[01:29:57] <jmkasunich> http://www.linuxcnc.org/compile_farm/festcam.avi
can you guys veiw that?
Mplayer works, Totem doesn't, and I don't know if doze boxes can see it
this is a quick test version, before I encode the whole thing
jmkasunich: I'll try my xp in vmware without a lot of media stuff installed
but I'm betting if it's even as exotic as mpeg4 it won't play...
jepler, "you can't configure linux and emc unless you have a degree in comp sci" ah yes, they should have a pleasant surprise booting the livecd
you need a degree in computer science to burn the CD though, don't you?
(at least, it seems a lot of people can't handle ISOs)
jmkasunich: it doesn't play on my windows xp
it didn't offer to download a codec, but it's possible I've deliberately configured it this way
nope, it's set to automatically download a codec but it didn't even offer to on this file
it is a statement on the quality of a CS degree ;) too bad the paper is so thick otherwise, it may be useful as TP
jmkasunich: I must be an incompetent windows user because even after installing the codecs from divx it still doesn't play
did you right-double-plus-unclick it?
SWPLinux: yes and yes
oh. then I'm no help there
has anyone any experience with storage oscilliscopes/chart recorders/paperless dataloggers
if you mean digital scopes. I haven't used an analog storage scope in 20 years or so
i'm basically looking at setting up a datalogging rack
but i don't know anything about that kind of equipment
so i'm not sure what features I am looking for, or even how they work
how many channels, what speed, what resolution?
i just have data that needs logging
and how do people want to review the logged data?
only a couple channels to the scope, say three or four
usually one, or just two
basically it will be just me, but i need to be able to print it out
which is why i want the paperless datalogger with the hardcopy printer
how do you want to be able to change the number of recorded channels?
it doesn't matter
(ie, load a config, or turn on/off channels ...)
turn on and off channels
this will be set up on a per-task basis
(1 MHz, 1 sample /sec, ...)
speed, very slow
basically, 3000 samples or so per minute tops
but it needs to go way down to like 20-30 cycles per minute
and still show a useful graph of some sort
automatically or under human control?
it's for spindle analysis
so sometimes the motor will be used, sometimes the spindle will just be turned by hand
well, I don't know of any system that can automatically change the sample rate or number of channels
basically i'm just looking for a history
jmkasunich_ is now known as jmkasunich
and also min/max values
what's the budget?
i'm trying to keep it to used equipment
because it's going in my personal toolbox at work, not the company's
like, i can sacrafice features, i would just like to know what i'm looking at
and know what i am sacrificing or buying
my cap is probably around 800 usd, used
well, you should just look for a strip-chart recorder, I think
i also have bizzare data analysis problems i don't think i can solve with a scope
or a strip chart recorder
like, i need to measure change in angular or linear acceleration
you won't get any analysis (min/max), but that'll probably be the least expensive route
well i'm going to be getting both
i'm going to be setting up a data aquisition cabnet, and a data logging cabinet
as long term projects
if you need to do other analysis, you should look into a generic A/D card and some PC capture software
i've looked at PC based stuff
and they don't have the voltage range, and frequently, the voltage requirements
plus, a lot of this involves chaining/looping stuff
and i'm kind of afraid the PC route will put a cap to that ability
I'm talking about A/D capture cards from MEasurement Computing or NI
not PC-based scope cards
analog to digital
you can get them from MCC for $200 and up new, 12-16 bit, 2 channel (or more)
i'm still kind of partial to the real instrument versus a pc solution
but that might be me being naive
it's probably better, but those kinds of instrument are $many thousands
yeah, but i am looking at new/used prices
and it APPEARs (again, based on my inexperienced view) that there are good units at reasonable prices
used it would be $a few thousands ;)
I haven't looked though, so don't taKE MY WORD FOR IT
gah - stupid laptop keyboards
but in terms of actual features
can dso's do polar graphing as well?
or is that sort of like a per-unit thing
you can do X-Y graphs
but not R-theta
even my $18000 scope doesn't do polar plots
see, i have it on authority from my father, who does have several scopes
that there are units that do polar graphing, and that they aren't too expensive
the only ones I've seen are vector spectrum analyzers
and those are now down to the low low price of $9000 each (new, base unit)
maybe my good scope can do it, but I haven't noticed the option anywhere
are you sure he wasn't talking about X-Y plots?
i've seen screenshots of true polar crap
but, nothing firsthand
I've never noticed polar plots on scopes, unless it was a feature in some spectrum analysis package
the plots then look polar, but are actually Smith charts
it is proving difficult to find what will do what i need
I'm not entirely sure why you'd need apolar plot, but then again, I'm tired :)
for spindle analysis, mostly
where on the spindle certain errors are occuring
ah. so theta is the time axis, and the channel(s) are plotted as the radius?
and i'd like to even be able to replace time with a rotary encoder
it's sort of a three channel operation
I'm reasonably sure that there's no scope that will do that out of the box
but I could be wrong
i dunno man, it's kind of frustrating
thank you for clearing stuff upk though
for the speds you want, this would work: http://www.measurementcomputing.com/cbicatalog/cbiproduct_new.asp?dept_id=540&pf_id=1757&mscssid=4V8CTWLSCJWN9J7ALVEKM3EFUE5NCSX6
i will bookmark this
save the data to a CSV file and use a spreadsheet to do the plotting
i will probably fall back on this if i can't find any analog equipment that will do what i need
the 14 or 16-bit A/D units actually have better resolution than most digital scopes (they're almost all 8 bits)
the problem is that i don't need resolution, i just need flexibility
and you can do a lot of analysis later, since you have all the data in the PC already
PC = flexibility
not as flexible as i'd like while trying to do the analysis
a scope is super convinient to use
all the functions are knobs, right there
waaaaay more flexible for analysis
and i can always run another patch cable if i want another device in the loop
capture is easier on a scope - you're right about that
i will never have my cake and eat it too.
nope, not for $800 ;)
the problem I've always had with PC-based "scopes" was that triggering is crap
the paperless dataloggers look amazing though
they MIGHT do it
could be. I'm not too familiar with them
maybe i can use a scope and just loop the pc into the whole deal
there's got to be a way to do this on the cheap
I AM DETERMINED
(to fail miserably)
everyone made it home safe and sound from fest?
I'm in Ohio instead of Vermont
SWPLinux: Ya, but you wander around the planet, so...
well, I'm safe and sound, but not home ...
SWPLinux: Got Cell?
yeah, but it's off for the night
and I'm almost there too :)
SWPLinux: ok, as long not in the back of a cab somewhere
no, I drove myself this time
no van crashes?
though I drove the Jeep this time ;)
good to hear
(no Jeep crashes either yet)
well, I think I will turn in. good night all
see you SWPLinux
I got my ebay drives today, gonna clean em up now =)
Jymmm, what kind of drives did you get?
The_Ball: Parker OEM-750's
Jymmm, looks handy
lerman_ is now known as lerman
The_Ball: I hope so =)
jepler, the pluto documentation says that if anything goes wrong, the pluto i/o's could be in a undefined state, it sounds like a good idea to have the enable pin work as a watchdog with a emc generated pulse train, in case anything stops responding. Does that make sense?
If you have the deaband set too small in a servo driven system, do you run the risk of overheating the servo motors and drives/
Oh, morning all :)
Gonna have to contact PArker tomorrow and see if they have any older revisions of the manual =(
Older revisions of the manual?
manual says 16 dip switchs, this drive has 21 jumpers instead
like Rev A,b,c,etc
I have a problem with my drives tripping out. I need to figure out why, but the motors seem warmer than they should do, given that I have been moving the axes only mms
Hence wondering about the deadband, which I guess in the ini file is set in user units.
Actually deadband isn't set in this example I have been configuring from. It has a DEADBAND variable in the ini file, but this is not then set as pid.0.deadband
Cool! I've got X and Y motion set up! A bit slow, and not too well tuned yet, but progress!
The tripping of the drives seems to have been electrical noise on my board, which is reduced when I close the cabinet.
Can anyone tell me what the logic for the setting of motion.motion-inpos to high is?
hcseb: no, but your system might oscilate..
hcseb: what drives are you using?
hello, i would like to install the apt360 (described in wiki), which also needs a c2f library. I downloaded this library, but i don't know where to copy it, so that ./configure will find it
kwajstabo_: did you install ./configure; make ; make install it?
i ran configure, and it writes something on the screen...so it seem it is installed
read read and read.. "something" might be anything..
you need to do make and make install afterwards too
unless you already did
i know...i read the instructions, but i get an error during ./configure
i get "checking for f2c library... configure: error: could not find f2c libraries" error
i downloaded this library to desktop
where should i put it for configure to find it
You need to configure and install the library.
how can i do that, there is no make install in libf2c
and also no configure
kwajstabo_: Is there an INSTALL or README file to look at in the library source?
anonimasu: These are Baumueller Servo drives. BUS21 drives powering Baumueller DS56M motors
The_Ball: yes, I want to add a watchdog but I am not sure I have the chip resources left over
The_Ball: if I am able to fit it, the watchdog will tristate all outputs
The-Ball: or maybe I should address myself to you. ^^^
closed the loop last night
skunkworks: good morning
I saw something about your PID in my scrollback
jepler: good morning to you
don't spend too much time on tuning without the table attached
I was screwing around with the pluto lathe setup, Because I am lazy
jepler: mainly to learn.
skunkworks: nothing wrong with that
i would still like to install libf2c (for apt360). I.m following the instructions in readme file. It says i should "make check", but i get "/bin/sh: xsum: not found" error. I have xsum ( http://www.pastebin.ca/576197
) i got from netlib which generates xsum.c, but i still don't know what do do with it. Any idea (i copied these to files in libf2c, but still no luck)?
kwajstabo_: I have never tried apt360 or any of these libraries you mentioned, but I'd try things like: "make install", "sudo cp xlib /usr/local/bin" or reading the documentation with netlib
kwajstabo_: are the packages libf2c2 and libf2c2-dev not the right version for you, that you are forced to compile it yourself?
i don't know, the manual for apt360 talks about libf2c...i don't even know there is also libf2c2
is this libf2c2 easier to install?
kwajstabo_: it is an ubuntu package, you can install it with apt-get or synaptic or whatever you prefer
i see....will try this
EMC is getting a lot of folks to try Linux out :)
03jepler 07TRUNK * 10emc2/src/emc/motion/control.c: SF#1734299: it is not a bug to get beyond the soft limit, though it sometimes requires a crafty user
jepler: thank you, this solved the problem with libf2c
skunkworks, that is a MASSIVE capasitator
Thank you :)
[13:30:26] <skunkworks> http://youtube.com/watch?v=l5q1rVKYLTQ\
It is only 1900uf at 350v iirc
* archivist_win gets The url contained a malformed video id.
archivist_win: remove "\"
[13:35:03] <skunkworks> http://youtube.com/watch?v=l5q1rVKYLTQ
The time stamping really is great
too much of me in that video
Next year - you will have to take jmk's seat ;)
missed boris the shop cat on that, I thought I saw it (gostly image) one day
there are at least 2 frames with a cat, but it wasn't boris
was the machining done _behind_ the camera?
The-Ball: the lathe is at the end of the table on the left and the milling was done on the big blue mazak in the background
behind the camera is the wall
hehe, i was trying to be funny, lot's of typing little hanging over the mill ;)
the mazak may not have moved until thursday or so -- past the end of the video
wait.. It thoought that was the whole video..
Huh - I guess it isn't.. I though I remember seeing a post by jmk that the whole video was up.
skunkworks: I think he was trying to find a codec that would work on most windows machines...
ah - he did re-upload a video though as the one on there now is only 38mb vs the 78mb that is listed
that must have been the test one he was wanting people to try\
oh well - now that i have a few options - re-sending it to youtube shouldn't be a problem
can you take down the one that was too short?
yes - I can delete it
Is there any limit to the memory that the kernel can recognise on the Ubuntu livecd?
I have 1024 MB of RAM, but am seeing only 900 MB. I don't believe any memory is used for a graphics device on this machine.
err maybe it's 900M
Ok, I figured there might be a limit. Is that related to the RTAI extensions to the kernel?
rtai breaks around 1G of memory
they will probably fix it "someday"
No big deal. Just so I know ther eis nothing wrong with the hardware.
900MB is enough RAM.
you could try a regular kernel to make sure
yeah, it definitely is
Guest974 is now known as skunkworks_
skunkworks_ is now known as skunkworks
JymmmmEMC: get any drives yet?
ooh - nice
skunkworks: Well....... LOL..... Here's the thing.... I bought 7 of them + 5 motors with US dogotal encoders (used of course) for $500. I received the order yesterday. But... only 3 of the 7 are OEM750's, the other f4 are OEM650's. So, I'm trying to work out what to do with the seller. I think it was an honest mistake on his part, they do all look the same. externally.
Well - atleast you got 3 :)
I honestly think you will not be able to tell the difference but that is just my opinion :)
skunkworks: Yeah. but I could have gotten geckos for the $500 too brand new, and I really can't use the motors of the encoders.
skunkworks: 750's have min-band compensation, 650's don't.
already confirmed that with PArker.
right. but like I said - we never had problems with the 650's at 1000steps per rev.
I was planning on keeping one spare, then sell the other 3 750's + spare PS + spare caps I have as a "complete system"
skunkworks: stalling you mean?
Yes - stuff selling in '3' really go on ebay..
skunkworks: Yeah, or keep them and make another system
He has one other 750+ motor (torque too small for my use) that he's willing to send to me. Just not sure if that'll be enough. JUST got his email 20 seconds after you asked me if I got drives.
at the cnc fest we picked up a bunch of encoders that where 2540 line with index. A good deal. 1/4 thru shaft with spring mount.
These encoders I got are only 500
na. work great for gecko servo drives... ;)
the encoders I have you mean, or yours?
actually that would be 2000 tranistions per rev.
that is quite a bit. and low enough that you could do printer port counting
I lost ya there , still asleep too, and on 1st cup
you could spin them at 600rpm with a base period of 50us. or 1500rpm with a base period of 20us. reading the encoders directly into the printer port
if I did the math right.. I have been having problems with that lately
iirc my BASE_PERIOD is like 18000
so 1666 rpm
but while having encoders on steppers are nice, as jmkasunich says, "and what do you expect EMC to do if it catches a stall?"
Though, I think if they acted like or do what servos do would be nice, but nfc
right - I was saying for other projects..
I'm just "bummed" in a way, becasue at 7 OEM750's that would have been awesome and I could have made back what I've spent rebuilding the controller with all the spare parts I've bought. But at $500, I could have bought Geckos for the same price.
although it would be a nice test bed to see if it was posible.. doing some sort of adaptive feed.. If the commanded possition started to lag encoder position - slow feedrate down...
skunkworks: I'm not a coder, so not much I could do in respect to EMC
Just curious, what were the encoders your bought going for?
I would try to do it without changing the code. strictly in hal.. I can see it in my mind anyways. adaptive feed is already there.
let me look (i can give you comparables)
Remind me of the command line version of emc that you can use to just run some Gcode script
We got them for $30. You can get absolute cheap ones for $60 but I am not sure if it has index pulse. So for sure $133
hcseb: maybe you're thinking of this? http://axis.unpy.net/01167419757
if they are toally enclosed encoders with shafts - you can get more.
like an explosion proff styel
skunkworks: I'm not in any hurry to play with encoders for now.
Nice that they were there on the auction, but no cables either.
panasonic 2bit gray encoders in audio pot packages #1
(but thats prob not what you meant..)
a-l-p-h-a2 is now known as a-l-p-h-a
There's no such thing as a bi-directly opto-isolator, is there?
no, I don't think so
What about the ability to use some kind of jumper configuration to make it input or output (not necessarily both at that same time)
"user choice" so to speak
I think that if you tried to hook up a standard opto in both directions, you'd end up "sticking" in the conducting state
yeah you could probably do that
you'd need 2 optos, 2 2x2 headers, and 2 jumpers
I was afraid you were gonna say that =)
er, I guess 2 1x3 headers
on the transmitting side you'd connect 1-2 and on the receiving side you'd connect 2-3
JymmmmEMC: are the motors you got bigger than what you have?
on the board layout you'd connect the pin to 2, the opto LED to 1 and the opto OUT to 3
skunkworks: Actually, they're have the torque of what I have now unfortunantly
they look nice anyways :)
jepler: So, I could do it with one opto, but more jumpers?
there may be some lesson to be learned in bidirectional level-shifters that could be applied to bidirectional optoisolation
JymmmmEMC: I think that with good opto design you treat the area under the opto as a "no go" zone for traces, so having jumpers that can route the isolated signals to pins on either side violates that
Ok, I can understand that.
oh - oops. I think I have traces running under the isolator.. Have to look
JymmmmEMC: hm, take a look at this: http://crio.mib.infn.it/wig/electronics/Componenti/HP/Accoppiatori%20Ottici/hcpl0560.pdf
bah - not worried. http://www.electronicsam.com/images/KandT/servostart/top.JPG
8 dip package on the left
hm or is that just one channel per direction?
jepler: Looks liek two-in-one at first glance
jepler: Crazy thought here.... What about "somehow" just "flipping" the opto around to choose if it's input or output?
some sorta socketed configuration.
there is usually support circuitry that you need to worry about - resisters in series - pull ups - plus most are not symetrical.
Well, take an 8pin DIP opto, in a 16pin socket as example. Where you plug in the opto determines if it's input or output.
kind like using a DPDT switch ti reverse polarity
seems like you could make it work
It seems "reasonable" at least.
not sure if it would voilate the "no go" zone you spoke of though.
jepler: Actually, if you look at that datasheet, if you flipped it around, the singals still match up.
[16:28:44] <jepler> http://www.picbasic.co.uk/forum/attachment.php?s=98865b9d1556d757b9fe3d0adb9d40a8&attachmentid=253&d=1112646124
I never looked before... On some sockets, can you take two 8pin stacked edge-to-edge and make a 16pin socket that would align up?
err, one on top of the other I mean.
on a board I recently built I used 2 14-pin sockets and a single 28-pin chip and the sockets fit fine
jepler: ok, cool. I thought so, but it's been a while =)
(a narrow 28-pin chip that is)
if an 8-pin dip is under .400 inches long then you can stack them end to end (since the distance from the first pin of socket A has to be .400 from the first pin of socket B)
Yeah, it just might work for this application if you only had to "shift" the IC other 2 pins (as example)
this is interesting, I'll have to look into this some more. maybe I'll grab eagle when I get more ram for my computer.
are most opto's two per pkg?
heh, I just set msg_level=RTAPI_MSG_NONE and now xemc is more responsive to input
i'm not sure if anyone is here
but i would like to discuss motors
seeing as how i know nothing about them
toastyde1th: any motor from Radio Shack will do what you need. No need to talk about it
* archivist used a car starter motor for a drill once
0 to 5 secs for a burnt out drill bit
toastyde1th: No, no, I don't care what you think you know, it WILL work. If it doens't, that means you're doing something wrong.
well, i'm more concerned about smoothness and accuracy of positioning
toastyde1th: (how's that for a fsck off and leave us alone =)
i'm looking at servos, and i've heard of stepless servos
except this application will be using external scales
so i'm not convinced I need to use a servo?
i just care about having smooth operation, accuracy in angular positioning, and holding force
toastyde1th: Ok, so you're between stepper and servos, right?
see, i don't know
i don't know about motors and motion control
i'm not sure what options exist
toastyde1th: stepper system = $, servo system = $$$
it would be very easy to pick an option, because i know what my requirements are
well, considering i don't need error correction
in the motor itself
will a stepper work?
worm drive from a stepper
it's going to be belts and pulleys
from a stepper
toastyde1th: ***ALL*** stepper motors can stall due to mid-band. servos don't.
what angular accuracy are you after
that's another problem right there, we haven't picked a leadscrew
the generic answer is "the more accurate, the better"
toastyde1th: The BETTER question is, how much are you willing to spand
toastyde1th: it's real simple... hundreds of dollars, or thousands of dollars
oh, a thousand or two
but i'm not really looking for specific motor reccomendations
just a discussion of the different aspects of servos and steppers
so that when/if the time comes to do our refit, we can pick a motor intelligently at that time
when we have specific numbers, like required angular accuracy and torque, in hand
if you are willing to spend that kind of money, then go servos.
servos won't help us
sort of machine?
diamond turning lathe
toastyde1th: actually call up a sales engineer for a company
a motor company?
that's probably a good diea
toastyde1th: If you want accuracy AND no headaches, go servo. And what I mean by headaches is the constant WHAT THE FUCK, it's doing this or that and we have NFC why and has ruined the workpiece and have tod o it all over again.
i am concerned the servo will try and do some sort of error correction.
when i really, really do not want the motor to do anything except what it is told
in a generic sense
the glass scales (or interferometer if we go that route) will be providing the positional feedback
and you want that to tell the servos where to go?
toastyde1th: ok, and what do you want the steppers todo if they stall?
toastyde1th: a SERVO always does error correction to be where you want it to be.
stall, as in just stop?
toastyde1th: Yes, at least momentarily
i don't want the motor to know where it is
toastyde1th: do you understand what that means?
if i tell it to rotate x number of steps, and it misses one
i do not want it making up that step internally
there are no steps with a servo.
and there are no steps to lose.
it will always be within the 0.00001 window where you want it.
i am very glad you guys are here.
toastyde1th is now known as toastydeath
no steps is very, very good
i'd prefer smooth motion
a servo bounces between encoder counts to keep position..
does that induce vibration?
that sounds like a great option then
actually it does, but that's very very very very small..
toastydeath: Now, just STFU and get your ass to Radio Shack like I originally said damnit!!!!
divide 360 with 2000 or something..
well, uh, three millionths of an inch
will ruin the part
so it's crititcal the motor just stays put
and three millions vibe is "barely acceptable"
get better encoders for your servo..
however, if the motor vibrates that much but the machine soaks it up before it gets to the tool
toastydeath: i think you REALLY need to look at a "motion control" catalog and look at things like dampeners etc
there are 10 000cpr encoders..
are the motor and the encoder seperate parts?
no.. usually the encoder is monuted to the motor..
but there are all kinds of fancy options..
fancy options are good
that gives something like 360 / 4 x 10 000 if I remember it right positions for the servo to be at..
toastydeath: actually you should be looking at linear motors..
linear motors will not work
they're not accurate enough and apparently have a ton of vibration
they use them for edm machines..
edm is nowehre near as sensitive as a dtl, though
i wish this was easier
most of the lathes i've seen/read up on
well, it's not really a problem.. call up your servo supplier..
toastydeath: is this machien of yours mounted to the floor usage dampeners?
use hydrostatic leadscrews
yes it is
or really really finely lapped ballscrews
toastydeath: you can forget about steppers.. I guess :)
yeah, i gather that
but i am glad i have at least eliminated some options
that brings me closer to what i need.
toastydeath: one example http://www.motioncontrol.com/
toastydeath: though your app will require really big gearing..
or really expensive encoder..
gears are unacceptable
it's gotta use belts
but yeah, i'm also figuring on that
"GEARING" is the term for that.
very slow rapids, too.
that's all okay, as long as it's smooth
[17:38:13] <anonimasu> http://www.baumerelectric.com/en/shop/catalog/Winkel_und_Positionsmesssysteme/page1.html?filter=all
well steppers are not smooth.
but you've sufficiently convinced me to use servos
toastydeath: Realistically, what you are looking for is the same kind of motion system that is used for wafer fabrication.
are these motors?
or just encoders
thoose are just encoders..
look at the highres ones
[17:39:53] <anonimasu> http://www.baumerelectric.com/en/shop/catalog/Winkel_und_Positionsmesssysteme/Drehgeber_inkremental_BHF_HighRes/
wafer fab deals with microns, so there can't be any vibrations that effect their usage.
320 000 pulses per rev
is that the 16 bit multiturn?
thoose are damn pricey.
where are the prices
I dont know..
but they are probaly a few thousand dollars..
we are probably not going to try and resolve microinches
accuracy is never cheap
what kind of screw do you have?
but we need vibration in the sub-microinch range
we haven't picked that
give me a number.
right now it's a lapped leadscrew at 10 tpi
and I'll give you what kind of encoder you need..
it APPEARS we'd like a minimum linear resolution of .00002
I hope that's inches
teastydeath: is taht in a temp controlled room?
we are discussing adding an oil shower as well
but you say you have an external scale?
well, let's see if this is right..
the scale will be internal, but yes
what kind of scale? can you use it as your feedback?
glass scales are what they're looking at
and yes, we're going to use it as the axis reference
this is cnc right?
each degree will be 10/360
you can't use both feedback - encoder + scales.
you need to pick one or the other.
we don't care about the motor
0.02" per degree
you do though
like i said earlier, we're going to use the scales
as our feedback for everything that we can
but the scales need to control the motor
this was my concern earlier
err that's wrong..
if you have no backslash you can use linear encoders.. I guess
which I hope they have
it will be zero backlash
anyone want to see a cool part..
right now we're turning in one direction
so it doesn't matter for us
aren't scales 'slow'?
slowness doesn't matter
the machine turns at 340 rpm @ .0001 or .0002" per rev
So you need scales that ouput something that can be read fast enough for feedback for the speed you need
ds2: not extremely
ds2: even my shitty china scales are pretty fast
heidenhain scales are pretty cheap on ebay
how does EMC handle that, by the way
our current control is shitty and i proposed looking into using emc for the machine
[17:48:11] <anonimasu> http://pastebin.com/932039
anonimasu: do you have a screenshot of the preview? I don't have emc here.
depends on the interface. Quadture - or maybe a absolute number.
It does not do serial or usb feedback... yet
well what i guess i am asking
even though this is a retarded question
does emc take the input from the scales
[17:50:34] <anonimasu> http://imagebin.org/9040
anonimasu, !! wow!
I thought it was pretty cool :D
whats that other girl holding, some sort of vase?
it's a ramp in all directions :D
oh, wait, I mis-typed the link
right, thats better
toastydeath: I would look at the scale you want to use and then ask 'can emc interface with it'
I want to see the picture robin looked at..
I wonder how long that part will take to cut
anonimasu: what is it for?
hey guys, i really appriciate the help
i will now go do more research!
you should call up a company specializing in motion control and stuff and tell them your requirements and ask them for a motor setup
for a "quote"
if work doesn't do this refit
i'm going to scale back the precision and build one myself
it'll be expensive(just let me tell you beforehand)
precision = $
that's the universal rule of all cnc:ing
hahah, not from what "the experts" are saying
i have a lot of research to do still, though
i got to talk with a guy who was one of the pioneers in diamond turning
his opinion is that precision is expensive because people don't do the proper research before trying
that's not really true
and wind up spending tons of money correct inherent errors in design
err, I mistyped it as http://imagebin.org/9041
ye gods that image is persistent
toastydeath: precision is expensive because precision requires parts made to even higher quality
stragnely enough, the webmaster who ran the original goatse.cx used to frequent this channel
right, but only certain parts need to be bought
at that precision
IIRC he was the sysadmin at Sherline for a while
most high precision gauges can be made by the shop that desires the machine, if they've got the people who know how it works
the motors and scales are going to be the expensive part, and the majority of the material cost of the machine
are you building one from scratch?
robin_sz: err, surely you jest
the existing machine was, yes
parts from other machines, and rebuilt
well, I guess you will see about it when you get quotes :)
the only expensive thing is the linear accuracy
the servo systems I asked about were 1600eur per axis..
surface finish is relatively easy and inexpensive
that's not so bad, dude
compared to a brand new machine
or a custom built one by someone else
that's just motors..
expensive is relative ;)
haha, yes it is
lerneaen_hydra, surely I jest about what?
the goatse thing
I forget where he was sysadmin
im fairly sure it was sherline
this is a classic example of TMI
or rather WTMI
perhaps relativistic tmi
such a vast quantity of too much information, it can only be measured relatively against other frames of information, without an absolute reference
more like chronical tmi
skunkworks: I think you answered this for me once, but what are the two pads on the right center for? (http://www.electronicsam.com/images/KandT/servostart/top.JPG)
what's the PCB for?
I have a problem with apt360... i installed it, and when i test a simple code, everything i get is this ( http://www.pastebin.ca/576930
) and an empty .tap files apt360 generates. Why idea what could be wrong?
[18:27:40] <maddash> http://www.sfgate.com/cgi-bin/article.cgi?file=/chronicle/archive/2007/06/18/BUGGGQGBSS1.DTL
there is precision and then there is repeatability
do you think a 1/4" al plate would work as a heat sink for ~300W worth of drives? If so, you think I could cut it down using a table saw with carbided time blade going REAL slow?
how big of a plate?
I need to make one cut ~ 12" long
tablesaw is too fast
P4 Xeons were putting out nearly 60-70W each
you need to gear it wayyy down
even for carbide
12 long and how high/wide?
ds2: one cut is all
JymmmmEMC: what is the target size of the heatsink
roughly 12" x 10"
passive 60W range heatsinks on the P4 Xeons were around 4"x4"x1" with lots of fins so...
yeah, no fins here...
if you don't have a band saw, I recommend having some Wheaties and using a hacksaw
fan yes, no fins though.
seems iffy, 120in^2 of flat surface compared to 80in^2 w/fins
it'll be safer
the Xeons used case fans... 12KRPM if I recall
cradek: gave up wheaties for lent =)
the machine shop here cut some plate aluminum with a circle saw and lots of wd 40
maddash, interesting, but not new
maddash, pulsed lasers have been around for a while. many ways of creasting short pulses, from basic q-switching to pulse compresion techniques, if they can get enough power out of it to do useful work, then it becomes interesting
* archivist used to cut aly at work 3000 rpm carbide blade over arm saw , must wax the blade though
what is q switching?
and clamp the aly
basically, you put a blocking element inside the cavity
archiist: what size blade?
its RF driven, and can go from blocking, to clear, in $noTime
was cutting truck tipper sections
robin_sz: so it is basically a serious detunning of the cavity?
energy builds inthe cavity
you let it out with a bang :)
skunkworks, the guy in the next factory to me cuts aly plate
big table saw, 500mm + blade
iirc it was a 3 horse motor on the blade
oh, baby one :)
I think this one is 30hp or so
200mm thick slabs
and the bastards would not supply ear defenders (circa 1970)
for 1/4" Al, why not just cold chisel it out?
[18:40:28] <jepler> http://www.nvnews.net/vbulletin/showthread.php?t=87541#post1216363
(short story: maybe around driver 9755, nvidia fixed their driver to work [better] with realtime)
this guy actually stocks and cuts 600mm aluminium round bar
a 2ft round is still considered a bar? =)
yes when its starts out 20ft long
i really need to get a tour of those extrusion machines used to make that
i think they use continuous casting to produce that size
oh so it isn't an extruded alloy like 6061?
you can't cast 6061?
it seems 6061 is considered a wrought (sp?)alloy which is why I assume they are extruded
they need to produce it in the first place
and it can be wrought from a 2ft size on a sizing line
hmm .. but on my x86_64 I'm using an older version than 9755 and it seems OK. I wonder if the bad characteristics of wbinvd are "not so bad" on x86-64 for some reason
er, sizing rolls
but then that's like rolling chunks the size of a truck!
* robin_sz thought they only cast/extruded alloys ending in 0
how do they make it in the first place?
03jepler 07TRUNK * 10emc2/src/hal/drivers/pluto_servo_firmware/register-layout.txt: describe the register layout of pluto-servo somewhere besides the source code
I thought the last digit was a post.casting heat treatment code? no?
i'm far from an expert on the subject, but it would seem to me they'd need to make some sort of ingot or continous casting to alloy it
no, not that i'm aware of
the T is the heat treatment
ah .. 6061 T3 ...
* robin_sz slaps head on table
forgive me .. that was a silly mistake
* robin_sz slaps head on wall a few times
no dude i do dumb stuff all the time
if you are near Wenatchee, Alcoa has free tours
no apology necessary
how could I be so crazy
* robin_sz cuts off left hand in repentance
so that would make you less sinister? ;)
i believe the first two digits are the major alloy type
and the second two digitis are the specific compsition
I know 1000 series is pure aluminum
followed by heat treatment if the alloy can be treated
yeah, no alloying
no silicon or nickel
or anything else
20xx has some copper, IIRC
for some reason best left unexplained, I spent half today reading about Silane
is Silane flamable like methane?
oh is that the crap that's 6 times heavier than air
its the silica equivalent of methane
interesting, so you wind up with sand all over?
Just for palying around, how could I hook up an encoder to EMC via the paraport?
or you wind up giving everyone silicosis?
you need to heat it to 420 degrees to dissociate the H from the Si
JymmmmEMC, i dont think you can
hmmmm thought it would do SiH4 + O2 -> SiO2 + H2O (unbalanced)
robin_sz: not even via hal scope?
nah, you nitorgen flush the chamber and then hard vacuum it before introducing the gas
can a person use HAL to do error correction and control active way calipers
sort of preprocessing
"active way calipers?"
"active way calipers"?
they move the machine based on some reference standard
they're a secondary axis on top of another axis
that cannot be controlled by the machine
"some refernece standard" .. you mean like a rennishaw encoder onthe machine?
like an optical flat with a capacetence sensor
to measure error induced in the X axis of a lathe produced by movement on the Z
you can only really have 1 axis reference
well what it does
is tell the calipers to move
you could build a mapping ...
and correct the error to the main axis
without the machine ever seeing that
skunkworks: but how to you connect it to the paraport, directly???
however you need some sort of signal processing to do it
JymmmmEMC, you dont
hal seems to be able to do that kind of thing
EMC would never even see it
hal would get the value from the sensor, and tell the calipers to move automatically
JymmmmEMC, afaik, the pport is not fast enough to read a motor encoder
JymmmmEMC, you might be able to do a handwheel with it though
JymmmmEMC: depends on what kind of encoder - if it has ttl out - you can hook it directly into 2 input pins.
robin_sz: it's only 500 cpr
JymmmmEMC, on a motor or moved by hand?
just wanted to test to see if these encoders work
with 500cpr and 20us period - he should be able to aproch 1500rpm.
re: earlier w/ 10 tpi leadscrew
you can do a handwheel, AFAIK, look for references to handwheel .. I think JMK had it working on the bench with a handwheel
i believe it's actually .00027" per deg
also bbl, in case you try responding
JymmmmEMC: do you have an 2 channel oscilloscope?
Dallur: Are you there?
Dallur: I have some feedback for you
Ill try to be around for a while
toastydeath: then you multiply that by the encoder count x 4 in quadrature (I think) then you have the resolution of your system
err calculate how many degrees one increment are
skunkworks: I have a ZERO channel scope =)
robin_sz: it's the 6061 - XX(heat treatment)
anonimasu: Hello, how is things?
but im managing
I try my best
Right now Im trying to learn how to debug the THC config for emc2
anonimasu: It wou;dn't be so tough if you just listened when we told you to stay out of the signles bars where the chicks have bigger arms than most men!
Dallur: still not around?
wth is an "indexer"?
on the oem drives?
no, a seperate box
it's for doing positioning
like on a rotary table..
It is probably a programmable controller - similar to the X in the oem650x
for making a positioning program - as anonimasu said ;)
I saw one at a surplus store here... it have thumbweheel switches on it
nfc what it is/does
then I don't know..
well, why the hell not damnit! ;)
his mind reading module failed
I made a thumbwheel switch once
it was a small wooden wheel, about 300mm in diameter
with thumbs nailed on to it in about 20 or so places, arranged so as you rotated it, they pressed on a pushbutton
getting the thumbs was hard, nailing them on was even harder
should have used U shaped nails I think
JymmmmEMC: it indexes stuff..
JymmmmEMC: the switches might set the degrees
[20:27:10] <anonimasu> http://www.cnczone.com/gallery/data/500/P9230080a.JPG
how long time do you think that took to cut?
nice surface finish :D
looks nice but I decline to guess
H.D. worship is tiresome
I posted a comment asking
lol, nevertheless it's a nice piece of work
they sure have great marketing
yes it is nicely done
cradek: it's as bad as the ones that worship god/satan/whatever :D
flying spaghetti monster
hey, don't insult the FSM
[20:31:13] <lerneaen_hydra> http://upload.wikimedia.org/wikipedia/commons/archive/6/6e/20050831232441!Touched_by_His_Noodly_Appendage.jpg
[20:31:42] <anonimasu> http://upload.wikimedia.org/wikipedia/commons/archive/6/6e/20050831232441!Touched_by_His_Noodly_Appendage.jpg
err wrong one
[20:31:50] <anonimasu> http://www.cnczone.com/gallery/data/505/P1020289c.jpg
another nice thing
nice resolution :)
didnt manage square edges though
archivist: nice comment :D
hah. "one transistor receiver" -- I see a whole DIP IC package on there http://www.makezine.com/blog/archive/2007/06/how_to_make_a_one_transis.html
one DIP IC receiver ;)
plus 10 in the ic and 2 diodes in the ic
* archivist sneaks a peak at the national data sheet for the lm386
jepler: But.... only ONE transistor =)
LM386 IC Audio Amplifier
god - I played the the lm386 so much when I was a kid.. It had enough output to run a speaker.
JymmmmEMC, I cheated and looked at the lm386 data sheet and countrd the internal transistors
someone needs thumbwheel switches?