[01:26:43] <giacus> http://www.giacus.org/files/gte01.png
a more complex scene :P
giacus: if you want more "things" you could toss in an ubuntu logo.
giacus: you might try making the laptop screen brighter & "glowing"
giacus that's cool! where's the hexipod?
jepler: yeah, have to wotk on it yet ..
giacus: So, when will the movie be released in theaters?
would be nice
giacus: And... are you working on making this available in g-code?
Jymmm: there are a bit of blender resources online
Hey, I have issues with 2D, much less 2.5D or 3D
Jymmm: I have to try yet to import some object to see
I'll try with tux in the next days ..
giacus I was joking, though it might be cool
jepler ever heard of Sudoku ?
Jymmm: I've played a few games of it
why do you ask?
jepler was wondering if there's any theory behind it math wise
I'll just do some research =)
Jymmm: You can make anything into a subject for a paper in a mathematics department. For instance, I know there's a paper about the exact number of different (filled-in) sudoku boards, and one open question is: how few clues can be given so that a board has exactly two solutions
the wikipedia article has some information about rules people use to solve sudoku but if you play you've probably aleady found some of them for yourself
I'm readinhg wikipedia now
and this... http://www.sudokusolver.co.uk/index.html
a page about two-solution puzzles: http://www.dailysudoku.co.uk/sudoku/forums/viewtopic.php?t=876&sid=a6abc33881c91eb606abd8ff1f9140f9
this occasionally links to sudoku-like puzzles: http://www.mathpuzzle.com/
-- for instance, one recent link was to "knight's sudoku" which involved filling in the numbers 1 through 8 and a knight in each 3x3 square, so that the knights didn't attack each other (I think it was something like that)
also, don't miss http://en.wikipedia.org/wiki/Mathematics_of_Sudoku
Ah, that's what I was interested in =)
"The number of ways of filling in a blank Sudoku grid was shown in May 2005 to be 6,670,903,752,021,072,936,960"
There never could be a 3D one, could there?
umm I don't know. there's this: http://www.mathrec.org/sudoku/cubedoku.html
I know that's not quite what you meant
I don't immediately see why you couldn't find 9 sudoku puzzles that you can put on top of each other, and have each "slice" also be a valid sudoku board
there are a lot of boards, after all
this one does seem to be a 9x9x9 puzzle: http://126.96.36.199/search?q=cache:eXblacvYhPkJ:www.sudoku.org.uk/discus/messages/17/102.html%3F1141697499
cool, I got lots of reading
Hey alex_joni, are you by any chance there ?
he's long asleep
ok, I was just going through how to implement the second part of my Torch Height Control and when trying to find a pay to pause the running G-code I disovered the halui things that alex has been doing
halui is not realtime, it should be used only as a human-interface
you can pause the motion in realtime by setting the feed override to 0
it's a hal pin, let me find it
no prob, I can find it ,
just needed a prod in the right direction :D
it's a float: 0 is stop, 1 is 100%
this is typically used for feed control on EDM
thanks for your tip, I will try to implement something with feedrate, although I might also use halui to set the gui button to "pause"
yes you could do either, depends if you need it to pause NOW or "soon"
this way the user will start the gcode file, it will run untill it hits the spindle_on, the torch will home itself, and set up for the correct height and wait for the user to "resume"
cradek: did you get a chance to try my latest axis checkins (from back at 5pm or 6pm)?
couldn't you just put m3 / m0 in your program?
the best way will probably be to use feedrate and halui pause but release the feedrate after
jepler: nope, is that jogwheel stuff?
cradek: not really, since the height is unknown and is dynamic
cradek: and the standoff changes from the initial pierce throughout the cutting time
Dallur: ok, I'll leave it to you!
cradek: yeah, it is another version of the "set hal signal values when choosing what to jog" but I think it catches all cases
cool, I'll test it soon
if you'll be up for a while I'll go do it now
cradek: not more than 20 minutes
ok, maybe tomorrow then
I used different pin names than you did
in the medium term they should probably be pins rather than signals
ok, I don't remember if I was using setp or sets
the code you pasted had sets, I think
but that's not in my /lastlog anymore
so you checked in everything that I need for the jogwheel controls?
I think so
mode switching and halcmds
you'll need to create the signals in a .hal file (I put an example in place)
but none of it is tested on hardware yet
look at configs/common/core_axis.hal
I thought about putting jog increment in there but if you can really get from the minimum jog you want, all the way up to top machine speed, just by turning the wheel faster or slower, then it doesn't matter.
well it might be nice to have .001/.0001
for touch-off I typically use .0001 at the end
but it seems like that's best done with a switch by the wheel
a simple one-bit toggle switch would work
yes, and one hal mux
hey I have that (the old index input)
I'll have to do that.
could we do one halcmd invocation to set all six signals?
echo "..." |halcmd -ksf
the equivalent of course (popen?)
but that would make it harder to replace with a real hal interface later, forget it
* cradek argues with himself
I'm not worried about 6 'halcmd' invocations
it's a silly thing to worry over
especially since it will become quite efficient when enough of hal is moved inside of axis
it's them changing at different times that bothered me
different enough to notice? hm
I'll let you know
"don't do that"
I could call 7 of them. 6 to turn off, one to turn on.
that would do it
but don't worry about it until/unless it's a problem
or it could be something other than a bit, translated by a bunch of blocks
there are options
* Lerneaen_Hydra pokes #emc
none of the commersial stepper motor drivers are designed with a microcontroller for the controller part
at least i haven't seen such yet
this limits the possibilities to create a real smart decay scheme for the current through the motor coils
in fact i've got a design that would kick ass if it was equipped with anti-resonance measure
so why am i bothering with this
well the answer is that this is a protest against poor and expensive commersial stepper motor drivers
i'm claiming that the only thing that stands between my design and the commersial ones is the midband-resonance compensation
so i'm looking for a partner in this issue
what type of stepper driver are you planning to make? if you have resonance compensation you must be talking about super-critical speeds for the motor
nah, i'm just trying to catch up with the commersial drivers on that detail too
and they say that they have this compensation (the good ones)
so besides this resonance issue my design already kicks ass
but actually adding that compensation only takes some math and a microcontroller?
is super-critical speed really needed?
well the thing is that i don't know
oh, is it a full/half stepping system or microstepping system?
well it's nice if they can go quick i guess
it's a microstepping driver
so there are lots of components?
the microcontroller does most of it
a programmed IC?
i'm gonna use ATmega8515
what's the estimated cost for all the components currently?
(It takes in step/dir I take it)?
the cost is right now around 40$ per axis
more like 300kr
that would be from input signal to high-current output I take it?
pretty much yes
but i haven't carefully measured the cost of the power stage really
ok. what voltage does it drive the motors with?
i'm hoping to aim for at least 50 volts hopefully even higher
oh, that's quite a bit
the best about my design (1/32 microstepping at best) is that it has a pretty clever adaptive decay scheme for the motor coils
I take it that you have an advanced chopper drive then
decay scheme? you mean for the inductance?
oh, does it need to be calibrated for each motor?
i'm claiming that it would be somewhat difficult for a stepper driver designed w/o an mc to catch up with this
or is it a generic-type thing?
i think it could go as a generic in fact
but it's too early too determine at this point. but at least i think it wouldn't need to be re-calibrated for another motor
that counds really nice
well it would have been strange if mine wouldn't win when an mc is used in the design
don't the expensive controllers use a mc?
you can't catch up w/o an mc for stepper or servo drives or whatever
an mc will pretty much always kick regular ttl designs in the butt
you've got a lot more flexibility to do whatever you want
that is absolutely true
ang. ATmega8515 finns det tillräkligt många timers (lediga) för att mäta in fast decay?
det ska finnas 3 men går några åt till att ställa pwm?
eh, du behöver väl 2 lediga timers eller?
chinamill: 8515 har bara 2 timers
chinamill: en för pwm och en för blankning
chinamill: kanske måste jag har en extern timer också
man beöver bara en timer för att sätta de båda pwm kanalerna?
chinamill: jo det är sant
chinamill: men jag kom på att jag kanske behöver två timers för blankningen
chinamill: men jag är inte säker på det
det räcker väl med en?
chinamill: jo du kanske har rätt
= samma blanknings tid när båda bryggorna ska blankas
chinamill: men då blankar man bägge kanalerna även om det bara verkligen behövs för den ena
bara, när det ska vara så... man får väl sätta upp några regelpinnar
chinamill: och det är onödigt att blanka bägge samtidigt om det bara behövs för den ena
det skulle man faktiskt kunna göra
chinamill: regelpinnar! smart!
chinamill: mycker smart till och med :)
Idag är det svenskt på emc :)
chinamill: helt klart
ja, dom andra kommer bli mycket fundersamma när de kollar loggen
chinamill: hoppas vi inte blir kickade :)
Bo^Dick: har du databladet på den där 3955 för att visa Lerneaen_Hydra om blankning?
[09:47:24] <Bo^Dick> http://188.8.131.52/lgh/datasheets/3955.pdf
Lerneaen_Hydra: I det där dokumentet står blankning förklarat... jag kommer inte ihåg vilken sida...
chinamill: jag kanske borde satsat på den andra av atmels kretsar med 3 timers istället
chinamill: men den hade färre I/O pinnar
Det har i alla fall med att göra momentant blanka ut senseres signalen för att undvika att kontrollenheten jobbar mot en spik på senseres
det funkar nog vilket som..
chinamill: en snubbe på #microcontrollers sa att man inte förlorar prestanda om man kör D/A omvandlarna seriellt från mc:n
den IC kretsen gjorde väldigt mycket, är den en stor del av kostnaden?
Lerneaen_Hydra: den ska inte användas. den är för svag
Lerneaen_Hydra: jag vill ha åtminstone 5 amps på utgången
Bo^Dick: ja, 1.5 var lite väl lågt... och att förstärka är ur frågan eftersom den har aktiv-avkänning av spänning/ström...
men om man ska köra seriellt till d/a:n måste man väl använda 2 rätt mycket dyrare d/a:s?
egentligen skulle man kunna parallelkoppla 3 stycken L298 på utgångarna för varje lindning
chinamill: även det ja
finns det liknande men med större uteffekt?
under förutsättning att de går att få tag på super-billigt
Lerneaen_Hydra: vet inte
går det att parallelkoppla L298?
chinamill: borde vara möjligt
chinamill: rent praktiskt alltså men jag bara gissar
Lerneaen_Hydra: Det verkar skralt med hög amps IC:s
3955 gick ju inte att p-koppla
5-10 ampere är väl inte så mycket?
Lerneaen_Hydra: för att sitta i en IC är det rätt mycket
iofs, den hade inte ens koppling för kylfläns
det var en vanlig dip kapsel va?
kanske är bäst med vanliga IRF530N
I think I've made my final version of the new chips.ngc program (at last). ftp://basic:firstname.lastname@example.org/chips_final.nc
fairly small too, only 22k rows
total production time of 30 minutes
it's quite big though, 100x70x40mm, so it's understandable
looks like I scared all the borkborkbork peoples away...
eh, hur går det med din svarv?
den är nästan klar, funkar bra
det enda jag har kvar är mina jogwheels och encodrarna för gängning
Jag blev lite lätt avis, när jag såg din film :)
asså, har du sett filmen?
inte alla bilder
vill du ha?
eller, jag kan ta bättre senare idag
jag har bara gamla nu
du har en fräs med EMC nu eller?
ingen fara... jag är avis ändå ;)
en sunkig fräs & ett plasma bord
mycket glapp/snea gejdrar?
fräsen är en lågbudget från china :(
Men bättre än inget alls
jojo, det är klart
själj letar jag efter en liten prescicionsfräs nu
jag har fixat bort glappen med fjädrade muttrar
* Lerneaen_Hydra känner sig dum nu, jag fattar ditt namn först nu
eh, jag måste dra nu, bbl. Den volvo ocean race går i mål snart, ska ut o kolla med familjen
* Lerneaen_Hydra is away
03alex_joni 07HEAD * 10documents/lyx/emc2/hal_drivers.lyx: modified SubSection and SubSubSection, along with some touching by Lyx from Ubuntu Dapper
03jepler 07HEAD * 10emc2/src/rtapi/rtai_rtapi.c: (log message trimmed)
nano2count(arg) seems to return the number of counts that is nearest
in duration to arg, not the largest one that is less than arg. As a
consequence, many small round-number frequencies (e.g., 14000ns) would give
errors at the HAL layer, since the resulting period was actually 14248ns.
This affected short times more than long times, because the HAL layer accepted
an interval up to 110% as long as the one requested.
Good job :D I just ran into that yesterday :D
Dallur: nice webpage
thanks, I wish I had more time to update it though
Dallur: yeah, I thought I saw you say something about that
Dallur: I'd run into it too
Dallur: you don't mention emc in there though..
alex_joni: Nope, not yet, I only started using it about a month ago and I am still getting everything up and running
I have not updated it much for the last 2-3 months, but once I have the Torch Height Control ready I will put everything there
The funny thing is that I am using PCBs from Bob Campbell, which were co-designed by Art of Artsoft :D
Once I am done the THC from both candcnc and bob cambell should work out of the box with emc
I poking around a bit yesterday, I was going through the source and I noticed that you have added the pause pin to halui but it is not enabled yet
let me refresh my memory
since 2.0.2 you added the halui.program.* pins but they are not hooked up to anything at this point it seems
2.0.2 ? you mean CVS Head I think
they are in cvs head
but not in last release
2.0.x is feature frozen
they will be in 2.1.x at best
:D do you know if they have a date for it at this point ?
2.1 that is
around september ;)
and it's not "they"
not me either
that's more like it
I see pause is in there, and should be working
hmm want me to report a but on sourceforge ?
err a bug
if it's a bug yes
did you load a program?
from another GUI?
setting it has not effect on the execution of the g-code
hmm.. let me try it too
there is a halui_halvcp config
I use that usually
Auto starts it
Pause stops it
Resume makes it go on
even step works
let me check again to make sure
I'v seen you are talking about the campell THC
I got THC up/down movement already done
Have you made it work yet?
I have some code allready
:D we can share and merge it
its in spagethi mode though
I took what you had on the wiki
and used your naming and such
Dallur: let me know what you're seeing
and also set DEBUG pretty high, to see if the NML gets sent
alex_joni: will do
Dallur: it works as promised here
I made a hack out of an excisting component so some of the naming is not good
chinamill: let me just finish trying to reproduce what I was telling alex about and after that I can send you what I have and hopefully see what you are working
I'll be around...
Dallur, chinamill: maybe it's usefull to have plasma/ as a sample config in emc2
but first, nice code... not like mine...
chinamill: we'll work it out ;)
I think we also need to add a couple of things to one of the UIs :D
it's woking like it should, (I thnink)
Dallur: is that you? http://www.dallur.com/index.php?id=44&tx_lzgallery_pi1[subg]=10&tx_lzgallery_pi1[showUid]=41&tx_lzgallery_pi1[old]=5x5x1&tx_lzgallery_pi1[pic]=12&tx_lzgallery_pi1[colrows]=1x1
alex: nope, this is the guy that I build everything with
alex: we have been friends for a long long time and we work on our crazy project togeather, like the plasma table
or the sailboat ;)
thats me :D
Btw, I got the pause to work now,
Dallur: was anything wrongß
I will try to find out why it was not working before but If I don't I will assume PEBKAC (problem exists between keyboard and chair)
alex: It looks like it was just a user(me) error, sorry to trouble you
Dallur: no sweat
Dallur: my THC version is functional and it uses M103 and M105 instead of M3 & M5. How far did you get?
well I did this:
it is on the wiki
and I will put the config in pastebin in just a sec
too bad, it seemes like double work
I pretty much implemented it in HAL
So, now, is there something that does not work in your version that we can use from mine?
giacus: seen it already :P
giacus: doesn't seem like that
oh.. Ubuntu instead of Debian ;)
also the laptop brightness
going to add some other stuff
few tools around :P
giacus: one thing I noticed
that server farm? on the right
in the back
doesn't seem to be sitting right on the desk
ok, I'll check it
looking for a server rack, should be nuch better ..
have to open the door at 30 degrees and use an image for the outside of the room
and add more stuff, neon, light on the top of the wall ..
don't forget all the chips and coolant around the milling machine ;)
oh, and an endmill in the spindle (I can't tell if there is one)
I like all the finished penguins on the table on the left
and the blocks of stock
it's very funny
and the tape measure, for fine tolerance checking ;)
and the robot putting in the brains (or whatever it's doing)
what you don't noticed maybe is the camera alignment ..
if you look the bricks on the top side of the wall you notice that
does anyone have a clue
what threading inserts are internal or external threads?
I cant remember if "ER" is for external or internal
I cant seem to find it on the web either :)
me is getting tired
seems like that
I really dont feel like going to work to look what insert is the internal/external ones
giacus: still haven't fixed those PC's on the right back
they seem to be hanging out of the desk.. somehow strange
[13:56:19] <anonimasu> http://www.iscar.com/ProductLines/PDF/MainTypeInserts2_33.pdf
alex_joni: is that what you meant ? http://www.giacus.org/files/0001.png
they seem to go more to the back then the desk
because is dark there
you notice it on the one that's on the left
are they too bis as size ?
they are too far to the back
the desk should be close to the wall.. right?
[14:06:54] <alex_joni> http://dsplabs.cs.utt.ro/~juve/emc/modify.png
giacus: is that more clear now?
Rugludallur is now known as Dallur
alex_joni: how you see it now ? http://www.giacus.org/files/farm1.png
[14:30:29] <giacus> http://www.giacus.org/files/farm2.png
[14:35:50] <giacus> http://www.giacus.org/files/gte03.png
I think the pc case aligned to the desk border would be not so good, did you mean that too?
giacus: maybe put them under the desk?
* alex_joni goes away for a while
tried .. I don't like it
I'll try to build a rack instead, and remove the desk
that's not the right place for the server farm ..
cradek: is there any other way to enable adaptive feed than to use G50 ?, a pin or parameter ?
Rugludallur is now known as Dallur
Rugludallur: no, but I think you can specify G50 in the startup codes in your ini if you always want it on
I am having problems linking hal pins during initialization from the ini file, is it recomended/needed to put in some sort of delay after HALUI=halui ?
Dallur: right now that kind of thing (adding a delay while a userspace hal component is set up) is the only solution.
It seems that I am unable to link halui pins to signals after starting up the UI, but from a .hal file it fails
Dallur: I assume the problem is a possibly-intermittent message that a halui pin doesn't exist
jepler: so the halui pins don't become available untill after the UI has loaded ?
something like that. the pins are created by hal API calls inside the halui program
jepler: ok thanks, I will try introduce some sort of delay before the other .hal files are executed
hi all...scorchin' hot here today...
03jepler 07HEAD * 10emc2/bin/wait-for-pin:
use new script wait-for-pin in scripts/emc, instead of sleeping a fixed time and
hoping for the best. wait-for-pin waits at least 2 seconds by default, but
usually finishes much earlier than 1 second, as soon as the named pin is
03jepler 07HEAD * 10emc2/scripts/emc.in:
use new script wait-for-pin in scripts/emc, instead of sleeping a fixed time and
hoping for the best. wait-for-pin waits at least 2 seconds by default, but
usually finishes much earlier than 1 second, as soon as the named pin is
jepler: Was that message from CIA indicative of a fix so that you can link to halui pins ?
mess here on the ubuntu live cd ; )
jepler: during init from the .hal files
Dallur: well, I don't use halui so I'm not entirely sure it works
Dallur: but that's the intent
Dallur: it replaces hardcoded 'sleep 1' with a polling loop that waits up to 2 seconds, but may wait less
I'm a little surprised that halui isn't set up after 1 second, though
jepler: just did a build, still broken, I think I will report on sourceforge but it is not critical for me, just a "nice to have" usability feature
jepler: it might be that it actually waits for the UI to load first, in which case linking to pins during init will not work :P
it looks like specifying halui in the .ini file does load it after all the .hal files, which makes no sense
jepler: hmm I figured doing HALUI=halui would load it in the order specified in the .ini file, instead of running it after the rest of the .hal files
I don't think te order of the sections has any effec
maybe you should just try putting 'loadusr halui' in your .hal file
maybe followed by a 'loadusr wait-for-pin halui.program.step' to wait for it to settle
I will try that, thank you :D
jepler: halui fails to load with loadusr, missing shared memory so I figure it depends on the UI, I will report this as a bug on sourceforge and see what happens
jepler: thanks for all your help
maybe there's no safe order to do it in :(
or maybe it just needs the "-ini" argument
jepler: that is kinda what I am leaning towards, there might actually be a need to have hal files that are load after the UI has been initialized
well, I guess I better go and get some cream for it, before it starts to ooze.
oops, wrong window
is it well with you?
not if he's gonna start to ooze...
* robin_sz observes his ebay bargain
nice Xerox phaser for £39
oh yeah. Had a breakthrough yesterday. And primary subject of discussion at the last meeting was a manufacturing facility.
les_w 7th hole?
but the thing already made 4KW from just breathing through a straw
today, I'm just detailing cars and mowing grass. Steak cookout later.
* robin_sz is eating pizza and chillin'
after steak, a glass of bourbon.
les_w: i have got a generic servo motor question for you....
well, anything to take the taste away i guess
I have a 12 volt motor that I am going to attempt to use as an el cheapo servo motor....
few hours for the big match
go USA go go :D
would it make sense to use a servo drive with 25 volt supply on this motor to improve the response?
a young guy wanders into a bar and orders a double whisky ... "celebrating?" asks the barman. "sure am ... I had my first blowjob today!" .. after which , he downs the whisky in one shot. "another?" asks the barman. "nah .. if one doesn't get rid of the taste, I can;t see a second one helping .. " ;)
lawrence is it a dc brush motor? shunt or pm?
les_w: a linear current drive design using a 10amp power op amp... motor is pm ( actually a heater fan motor from a car )
ok. Yeah drive it with a current source. If the thing can take the rpm 25v max is fine.
one word of caution ... are the brushes symmetric on the shaft? those motors are often designed for one direction only
les_w: under control, I dont see it ever getting close to max rpm..... I could run the rails at 12v, but I thought due to the motor inductance that having the capacity to hit it with a faster risetime current pulse would make it perform a little better ( at the expense of extra heat in the driver )
basically, volts for rpm, current for torque
adding volts wont help if you need mroe torque etc
and a little inductance is fine for integrating the pulses
no linear drive.... and the motors seem fine bidirectional.... loads of torque in them... looking at direct drive onto a fine pitch lead screw.... screw whip will be the limiting factor
even with 12V the electrical rise time is way faster than the mechincal time constant
I think the inductance is a non issue
LawrenceG: any feedback to the amp? current?
or is the amp strictly voltage in - voltage out
robin_sz: cool.... I havent measured and characteristics of the motors other than hook them to a supply and try and stop rotation to measure locked rotor current
the electrical time constant will be way way shorter than the mechanical one
LawrenceG, for maximum performance make sure your amp can supply at least the "locked rotor" currnet
that motor will have lots of rotor inertia
* robin_sz nods
yea... I realize they are not the best servos, but I am going for a servo system that is less than $50/axis including drive, motor and encoder!
well, treadmill motors are also good and cheap
some of them are pretty good servos really
yea... I will try driving one of those as well..... I am using a split supply +-25 volts to feed the op amp in a current feedback configuration
LawrenceG, usually you'll need lots of torque, not so many RPM, fan motors tend to be wound for rpm, a treadmill motor is wound for torque and will accept lots of torque at low rpm without buring up .. a fan motor might not like being stalled for long
* robin_sz looks forward to going flying tomorow
thats one of the first tests..... see if the smoke leaks out of the motor or the drive first
its a bugger to get backj in
the company next door to me does that ...
I was going to put a pulley on the motor and have it run a pail of weight up and down for a few hours to fine the weak spot
easy when you know how
well I'm gonna hop in the oldmanmobile and go to the butcher and get filet mignon now....
mighty fine idea.... just past breakfast time here
heh, dont excite yourself too much granpa
is there is web site I can paste a picture without having an account created?
go big Les... the whole strip
LawrenceG flickr.com (if you have a yahoo acnt already)
thaks Jymmm... will try and post a couple of pix of the servo board
cool, got the motor spinning yet ?
[17:54:22] <LawrenceG> http://www.flickr.com/photos/64111811@N00/?saved=1
try that link
* Lerneaen_Hydra_ meeps robin_sz
the opa549 is a similar powerstage as the irams ?
* robin_sz meeps back
no... its just a 10 amp +-30v op amp
Lerneaen_Hydra_ is now known as Lerneaen_Hydra
hmm. so this is for brush motors ?
LawrenceG, are you familiar with "safe operating area" as a concept?
I am just looking for a cheap drive for some brush motors that can handle around 5 amps
robin_sz: oh yes.... I intend on running it at much lower than rated specs
the 10amps is like a for a short pulse.... they say 8 amps continuous, but that is probably staurated output stage hard to one rail
remember, a motor will pull max current and lowest volts in a stall
ok, so .. 8 amps ...
say the motor pulls 8 amps at 3V in a stall, right?
you have 25V rails, so ... 23V x 8 amps to be dissipated in the op amp
175W ? roughly?
thats one of the reasons I wanted a current feedback drive.... I can size the resistor tho keep the drive from melting down.... If I only can expect 3 amps in servo servo, I can live with that
well, all I can say is this. even our early 70s "boss" based bridgeport had a switching supply to the motors
pwm has a lot to offer for servo drives, linear is problematic
what is Ptot for that amp?
pwm has its own set of design issues.... I have a honking big supply already and a little heat during operation is ok..... this is just a little step above the l298 2 amps limits
what is Ptot for that amp?
[18:04:21] <LawrenceG> http://focus.ti.com/lit/ds/symlink/opa549.pdf
looking... depends a lot on how cool one can keep it
the nice thing is that it has temp shutdown and current limits built in to protect itself against bad designs!
they are expensive little units, but ti will ship 5 samples for free
so ... 25W with a 5 deg c/watt sink
these will be bolted to a rather large chunk of heatsink
fan cooling if needed
* Jymmm hands LawrenceG a bottle of LN2
I could run it in the aquarium an make fishy tasting tea
good day gents
the motors I am playing with draw about 1amp no load and about 5 amps peak... I wont know what the real demand will be until I get them turning on a machine ( a small mill bigger than a maxnc, but smaller than the desktop minimills
I think this drive should be a reasonable match to the motors
what are you going to use to send to the amps - filtered psudo pwm out of emc - or do you have a interface of some kind
gents i've missed last weeks sunday, did any proposals on trajectory planning occur?
giacus: what program did you use to make the dxf to stl?
skunkworks: quadrature drive from pc..... with the pid loop in the local microcontroller
the power portion takes filtered pwm from the dspic and drives the current feedback power op amp
the pic could be pulled and the encoder and drive could be feed to the pc parallel port
the pcb was milled by emc2
I am doing a ir2111 controlled h-bridge - but at this point I wasn't going to do any current limit. wondering if I should
Guest138: what do you mean proposals?
I've seen some details on PID trajectory upgrade, lookahead. if I may say..
improved lookahead would be nice. now it's two segments I think. someone working on BDI-EMC claimed it had more
Guest138: there was some talks, but nothing tangible
me boy is off to reboot ubuntu on his machine... : )
.. so no lookahead feature anytime soon I guess..
guest138: if you write it we can have it tomorrow :)
jepler had looked at the emc tp - his comment was it would probably take a good weekend to test it out.
or something like that :)
* giacus goes to watch italy-usa :D
USA can't play football... why bother watching...
I think can win ;)
because losted the first match
czech rep. 3.0
did anyone try EMC2 on a dual CPU?
A-L-P-H-A: they can't play soccer, they can play football
Guest138: RTAI doesn't really have good support for dual CPU
Guest138: so it should work, but nothing impressive imo
A.Joni, I'm just looking on some timings of other peoples setups, what # incicate the level of data starvation in the traj planner?
uh... that's football, and american football. foo. :P
Guest138: the problem (from what I understand) is how the data gets to the RT part
Guest138: right now there is one command / message sent to motion (through SHM)
easy saturday reading http://www.toolmonger.com/
so if you have lots of small segments, you might end up with problems
[18:42:33] <A-L-P-H-A> http://toolmonger.com/2006/06/15/how-to-build-a-game-chair-with-a-380-wire-welder/
that is the case.. so the solution for now would be faster cpu and memory?
that's why it makes sense to blend segments in non-RT and send blended segments
Guest138: probably so.. not sure how far this can be pushed
I have to try the g64 with the p params to see what happens..
Guest138: sounds like a plan ;-)
still would be nice for some good sould to implement the lookahead .. :D
Guest138: I couldn't agree more ;)
Guest138: do you have a good soul? :D (lol)
ya...but bad programmer..
giacus: you there?
ok gents ciao..
03alex_joni 07HEAD * 10emc2/src/emc/task/emctaskmain.cc: removed some old unused code
* jmkasunich is recovering from a week of listening to stupid stuff
* alex_joni is working on task
jmk: have you tried the new motion jog-pins already ?
the jogwheel ones?
were there some problems with halui.jog-wheel ?
halui.jog-wheel was too jerky
because its non-realtime
jmkasunich: mmm.. marketing. wonderful isn't it?
I think the halui jogwheel stuff should be removed
we might want to put jog pins on halui, but just simple jog buttons
jmk: any idea how to route a jog-wheel count to the different axes ? would need a demux HAL block ?
push button, it jogs, let go, it stops
there are +/- jog buttons already
etla: route the count to all axis
then just enable the one you want
etla: jmk wrote those (the rt ones), so I guess he did test them :D
etla: btw, I uploaded new pdf's today
yeah, I tested them
alex: OK, great.
etla: I noticed you are running 1.4.x (lyx)
jmk: OK, so there is a bit input that enables jog. is jogging allowed only in manual mode like through halui ?
alex: that is the latest fow winXP
etla: I know..
dapper has 1.3.7
which causes some diffs when saving, but nothing serious
what about breezy? I think it has 1.3.4
* jmkasunich checks
1.3 is ok too
only 1.2 can't open 1.4 docs
etla: one more thing I noticed..
ok, no prob then
you used SubSection* in hal_drivers (m5i20 pinout something)
hmmm.. I will get a 3-bit binary value indicating the jog-function, I wonder how I route that to the jog-enable pins
if you use the ones with * at the end, it won't number them
etla: what do you mean 3 bit?
0-7 indicating what axis to jog?
000, 001, 010, 011, 100, 101, 110, 111 ;)
I envision that the jogwheel will have 6 states, X, Y, Z, FO, RPM, off
they will be input as alex suggests above
I'd suggest classicladder
so I guess you need a bit->int converter, then demux
which gets messy in hal ;)
I guess mux4 could do it
for X: ---|\|---|\|---|\|---(Xen)---
for Y: ---|\|---|\|---| |---(Yen)---
for Z: ---|\|---| |---|\|---(Zen)---
jmk: I haven't tried classicladder yet...
maybe I should
etla: might be a good reason to do it;)
it gets easy after you start to
I really do think its the right way to do things like that
* alex_joni loves synergy
like a KVM, but you need multiple monitors....
jmk: another thing I was thinking about, how to use a servo as a spindle ? i.e. normal PID is position-command, position-feedback and speed output
I'd rather have one very good monitor
jmkasunich: I have a very good monitor.. and a laptop :)
but this would be speed-command, speed-feedback, so accel output from the PID ?
etla, more like torque output from the PID
Lerneaen_Hydra: blender 2.4
jmk: ok, but my drives want speed. so I need to put an integrator on the output ?
not sure what you are doing
you have a drive that wants a speed command (and closes the speed loop internally)?
I just want to run a servo at constant speed (iocontrol.0.spindle-speed-out)
the drive does not close the speed loop
you said "jmk: ok, but my drives want speed"
they expect a speed command?
well, the command that is input to the drive translates into the motor voltage, so it would be speed under constant load
so your drive works in voltage mode....
and you want the HAL PID to correct for slowdown under load and other such things
so much drama *sigh*
I expect this will work just fine for XYZ, but the spindle would have to run at constant speed
hey jmkasunich, ltns
I'd call the signal to the drive "motor_voltage" or something like that
Jymmm, I was at a stupid work thing
input to the PID is speed
jmkasunich: Sorry to hear that... couldn't call in sick with milaria huh?
feedback to the PID needs to be speed as well, so you take the encoder position thru a ddt block (to convert to speed)
i look match italia usa
unfortunately speed from a ddt will be very noisy, because of quantization
maybe the spindle-PID could run at a lower update rate than for XYZ, to alleviate quantization
yeah, you could do that, it would probably help a lot
you need to do the math and see how bad it will be
I'll get 8000 counts per rev so should get plenty of counts ...
min and max spindle speed, encoder ppr, etc
thats a lot for a spindle
are you counting in hardware?
how will ddt react when the counter turns over ?
yes, m5i20 card
I mean counter overflow
that might be a problem
Jymm: you made that?
jmkasunich: Yes Sir.
hows it colored? mill, then stain the recessed parts, then sand the flat parts?
etla, one thing I'd like to add to the encoder drivers is a speed output
jmkasunich: Mill, stain, sand, clear coat
that would solve the rollover problem
jmk: sure, would that be as simple as a delta from the last call to encoder-update ?
jepler said it was technical nice, but not somethign he's put in his home. alex_joni said he loves the clock, but hates the wood thing =)
well, it gets interesting
deltas on the floating point position (which is what ddt does) will have problems when position gets large
jymm: it's something that would be sold in a souvenir shop :)
cause a spindle turns the same way all the time, the position gets very very large after a while
Before I design simular items, I wanted to get ppl's feel for it... would you buy it? What you put it in your office? Game room? garage? etc
deltas on counts doesn't have that problem, if you do diff = new - old with each value an s32, it handles rollover just fine
Jymmm, thats a matter of what people like/don't like
counts are probably better then
I was commenting on the quality of the work
am I right in thinking that "emcstrip" is not a part of emc2? I see there's still support for it in scripts/emc.in
Jymm: as I said, sell it to japanese and chinese tourists visiting lapland or similar ;)
jmkasunich: I understand, but if ppl want puppies and kitties instead of deer, fish, and bear....
totally depends on the potential buyer
heh, thats that marketing crap that I was hating all week
jmkasunich Ha, I hate marketing all year long.
I hate it passively most of the time... because I can ignore it.
when its shoved down my throat like last week, I hate it actively
I dont have the luxury of ignoring that.
but seriously, marketing is what you need...
and I seem to suck at marketing too =)
go to wherever you intend to sell these things and see what people are buying
Jymmm: it's good quality
but I wouldn't buy it for me
jepler, to my knowledge, emcstrip is no longer used
probably would buy it as a gift for someone else
that's kinda what I'm trying to do... narrow down what ppl want and/or are willing to buy.
halscope pretty much replaces it
jmkasunich: so a checkin to remove emcstrip from scripts/emc.in would be appropriate?
alex_joni thanks. But that's the feedback I'm getting so far.... "It's nice, but not for me" sorta thing.
Jymmm: I guess we're not your target audience
try some housewives :D
alex_joni: That's what I'm trying to do, FIND the target audience.
you won't find it among people working with CNC's
most can do their own if they ever want to
jepler: seems so to me
alex_joni: Right, but you and jepler both have said "not for me"
Jymmm: I prefer the really old stuff, if it's supposed to be something nice, or very modern
Jymmm: almost all the stuff hanging on the walls of our house was picked by my wife
I gotta find some redneck hunters with a shitload of money =)
find the people who bought the singing fish plaques that were a fad a couple years ago
couldn't walk thru a store without hearing those damn things
jmkasunich yeah! there ya go
I think I'll try the local sporting good store... who knows.
03jepler 07HEAD * 10emc2/scripts/emc.in: remove references to the defunct emcstrip
alex_joni so if I aged the wood... it would appeal to you?
alex_joni that weathered look
jepler: that will definately give me mergin errors now :(
I'm reorganizing the runscript a bit, to start halui sooner
you were right here 5 mins ago when he asked if he should do it?
I didn't see it..
missed it with all the clock talk from Jymmm
jepler: no biggie
alex_joni shush you
so you guys (alex at least) saw my post to emc-devel?
but please ask in #emc-devel the next time.. this is too crowded :D
starting halui earlier solves the original problem (actually, I'd just start it with loadusr from the hal file)
03jepler 07HEAD * 10axis/setup.py: change default nmlfile for emc2: now axis will start for "run-in-place" without a -ini argument
03jepler 07HEAD * 10axis/scripts/axis.py: other changes necessary for iniless startup
giacus: do you know how to do heightmaps of images in blender? I can't seem to find a tutorial for that that's not outdated
03jepler 07HEAD * 10axis/i18n/Makefile: reenable fuzzy messages in msgmerge
Lerneaen_Hydra: nope, sorry, I'm not much familiar with it
reading at blender wiki right now
giacus: I want to be able to make an stl from an image that I use as a height map
blender use python scripts for some advanced feature
have to check what is available
right score until now
who is winning?
K4ts: watching ? :D
annullato per gli usa
giacus: you think there is a script to do just that?
dunno, but scripts are powerful and it could be a way to get it. I've to play with that yet, I'll tell you if found something
end match: 1-1
donovan is a good player
giacus: seems italy is very down :)
if all they can do is a tie with the US
poor suckers :D
now what? do they do overtime or something to break the tie?
* jmkasunich doesn't know much about soccer
just ended at 90°
alex_joni: btw I seen a better US team than years ago
jmkasunich: no, this is still in the early stages
they get 1 point each for a tie
3 points for a win, 0 points for losing a match
and there are 4? teams in one group, and they play combinations
when they finish with those, the first two teams will qualify
then they start a piramid like (final on top)
on each match there needs to be a winner then (overtime, golden goal, then penalties)
* alex_joni goes back to NML
and the ball is rounded :D
anything is possible
03alex_joni 07HEAD * 10emc2/src/emc/task/emcsvr.cc: remove/#if 0 out/ some old forgotten cruft from emc1
03alex_joni 07HEAD * 10emc2/scripts/emc.in: run emcsvr by default, as the first one to run, as it owns/creates all NML buffers/channels
03alex_joni 07HEAD * 10emc2/configs/common/emc.nml: remove cruft, make emcsvr own the buffers
kerry_ is now known as rayh
03alex_joni 07HEAD * 10emc2/configs/common/server.nml: remove cruft, make emcsvr own the buffers
Lots of good food
did you walk the wall yet?
No. Doesn't look like we'll get to this trip/
Saw the forbidden city and a few other landmarks yesterday.
Show is done.
ooh.. nice ;)
off now to visit factories
rayh: gonna visit the factory that makes the HF minimills =)
I am amazed at how many China companies are making cnc.
Yes we will visit Seeg in Shanghi the end of the week.
italia usa 1 -1 sigh
I thought to the rest of the world it was football.
yeah, it is ;)
but it's not safe to use that word when talking to amis
beautiful sunrise here
gotta run to the airport. catch you later.
ok, have a safe trip
03alex_joni 07HEAD * 10emc2/configs/demo_step_cl/demo_step_cl.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/max/max.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/hexapod-sim/minitetra.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/sim/ (6 files): remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/etch-servo/etch.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/stepper/ (sim_inch.ini stepper_inch.ini stepper_mm.ini): remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/nist-lathe/inch.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/stepper-xyza/inch.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/univpwm/univpwm.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/univstep/univstep.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/vti/vti.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/halui_halvcp/halui.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/stg/stg.ini: remove EMCSERVER from all the inis, this is started by default now
03alex_joni 07HEAD * 10emc2/configs/halui_halvcp/.cvsignore: added .cvsignore
03alex_joni 07HEAD * 10emc2/configs/m5i20/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/demo_mazak/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/demo_step_cl/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/univpwm/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/univstep/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/max/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/motenc/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/nist-lathe/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/stepper-xyza/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/sim/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/vti/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/stg/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/stepper/.cvsignore: silence CVS about core_axis.hal
03alex_joni 07HEAD * 10emc2/configs/hexapod-sim/.cvsignore: silence CVS about core_axis.hal
alex_joni is on a streak!
Lerneaen_Hydra: you think?
alex_joni: indeed, nearly even trolling ;)
weather report warned there's a flood coming up..
*ba-dum-kish* - Seriously folks, he'll be here all night
Lerneaen_Hydra: only testing my latest fixes.. they'll follow soon
alex_joni: Another flooding? Think of the children!
03alex_joni 07HEAD * 10emc2/configs/common/emc.nml: I meant tool not to be the master
03alex_joni 07HEAD * 10emc2/configs/halui_halvcp/ (halui.ini halvcp.hal): start the vcp stuff from the hal file
03alex_joni 07HEAD * 10emc2/scripts/emc.in: start halui earlier, this is possible now because emcsvr already runs
03alex_joni 07HEAD * 10emc2/src/emc/task/emcsvr.cc: removed the unused stuff, fix reading the DEBUG from the ini
03alex_joni 07HEAD * 10emc2/src/emc/usr_intf/halui.cc: moved the HAL stuff earlier, removed the peek() because sometimes halui starts earlier than task, and there is no status to be read yet
ah well ..
almost time for bed
so go... it's past midnight for you
02:22 < A-L-P-H-A> so go... it's past midnight for you
03alex_joni 07HEAD * 10emc2/src/emc/usr_intf/halui.cc: call hal_exit() on shutdown
[01:22]<A-L-P-H-A> so go...
one hour exact of difference
38 parallel NE
to be honest, now I can start doind something :(
during th day is an hell
I prefer the night
huh.. monday will be HOT over here:
jepler: still there ?
omg that's HOT.
A-L-P-H-A: weather is crazy lately
I blame the americants...
we had like 12°C max one day last week
their stupid SUVs, and industries.
and weather control devices
that's that Japanese, and North Koreans...
and russians and god knows who else now
The russians can't afford it.
heh, the "industry" isn't the americans anymore, we import everything now
the russians sold it to the japanese yakuza
NK, just threatens the scientists families.
Why are you blaming us? China has most of our industry now
jmkasunich, steel? :P chem plans...
oil guzzling suvs and such
I blame americants.
cause I just like the new made up word, americants.
hard working, selfless americants
damn it... http://americants.com/
ALPHA: You're in Canada, aren't you?
A-L-P-H-A: try http://americunts.com/
Heh, I'm sure every variation on the word is already taken
i'm Canadian with the right to abode in the following countries: Canada, Hong Kong, and United Kingdom.
[23:46:03] <A-L-P-H-A> http://www.cantadian.com/
isn't taken... but it doesn't have the same ring as americants.
some americant registered americant.com
ohoh... expires soon... maybe i'll be able to get it.
I was sayng .. <giacus> to be honest, now I can start doind something :(
might be note true