never mind, the answer was find in the rtai patch names..
225 lbs casting is now in the basement
still gotta get it onto the bench
225 lb - is that the table for you next machine?
no, the base of the benchtop surface grinder I bought at HGR surplus
almost identical to this: http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=16720603&PMAKA=619-2879
heh - http://www.cnczone.com/forums/showthread.php?t=61437
crap - have to go agian.
casting on the bench
rope is a wunderfull thang
tempus done fugited while I was doing that
must find food
maybe i'm done clicking the wrong button for a while.
grinder move pics: http://jmkasunich.com/cgi-bin/blosxom/shoptask/grinder1-07-19-08.html
jmkasunich: that looks like it was fun! :p
jmkasunich; nice transportation of the grinder :)
I'd have probably called friends over and tried to lift it up & carry it down
* fragalot imagines the mess
hehe yeah, with three people that's only 30kg/person
and we all know three people carrying stuff over stairs is oh so fun
one at the bottom gets all the weight
* fragalot remembers carrying the 185kg mitutoyo granite stone down the stairs
well,... carrying.. we ended up extending the forks on the forklift and just lower it :p
yay just finished making up my first stepper motor driver circuit :p
now to design the PCB..
[12:31:06] <fragalot> http://omploader.org/vbWEx/Stepper%20driver2.pdf
<-- any design flaws in here that i should be aware of?
where is U2 connected to Vdd?
*cough* now on pin 16..
can't spot anything else atm, but I don't know these chips
do they do half stepping/microstepping?
U2 is a BCD up/down counter, and U3 is an decoder BSD-DEC
hmm.. then maybe you are better set with a L297/L298 combo
* fragalot googles
don't think i'll ever need halfstep
full step now should give me a displacement of .005mm per step
full step sucks balls
and since the rest of the construction won't exactly be accurate,.... :p
look at the torque curves.. you'll lose steps like crazy at the resonant frequency
[12:39:26] <fragalot> http://www.robotshop.ca/PDF/rb-soy-07-unipolar-stepper-motor-specs.pdf
notice that the curve is given for half-step operation
anyway the x-axis resolution is way too low to see resonance phenomena
on the graph (pps)
all i'll really use this for is making PCB's, it's tiny
fragalot, bipolar is better
but double the cost
fragalot: motors moved with half step run a LOT better than with full step
even if you don't need the resolution
how many axes do you plan?
you could output 4 lines / axis from emc2
often unipolar motors can be run as bipolar
and interface that directly to your transisters
then use emc2 to output half step phases
[12:44:20] <alex_joni> http://www.linuxcnc.org/docview/html//hal_rtcomps.html#sub:Stepgen-Step-Types
alex_joni: so drive the FET directly from the parallel port?
step type 6
yeah, but you need 12 outputs for this
that is the maximum you get from a single parport
so you won't be able to drive anything else (like spindle start/stop)
unless you add a second parport
don't need the start/stop
i'll just cram a dremel onto the Z-axis :p
fragalot: you might want to take a long hard look at http://reprap.org/bin/view/Main/Stepper_Motor_Driver_1_1
* fragalot puts on his reading glasses
unfortunately there isn't much explanation of 'how it works'
but its a good design that is about right for what you're trying to do
fenn: cheap too!
well, the board is :p to have a single-layer board etched here i pay ~15
yeah $28 is not too bad
alex_joni: so, parallel port -> darlington -> FET -> stepper would work?
fragalot: yes, probably not as good as a 298
you don't have current settings in that scenario
so you need to use low voltage in order not to damage your motor
alex_joni: that's why i put a ballast in my other schematic :p
* fragalot gets crackin'
"Computers are incredibly fast, accurate and stupid. Human beeings are incredibly slow, inaccurate and brilliant. Together they are powerful beyond imagination." :)
it's pins 2-9, 14, 16,17 that i can use as outputs, right?
*stare* why can't I ever find those things on your site :p
fragalot: ask BigJohnT
he's the big docs guy these days :D
fragalot: what were you looking for and where
BigJohnT: gathering info for my lil' desktop sized 3-axis mill
on how to drive the steppers in the most cheap way :p
we call that cost effective at my shop :)
I have been working on the docs a lot trying to make things easier to find...
BigJohnT: reckon hooking up some darlingtons to the parallel port, and using those to drive FET's would work? :p
(and have EMC2 do all the hard work :p:p)
where is SWPadnos_ when you need him :)
I use a cnc4pc breakout board with geckos
any1 know a site that lists FET's by the current & voltage they can handle?
google does :)
makers sites can
google doesn't give much usefull stuff out atm :/
with EMC2 I can invert the output signal, right? (eg. for the motor control)
setp parport.0.pin-xx-out-invert TRUE
morning all. I have a physical momentary button that i would like to work like the 'toggle Machine Power' button (the F2 key). any suggestion how to do this?
Roguish_: check for nist-toggle
it's a comp
it's called toggle2nist
awallin (iirc) had some description about it on his website
i've been all through the 'toggle' (w/ and w/o reset) and flipflop. what signal or pin do i 'toggle'?
[14:22:03] <alex_joni> http://www.anderswallin.net/2008/04/toggle2nist/
something like this?
i see 'toggle' in the docs but not toggle2nist. where is 'toggle2nist?
Roguish_: might be newer .. (it's in the latest CVS), but you can also take it from the site I pasted
and install it with "sudo comp --install toggle2nist.comp"
as you spotted an error in my last try, hows this one (BC517 = darlington NPN, IRFZ44N is an "overkill" 49A N-FET)
irfz44n is a good lil' bugger :)
fragalot: what are all those squiggly lines?
fragalot: make sure you don't need pull-up's on some of the parport pins
when the darlington switches off that will turn the FET off near-instantly, causing flyback voltage
so i think you need a resistor between the darlingon and gate of the FET
to slow down the switching action
BigJohnT: those are resistors. :p
also, ultiboard/multisim sucks as a combination,.. i export the netlist and notice lots of the nets are just.. missing
fenn: and for the flyback voltage, i'll just put some diodes to catch that
hm this is interesting -- if I probe the same spot repeatedly, the first probe is high, the second probe is a little high, and then the remaining probes are less than 3 microsteps away from the same value (tool length probe from cheap microswitch)
fragalot: the fet already provides a diode-like effect, and any diodes you add have to be much faster to have any real effect.. i think a resistor is a cheaper and just as effective
k, which value do you suggest
i dont know :)
probably somewhere between 10-100 ohm
but i think you just have to determine it by trial
that would slow down the FET's tho, which is not exactly something I want. :p
at slow-ish switching speeds (some kHz) it doesnt add much to the overall losses
alex_joni: and asfar as the pullups, don't think i'll need those with darlingtons
jepler: do you think you have to loosen up the switch to get a consistant reading?
BigJohnT: I don't really have a theory yet
* BigJohnT goes back to mowing grass
jepler: tried both ways probing?
switch trips/moving off the switch?
alex_joni: can't do that in 2.2; there's only G38.1
cherry-pick in emc2-backport-modules ?
* alex_joni is kidding..
I have a theory why, though
I am probing in a loop. First I probe from top-of-travel at a high speed. then I back off by Z.1. then I go into the loop, probe slow, back off by Z.05
so the probes are different because the starting position is different and the machine is accelerating ..
after the first two moves it very quickly settles down to backing off to exactly the same point, then probing to exactly the same point.
bbl.. going outside for a while
and indeed, just by fixing the first and subsequent back-offs to be the same distance the effect almost disappears
see yuo alex_joni, BigJohnT
nice clean signal from the switch even without any debounce
when going slow?
is it a micro switch?
yes, radioshack special
most I have used have "snap" action
[15:41:11] <jepler> http://www.radioshack.com/product/index.jsp?productId=2049718
the machines we are finishing up on test a micro switch for the electric starter on a lawn mower engine.
similar to the rat shack one
once toggled it takes quite a distance to unlatch.
yes that's true
that's no trouble for this application
after measuring the tool, lift up .1 inches, no big deal
actually I'll have to lift a bit more than that to clear the whole lever
but .5 inches, no big deal
I didn't know how good they were till I used them for my home and limit switches on my plasma table
I am planning to use them for home switches as well
still struggling with where to install them, though
I used a single switch for each axis. have you seen the pics?
I thought I had a good plan for X and Y, but now I see that the screws to attach the switches would interfere with the screws that assemble the table
have a link?
actually it's Y that's the problem, and I see a way to mount it that would only lose me 1/2" which I can afford on Y. X I can still mount near where I was first considering, and not lose any travel.
this is Z and the floating head one switch for each http://i47.photobucket.com/albums/f163/johnplctech/Plasma%20Cutter/Plasma012.jpg
this is Y and X is similar http://i47.photobucket.com/albums/f163/johnplctech/Plasma%20Cutter/Plasma006.jpg
well then, time to design the PCB... should be fun, fitting all this onto a single layer PCB without all too many bridges...
if you get bored design a THC circuit :)
Torch Height Control
it takes a 0-5 or 0-10vdc signal and converts it to up and down inputs to EMC
[15:58:33] <fragalot> http://members.aon.at/mittelmann/kytrondocs/circuit-K3.gif
there you go.
BigJohnT: i'd probably do that with a PIC with built-in AD converter :p
I think I have a pic programmer around here somewhere
I've got ~3
one doesn't work, the other gets corrupted if you unplug while the pc still runs, and the 3rd works with all types of PICS, apart from the ones I have.
[16:01:15] <fragalot> http://forum.velleman.be/viewtopic.php?t=1674&sid=c237cc45f5b8725f6f0f505924197ed3
<-- this is the 3rd one
the previous posts got deleted, but, they said "talk french or english",.. so i looked up the documents in english, and in french, only to get told that those aren't from them, (however they are hosted on their own server), and then get linked to a dutch one ;p
got to find a better client... will be back later... :)
anyone have cnc lathe? can make me one of these? http://omploader.org/vbTV4
didn't I see that on star trek?
it's all done with smoke and mirrors
its been waaay too long since i used irc...
anyone here knowing something about the emc internals?
or should I go to emc-devel?
lots of people do, but they are out mowing their lawn :p
i should do too probably... :)
is this the best place to ask questions or is the mailing lists better?
well I have a problem understanding the dataflow from g-code to motion...
g-codes get processed by the 'rs274 interp' into 'canonical commands' which get queued in the user/rt 'shared memory buffer' that feeds motion
looking at some of the wikipages there seems to be a shared memory area for comm between emctask and motion
ok, regarding the queing; can you point me to where it is queued in the shared memory buffer?
the reason why i'm asking is that i would like to know how much is actually queued
and a lot of other stuf i would like to know... :)
i would guess that it would be emcmotCommandhandler i should be looking at
hmm i havent really looked at this too hard, but see the emcmot_command_t in motion.h?
yep, doing that right now... :)
but there seems to be only one of these
so basically no buffer
right i'm looking for the queue
ah see this usrmotintf.cc: shmem_id = rtapi_shmem_new(SHMEM_KEY, module_id, sizeof(emcmot_struct_t));
yep, got it...
so just after that is how you get the pointer to the start of shmem
now /me looks for where that gets initialied
still, emcmot_struct_t does not seem to implement any buffers, except for an error msg buffer
no, i think the buffer is made up of emcmot_struct_t's
wouldn't the above line be something like rtapi_shmem_new(.....N*sizeof())
i have no idea though... :)
btw, is there a best time for visiting this channel? i thought i saw something like 17:00 gmt somewhere
the weekly meets have sorta fizzled out
IIRC, the queue depth is 200
the queue itself is not in shared memory
one command is placed into shared memory by user space code, on the next execution of the servo thread, command_handler() takes it out of shmem and either executes it (if it is something immediate like abort, etc) or queues it
once user space sees that the command is no longer in shmem, it can place another one there
* fenn greps for command_handler and fails
but if the queue is not in sharedmem then there is a possibility that userspace code will not have time to put a new command in shmem?
say, if a user starts up a browser while running emc... :)
if there is no command in shmem the next time commandhandler runs, it simply does nothing
most movement commands take many milliseconds to execute, and there is a 200 deep queue of them within the motion controller
(commandhandler puts them in the queue one at a time as they are recieved from userspace
the trajectory planner takes them out of the queue as each move finishes
let me just see if i understand this... :) From userspace we have emctask putting commands into the shared mem
during movement, for motion this may happen say every 10 ms
it doesn't happen at any regular interval
ok, i was simply under the impression that the motion controller ran at regular intervals
the motion controller does run at regular intervals
but this may only be some parts of the motion controller?
the motion controller is on the realtime side, and it runs at regular intervals
if there is a command waiting in the shmem buffer, motion will process it on the next interval
but there often is no command waiting
* jmkasunich tries to figure out what happens when the queue is full
queueFull var maybe?
hopefully emctask stops doing anything... :)
but I don't see any code doing that test in command.c
that is in userspace...
oh, maybe userspace doesn't put a command in the buffer if there isn't room in the queue?
ahh, youre talking about rt
yep, that would be my though
* jmkasunich <-- knows about the RT part, not so much about the userspace part
actually, it has to work that way
I was thinking that if userspace put a queueable command into the shmem, rt could just ignore it until there was space in the queue, and userspace would do nothing until the command was removed from the shmem buffer
but that would prevent userspace from issuing any other command, including unqueued ones like abort
mm,.. i just thought of putting a lil' oily sponge onto one of my whatever you call them.. those threaded rods that move the X, and Y axis
to keep it oiled :p
emc/motion/emcmotcfg.h:#define DEFAULT_TC_QUEUE_SIZE 2000
someone must have thought about that... :)
looks like I was wrong about the queue size - it is 2000, not 200
memory is cheap... :)
cheaper now than in 1995 when the code was first written - I believe it _was_ 200 ;-)
even though abort will then first be executed after 100 us (or 1 us or whatever) it will still be executed fast
btw, #emc-devel is probably a better place for code discussions
so commands are only queued with the trajector planning interval?
i would have thought that it would be at servo rate
i think they are the same now
CommandHandler runs at the servo rate
user space is what queues things, it doesn't run at any rate - it is user space
well the motion tasks also queues stuff you've just said...
oh, sorry, I'm being unclear
2000 of them... :)
there are two places where commands are buffered, and I mixed them up
userspace -> shmem command buffer (1 entry) -> CommandHandler -> motion queue (2000 entries, not shared memory) -> trajectory planner
so you are correct, transfers from the shmem buffer to the queue can happen no faster than one per servo cycle
(and can happen slower if there is no command in the shmem buffer)
which is the typical case i would think, except during startup
is the servo rate and trajectory rate the same now?
you can set them the same in the ini file
in the next major rev, I think they will be the same
ok, that was also what i thought
back in 1995, cpu cycles were also much more expensive, and the traj planner/kinematics may have been too complex to run at the servo rate
so there was a simple interpolator to take the traj output (every 10mS typically) and generate positions at the servo rate (every 1mS typically)
that just adds unneedded complexity and delays today
a stupid question: if the servo rate is 1 ms hwat is the typical rate at the output of e.g. a parallel port?
and who takes care of this updating at this much faster rate?
that is set by the "base thread"
emc typically uses two RT threads, servo and base
the emc motion controller doesn't have any code that runs at the base rate
base is used for software step generators, software encoder counting, etc only (HAL components)
ok; i have only run it simulated, so i don't know anything about the base thread
wrt expensice cpu cycles my hexapod controller currently runs 3xrealtime with 1 ms outputs on a 2 GHz pentium; so there is not much margin... :)
* fenn perks up
did someone say hexapod?
Guest270: do you have a real name?
is that a bad thing (TM) here? :)
well, the next time you drop in you will be Guest199
i've only just bagun doing emc stuf...
i'm quite interested in any hexapod development you may be doing
it is nice to know you are talking to the same person later
(or anyone else for that matter)
especially when that person isn't just a casual user - you obviously understand machine control and code
i know... :) i couldn't figure out how to set up mirc, so i logged in using the webinterface three times or so, and the last time i forgot to write pmb...
I believe even when using the guest account you can set your nickname: "/nick somename"
Guest270 is now known as pmb
does that mean "pmbdk123" in the _dev channel is you?
man an IRFZ44N takes a beating
I watch both channels, but I didn't respond to "anyone here". because I was in the middle of something else
fenn: i have a hexapod running and i have a very crude controller (it interpretes a homemade language and have no gui)
when I saw you and fenn having a discussion, I jumped in
it is usually better to just say what you want to talk about
well, mirc kicked me out; have to find an irc client i can figure out under linux
ubuntu has xirc
not sure if it is installed by default or not, I think it is
thats what I use
I use irssi + screen + ssh
i'll use that the next time... :)
well as you may have figures out i dont like mirc now... :)
I use chatzilla plug in for firefox
and it has been a while since i used irc
so i dont remember any commands...
are you guys always on or? :-)
I use xchat on ubunt
I leave the IRC window open all the time (this computer runs 24/7, serves my website, etc)
from which country?
doesn't mean I'm in front of the screen all the time ;-)
nice to have a lastlog of what people said to you while you were asleep :)
im from denmark
usa, eastern timezone (it is almost 2pm here)
ok, its 8pm here
on a standard PCB, if you need to run approx 1A trough a lane, how thick would you make it (minimum width)
i do have a lot more questions about the source, but i guess i'll try to get another irc client i can use (also under emc-devel); the web-client automatically loggged me into emc
[17:52:52] <jmkasunich> http://www.desmith.com/NMdS/Electronics/TraceWidth.html
jmkasunich: *bookmarks* :)
fragalot: first google hit for "pcb trace width current"
i couldn't find it with google :(
fragalot, bot in #electronics has a calculator built is iirc
mm,.. Cu thickness.... how much is that on a default eurocard, in mils :p
depends on the card... )
[17:56:55] <pmb> http://www.daycounter.com/LabBook/PCB-Copper-Thickness.phtml
last time I did a protoboard it was 35 um
just a question more before i go look for another client: In simulation mode how is the motion controller doing the timing?
hmm, jepler is the expert on that
I believe he uses pthreads to emulate the RT threads
ah, just looked in motion.c
sod this, i can't fit the lane width I need on here.. i'll just slap some solder on :p
you won't see anything in there that will help
it seems to to a usleep(10000)
that seems astonishing
it the last part after #ifndef RTAPI... But that is probably the wrong place... :)
RTAPI is defined during a realtime build
so #ifndef RTAPI is code that is sim only
maybe that is it
yes, and thats what i asked about, so thats ok... :)
I'm a little surprized - since I thought jeff managed to do all the sim stuff at a lower level
maybe i am just worng... I can try to change it to 1 sec and see if it goes slower... .)
i'll have to leave; cu later... :)
general machinery question: this surface grinder has dovetail ways for Y and Z (X is ball-bearing)
there are zerk fittings for Y and Z, drilled thru to grooves in the dovetail surface
grease or oil?
I'm guessing oil, even tho zerks are usually for grease
not oil but way lube
so I need to figure out a way to pump way lube into a zerk
is it a grease type of zerk or the kind that is flush with a ball in the middle?
it might take grease then
I'm tempted to remove the fittings to clean out the passages
it would be enlightening to see what is in the passages
although that won't tell me what it is supposed to have, just what the former owner used last
Y has a fair amount of wear, I think I'm gonna be doing a bit of scraping
* fragalot starts over with his PCB design
do the fittings look like factory standard
seems that way
the passages must be - they are 1/8" or so diameter and 8" long - nobody drilled that at home
not straight anyhow :)
zerk on one side, passage goes under both ways, short side passage to each way
come to think of it I had a small manual grinder at work a long time ago that used grease for the Z and Y as they didn't move very much compared to X
this is a "small manual grinder"
dunno if you saw the pic yesterday
looks like this Enco one: http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=16720603&PMAKA=619-2879
re: zerk oiling, i remember reading somewhere about making a notch on the end of an oil can so you could depress the ball and squirt oil through the notch at the same time
unfortunately Enco doesn't have the grinder manual online (they have many other manuals, but not the one I need.... bastids)
it wouldn't be pressurized or anything, just drip in
in that enco pic, you can see the Z zerk - it is just to the left of the nameplate
yep, smaller than the one we had but similar
hmm, sideways makes it a bit harder to dribble in :)
fenn: you need pressure
can't crank the ways back and forth to draw oil in by capillary action?
for Y, the cross passage and fitting are below the ways, you gotta push the lube up
yay,.. just as i finished my PCB i found a design flaw... forgot to limit the current going into the parallel port from the homing switches
remove zerk, add compression type lube connector, run all to a stock way lube distributor block, feed from bijur
tomp: I thought about something like that
but a bijur oiler would cost as much as the machine did
check that used machine house in cleeveland. gotta be some used cheap
yeh, ~150$ manual bijur vs 10$ trigger style oil can with hose and clamps.
hmm.. automatic oiler's are so nice though :p
except when the machine is serviced... on/off/test/on/off/test.... makes pools of lube ;)
[19:29:08] <fragalot> http://omploader.org/vbWFx/PCB.JPG
-- what do you guys reckon sofar of the component placement of my controller
should i continue that pattern, or is there something bad about doing it like that :p (apart from the ones still next to the pcb, ofcourse)
line the fets up coplanar so you can bolt them to one heatsink (or two)
also, screw terminals should be pointing down so chips dont fall in them (not sure which way is down)
mm, good point @ the terminals
and i don't think i'll need a heatsink for my design.. i've had a FET running behind me for about half an hour now (still going) at a 60% duty cycle at twice the load it 'l have in the end,.. and it's still fairly cold
if i rotate the FET's on that PCB, i'll have too many lanes crossing that i can't avoid,.. trying to keep it all on the bottom layer
how many amps is that?
the steppers i'm using are rated for 0.8A
well, 0.4A per phase
guys, can't find this in the docs, but does EMC2 support GERBER files?
or would i do the image-to-gcode thing
GERBER is not rs274ngc
fragalot: if you use eagle to generate the PCB
there are some eagle scripts that output g-code
which emc2 takes
alex_joni: multisim, 'cos i can't get used to the interface of eagle & others
then you're on your own :P
image to gcode it is!
* fragalot makes the CNC controller on a breadboard and lets it mill out it's own PCB
eagle isnt too hard
no its' not, it's just.. I can't get used to it
multisim got the interface right, but everything else sucks
eagle doesn't have a (for me) usable interface, but it rocks
same goes with kicad & others
gcad supposedly will trace around the lines in a gerber file
gcam, not gcad
I got the blues....
prussian blue that is, all over my fingers
Y ways on my grinder
* fragalot stares into the distance, proud of his work
* alex_joni is off to bed
good night all
* fragalot decides to install EMC2 on his Gentoo laptop and play arround before he can go to the store to pick up parts to make his machine
* skunkworks doesn't think he would have the patience to scrape ways...
Currently an x86 PC is required. ... amd64,... *crosses fingers that x86 compabillity kicks in*
patience I have - energy is flagging tho
lifting and flipping the ~70 casting gets old fast
boo, only getting 1MB/s download speed for EMC2.2.2-&
* fragalot impatiently walks arround in his room
you've GOT to be kidding me
it won't boot on my laptop :'(
lol, Acer reset my bios when i sent it in for repairs
10 rounds completed - if I had to guess, I'd guess another 20 are needed
hmm 10 rounds of scraping?
how much are you taking off
fragalot: not much
great, the CD I downloaded won't load X
* fragalot downloads the other one and crosses fingers
what is your video card?
8600M GT in the laptop, 7950GX2 in my other box, ATI X800XT in the other one, and can't remember the other ones from the top of my head
when I fixed the xorg.conf file, it refused to start because supposedly, I didn't have a mouse >.> thus it must have recognized the video card that time,.. just not the mouse
I should probably just install it in Gentoo, and hope it works in here if the other cd also fails
:p I hate waiting
bbc7 is playing 'Hordes of the things' (parody of Lord of the rings) http://www.bbc.co.uk/bbc7/listenagain/saturday/rams/0230.ram
can't make out what they're saying >.>
the2nd cd failed to write @ 99% done
happy monday people!
well, thats 3cd's ruined.. *tries his luck in Gentoo*
yay it's working!!... I think..
who was it here that said that http://omploader.org/vbWFh
(PDF) this circuit would work?
Hardware that doesn't work
* USB devices, including USB parallel ports.
guess that I won't be using this :(
fragalot; that circuit doesn't look all too good
lerneaen_hydra: apart from a missing resistor for the sensors, it would work tho ^_^ -- yet, as i just figured out that USB parallel ports aren't supported, and I nolonger have any PC that has a spare PCI slot,.. :(
it will work, but not very effectively
the primary fault is that it will take 2 base periods to switch between the upper and lower mosfet, as you can't switch them at the same time
additionally a mosfet driver would be good to have though that's trivial to add
ideally there would be some circuitry that automatically switches the upper and lower mosfets without needing to waste 2 base periods
i'm just trying to find something cheap that 'l work :p
also you can't do a nice PWM'd drive that applies lots of voltage initially and lowers as the current starts to rise, you'll be limited to low rotational speeds
it will work though for a test rig
meh, wanna get it right the first time. :)
have you considered looking at any of the many freely available chopper drivers?
yes, couldn't find any under $150 :p
hmm, that shouldn't be all too hard to find
[22:42:17] <fragalot> http://www.instructables.com/id/Easy-to-build-CNC-Mill-Stepper-Motor-and-Driver-ci/
-- could this work?
that's for unipolar motors
I'm not sure what you have
if that works,.. great!, then theres the issue of the lack of a parallel port that's not hooked up to USB :p
oh, wait, I thought the first schematic was for a bipolar motor, I just glanced at it
for driving a unipolar motor it's simple but it will work
yeah, but i don't have any outputs left, which is lame so i can't trigger anything else via EMC2
though you should consider raising the 100ohm pullups to a few kohm and using some real mosfet drivers
parports are cheap though
I also can't figure out how to get EMC2 to control the thing I made
yes, they are cheap
but i don't have anything to put them in
I think there's a steptype that outputs what you want
all the PCI slots are either covered by a huge graphics card, or taken
one of the devs would know which file thats located in stepgen.c maybe?
so i'll prolly need another pc "on the cheap"
or install X on my server,... but i don't feel like doing that :p
in general it's probably a good idea to have a dedicated cnc computer
agreed, need to find one on the cheap
* fragalot curses self for donating the 200Mhz box 3 months ago
as you can run realtime well on older hardware that shouldn't be too hard to find
I haven't had trouble finding 700-1.5ghz machines for free
I haven't looked :p
anyways, i'll figure that out... -- back to the driver
which one would you suggest that works the best in price/performance (very small desktop sized mill, :p)
.4A per phase on the stepper motors, 1.8° steps
the instructables one is much more expensive (or rather, it costs money, while yours is free), though it only uses half as many IO pins
oh, that's... small
and mine isn't "free",.. I have the parts laying about,.. but they still cost money :p
if you do microcontrollers then making a step/dir or quadrature to mosfet driver shouldn't be hard
shouldn't be more than $10 per chanel though
probably more like $5
the PIC's are all donated nicely by microchip \o/
sampling is always nice :)
gotta do it while you still can (student)
16F877A should do
I've never worked with pics myself (atmel all the way :p )
hehe, I started out with PIC's, and,.. well,.. it's all i have & know :p
2 homemade programmers and 1 sucky (doesn't work, thanks to the parallel port issue thing again) commercial one
homemade ICD2 ftw :p
Thanks for your support (Atmel has a clean room assembly plant just down the road.)
guess i'll continue tomorrow
just finished listening to the Pikes Peak Hill climb - Kinda wanted to go, but glad I didn't it was in the 90's up on the mountain.
I'd like to go up the hill.. not watch :p
in the unlimited class ;)
Me too - but I don't have that kind of money
err - Unlimited quad.
entry fee's for bikes and quads is only $500
but you will burn more than that in fuel and tires in the practice runs.
fragalot: you wrote something about PCI slots being covered by huge graphics cards
that is your problem!
why do you have "huge graphics cards" in a CNC control computer?
you don't need a gaming box for CNC
SkullWorks_PGAB: im thinking unlimited car..
awd + 1000hp or more
$500 for turning the ignition key.. :p
I've driven some of these cars - they are scary fast
Wells coyote - open wheeler & a Subaru WRX concept test car with huge blower drive.
after that I decided I would only try it on a quad.
JMK - have you tried rigid tapping in your lathe spindle?
nope - its a single phase motor, won't reverse unless it is stopped
cradek: how is the lathe
skunkworks: in the driveway!
unfortunately it's 1' too high in the air to roll into the shop (on the trailer still)
aww. = take some air out of the trailer tires>
or tall trailer, or both
it's 6' tall with the guards. the control is 6'6 I bet
* jmkasunich envys cradeks garage
did you see the adventure I had to get my small grinder downstairs?
[23:25:42] <cradek> http://www.united-machine.com/hardinge.jpg
and yep this one has the same control
is that yours, or one like it?
[23:26:19] <jmkasunich> http://jmkasunich.com/cgi-bin/blosxom/shoptask/grinder1-07-19-08.html
you got a surface grinder? that's slick.
small, but hopefully usefull
you can make flat things now
also got a 400W 50,000 rpm spindle
only problem is that the only collet is 6mm
how flat can you typically expects parts to be from a surface grinder?
0.0001 to 0.0005 inches I thing
does the grinder have ball ways?
whatever that is in mm
for X anyway
parallelity I assume is dependant entirely on the mounting
Y and Z are dovetails with acme screws
Y is pretty worn, I'm taking a break from scraping it at the moment
not half bad
I'm starting to think about CNCing it
grinder operation: crank X one way, crank X back, move Y a tiny bit, repeat till Y covers entire workpiece
crank Z down an even tinier bit
repeat entire process
it's not automated?
no, its a little, all manual one
like this: http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=16720603&PMAKA=619-2879
a power feed would be almost as good
well, a power reversing feed
with a ratchet to advance Y at each stroke
all the ones I've seen are mechanically automated. X is astable, when it hits a limit it reverses, one of the limits hit's y a little
given a couple motors, HAL can do that
mosquitoes with a cannon.
I'd actually be more tempted to use a µC, an entire pc seems overkill and unnecessarily complex
one step at a time tho - got to get it scraped, reassembled, and working manually first
lerneaen_hydra: I have PCs laying around gathering dust
its the drives and motors that will take effort/money
even a microprocessor is overkill but 'easy' is more important that 'simple'
right - and hal + parport on a computer that I already have is easy
fair enough, it still seems a lot more bulletproof with a uc, for me that would be easier, though I don't doubt that hal would be far easier for you
in the long run a uC would be nicer - less space
but its not as simple as you think
there needs to be a UI
why do you need a UI? I can't seem to see the need
set length of X travel, set X speed, set Y advance per move, set Y endpoint
oh, you could use a ui for that, true. I thought of hardware stops but software stops are easier
hardware cams that hit a microswitch would work for travel
and a pot would work for speed
and a stepper for y..
and a normal motor for x..
that's about what I was thinking, maybe a dc motor for the X axis and stepper for y
a pot could work for Y stepover amount, but it needs a wide range
well, only the width of the wheel..
some work you step over half the wheel width or more, other work you only step over a few thou
log taper pot ;-)
or an lcd attached to the uc with buttons to adjust the stepover
with the current stepover displayed
stepper as not go have ~
yeah - lots of ways - but the point is, there _is_ a UI
true, you have quite a bit of input data
I think when grinding you dont wnat any hunt for position in Y direction..
design and build a PC board, and do firmware, or use the parport and write HAL and/or a little C
regardless of whether it's via a mouse/kb or mechanical cams
anonimasu: agreed - DC motor for X, stepper for Y
the X drive is by rack and gear, so it only takes a few turns to travel the full distance
actually, DC motor might be too fast, unless its a small gearmotor
hmm... that's no good
have you got a scale for the rack or just a resolver
limit switches :p
its a simple all manual machine
would work for that kind of thing(I think)
oh i see
I've got to sleep
maybe a hydraulic motor would be better?
dmess: hell no
JMK - I ask cause I was thinking many small machines are that way. - You may be able to program/control on/off and speed but most require a DPDT switch to reverse direction with power off. - I was wondering if a pause could be added while tap is at depth to allow operator to flip the reverse switch - then off again to return to the forward position. Maybe call it G33.2
hydraulics = loud, messy, etc
ok maybe NOT then
jmkasunich: that's not true
dmess: this is not that class of machine:
[23:42:41] <jmkasunich> http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=16720603&PMAKA=619-2879
jmkasunich: only for the most part ;)
SkullWorks_PGAB: that sounds like broken taps for sure
on my machine at least, the coast time is many revolutions of the spindle
you'd have to hit stop before the tap even started cutting ;-)
hmm, the table pinion is 2.25" diameter
= 0.588 feet
hmm - mine is a DC spindle and in LO gear it stops real fast (even without a breaking resistor.)
so for 100 FPM, I'd need 169 RPM
SkullWorks_PGAB: mine is single phase induction, 1750 or so RPM. two stage V-belts with step pulleys
one of our grinders has hydraulic feed/advance
ah - more like a standard drill press motor.
works nice - set hard limits and away you go.
yeah, the more I think about it, the more I like cams and limits for X
knob for X speed, and a knob for Y advance
there are stops on the table, can't tell from the pic if they're adjustable or not, and the real table is out in the garage
bah - neither adjustable or removable - dowel pins in the table casting
I could probably make a cam rail that attaches either to the pins or just to the side of the casting
heck, the actual oscillation part doesn't even need a micro if I use a DC motor
couple relays would do it
enough procrastinating - back to scraping