giacus is now known as giacus_afk
skunkworks: interesting chip. I'd never heard of an R/D converter before.
jepler: It is interesting - maybe we can use it for our accupins
they are only 21 dollars from digikey per axis. then we would have accurite scales on the big machine.
"TunaPie is a directory browser for Internet radio and TV streams" -- freshmeat.net project description
I can see why they gave it that name
[chicken little] The download is failing! The download is failing!
I've sent a message into the bdi4emc list...the 4.40 iso is no longer present on either of it's listed mirrors
[snaps fingers in frustration] Nuts.
asdfqwega: if you were going to upgrade anyway, that makes it a great time to try emc2 on ubuntu.
Oh, right now either I apg-get the upgrades, or when the BDI fails to install properly, I use my Debian Sarge installation DVD
I second that - emc2+ubuntu for the win
I wonder if there's a problem with the bdi 4.40 release, or just a problem with the mirrors.
I'm in the process of wrecking a BDI install right now...just as a point of curiosity, I've been trying to get accellerated openGL working
Aw heck...it looks like I'm going to be having guests...better shutdown the computer(s)
asdfqwega: see you later
Guests - Lightning and Thunder :(
asdfqwega: good luck on your upgrade when you get back to it
* asdfqwega pulls the plu-#@#%!...NO CARRIER
anyone have any good ideas for a board that can control three steppers, and one PWM servo?
magician look at http://pico-systems.com/PPMC.html
if you're buying
oh crapsky he left
oi listen to this meshugganah
I thought this channel was in english
nope mostly italian and techno-babble
jepler: seen what cradek said in emc-devel ;)
I'll try not to interrupt any cvs activity anymore :D
but.. if the connection fails, it'll be the same thing..
yes, I think I fixed it last night
you mean that file? or the problem with interrupting cvs commits?
the problem with interrupting is probably an cvs bug and I'd rather not have to dig into it - let's see if we have an ongoing problem.
maybe I should make a ssh keypair with passwd
you can set a passphrase on your existing key if you want
that way I can still stop it, before entering the passwd
or, you could think before you type
usually it works without thinking
IF I would think before I wouldn't have to stop it :D
-p Requests changing the passphrase of a private key file instead of
creating a new private key. The program will prompt for the file
containing the private key, for the old passphrase, and twice for
the new passphrase.
* alex_joni goes back to carrying boxes around
have fun, I'm going to go drink coffee
sounds just as bad to me :D
you'll appreciate coffee when you get to be my age :-P
heck, just by not using "-m" you get a chance to see what files you're committing, and think it over.
yeah, I do that when in doubt.
jepler: I usually use cvs diff
before I commit, to check it's all I mean..
but sometimes (usually late at night) I forget :(
cradek: I really doubt I'll ever get to appreciate coffee
* alex_joni goes back to the damn boxes :D
ok - so maybe I am starting to like eagle ;)
i'm learning eagle too
Slowly getting used to it
btw, could someone give me a hint on how one could use an opamp for an and-gate
first, sell it on ebay, then buy an and-gate with the proceeds
or, find someone with an and gate, and say "hey, wanna trade for this op amp?"
it's not an economical issue
it's about lack of space on the board and since i'm gonna have to use one opamp for a buffer i though maybe i could use the other one for the and-gate. why add a quad DIL14 and-gate too if it can be done with the other opamp in the DIL8 package. DIL8 packages contains two opamps don't they
this all assumes an opamp can be used as an and-gate!
Let me preface this by saying I know jack about analog.
but what about this avenue: Create something like a combination of a comparator and a summing aimplifier
so the output measures whether V1*k + V2*k > Vthresh
cradek: what do you use for a cutter to mill circuit boards
how does a "summing amplifier" work?
cradek: btw, how strong are your stepper motors?
Bo^Dick: 1xx oz-inch, I don't remember exactly
Huh. Do "I know jack" and "I don't know jack" both mean "I don't know anything"?
what a stupid language.
yes I think they mean the same thing
we found a resolver to encoder chip that is still in production.
jack == 0 so "know jack" and "dont know jack" are equivalent
Jack shit has the remarkable property that its absence and presence are identical.
giacus_a1k is now known as giacus
giacus is now known as giacus_afk
emc stepper pulsing seems to make clitches when hard disk operates.. is it always like that?
that means something is misconfigured, or your hardware is incompatible with realtime
are you using Ubuntu or BDI?
hard dsik dma seems to be disabled. could this be the reason?
i tried to enalbe dma with hdparm but it didn't enable it
you may have to ask on the bdi4emc help mailing list for specific things to try, we aren't experienced with BDI here
let me find you the link
[15:31:34] <cradek> http://lists.ourproject.org/cgi-bin/mailman/listinfo/bdi4emc-help
but bdi just only linux distro, i think they don't code that realtime stuff
no, the realtime configuration is part of bdi.
it has rtai kernel
i'll try ubuntu. it might be hardware "feature" if it behaves the same way
the ubuntu install gives you the RTAI tests which can give more information - those are left off of BDI, probably to save space
btw, is there a handy way to monitor follow error while operating emc?
when running EMC2 in ubuntu, which dependancies are there (I'm trying to remove all uneeded programs).
I don't know, but if the dependencies listed in the .deb are right then apt-get will prevent you from removing packages necessary for emc to run.
When I wanted to get a "lean system" I did the Ubuntu server installation, then followed instructions from the Ubuntu wiki to add the minimum packages needed for X -- in particular, this didn't install gnome.
approx: You can use halmeter or halscope for that
jepler: Ok, right now I'm just going to remove the unnecessery stuff via the synaptic package manager (openoffice, gimp, games, gstreamer and so on)
Lerneaen_Hydra: I followed basically the instructions under the heading "Install a Lightweight System" here: https://wiki.ubuntu.com/Installation/LowMemorySystems
jepler: I'm not really low on ram or CPU (384mb ram and 800mhz) but I might as well remove stuff that I will never use.
jepler: And I'll also leave the lib* section untouched.
approx: For emc2 stepper-based systems, "following errors" aren't that useful. If you've specified the proper (5%?) "headroom" for STEPGEN_MAXxxx, then following errors generally represent bugs.
approx: Only in servo systems with feedback are following errors "useful" (they indicate that you've specified too high an acceleration or velocity, or that your PID is poorly tuned)
ok, i'm trying emc2 real soon
i'm not happy with emc1's stepping pulse strain "smoothness". hopefully emc2 is better on that
it sounds like emc1 pulsing has some kind of ringing when running constant speed
what step rates are you trying to attain?
I have 1333 steps/mm . ringing occurs when running over 10mm/s
i have tried P=100..3000. it's always same
emc2 doesn't have an imaginary pid loop that you have to "tune" like that
If your version of emc1 has it, you should try "quickstep". It's a better step generator for emc1 (in fact, it can do a few things that emc2 can't). Unfortunately, I don't know which versions of emc1 have "quickstep".
google didn't find much on quickstep
I don't think there is any documentation particular to quickstep
it's very new
so, should it work if I just replace freqmod with quickstep?
any new parameters needed?
hold on a second, I'll find the mailing list message the quickstep author sent
I really had issues with emc1 and steppers. emc2 after a few bug fixes really runs nice. Smooth
[15:59:21] <jepler> http://lists.ourproject.org/pipermail/bdi4emc-help/2006-February/000119.html
despite paul saying he put this in an update on his website, my updated bdi4 machine (installed as bdi 4.29 or so) doesn't seem to have quickstep available
if you can't figure out how to get quickstep, you might post to the bdi4emc mailing list asking about it.
they haven't put much documentation their wiki yet
but i'll try emc2 first. bdi still has that hard dsik interference problem and all :|
jepler that logic analyzer is ~$50 to $75 with components on a soldermask/silkscreen board *tempt tempt*
er, the components arent soldered or programmed btw
fenn: I still don't need one
* fenn sulks
jepler: I'm not sure but I think there may be a typo in the english translation of axis, in message 163
SkunkWorks: my steppers also work a lot better in emc2 than emc1: I don't know if I ever had it configured right, it was complicated.
Lerneaen_Hydra: I haven't looked at it in detail yet
I am new to this IRC stuff. But I'm looking for help configuring EMC2 the puppy-linux distrubution.
wholepair: ask away
I put a thread here: http://cooltool.he.fdread.org/cncforum/viewtopic.php?t=24
wholepair: hi, I think you emailed me
yes - cradek - you made the cool nixie tube clocks!
so this looks like mostly just hooking up hal pins
btw, has anyone tested emc2 package on kubuntu?
wholepair: hope you aren't expecting uch mor than 60ipm
OK - I have a step and direction based 3 axis mill - and I think I need to make a modified emc.ini file and change another *.hal file for accomadating my pinouts.
Im fine with 60 inches per minute...
approx: yes, it works fine, the only problem is EMC shows up in a funny place on the menus
Im on a XP computer now(at work) the machine and the computer I use for control is at home.
Can I edit thie files here that I need and then put them on a disk to bring home?
wholepair: I think that should work, but I've never used floppies with puppy.
I've got a thumb drive to - and I could email them to myself or put them on my ftp to pickup from home.
any ideas whether kubuntu or ubuntu is "lighter"?
neither is light...
approx: if you want "light" then install ubuntu using the instructions here: https://wiki.ubuntu.com/Installation/LowMemorySystems
but you get a fairly "barbaric" system that way -- not many applications, no desktop icons, etc.
3 files look like they need editing - emc.ini - emc.var - & - standard_pinout.hal - Im looking at the screen shots in "cooltool first steps" pdf file now
wholepair: usually emc.var doesn't need to be hand-edited.
wholepair: to change the pinout you'll change the .hal file, and to change the stuff like steps per inch, acceleration, etc., you change emc.ini
Ok - good to know - where can I get these two files to start making edits here now so I can try them at home later.
wholepair: that's a harder question. The source code (which includes the configuration files) is available to browse on-line, but some incompatible changes to the configuration files were made since puppy emc was released.
wholepair: so you would have to get an old version of the files, I'm just not sure which old version is the right one.
I guess it's best to use the files that are on the bootable CD I have. - which means I should get back to work and wait till quiting time.
anyway, the place to view and download the stepper configuration files is here: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/stepper/
i guess both distros are fine on 512 megs ram
For standard_pinout.hal this is probably the version on your cd: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/stepper/standard_pinout.hal?rev=1.1;content-type=text%2Fplain
approx: yes, 512 megs ram is just fine for regular ubuntu.
wholepair: and this is probably similar to the "emc.ini" on puppy: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/stepper/stepper_inch.ini?rev=1.3;content-type=text%2Fplain
jepler: thankyou - so, standard_inch.ini has leadscew pitch and velocity accelertaion variables, and standard_pinout.hal has the pinout variables.
I will download them and edit them while refering to the wiki - I will do as much as I can and then come back later.
wholepair: yes. If you have questions about any of the specifics, don't hesitate to ask here on irc.
wholepair: our documentation of the .ini file could be better than it is
what do you mean by "your" documentation - the readme file in the directory where this other stuff is?
wholepair: the wiki, mainly. When it comes to hal there's actually a fairly complete document about it in pdf format: http://www.linuxcnc.org/Hal_Introduction.pdf
wholepair: the wiki is http://wiki.linuxcnc.org/
but it sounds like you already found it
yeah - I looked at Hal_Introduction.pdf - but it lookes like it was written for developers - I just want to know how to edit configuration files.
wholepair: if you just want to change the pinout, once you know which file to look in, it's pretty evident - there are just pin numbers like you would expect.
I'll dive in - be back later!
eagle: how do you make it reroute after moving a part?
Does/will emc2 support microstepping?
micro stepping is a stepper drive option.
so if the driver supports it one increases the steps/inch parameter?
emc would output the pulses - if microstepping is enabled in the drive - then emc would output more pulses for the same distance
this would be setup in emc's ini file. (steps per inch)
steps per unit
eagle: think I figured it out - I can turn a trace back into air wire then re-route
eagle: I wonder if you can force it to reroute everthing automatically
in the hal_introduction pdf file, I cannot get any of the commands working. Which directory should I be in when running the commands (such as scripts/realtime start)
That is one of the many things I have not played with yet.
SkunkWorks: you can rip up all the traces by typing "rip;" and then confirming.
SkunkWorks: "auto" will of course re-route all airwires by default
Jepler: thanks again. that is what I wanted. thanks
jepler: You don't have any idea of why the HAL bash commands don't work?
Lerneaen_Hydra: no; do you get an error message?
Ok ok ok. really starting to like eagle :)
jepler: I can't run any of the commands in the HAL documentation, for example "scripts/realtime start" doesn't work, nor does bin/halmeter
Lerneaen_Hydra: You're using ubuntu with cradek's packages, right?
Lerneaen_Hydra: try /etc/init.d/realtime start
Lerneaen_Hydra: The paths to some of those commands are different with the ubuntu packages. For instance, "realtime" is /etc/init.d/realtime, and you should just be able to type "halmeter", not "bin/halmeter".
and you need only 'halmeter' not bin/halmeter
myabe it would be good to add that to the hal_introduction file, as it is not very obvoius that the paths are different
Lerneaen_Hydra: the hal_introduction.pdf is intended for run-in-place (maybe the doc should say that..)
Lerneaen_Hydra: we'll change it, thanks for letting us know
Lerneaen_Hydra: if you get the sources and compile for testing, you can either run them somewhere in your home directory, or install to some default location (like the files you are using now)
but you shouldn't worry about this..
strange, etc/init.d/realtime start complains, no such file or dir.
you are missing a leading /
it's '/etc/init.d/realtime start'
what is it that is started with that line?
something that emc2 starts as well I presume?
the realtime subsystem
when you run emc2, the first thing it starts the realtime part
ok, is there some type of self-diagnostic script that comes with emc2, or soem command that will test the subsystem?
the realtime subsystem, say output a signal and check that it seems to be output correctly
yes and no
the realtime subsystem is actually RTAI in your case, that's a RT patch to the linux kernel
if anything is wrong with that part then /etc/init.d/realtime start will fail
in my case you can test rtai with /usr/realtime/testsuite/user/latency/run
yup, that sounds like a good test too
actually on ubuntu you probably want /usr/realtime-2.6.12-magma/testsuite/kern/latency/
in the terminal, change to that directory and type "sudo ./run"
fenn: you should use kern not user to test RT stuff
but I'm not sure that actually works (I get an error about /dev/rtf, but it might be something I screwed up on my own system)
does kern have a higher priority or something?
user uses LXRT iirc, different architecture
thats all beyond my expertise i'm afraid
kern uses kernel mode real time (the one emc2 uses too)
i'm sure it would be really simple to understand with good documentation
LXRT is a way to run realtime stuff from userspace applications (a little bit more latency that kernel stuff)
alex_joni: are you sure there's more latency?
so the gui/base part of emc2 sends commands through lxrt?
I also get an error, "error opening /dev/rtf/3"
fenn: I suspect there's more max_latency, the average latency might be comparable
Lerneaen_Hydra: 1st question : emc2 uses shared memories to talk to the RT part
Lerneaen_Hydra: try realtime start first
the realtime module opens a SHM, then the userspace opens the same SHM and they talk by sending commands through a buffer
Lerneaen_Hydra: what are you trying to do? I missed the beginning of this conversation
test the realtime subsystem
fenn: the realtime test suite doesn't depend on /etc/init.d/realtime
but Ubuntu uses udev, and the /dev/rtf/* are missing by default afaik
ah yeah i remember adding those to udev
i dont remember what they do exactly
Lerneaen_Hydra: do you have reason to believe it's not working right?
nothing, just some mknod iirc..
cradek: no, I just wanted to test it. originally I had some trouble with the bash commands in the hal_introduction.pdf, it turns out many of them were outdated. Ideally I want to be able to look at the signal output that goes out through the paralell port
cradek: By outdated I mena that they were made for some system other than your ubuntu script
Lerneaen_Hydra: I see. If you just run emc, and then at the prompt run halscope, you can view those signals
Lerneaen_Hydra: the current pdf is outdated
in more than one way..
do I want to look at traj-thread, servo-thread or base-thread?
Lerneaen_Hydra: If you want to look at the parallel port, then it's base-thread
for fast parallel port signals, you probably want base thread
I don't seem to be getting any output at all
Lerneaen_Hydra: It may not be "triggering"
Lerneaen_Hydra: did you select some signals using the numbered buttons near the bottom of the screen?
I set it to normal and forced trig
if you want to look at the value of a parallel port pin, it's under "pins", starting "parport.0."
this assumes your emc configuration has loaded the parallel port driver, of course
I still dont seem to have any signal
when you change signals, it sets back to "stop"; click "normal" and "auto"
Run Mode Normal, Trigger Mode Auto
bon soire dmessier
ah yes, now it works fine
I ran emc with stepper_inch.ini, selected parport.0.pin-03-out, "normal", "auto" and now I see pulses when I jog the X axis
sa va Alex??
I see pulses when running as well
dmessier: oui, c'est bonne, mais je suis tres fatigu.e
im a free agent up on waivers this weekend... LOL
Lerneaen_Hydra: then by clicking "Source" (lower-rigth) and selecting channel 1, and setting run/trigger to "normal" "normal", I see pulses when I jog, and when I stop jogging it stops updating the scope (because it doesn't trigger anymore)
oui aussi.. stresser
j'aim mon travaille mais je veux de la $$$$
Lerneaen_Hydra: feel like you're getting the hang of halscope now?
Is Jepler and cradek here? I have [AXIS_#] Section questions...
Wholepair: I am here
Wholepair: a few others around too, ask away
jai venu a savoir q'est mon contracteur charge pour moi ...
INPUT_SCALE - I have 10,000 steps per inch, how do I calculate its value
INPUT_SCALE is the number of steps required to move one inch
that's a rather simple calculation then.
well, it's steps per UNITS
UNITS = 0.03937007874016
if you have this, then UNITS are inches
58% pour moi et 42% pour eau... non j'un pense pas... ; )
jepler: Yes, although the update speed is rather low (10 updates/sec I would guess)
yeah - so UNITS is 0.03937.... - so in INPUT_SCALE I pu 10000
dmessier: 58% sounds ok
10 % is std
Lerneaen_Hydra: The samples are gathered at each time the "thread" you selected executes
Lerneaen_Hydra: it's true that the onscreen display doesn't update that often. In fact, it will never update more often than the time it takes to fill the buffer of samples
however the raw dat output is more often..
jepler: so if I lower the buffer it will display more often?
Lerneaen_Hydra: yes, up to a point
eventually the limiting factor will be how fast the screen can update
are you using halscope to display a value?
you know about halmeter right?
yes, although for this purpose I found halscope to be better, I wanted to look at the paralell port signal to check that everything was ok (not that I had any reason to think otherwise).
Lerneaen_Hydra: For instance, my base-thread is 50.3 usec and I selected 4047 samples. The buffer will fill every .20xxx seconds, an update rate of about 5Hz.
thats pretty fast
in practice it seems to be updating the screen at about 2Hz
It's interesting to run this and watch one of the step outputs in halscope: g3i1x0y0
you get to see two accel-decel cycles
For STEPGEN_MAX VEL and ACC - I assume this is related to the speed of the computer Im using - lets say Im ok with a max of 60 inches a minute and my motrs are capable of high accelerations, they are 1/2HP servos not steppers, and the table is not very big. What are the units of these param, steps/ms?
Velocity is in units per second, and acceleration is in units per second squared
I'm no good at guessing values based on the HP rating of a servo
Wholepair: you have a motor which turns at a certain speed
Wholepair: you generally have to tune servos by trial and error
now you need to tell us what amount of distance the table travels at one rotation of the motor
The machine has already been tuned - I've been running it with TurboNC by Dakeng - the servos are driven by gecko320's.
how many "new people" do you see every week? don't these repetitive questions get annoying after some time?
Lerneaen_Hydra: they get sometime.. but then we look away, and there are others answering :)
we rotate doing this.. (not by a certain schedule..) but by availability
Lerneaen_Hydra: we figure one of you will update the documentation one of these days
jepler: it is a bit lacking in places, thats for sure
Lerneaen_Hydra: we usually don't do this stuff (learning the first few steps), so it's hard for us to put ourselves in your shoes..
Lerneaen_Hydra: that's why we benefit from open-source, because anyone can contribute.. either with code or with documentation
gotta run - ill come back
hello, i installed emc2, then i created the emc2/configs directory in home directory, this directory showed up when i run the emc, but now i recompiled the emc2 (i dod not change anything) and the home/emc2/configs directory doest show up any more....why is that?
fogl: where did you put the emc2 you recompiled?
random question #235923: how unsafe is it to set ubuntu to autologin (autologin to standard user)?
Lerneaen_Hydra: depends who has acces at your PC, and what you keep on it
do you have wierdo's roaming around your home?
should be ok for a home desktop
no to both of those, this is the EMC computer, and it will be attached to the network (afaik, if it's not I know this is safe, but at the moment I think i'll attach it to the network to ease transferring nc code).
fogl: did you run ./configure or ./configure --enable-run-in-place ?
yes to which one?
huh, scared him away :D
i did run ./configure
then make clean, than make
like described in README file
fogl: what system is this?
what distribution I mean?
ok, there's little need to recompile on ubuntu, unless you want to test some specifics of your own
you should either remove the apt installed package, then install one as you started (./configure, make & sudo make install)
or you could leave the installed (by apt) emc2 alone, and test the sources in another directory, for example home/emc2
for #2 you would want: ./configure --enable-run-in-place && make && sudo make setuid
that means emc2 will be compiled locally, to work in home/emc2
so you can modify the sources and recompile, without affecting the installed emc2 version (from the deb you installed with apt-get install emc2)
I work in a scientific instrument machine shop and students useing the facilities keep interupting my slacking off. Where was I? Oh yeah, people were saying my questions where annoying.
stepgen accel and velocity values?
wholepair: keep those 5-10% bigger than the normal ones
these are very strict limits
if the needed pulsetrain exceeds any of these limits, emc2 will error out
alex, i dont eve know wher i installed my preveous version of emc, becuse i used a script i found here: http://timeguy.com/cradek/emc/ubuntu
ok, that does the apt-get install emc2 for you
fogl: can you tell me what you want to do? so I can advise you of a decent way to sort things out?
because having 2 versions of emc2 on the same PC is error-prone
i would like to see the code and eventualy edit it
i would like to make a machine you did (with three steppers)
ok, then best it is to get emc2 TESTING into a folder in your homw dir
cvs -z5 -d:email@example.com:/cvs co -rTESTING -d emc2-testing emc2
currently the normal ones are set maxvel is 1.2 and maxaccel is 20.0, what is 1.2 in inches per minute? I want a max of 60ipm to start, I will increase it if emc2 is happy. Im running a PII 800mhrz. I have 10000 step per inch lead screws.
that will create a folder called emc2-testing in your home dir
1.2 * 60 = 72 IPM
1.2 is inches per second, and a minute holds 60 seconds (last I checked)
fogl: now that you have the emc2-testing dir, go to emc2-testing/src and run './configure --enable-run-in-place'
if that causes any errors, then you want 'sudo apt-get build-dep emc2'
fogl: apt-get build-dep emc2 -> gets all packages needed to compile emc2
i did that
i did what is described in wiki: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2
that page says './configure --enable-run-in-place'
if you run the newly compiled emc2 (cd ~/emc2, scripts/emc) does it work?
make sure you ran './configure --enable-run-in-place', if in doubt do it now, and then 'make clean && make'
I will use 1.2 and 20 for vel and accel - and a small % higher for the others like the comment in the ini file says. Now, where do I put info on the limit switches. I edited the hal pinout file for step and direction signals but I did'nt see places for limit switches. Limit inputs are wired parallel - normally open - they go through a cambell breakout board - Inputs are wired parallel - Normally Open -
positive limit is on 12 - negative is on 11
wholepair: you need to add some hal connections to your hal file:
newsig minlim bit
newsig maxlim bit
linksp minlim <= parport.0.pin-11-in
linksp maxlim <= parport.0.pin-12-in
linksp minlin => axis.0.neg-lin-sw-in
linksp minlin => axis.1.neg-lin-sw-in
linksp minlin => axis.2.neg-lin-sw-in
linksp maxlim => axis.0.pos-lin-sw-in
linksp maxlim => axis.1.pos-lin-sw-in
linksp maxlim => axis.2.pos-lin-sw-in
that should be about it..
alex_joni: thanks for figuring that out .. I wasn't sure of the pin names
err, except I really meant 'minlim' not 'minlin'
wholepair: I can explain what is going on there if you want
but if you don't care, just put those 10 lines in your standard_pinout.hal
(after changing minlin to minlim :) )
jepler: I wasn't sure either.. looked them up in simulated_limits.hal (configs/sim/ )
Alex - not right now - I will just cut and paste - can I put it in right after the 6 step and direction lines - got the minli'n'
wholepair: yup, that seems ok
please ask jepler or SWPadnos for help if it doesn't work..
* alex_joni is dead tired..
* SWPadnos knows the feeling
need to head to bed.. goodnight everyone
SWPadnos: been a 10-12 hour day again (the 4th this week)
SWPadnos: 2 more to go ;)
yep - did that over the weekend, then got no sleep on the overnight flight from LAX
finally got to sleep last night
* robin_sz meeps
is it going well with you?
* robin_sz is tired out
yeah, things are good. Plenty I should be doing though.
just moved factory ...
so exhausted, but happy to be moved
monday was the big day, 2 18 tonne trucks and a 26 tonne truck :) .. and we filled them
moving within a city, or to a new city?
from a small village to the local town ... 10 miles or so
jepler: you there?
* Lerneaen_Hydra wonders why there is so low activity now...
giacus_afk is now known as giacus
supper time i suppose
yesterday someone (IIRC jepler) wrote that the average stepper+drivers+ballscrew cost around $1000, what is it that is so expensive? As it is now I can see my components costing under the $100 line...
(excluding the mill of course)
whiat kind of machine?
just the steppers and drivers
say for a smaller mill
really? what drivers?
medium/small benchtop size
various electronic components
ok nevermind, and good luck
an XOR chip and another cheap digital component
* fenn is sick to death of stepper driver design
... do most people buy premade things?
[21:53:01] <Lerneaen_Hydra> http://www.edn.com/archives/1998/032698/07di.pdf
anyone who wants good performance usually does
yes, but some people here have designed or built their own
by good perfomance do you mean a large current?
all that's missing from that design is the actual power electronics
performance is performance
if it has lots of current but doesn't work right, that's not performance
and you don't need that logic at all if you have a second parallel port
yes, however some basic FET shouldn't be that expensive
(emc can be made to output 4 phase waveforms)
there's more to motor control than a FET - ask bo^dick :)
can you set EMC2 up to output the correct 4 pinned waveform?
ah!, now that is interesting
use the right stepping_type (not sure which one it is) when loading stepgen
a simple FET design will only work with unipolar motors, which I think are generally quite low power
wel, you'll need som H bridges, as it says in that pdf
I have some low-power steppers I'm going to test with to begin with
Lerneaen_Hydra is now known as Lerneaen_Hydra_l
Lerneaen_Hydra_l is now known as Lerneaen_Hydra
so most people here haven't had a good experience with homemade drivers?
homemade drivers are OK, but they have two disadvnatges
1) they dont work as well as geckos
2) they cost more than geckos
* fenn rolls his eyes
#2 is only true if you end up replacing them with geckos ;)
I dont see how a commercial product can be had for under $2,,
robin_sz: what about a pre-designed design such as on pminmo
if you count any value for your time, then they cost more
* robin_sz nods
my time is cheap, for now ;)
only for you guys that make $100/hr
nah, geckos are what? $99 ?
even at my rate of only $75/hour, geckos are a better deal
close enough ...
unless you buy a lot
bleh not when you need 6 of them
well, I tried to build drives .. they cost me three times what geckos cost, were shite, and then I binned them ...
...as my time is $0/hour, it is rather cheap
but, feel free to prove me wrong :)
Lerneaen_Hydra: your time is 0$ ..??
i still dont understand how you can spend so much money on components
fenn: very easily ...
i'm getting fets for $0.37 each
robin_sz: I'm building this for fun/hobby project
where's the expense then?
Lerneaen_Hydra: take my advice .. the software and mechan ics will be enough of a challenge
or did you really blow up 310 mosfets?
mosfet drivers I blew a few ...
waste dhours makign pCBS
the software seems to be working, at the moment. the mechanics however will probably be real hard to get to a good level
fenn: you can make your own monitor from a old TV tube .. ti would be cheap ... but non-worhtwhile. stepper drivers are in the same class of project IMHO
I don't see the difficulty in it though, my dad does electronics as a living, and as so I don't beleive that there should be much of an issue
SWPadnos: is it difficult to set up a second paralell port?
well, fair enough ... I did electronics as a living too ... but, feel free to prove me wrong
but it varies
I'll write something up if it works well
SWPadnos: It varies?
you'd probably be best off getting a kit or hobby-style stepper driver. there are several available
it varies depending on the bus (ISA, PCI, PCI express), and whether or not the board needs a driver
if it's just a standard paralell port adapter
then it should be pretty easy. all you have to do is specify both ports on the HAL load line (I don't remember the exact syntax for that)
the lines that go something like parport.0.pin.0 - pin.x, and also then parport.1.pin.0 - pin.x?
how do you specify outputting as a 4 phase waveform?
not going to be microstepped?
* fenn flings poo at robin_sz
I would be content with half-stepping
fair enough ...
the ratio in teh end will be so low anyway
err.. high I mean
personally, I love the smoothness of microstepping too much to go back to old clumky now
anyone know what this "electrical damping" is all about?
by microstepping do you mean full-sine wave or quarterstepping?
full sine wave .. or 10 discrete steps per normal step anyway
nearly sine wave
"sine approximation" at any rate ... of course you transition to full sq wave at high sopeeds anyway to get max torque at speed
that doesn't make sense
yes it does ...
to reduce switching losses?
to bring the I up as high as possible as soon as possible ...
the average "push" of a sine wave is sqrt(2) of the square wave
the motor power output is the integral if the area under the I curve
wouldn't you want to phase shift it?
or maybe it would look more like a backwards sawtooth
at high speed, it probably makes no real difference
the phase shift that is
at most you woudl be 1/2 step out from the "REAL" position
what size of motors do most people run (and how large a mill)?
typically, drives go sq wave at about 600 rpm
Lerneaen_Hydra: i'm using nema23 on an overgrown dremel tool
model numbers dont say much, how many watts?
Lerneaen_Hydra: you can run a Bridgeport mill from Nema 34 / 6Nm motors on 4:1 gearing quite nicely
4:1 gearing! :|
thats rather... fast
4:1 onto the ballscrews
or do you mean 1:4?
that sound more reasonable
so the motor spins fast, the scrw goes slow
getting about 20Nm onto the screw basically
Lerneaen_Hydra: nema23/nema34 are bolt hole patterns i think.. about 2 inch and 3 inch, respectively
the thing is to avoid those crap old 110 Oz/In surplus motors like the plague and get decent, modern, hybrid motors
those motors are still rather large/powerfull, these mills that you're running, are they fullsize or benchtop?
you basically get about 3 times the perfomrnace from a modern motor as the old crud, which really is pretty useless
ah, ok. that would explain the rather large motor
nema23 like fenn said is perfect for a benchtop
old crud, what differentiates the new stuff from the old?
mainly th emagnets and the indctance
you get about 3 times the torque from a modern motor
and it holds it to 3 times the rpm
the old motors out of printers are crap
well, thats to be expected
well, its been quite a dramtic improvement
there really is no comparisom between a modern hybrid stepper on a good microstepped drive to an old clunker on a L297
robin_sz: where do you suggest getting motors?
I don't know if the steppers are "old clunkers" but with l297 half-stepping cradek gets a good .6IPS on his maxnc
32? 36? ipm
is that "and can do stuff" or "once got ti to do that without missing steps for a bit"
"and can do stuff"
"never loses a step"
my router does up to 600ipm on Nema34 and geckos ... direct drive onto a 10mm pitch screw
* fenn yawns
time for bed methinks
but I guess 36ipm is OK on a small mill, its not that big a travers from one end to the other
the maxnc is a hardware stepgen no?
robin_sz: the original maxnc drivers are terrible---winding activations in, unipolar, big power resistors
robin_sz: that was tossed out and replaced with l297/l298 bipolar
600ipm! thats a bit
I tend to run G0s at 400IPM just for my own sanity
the large & meaty CNC mill at school does 950ipm
its rather frightening
you dont want to know how fast my laser goes ... ;)
especially when you have a g0 to z1 and the material is at z0
the laser G0s at 84m/min ... thats over 3000 ipm
thats quite fast
the laer beam however, a bit faster ;) 708,661,417,322 ipm
if you do build drivers, try the trusty L297/L298 combo
probably the sanest choice
those actually look quite good
although I may not need the l297 if I have two paralell ports
is it just me or did many people disconnect and then reconnect?
[Global Notice] Hi all. We're experiencing serious problems with connectivity to our European hub. Investigating now.
[Global Notice] Hi all. We momentarily lost our connection between our European and US hubs. We're told there are reports of ssh and http connection problems, so this may have been a large backbone glitch. If you have any information, please let us know; we're looking into it.
[Global Notice] Everything seems back to normal. Thanks for your patience, and thank you for using freenode!
[Global Notice] Hi all. Looking at http://www.internetpulse.net/
(internet pulse), it does look as if there were fairly widespread problems. So far things seem to be stable. Thanks again for your patience, and for using freenode!