidewalks all rolled up?
for tonight, yes.
is the XP box capable of doing connection sharing?
even the red light area has closed early.
those were brake lights
I think Les has just the one nic in this box, so no chance of doing NAT
OK - pppoe support is included in kernel 2.6, so you might not be totally screwed
but then again, you could use it as an excuse for a beer and some sleep :)
first question, is the ISP running PPPOE or just DHCP?
nah - I'll just mail the files to you for comitting.
OK by me.
but then, how can you get email and not have CVS access?
floppynet from Linux box to XP box, mail with Les's account
jmkasunich: I have no idea what the ISP setup is...
that's so '90s
SWPadnos: It's still 50 years ahead of the rest of Tiger.
accepted as fact
mail sounds easier...
heck - Tiger opened up their first drive in movie theater last year....
just use WinCVS
they have cars?
I think our last drive-in closed last year
Flintstone types, yes.
gives new meaning to "foot-pounds"
speaking of mail... the dumpster spat up some 3 port SCSI cables with terminators on one end. You need? I could snail-mail to les's place
how long would the post take ?
priority mail is usually 2 days
how much longer are you there?
probably three or four days....
Les has some meetings lined up with a local company.
Gah..... Looks like cia is down
poxy server error
I get nothing here (FireFox on Linux)
let's see if I can get sneakernet set up....
OK - time to go. See you all later
SWPadnos is now known as SWP_Away
paul_c: still there?
I'm gonna be calling it a night soon.. if you want that cable, speak now, and I'll mail it tomorrow
(speak a mailing address)
just packing up here...
paul_c is now known as les_away
Jymmm: Look it up in the jargon file.
in the wha?!
You claim to be a programmer, and you've never read the jargon file?
your words, not mine.
or jargon file
Hm...I thought you said you were
it was published as a book.
the best stuff is the appendix tho.
The story of mel.
I must have an old copy...I don't remember those
and the magic, more magic switch
THAT, I remember :)
the story of mel particularly is good.
is not (giyf that is)
nfc who mel is
he's a real programmer.
as opposed toa lip syncing one?
lip, not lisp =)
Well, I'm about due to re-read the jargon file - it's been about ten years or so since I last did, and I see they've added to it - I just had to look up LART XD
[04:42:54] <nevyn> http://www.isri.unlv.edu/~slumos/jargon/Table_Of_Contents.html
as in real programmers don't eat quiche
Back in the good old days-- the "Golden Era" of computers-- it was easy to separate the men from the boys (sometimes called "Real Men" and "Quiche Eaters" in the literature)
Hey, Did somebody here want dimensional specs for a 5-C collet?
A-L-P-H-A I think
Phydbleep But I'm not sure if it was a 5-C collect or another one
* Phydbleep sends a wheelie-bin chasing A-L-P-H-A.
A-L-P-H-A: Wake up! :)
Phydbleep I do know that whatever A-L-P-H-A was looking for, he found.
but he'd probably appreciated any drawings and such. He wanted the angle of soemthing.
[06:15:17] <Phydbleep> http://www.lathe.com/images/5-c.gif
There's a 4-C too.
[06:16:42] <Phydbleep> http://www.lathe.com/images/4-c.gif
A-L-P-H-A: You wanted 5-C collet specs?
nah... I wanted DA-300 specs. but I got them, and don't need them... I'm just gonna buy a ER-11 collet system.
Phydbleep the gratitude!
Ah.. OK.. Well bookmark http://www.lathe.com/images
* Phydbleep has to go pick up a PowerCraft (Logan) 9" lathe Thursday..
WTF am I going to put it?
Phydbleep next to your mill
in the dump
I'm looking for a nntp filtering program, based on headers. Not, subject, or poster, or body, or length. I want to filter from the header. Anyone know of a news reader program for win32?
filter what specifically?
* Jymmm pokes A-L-P-H-A
Jymmm: I don't have a mill, but I'd be tempted to trade my South-Bend for one now that this lathe has turned up.
alex_joni: Greetings. :)
hey Fido ;)
alex_joni: Somebody was looking for you earlier, but said he'd email.
Phydbleep: anyone familiar?
alex_joni: It was jmk.
alex_joni Yeah, they said they were from the free clinic with some test results?
Jymmm: I forwarded it to you
alex_joni asked if you allergic to antibiotics
alex_joni that's right, be in denial
Hmmm.. That means I should check because i got the sheep from Jymmm...
Hehehe... Same to ewe buddy. :)
Jymmm, I want to get binaries, but alt.binaries.warez.autocad has a shit load of spam.
* Phydbleep always wondered where they got all the pink bits for spam...
100K messages, 99% spam.
A-L-P-H-A Very nice program.
I mean what is it? S-urplus P-orn A-nd M-ore? Thats the only place I've ever seen that color occur in nature..
Jymmm, used to use that...
I'm looking for a way to filter via the IP in the header.
or originating server.
A-L-P-H-A I think it can filter on headers
A-L-P-H-A I've filtered on posters before, yeah.
re-doing the GPS interface for the machine..
although I don't have any urge for that :D
no, not filtering posters... filtering based on ip from the header.
there's this creep that just loves to send out crap loads of spam.
alex_joni: What's up with you?=
alex_joni: doing anything interesting?
don't feel like nothing else ;)
A-L-P-H-A, wanna write some gcode for me?
alex_joni, wanna write some gcode for me?
I'm not really good at gcode ;)
and I don't feel like writing either :P
bah, maybe I'll just throw it into mastercam, and see what it does.
G-code for what?
anonimasu gps for what?
* asdfqwega has been freelancing writing g-code
I'll post it somewhere.
Jymmm: a machine at work..
anonimasu details man, details!
Jymmm: lol, its for telling customers where they've been with it ,so they get a printout of where the machine's been running..
anonimasu What machine?
an0n: making probability maps of future woods?
Jymmm: tree planting machine
"Hello, is your mill running? Well, you better go out and catch it!"
or well, so the workers planting the tree's knows where to plant..
anonimasu and your writing code to do this?
I've got a seed planting machine prototype sitting on my desk.. also
yeah, although I am just interfacing the gps program..
anonimasu Why? just download
I started writing my own but using .SID files costs loads of money...
Jymmm: dosent really meet all my needs..
anonimasu APRS ?!
gps > map > printout
automate the machien to plant trees w/o operator
an0n: what do you write your code in?
care for some code?
some friends of mine did some GPS navigation
asdfqwega, :/ dunno what's happening with my dcc, should just work.
asdfqwega, it's all in mm.
Jymmm: positioning isnt interesting in that way, it's to help the people running the machines get paid
anonimasu : charge 100X as much and tell em no operator required!
but I'd love to ;)
anonimasu nothing is impossible
the trouble is that there's trees in the forest.
anonimasu what, you never heard of sensors?
an0n: get a bigger machine
then you'll not mind the trees
and.. you can always plant new ones
the machines just weigh 12-13ton's..
anonimasu So, what's another ton for the robotics?!
* alex_joni packs Jymmm in tin-foil, and hands the resulting robotics to an0n
Teach it to pluck them out of the ground in front and replant them behind?
an0n: seen those 6-legs stepping type of machines?
those are neat
or was it 8 legs?
alex_joni: never really
Yeah, The logging units.
the one I've seen was for cutting trees
got a image?
6 legs, 2 arms, stripper saws, cutoff saws, whizz, whir, stack of logs. :)
google logging hexapod
don't seem to find it
[07:57:59] <Phydbleep> http://www.plustech.fi/Walking1.html
yeah.. that's the one
That's an old one.. The newer one has a second smaller arm at the front.
well shit... I got 4 of those in the back... one gets me my groceries every week, another one does the laundry
Great, Send one down here to babysit so I can get something done.
yeah I've seen thoose..
but it feel like a hoax..
* Phydbleep is scared..
anonimasu: It works.. They just dont travel very fast.. <10kph.
Phydbleep: that's fast..
Jymmm: Hell no I'm not giving that child duct-tape.. I do not want to see what the cat would get grafted to.
Phydbleep: the machines go ~30kmh in high gear at flat road..
Phydbleep *SIGH* you dont GIVE the tape to the kid, you DRESS the kid WITH the tape!
Jymmm: Same difference.. The kid ends up with the tape and the cat gets the shaft.
Phydbleep fine... get the kid butt naked + bottle of crazy glue + smooth surface wall
Jymmm: there was a page about sticking people to walls
Phydbleep apply glue to kids butt, place butt againest wall. Any questions?
alex_joni you never saw the pics of that? kinda funny.
* Phydbleep would use velcro on the ceiling and a trampoline.
[08:08:23] <alex_joni> http://www.octanecreative.com/ducttape/topten/index.html
[08:09:41] <Jymmm> http://www.octanecreative.com/ducttape/walltapings/images/gal_door.jpg
Phydbleep -----------------> http://www.octanecreative.com/ducttape/walltapings/images/gal_door.jpg
Phydbleep There, now yo have illustrated instructions!
* alex_joni yawns
* A-L-P-H-A puts a sock in aj's mouth
* alex_joni finished swallowing it
* alex_joni yawns
* anonimasu yawns
the prototype I built works.
does it do anything?
or it simply only works?
but it does work..
what kind of seeds?
any mary jane related?
ganja seeds ofcourse
it's pine or whatever you call iut
I call it fun :D
[10:15:28] <anonimasu> http://linnaeus.nrm.se/flora/barr/pina/picea/piceabi15.jpg
heh.. wouldn't mind to get a few in front of my house
do you plant by request?
buy a machine yourself..
we just build them ;)
how about some testing?
get it over here, see how the GPS behaves :D
shipping a 14 ton machine costs a bit ;)
let it run over here ;)
would probably take a while
what does it run on? diesel?
make the next one solar powered
but the machines costs 600000$ ;)
steam powered, and burn trees for steem
that our equipment mounts on..
you could probably buy small trees for a couple of $ each..
but it takes 50-100 years for them to grow..
I'll buy some bonsai
fully grown ;)
now I need to machine another prototype, to see how easily manufacturing is..
* anonimasu cries at toolchanges
need to change 3 kinds of drills for the job..
but I might be able to mill all holes and get away with just one..
build a toolchanger
or what's it called
but I still dont have a drawbar that'll help out with it..
part of the support is already there
NML & such
it's the physical part..
how do I get the tool out of the machine that's the trouble..
make it spit it out
hmm.. small explosive charge?
how about changing spindles?
[10:27:23] <anonimasu> http://home.insightbb.com/~joevicar3/default.htm
maybe I could do with somthing like that..
not very expensive at 350
lol changing spindles is too expensive..
that change is too small anyway
but the concept..
maybe if you automate it twice
on the other end it places tools on a conveyor
a large one, with many tools
I was thinking about a small robot ..with a tool magazine..
that's too complicated
the control isnt that advanced..
but it needs to be pretty stiff since tools stick...
you probably need a few kg payload
to have the force to get them out of there
I dont need any reach though
since I can slide the whole robot to the magazine..
* alex_joni loves to listen to Supertramp ;)
yahoo launchcast is pretty nice
I am listening to "love like blood"
a bit dark but nice
love like blood?
goth rock or somthing like that
hmmm .. too dark for my taste ;)
I set up my own station
* anonimasu nods
that's ok for work ;)
do you use zee messenger?
msn sux ;)
it's got scary emoticons :))
I dont use/abuse them
what's up Fido?
Or was that >:)
Hell If I know.. This is just a text term, no icons. :)
alex_joni: I'm trying to figure out where to put a 9" or 10" Logan lathe.
try on the ceiling :D
or maybe vertical on a wall
hm, you could turn if it hung upside down..
only trouble would be gearbox oil..
Won't work..I'd have to sit on the floor to use it.
highspeed machines are often horizontal.. to help with chip evacuation
New they spec'd to 0.0005 runout so I'm tempted to try. :)
that's pretty much *grins*
I need to get it over here and then decide whether it or the SouthBend is going to go out to a friends farm.
my little crap lathe has < 0.01.. the dial ripples.. a bit..
but well that dosent help when the bed is ~
* anonimasu kicks the laths
* Phydbleep waits for the nerve impulse to reach the brain..
Wow.. This is scary.. Still no "OW!" :)
I'm wondering whether this 600 lb lump will sink the shop..
Hehehe... But I parked the SouthBend on the high side and it's leveling out at least. :)
what are you in a swamp
* Phydbleep needs to add ~ 2' of shim to the se corner now.
Rio Grand flood plain.. Mostly sand and every time it rains the terrain contours change slightly.
It doesn't help that the city needs to fix the leaking storm drains.
It rains and I lose part of the front yard down the sinkhole.. Oh, Well.. Park the beater over there and if the sinkhole eats it the city can buy me a another one. :)
Hehehe.. At least it will have to eat the 3 cars and the house before it gets to the shop. :)
Cheese it! We broke it..
I am building a manual
a manual ?? Shop manual? Owners manual? Manual labor? What?
Oh, For emc?
for the product I am working on
* Phydbleep tries to sneak a peek through anonimasu's laser mouse and sees nothing but spots.
OK, I give up.. What is it?
tree planting thingie ;)
Send me one and I bet I can break it. :)
Ot confuse it. :)
I bet you cant..
* alex_joni thinks Phydbleep confuses anything
Maybe both at the same time. :)
unless you apply a force > 1000Nm at the tip of it..
* Phydbleep has proved it is possible to fsck-up an anvil with a rubber mallet.
well 1000nm into a dead stop
a rubber mallet cant do that..
Of course we popped a mold of the anvil and cast one from crap-crete for the purpose.
* anonimasu dosent get what you mean
CRAP!.. This is going to suck!.. I still have to move the damn anvil out of the laundry room..
anonimasu: crap-crete.. Cheap concrete and skimp on the water...
yeah but anvil/cast/?
2 part plaster mold..
Looked great when it crumbled. :)
I wish I could get some of that rapid moldmaking epoxy stuf.f.
Tap, Tap! *SLUMP*
machine the pour then wait a but remove..
and cast stuff in it..
I want to see the UV setting tank/cnc unit come down in price and lose 99% of the toxicity..
I'm going to say "G'nite" because I need to go fall over and try to get my back to de-spasm.
G'nite all. :)
sweet dreams :))
* alex_joni yawns
good morning (or afternoon)
well we will switch gears a bit here
We have deemed SQ unworkable and probably unfixable
I think so
the latest mods just made a big mess
the ones cradek did?
sterling went berzerk
And we must recongnise that no one, not even the author, has ever gotten it to work
so what are our chances?
well .. :)
maybe one day
but a quintic planner is needed no matter what
we will now look closely at the cubic sub interpolation section
and move that over to the normal tp?
unlike SQ it works with non trivial kinematics
but it is sub interpolation at the servo rate
SWP_Away is now known as SWPadnos
it needs to be at the trajectory rate
we just had a split ;)
and it could easily be made quintic
so we missed everything before "but it is sub interpolation at the servo rate"
* alex_joni hands SWP the link
My feeling is that perhaps the lower the ratio of trajectory rate to servo rate....the more smoothing happens
shouldn't it be the other way around?
[12:43:10] <alex_joni> http://22.214.171.124/irc/irc.freenode.net:6667/emc/
should have a few logs :D
We could never test this because the original TP suffers from queue emptying
the violent motions mask anything else
* alex_joni hands les a pill against queue emptying
paul will work on that today
and also switch to the emc2 tree
cool - did you ever figure out how to get his Linux machine on the net?
basically ANYTIME the cpu is free and the queue is not full should used to fill it
now that's a CPU hog ;)
but it can't tell that the CPU is free when it's waiting for a timer event
SWP: I think my isp has some proprietary stuff in the adsl modem
(ie - after timer->wait(CYCLE_TIME)
les: how about a second CPU for queue filling?
it won't give him an ip
wouldn't surprise me
can you check your XP networking?
and a third for GUI & non-rt stuffs ;)
alex: perhaps...I will mention that to paul
les: was actually kidding ;)
We need to have a conf call with Fred if possible today
second thread, I think he thought :)
I fear paul_c will hit me with something if you do...
I may try emc on a dual opteron soon - we'll see what happens ;)
I do math...I don't know much about RT programming
SWP: dual 64 bit?
yes - and eventually quad (dual dual-core)
will actually run slower (my guess)
servo update cycle of 10 microseconds - heh :)
damn it's hot out there
27 degs right now
not out here
I have a blackboard full of cubic vs quintic graphs and stuff for paul
it's about 27 degrees (farenheit)
uh-oh - I hope you have some strong coffee for him
C and C++ here
SWP no I just drew graphs of pos, vel, accel, jerk
html, php & the like here ;)
ah - good. That shouldn't explode anyone's brain then :)
thank god for php ;)
no it just shows that quintic is the lowest order with cont jerk profile
otherwise I'll go nuts updating 200+ pages :)
oh also....emc does not blend adjacent arcs
exact stop only
yhat is bad...very bad
many higher end cam progs fit g2/g3 arcs to contours
so that needs to be looked at
A single servo cycle only deals with linear motion, right?
(since it can only output a simgle velocity command)
this ends up turning into a quadratic interpolation, since the beginning velocity and ending velocity may not be the same
(but not calculated as a quadratic, just possibly physically ending up as one)
well the cubic is a cubic position/time profile
so velocity/time is actually quadratic
accel/torque becomes linear ramps with sharp cutoffs
for a given axis (joint), there is only an initial velocity, and an ending velocity
sometimes the infinite jerk in cubic can be benign
when paul and I were driving to dinner I told him to depress the gas pedal slowly at a constant rate
so if the smoothing occurs on the end velocity of one segment to the initial velocity of the next segment, it should be simple(tm), right :)
i.e. liners accel ramp
then I told him to verry suddenly take his foot off the gass ( sudden accel discontinuity)
but because friction is low, the effects are benign
just a non math example
Math is OK - you can't get too hairy an equation with IRC :)
There should only be major problems when there's a discontinuity in the milling profile
(and a discontinuity of some proportion - like greater than a 45 degree turn)
or when your machine falls apart
or the power goes out
well I showed him an example of an infinite jerk move in x and y
two semicircles joined together to make an s
with a perfect alignment?
(ie, a bezier or other cubic would have no problem)
centripedal accel of course suddenly reverses at the join pont
then the planner is running at too high a level
it should be dealing with linear segments
I showed him how a quintic would actually warp the circular path a bit near the join point
this is just a graphic example
in linear-land, the two segments would have identical velocity vectors, and there should be no acceleration "transition" between them
The we drew graphs of with Mariss does with rectangular convolving a trapezoidal profile in the g200x embedded emc
* alex_joni is curious
well in position trapezoidal TP blends with parabolas
yay... I finished my sitemap
smoothing by convolution makes a bit of a mess
large assymetrical path errors
yet you don't hear any complaints :D
well allowed following error is large
it is smooth...
but in a way tht can sometimes make big chordal errors
for a plasma cutter or something...
oh SWP we will see today if your sleep thing has any effect on queue emptying
although I understand that wasn't the reason for doing it
It shouldn't - it only affects the rescans of NC files
However - it's pretty trivial to change the sleep conditional to not sleep if the queue is below some percentage full
* alex_joni goes away
But - I'm not sure my change is in the same place
in fact, I
hmmm I need a speaker phone here so we can talk to fred
usually use skype
in fact, I'm pretty sure it isn't - mine is in the interpreter, not the command issuing function
(well - the routine that calls the interpreter)
perhaps a run to wallmart is in order before we call fred
are you guys still working with EMC1, or have you started in with BDI4?
No we agreed to switch this to the emc2 tree if we are not working with segmentqueue
will pull the stg and plug it into pauls bax
just take a few minutes
but he has to hack up a quick stg driver
heh - no rest for the wicked
well he said he can have something written in half an hour
so by tommorow We should be going
hand will be on the big red button a lot I assure you
heh - you should have a dead-mans switch
ha as much as I hate em
like on lawn mowers
I disable em
yep - plus a heart rate monitor
too fast or too slow, and the machine goes off
yeah rate skips then goes up after the e-stop dive
Well gosh 500kg gantry mass with 5.6 kva peak servo va is just ideal for debugging stg motion programs right?
yeah - nothing like the maximum stress test to prove things out
'scuse me - time to make more tea (thanks, Paul)
aw I can drop the PS voltage at first
might be prudent!
OK - mostly back
I don't see any queue checking in emctaskmain.cc
the most recent changes to cubic.c involvethat
uh yeah that is what I am looking at
let me go to emc2
let me get paul
les_away is now known as paul_c
cubic.c (for bdi-4) is in emc/motion (I think....)
yep - and there are no important changes from EMC1 to BDI4
and the same for emc2 HEAD
no real changes there either.
I don't think that instance of the word "queue" is in the right context here.
cubic just breaks an end point in to way points for the servo loop to work with (as I understand it)
so it generates (for the want of a better term...) a ring buffer with four points in it.
paul_c, if les is there could you ask him if he minds if i steal (with attribution) his article on "scraping with an angle grinder" for my machine tool construction wiki
Les says "No problem.."
* paul_c hides
pc...usually a good idea!
gotta go and plug Les' STG card in to another box and run some more tests soon
We found segmentqueue has a few "issues"
any that aren't already in the tracker?
hope they can be fixed
Not this week I don't think.
any way to test the output/performance without a machine?
dave-e: sure, run it with the simulated machine
but can you spot the problems that way
depends on the problem.
sometimes you can.
you could also use the headphone trick and hear most problems.
that is probably a better way...the ear is pretty good at that sort of thing
I think the remaining problems cause discontinuity which would be very easy to hear
but ... I don't know what problems paul found; nothing has been added to the bug tracker
Have a problem where a couple of long segments are being linked in with short ones, and feed drops to a very low value
I've seen that too. Do you have a simple program that does it?
and the other problem is with a long series of very short segments that causes SQ to halt
I have *not* seen this lately. I thought I had all the fatal errors fixed.
I'm in the process of isolating a small section of Les' test program that demonstrates the "bug"
or, I don't understand, please describe "halt"
do you mean it stops completely?
There is a line in SQ that does a check for s->ID == 474 (as I recall)
and as best I can tell, this is the point where SQ is stopping.
no, all that does is print a message
as far as I can tell
well please keep me updated. I can help with this.
SQ doesn't stop at that point, but that message is the only one to be printed prior to the halt.
you have the latest cvs, right?
I have used it extensively and have *never* seen a halt
interesting that "diagnostics" is used there, but "diagnosticsOff" is used in many other spots around that one
from tail end of last week, yes.
and you're talking about emc1, right?
SWPadnos: Too many prints at the RT level is not a good idea - You end up overloading the syslog daemon
cradek: All the tests with Les has been with an EMC build
right - the diagnosticsOff routine causes fewer erros to be printed, no?
paul_c: ok, just making sure
Les has just reminded me... There are also some max accel violations with SQ that caused some pretty hefty joilts
and that only occurs with the latest SQ, not the earlier version.
apparently, at axis reversal events.
Today, I think we will look at using the emc2 tree, primarily because I have some additional code that can be used to aid debugging.
time to head out to the shop to move heavy iron..
Will report back later, hopefully with some detailed observations.
paul_c is now known as les_away
Jymmm is now known as Red70sShow
Red70sShow is now known as Jymmm
hm, I wonder if I should go out to the mill.
anonimasu pick me up a 55lb bag of all purpose flour if you do.
make that a 100# bag instead
[19:46:46] <anonimasu> http://www.mecsoft.com/Mec/
you can see the same trouble I have with circles.
so irls? what gives?
I presume paul_c is over at chez les fixing segmentqueue?
an news from that yet?
so, does anyone want to grab my refactored/recoded version of tkemc before I delete the CVS?
hows the queue?
bad an cureable?
or fundamentally bad?
I think not but don't want to discourage too much
recent changes did not help
it gets stuck in a loop
even the old tp has always had serious problems
the queue runs out
more specifically the tp stack
if that were fixed
and if the cubic sub interpolation were made to run "softer"
then we would have a reasonable cubic splined TP
that works with more than trivial kinematics
so we have diverted to that while perhaps chris looks at segmentqueue
but it would still go bad on the "lots of little lines" type interpolation?
It should not
it should velocity adapt to prevent aliasing
and what about lookahead?
thats waht killed the old tp
We figure lots of little lines really must function...because cam programs spit those out...even if they are not desired
it should have at least 4 pt lookahead
I think Im past caring now anyway ...
was just wondering where/if to upload the remnants of the GUI I rewrote before deleting the CVS
I suspect that if the sub interpolator runs at the servo rate and the trajectory/servo rate ratio is reduced more cubic smoothing will happen
but first we must fix tp stack starvation
smoothing doesn't matter much if the thing is stopping to refilll the tp stack
We will give it our best shot
since the alternative is an emc that is not useful for industrial aps
Which will make the "little project" of late dead in the water
History Channel (if you got it)
les_away: theres always the G200X
* anonimasu sighs
les_away hand tools... very cool
I did an analysis of rectangular convolving trapezoidal profiles...more on that later
I saw the hand tool one...it was good
robin_sz: will gecko push around a servo system?
les_away hammer with a tuning fork built insdie the handle
yup dynamic absorber
robin_sz: stepper servos?
les_away exactly =)
robin_sz: do you know of any recent comercial machines that use step servos?
les_away Hey, have you ever heard of explosive being made from bleach?
* anonimasu nods
most just use steppers
well some bleach...hydrogen peroxide...
not sodium hypochlorite
Potassium chlorate explosive
bleach + potassium chloride
anon: this are a myriad of low end light industrial routers with steppers
anonimasu: thats part of the reason im off, ive got bored crapless by the "if its not real servos then its not any good" along with the "if its not linux its no good" and the "if the compiler isnt free then its no good" stuff, people seem to be more interested in ideology than results
les: yeah, but that's low end..
We are not doing that because it would be a me too thing
there are enough of them on the market
anonimasu: you also forgot to ask "will it do real servos" to which the answer will be yes
robin_sz: actually I've asked you that before
well step servos are real servos
well, laggy ones
but we don't want to do that
well they work way better then steppers anyway..
no they dont
they can yes
well, the servos I have work better then the stepper I have..
no they dont
it depends on what you use them for
a step pulse train is just another form of an analog signal
a servo motor in a system that has to hold a constant force in a fixed position burns itself up
a stepper just sits there and hums happily
its like saying "diesel engines are better"
it depends on the application
anyway actual position on a display is a requirement of what we need
so thats steppers or servos then
well heh a stepper with a feedback device is technically a servo anyway
have one on the table right here!
guess so ..
a 200 pole AC servo heh
be intersting to see if Mariss gets the "unstallable" idea to work
les_away you want to see the rest of it?
robin_sz how dod your usb breakout work?
robin_sz how did your usb breakout work?
robin_sz fired it up full bore yet?
sure, its been spinning steppers for a while now
robin_sz cutting air?
robin_sz: I dont care really whatever that works, but I like emc and hope for it to be able to push a machine at industrial speed some day..
fixed the bugs with the analogue inputs and some of the IO
robin_sz Ah, let me know after you got soem sanddust or swarf goin.
les_away you was to see the whole document?
les_away /join #xyz and I'll paste in there
anon: we are trying as hard as we can do get emc working at industrial speed
unless ppl want to make explosive from bleach in here
I just want some nitric acid wood stain
les_away Hey, I already found you a source.
anonimasu: sure, whatever works. I like some bits of EMC, but other bits suck, and I cant see those changing anytime soon due to fundamental desing issues
$50/liter makes me sad
anyway, does anyone want to grab this tarball of the GUI before I hit delete?
[21:08:06] <Jymmm> http://www1.chapman.edu/~jipsen/mathml/asciimathcalculator.html
heh robin don't you delete that yet!
robin_sz: too bad.
well grab a copy quick then
* anonimasu takes a break
[21:09:37] <robin_sz> http://www.redpoint.org.uk/cgi-bin/cvsweb.cgi/tkemc/
tarball link at bottom of page
it was a rewrite of the old tkemc with some sort of structure before I grew tired of tcl
the devils own language
robin_sz I thought that was perl?
it was tcl
how i deteest that langauge
ok delete away...I have it
the next lang I'll learn will be c or python
not c++, ansi c
* Jymmm be old skool
I will go back to the shop and see what Paul is up to...we had some pretty ecxiting moments getting servo to work on the emc2 tree
or do Perl then C maybe
k, later les
Nah, not perl. maybe way down the road.
Perl is very C like, but without the compiling
I can't for the life of me find 1/2"-5 acme
I have a some ROA perl books.
yeah, the "camel" book
and the llama
I got lazy and learned php, so I want something with some balls to it now. I REALLY should lean towards C.
php is a really bad example
in all web stuff the good technique is to sperate your code from your presentation layer
PHP just mixes it all up
Well, if pascal was still mainstream as it used to be, I'd probably pick that up again.
well, try Perl .. its pretty hardcare
Way down the road I will. I'm goin for C to be the next one.
I need gui and electronics
PIC, uC, apps etc
well forget C then
There are some GUI toolboxes for ansi c
old and hard to use
and thats all I need
oh, well try ALGOL 68R then
if I cant pronounce it, I aint touching it =)
thats hard to use.
so is BF
all modern GUI stuff is object oriented, you cant really do it any other way without great pain
yeah, I know.
just sadistic like that
back for a bit
paul is just looking at a lot of diffs
les_away: Ask paul if I need to epoxy a carry-on handle to the top of a Compaq..
* Phydbleep will get rid of that #&^*$%^&* machine one way or another..
tomorrow is more testing of the prototype..
* anonimasu yawns
got to run it for the remaing hours.. of the week..
I dont know how the DC motor holds up to be pulsed..
to sow a seed?
so it does one turn?
it's a seed feed mechanism..
like a plunger?
it turns and feeds seeds..
nope, cant really go in that detailed on that..
sounds like a job for a stepper :)
how would I drive a stepper?
driver electronics would add too much stuff
same way you drive a dc motor with a drive?
Stepper or dc motor with a spring/release.
I have a DC motor with a gearbox there already..
this is on a mobile platform?
it almsot sounds like a job for pneumatics
Several tons of mobile platform from the sound of it. :)
pneumatics arent applicable..
So...gas/diesel engine power plant
this is a 8x8cm thing
I am just curious if the motor will live if I run it like - - - - - -
anonimasu: Ah.. Model, not full size?
it's full size..
Well, I was going to suggest using a RC servo
anonimasu: Sleeve or ball bearings in the motor?
lots of power inoput for accel, little revolutions .. it might =get ratehr hot
and a support bearing in the end of the shaft..
Should be fine as long as the dust seals survive.
it's 200:1 geared ;)
What's the duty cycle?
how high does it jump?
What's the duty cycle? OFF/ON times (approx)
So a 0.3 sec pulse every ??? seconds/minutes/hours?
heh...geek-trap AND engineer-trap
i installed bdi 4.20 yesterday on a computer that was in a grocery store dumpster of all places
fenn: Woohoo I call dibs on the touchscreen panel. :)
a hole takes <1,4 seconds to dig.. ;)
I dont really know since I'll be counting how many seeds I release..
strange things happen though... random programs stall and comes back from screensaver with colored static specklies over the gui
its just a pentium 2 desktop
I'll run until the "plant X seeds" condition is met..
or until it times out..
fenn: Blow the dust out of the video ram or change the card.
if the sensor is broken
Phydbleep, could it have gotten metal dust in it already from sitting in the shop one day?
So ~ 5 secs max.. What's the travel time tot the next hole?
fenn: Regular dust will give wierd effects if the vieo ram gets caked in it.
Phydbleep, will try that
or well 0,3 might be long..
fenn: You could try ripping it out of the case and putting it in a big tupperware if you're worried about metal dust.
a couple of ms might be enough..
Phydbleep, i was thinking of just putting furnace filter material over all the air vents
anonimasu: So you have a <5% duty cycle on the motor.
The batteries will be scrap before the motor is.
Phydbleep: it's a forest machine the equipment mounts on.
Phydbleep: as long as there's disel there's power.
fenn: Furnace filter is OK, but not the greatest.. Check the clearance racks at Mall-Wart and others for HEPA filters for humidifiers. :)
Phydbleep: I was thinking about how the internal of the motor will work, how long..
given the enviroment (mud, water, clag and grunge) i figure IP65 sealed is a must :)
anonimasu: Nice diesel to fit on an 8cmX8cm platform. :)
you could try using one of the bearing companies' bearing life calculators
anonimasu: As long as the seals survive the motor will be fine.. YOu can take a cheap-o motor and make it last a long time by putting it in a sealed box with good shaft seals.
robin_sz: the motor will be enclosed in plastics.. with O rings..
recycled milk jugs?
wont they get trashed when the oh so careful operator uses it to hammer down some brash?
cant remember what the plastic is called..
robin_sz 36 DD
it's white and cuts nicely..
Jymmm: can you recycle milk?
Phydbleep: cost isnt a issue, I could put a servo there and it'd be profitable.
That matches a lot of plastics.
right, enough of this .. im off.
so long, and thanks for all the fish
We knew that.. Are you leaving too?
Phydbleep: it cuts very good
Hehehe, G'nite robin_sz :)
anyone know of a source for 1/2"-5 TPI ACME rod?
so you could say the motor is sealed.. the only stuff that could come inside of it are seed parts..
anonimasu: Add a layer of seals to rpevent that and the motor should last the brush or bearing life.
I was thinking about brush life if you run it on/off lots..
anonimasu: Or just make it easy to clean/replace and say "Fsck-It".
Phydbleep: a hours of downtime costs too much..
Make it part of the seed hopper if you're really worried.. Reloading the system swaps the hopper/motor/feeder.
Plug in the control cable and off you go.
well one hour in 4 years is one thing..
but all downtime is bad..
1 hour a year as part of the annual maint..
yeah they still have mainentance on the other stuff..
.1 hour a year as part of the annual maint would be all it really needs to open/blow/close...
Jymmm, just use a delrin one :)
Phydbleep: compressed air?
fenn delrin rod?! delrin nut, ok. but rod?
Jymmm, yeah i know weird huh
actually I could put the motor housing under pressure.. but that's overkill.
anonimasu: Yeah, pop the cover and blow out the crap.
what s good size setscrew to use for shaft couplers?
Jymmm: I see 1/2"-10 and 1"-5 but no 1/2"-5
Phydbleep Yeah, be neither, except the delrin
why does it have to be 1/2-5?
Jymmm: 30% of the shaft dia for a set screw.
fenn Well, it's the 5 TPI I'm mostly interested in
Phydbleep ah, TY
Phydbleep what about thread?
setscrews always seem to use fine thread for some reason
or was that thread
Jymmm: That's to make sure it stays.. Use as fine a thread as you can for the holes/screws.
i use two setscrews at 120 degrees apart with coarse thread so i don't have to buy another tap
or 90 degrees apart sometimes
for 1/2", I'd use 1/4-28 or 5/16" fine.
I dont' mind buying a tap, I dont want to order shitloads of setscrews
Jymmm: Pick something you will use more of later?
* Phydbleep ended up with a 'Dorman' box full of various setscrews and keystock.
will a 5 -vs- 6 TPI make a big difference?
Jymmm: Depends on that you're trying to do.. It will add a little (~20%) precision and bork your nice X/5 math. :)
That's if you're using it for a leadscrew.
I would be, yes.
i gather he's not going for precision
they have 1/2-10, and then 5/8-6.
fenn ya think? =)
fenn I said say acme and not ballscrew =)
So X/5 becomes X/6 ,, Except stock acme is 5/8"-8
I see 3/8"-12 as a stock size.
Phydbleep I found 5/8-6 and 5/8-8
1"-5, 3/4"-6, 5/8"-8, 1/2"-10, and 3/8"-12 on this list.
Phydbleep your looking at mcmaster
hmm mcmaster had a 5/8"x6?ft ballscrew for something like $25 but i cant find it now
Bout a mile over from me. :)
yeah i was surprised too
Phydbleep oh, heh
[22:14:41] <Jymmm> http://www.use-enco.com/CGI/INLMK32?PARTPG=INSRAR2
They have 1-3/8"-4 if you want to commit assault with a threaded weapon. :)
Phydbleep Shit... I found 1"-2 TPI
make sure that it's actually a single thread and not multi start, unless you are into multi start threads
the 2 TPI ?
Jymmm: Why are you showing me a page where I cant even afford to pay attention?
well, any extreemely coarse thread like that
Phydbleep for available dimensions
Jymmm: Yeah, 2/3/4/5 starts.. That way the thread isnt a twisted flat-bar. :)
Jymmm: When I hit the enco page it was all motor kits.
i keep hitting reload on the enco page and it changes every few seconds
Cheapest thing I saw was $950+
[22:18:55] <Jymmm> http://www.use-enco.com/CGI/INSRAR?PMSECT=451
is that any better?
I hate when websites do that
that's actually not bad, $20 for 6 feet of very stout leadscrew
fenn which one?
$7 for a 6' 1/2-10
1"-5 is $17.19
i got some of their 1/2-10 and it is very pretty
I'm think fo the 5/8-6, fewer threads should help the overall speed
i can't vouch for accuracy
fenn first machine, so I'm going for cheap prototype
yeah i guess a heavy leadscrew can be more of a disadvantage sometimes
5/8-6 on the X, then 1/2-10 on the YZ - gantry router
Jymmm: ICK!.. Match your X/Y pitches if you want round circles.
i was surprised how fast steppers could go first time i saw my friend's gantry run
Phydbleep Ah, wasn't aware of that.
ok, 5/8-6 on all three.
Divide 3/8 then 3/10 then find a common factoral in your 200 or 180 ppr steppers. :)
It's a lot easier when the x/y pitch match.
i figured your error tolerance should be above the resolution of the steppers anyway
Well.. Match or are an even multiple.
probably easier to use the same thread anyway so you can just crank out whatever needs to be done
instead of fiddling around with several different sizes
fenn: That is why there are the doubles for the threads.. 4/8 5/10 6/12 :)
Phydbleep, i always thought it had something to do with the angles involved
I saw some 1/4"-20 acme somewhere a while back.
was it on something you made on the lathe? :)
fenn: Nope.. Commercial product.. Remco carried it for a while.
heh. if I machine FASTER I get smaller chips.
1" long ribbons of alu.
or at least thinner
much better than the 10" of the same thickness.
I think I'm going 180mm/min
on the lathe
it's a TMNG221 bi
i guess that means no chipbreaker... you can add one by clamping a little triangle of metal that goes almost to the edge and forces the chip back in on itself
a busted positive rake insert will work
it goes have a chip breaker.
[22:46:21] <A-L-P-H-A> http://www.carbidedepot.com/formulas-insert-d.htm
TNMG221 (this is the correct number now)
well i'm going to bed. maybe it will be warmer tomorrow and i'll actually get something done
Phydbleep when you said 30% for the setscrews, did you mean the length of the setscrews being 30% of the shaft diam? so, a 1" shaft should have 1/4" long set screws at least?
Jymmm: No, About 30% of the shaft dia for the screw dia.
Phydbleep ah, ok. glad I asked =)
1" shaft, 3/8" setscrew.
3/8" or 7/16"
Basicly you want a good secure lock.. use 2 or 3 if you feel you need to.
Yeah, I was considering doubling up if needed.
parallel to each other, or may at 90 deg
bbiab .. Going to dive in the shower and soak out some sore spots.
90 or 120 degree.
2@90 or 3@120 degree.
Thanks again for the switches and stuff