jepler says freqgen does work right - he must have been smoking someting odd before
is he gonna undo his changes?
yes I'm sure he will
we've almost got motion here
very exciting (we've never done servos)
we? you guys are working together on the etch-a-sketch?
his driver board seems sound, and we have feedback
whats he using for drivers? something like a L298?
I have my joystick sitting here from when I was testing the driver... I just had an image of a joystick controlled etch-a-sketch
no emc or g-code, just draw with the stick ;-)
holy crap, it's working
and it sings a little tune for us
that is the nature of the pseudo-PWM
its actually PFM (pulse frequency modulation)
now if we had any idea how to tune, we'd be in business
yeah, that the kicker isn't it
I use halscope
apply either a step or triangle command
I try to set the scales first, so everything is in reasonable units
encoder scale so that the encoder position is in inches
commanded position will of course be in inches too
freqgen scale so that the output is in volts at the motor
it's probably grossly untuned, but it seems to work
(for example, if you have period=50uS (20KHz max freq = 100% duty cycle) and a 12V supply. set the scaling so a command of 12 will result in a frequency of 20KHz
command 3 and you get 5KHz = 5KHz = 25% duty cycle = 3 volts at the motor
oops, echo echo
ok I see what you mean
then the gains become meaninfull
we're a little shaky on our scales because we don't know the etch-a-sketch internals
we'll have to measure and guess
for tuning being within 50% is good
modifying axis to show actual position...
re: meaningfull gains: if scales are right, a Pgain of 200 means that with an error of 0.01 inches you'll be commanding a correction of 0.01*200 = 2 volts
Igain of 100 means that an error of 0.01" for 1 second will result in 0.01 * 1.0 sec * 100 = 1 volt
Dgain of 0.3 means that an error changing at 2 inch/sec will result in 2.0 in/sec * 0.3 = 0.6 volts
and so on...
that sounds useful because if you change power supply, you know what to do
well we don't know why, but it seems to follow the path as-is
you guessed well on the gains
maybe it's not at all critical for our application
(what are you using?)
150 100 0.1
I don't recall what I used for my little motors
* cradek shrugs
he copied them from somewhere :-)
I did a bit of tweaking with halscope, I had the motor go from a standstill, turn 1/2 turn, and stop again in 40mS
you're not asking it to do anything like that I suspect, so you can get away with minimal tuning
how much does it change when you change the load on the motor?
we just want it to have fast but not violent action
we should hook it up the the eas before we tune anything
I aimed for very good (fast) response on the position loop and then used accel and velocity limits for smoothness
the the eas?
oh TO the
the eas load is almost entirely friction, right? (very little inertia)
and it's surely not even friction
so it'll be a bit sloppy at best
wow, it flies
cds at 1000%
you gotta bring that to the fest
oh we will!
it's jepler's only machine, he'll bring it
that's interesting - with deadband of .01, it lets you have -.01 to +.01
seems off by a factor of two
guess it depends on how you define deadband
its documented that if (fabs(error) < deadband ) error = 0
dunno if that is the usual convention for deadband or not
my way of looking at it is: "If you are within +/-<deadband> of the desired position, do nothing"
ok I assumed the other thing: there's a band |<---->| this wide, within which is "good enough"
yeah, I've seen both conventions used - I like mine better ;-)
more like a tolerance spec
hmm, it ferrors pretty easily when hooked to the actual eas...
probably the friction slows it down
you could either:
1) turn down vel or acc limits until it doesn't ferror
2) fire up the scope, see if the output is saturating
if it's saturating (putting out 100% duty cycle) then you have to lower the vel and/or acc
if not, you can probably raise gains
ok, that makes sense
working on set screws now
I think if you raise P and I by the same amount (multiply both by 1.5, or 0.3, or 10, or...) then you'll maintain the same time constant
changing only one changes the time constant
so meanwhile about threading: what does the encoder information look like? does the index pulse reset the count or something?
depends on the driver
but I'm aiming for some conventions for all HAL encoder drivers
there will be a HAL pin that enables the index reset
if that pins is true, every index resets the counter, and the position is a sawtooth
I'm not sure how else you would get absolute positioning
when you are ready for the pass to start, set the index-reset-enable pin false, and the position becomes a ramp
are you or jepler gonna post anything about the servo etch-a-sketch to a website?
I'm sure he will
ok we were saturating on Y, so we turned it down to 3ips
seems to run pretty well other than that
x at 4ips
3 ips isn't exactly slow
the screen is what, 6" wide? so 2 secs end to end
it runs pretty well actually
how does it compare to the stepper version?
the action is much smoother than the 7 degree steppers
yeah, I guess it would be better ;-)
yes there was a lot of stairstep that showed on diagonals
how many counts are the servo encoders?
so not a whole lot higher resolution than the stepper
with a 6-1 gearhead (emc counts 2400 counts/shaft rev)
that was 48 steps/rev
oh - gearbox makes a big difference
so its just a little bit finer resolution ;-)
err they were 7.5 degree halfstepped
so 96 halfsteps per revolution
and now we have 2400, cool
still, 2400 >> 96
no wonder it's smoother
where'd he get the motors?
[02:55:11] <jepler> http://emergent.unpy.net/index.cgi/projects/01142347802
link to servos on that page
interesting - you are using separate direction and PWM bits?
wow, those look like nice motors
with the loop closed, the shafts are hard to turn
and it whines at you when you try?
(we're using 12v)
[02:59:55] <jepler> http://emergent.unpy.net/index.cgi-files/projects/01142347802/03_093_e.pdf
[02:59:57] <jepler> http://emergent.unpy.net/index.cgi-files/projects/01142347802/03_200_e.pdf
[03:00:00] <jepler> http://emergent.unpy.net/index.cgi-files/projects/01142347802/03_215_e.pdf
datasheets for the motor, gearhead, and encoder
nice - I'm not used to having data for my surplus junk
* jmkasunich scratches his head and pulls out the calculator
dude, use units
units "nM n" "oz in force"
so 31.9 mNm = ~4.5oz-in
* 6 for the gearhead
* 6 in the gearbox = ~ 27 oz-in
not real strong?
perfect for this application
wow, very low current
max continuous about 1/4A, locked rotor current 1/2A
we're going to try 24v...
was just looking at the schematic
what caps do you have on the +12 (+24) supply?
it's a switching PC supply
I mean the ones right on the board with the L298
0.1uF, or electrolytics, or one of each?
looks like both
small electrolytics and some ceramic discs
I think the schematic is on that web page
I dunno what the voltage rating of the L298 is, but switching an inductive load makes spikes
it is, I looked at it
it shows two caps, but no values
there are those "fast" snubber diodes
the 5v regulato doesn't like 24v on its input side
so we're back to 12
which do a fine job of dumping the transients into the supply rail - you're fine as long as the supply rail has enough C to absorb them
got a bit warm did it?
yeah we didn't run it long...
well this is surprisingly successful
I'm not surprised
once he got hal right, it was smooth sailing
I wish I lived close enough to other emc folks to just drive over and spend an evening messing with stuff like that
I wish I lived close to cleveland so something would be interesting around me
what is there in Lincoln?
as far as techie stuff, not too much - we have the surplus center
which is kind of cool I guess
we have different kind of surplus
yeah a lot of that in your area (I explored there recently)
yep, I sure went there
you were in cleveland?
yes, last april I think it was
dang, should have said hi while you were here
I don't think I knew you then...
otherwise I sure would have
have you been to the early television museum?
I'm sure I've asked you that before
I recall you talking about it a few weeks ago
its near here?
yeah I try to tell everyone it's cool
columbus is probalby 3 hours
dammit I can't spell probably
3 hours is a short drive for me
probalby can't spell?
takes 8 to get anywhere from here
actually it will only be 6 to fest - I'm thrilled about that
about 10 for me
that's getting too long for driving all at once...
or is it 8?
I'm getting mixed up with other trips
my wife's parents live on the eastern shore of maryland, thats about 9-10 depending on traffic
I think the CNC fest is 530 miles
8-9 depending on traffix
traffic that is (and how heavy my foot is)
most of my drive is through IA which has a 65 speed limit - makes it even longer
part of mine is through the south side of chicago - one massive road construction project
but I'm not complaining - it's rare to have any gathering of technical folks that I can drive to
oh yeah, I80 around chicago is hell
both times I went thru there I got off course, one time by about10 miles
I try to go through in the middle of the night
I don't like to drive long trips at night
afraid I'll fall asleep (or just start bleeming)
I guess I never bleem - don't know what it is
when you kinda get a blank look on your face
I get goofy and start singing and crap
sleeping with your eyes open, or very bored
it's pretty cool, the etch FEs and stops when it hits the edge
not used to that
now I know why jepler added the "count" pin
that approach uses 3 parport pins per motor though, using the up and down would only use 2
wish we could get 3 axes on a port
you can if you drive the H-bridge differently
forgot about that
the 8 bit port can be in or out
there are 5 dedicated ins and 4 dedicated outs
yeah, either way it doesn't fit
well its only 10:45, but I'm tired....
will you be around this weekend?
I have a machining job to do by Friday
ok, see you then
I want us to work on threading sometime... I need your help with the hal part
we can do that - not sure exactly when tho
there's no hurry but I'd like to have something before fest
I'm kind of afraid I won't be able to work on it without some hardware, but maybe I can fake something up in hal (with your help)
we can definitely fake something up
I wish someone could loan me the cnc parts for my sherline lathe, but I'm sure nobody has that sitting around
jepler's little motors don't have an index pulse do they?
their controller is really expensive iirc
no, just quadrature
you already have a sherline lathe?
so all you need is steppers and drivers?
yes and some kind of encoder for the spindle I guess
well the motor mounting stuff too
I have an assorment of surplus encoders, I could probably come up with something
the motors and mounts are the tough part
Xylotex has a three axis stepper driver that works nice for NEMA23 motors, about $130 IIRC
(I bought mine 2+ years ago)
I could even make the drivers for the cost of parts
man sherline's website is bad
you have a CNC mill already right>?
(trying to find a price for the motors and mountings)
yes a small one
how slow can the spindle turn?
you have a pic or your mill somewhere on your website don't you?
pretty darn slow, it's a closed loop control
[03:58:32] <cradek> http://timeguy.com/cradek-files/cnc/spindle-mount.jpg
think you could mount an encoder on the spindle?
thats the head of the drawbar on top isn't it?
[03:59:27] <cradek> http://timeguy.com/cradek-files/cnc/milling-setup.jpg
see the threads where the nut used to hold the tool in?
a mill is just a lathe standing on its ear
at the bottom above the new style of toolholder
that's sure true
the sherline 3-jaw actually threads onto those threads
you have a good idea here
my dad was a machinist for 40 years, but he never had a lathe at home (something he regrets now) - but he has turned small parts in a drill press several times
I am not familiar with encoders - how much height do they take?
how big is a box?
well assuming you know your junkbox
depends on the encoder, some are 4" dia x 2" long, some are 1" dia x 1/2" long
so they aren't like a pancake
I could spare that height down at the bottom, but then I couldn't thread on that chuck
would be ok for experimenting, but not too useful after it works
since you need to keep the drawbar clear, the best bet would be a toothed belt
I'm looking at the pic of the motor and mount
that belt isn't toothed is it?
just a V
looks like a couple steps for different ratios?
I have a different (huge) back pulley now for high speed
the slow setting is about 1:1
threading would probably be done at a few hundred RPM
fast is about 3:1
I'm not sure it has any power at that speed, but it might be enough
the idea is that you reduce speed with the belt, not with the drive
can you change the ratio from 3:1 to 1:3?
not without entirely new pulleys
the 1:1 might be ok
duh, I was thinking swap spindle and motor pulleys, but they mount differently
for testing, you'd be cutting air first
yeah different sizes
then probably something soft
wax, plastic, something like that
that drawbar looks like an ordinary bolt head...
it is, with the back machined flat and concentric
actually the whole bolt is cleaned up so it's concentric
what size bolt?
bigger than 1/4" ?
* jmkasunich has no concept of scale from the pic
the bolts holding the motor on are 3/8 I think I think
waitaminnit - if you're using the lathe chuck, you don't even need the drawbar do you?
nope, I could take it out
I found a couple encoders here that are about 1.5" dia, 1.5" long, with a 1/4" stub shaft about 3/8 long, and ball bearings
I think that front pulley is 2" dia
if you could make a dummy drawbar that locks in somehow (threads into an empty toolholder?) and has a 1/4" hole at the top end
you could mount the encoder by its shaft (its very light)
use a little arm to keep it from spinning
there is not clearance above the spindle
the motor is actually somewhat over the drawbar - have to remove the motor to get the drawbar out
oh, the picture is deceptive, it looks like the bar could come straight out
it could mount to the side with another belt, but the "no teeth" problem is still there.
how is the pulley attached to the spindle?
it's a tight friction fit plus one big setscrew
looks like the spindle OD below the pulley is about an inch or so
yes something like that
I'm about to be kicked out of jeff's house
$12 for an aluminum 60 tooth timing belt pulley, 1.73 OD, .531 overall thickness
it has a 1/4 bore, but that could be opened way up
and maybe mounted on the bottom of the existing pulley
that distance from the bottom of the existing pulley to the casting looks like 1/4" or so
the actual pulley part is .276 thick
the spindle gets a bit wider there, but you're right I could mount to the other pulley
and you could probably machine off the flanges to make it thinner (the matching belt is only 1/8 wide)
it could go on the bottom
I'm looking at mcmaster-carr P/N 1375K53, if you want to look at it when you get home
if I could put those big threads in it, it would just thread on
these encoders (I have two) have five leads - I assume red and black are power, I'm hoping the other three are A, B, and index
I'll try to make time to test them, if they work I'll send you one
if we can rig this up, I'll bring it to fest for us to work on
I better go
we can also investigate rigid tapping...
I'll be around tomorrow
ok, time for sleep here too
if we're tapping wax or something :-)
well there's no reverse on mine
the sherline drive can't reverse?
thanks for all your help
well not as-is
simple matter of hacking I guess
nuttin's ever fsckin' simple
ok goodnight for real now, we'll work on it tomorrow
SWP_Away is now known as SWPadnos
SWPadnos is now known as SWP_Away
SWP_Away is now known as SWPadnos
are you home now?
trying to get back in shape
darn, anon CVS still not working.. I wanted to make a new testing package :(
well, drink lots of coffee, eat donuts, and read email. that should help ;)
guess I'll make one with devel checkout
I was just looking into making a cvs repository copy on dreamhost, but there's no anonymous access (as jepler had pointed out)
jepler: you there?
jepler: n/m, I'll bbl and we'll talk then ;)
alex_joni is now known as alex_joni_away
* jmkasunich thaws his fingers
cold when you're running a mill tho
cast iron doesn't heat up very fast
Wow. Colder than the ~40 here.
that's gotta be rare!
You guys are normally warmer that we are.
right, sometimes by a lot
I was expecting you to come back and say "28 is balmy, it's -15 here" ;-)
I've got a hal file sorter looking for AXIS_
couldn't it just look for '['?
With a stepper setup I'm seeing stepgen max accel
but not the parallel max accel
course they go to different places.
that one isn't used by hal, only by emc
Right it just seems that not being able to see the emc one means a shot in the dark for the one we do have control over.
you can change hal params on the fly, but can you change emc params on the fly?
Back in the day when we had acces to globals we could.
at least some of em.
what was the method? sending nml msgs?
There are still global PID they just don't do anything.
right - the PIDs should go away as part of a cleanup one of these days
but things like [TRAJ]MAX_ACCEL and [AXIS]MAX_ACCEL?
I'm not sure that we could ever change accel or max vel on the fly.
it would be interesting to have a table of all the ini file items, with info about: where used, change on the fly vs change at startup only, and probably a few other bits of info
Quick fix is to call this widget EMC-HAL axis Calibration
or just hal axis calibration
it would sure be nice if it could tweak all the modifiable items, not just the ones that go thru hal
Yes it would but I don't know any more how to handle the emc ones.
for the most part they didn't change from emc1 I think
some of course no longer work, like PID
I do have the old tweak code but didn't think much of it worked.
but the rest of iniaxis.cc and friends still work the same as they always did
all of the ones I used went through emcsh. I'll look there for a bit and see what I find.
well, fingers are getting warm, time to go outside again
jmk just went back out to the shop to make swarf
yeah he said he had some kind of task to finish
last might we were talking about fitting a spindle encoder to my mill for work on threading (aka rigid tapping)
seems like the mechanical problems are always the hardest part
What kind of mill is this?
(like no good place to mount it)
a mishmash tabletop thing
[20:00:45] <cradek> http://timeguy.com/cradek/cnc
the pictures are a bit out of date
[20:02:32] <staggerlytom> http://www.maxnc.com/
staggerlytom: yes, but it has little in common with their setup anymore
dunno, tiny pictures
You have to keep the top section of the spindle drive to unload tools?
yes, it has a drawbar
seems like it will have to be off to the side somehow
Okay. You could try the joe martin sherline approach and paint stripes on the wheel below the pulley.
vertical stripes and read them with two photo transistors.
I'd rather it be able to work with the lights on...
also don't I need better resolution than that for threading?
How did you get that driven pulley in the stack
the front pulley is made in one piece
joe had 4 quadrant pie shaped and got a pretty good indication of speed.
You made it?
yes, I made the pulleys (new ones since these photos)
I made a large one for the motor to get the speed up for milling pcbs
couldn't you make two of them there and drive an encoder off the side.
the dead space on the bottom of the pulley is for a big set screw, but yes there could be another groove there
jmk thought they should be toothed pulleys though
Well between the spindle and encoder yes.
ah and you just used a v pulley.
I wonder exactly how many pulses per rev are needed for threading
otherwise I couldn't make the pulley
the stripes idea is interesting
A few guys are using the one pulse per rev
and I recall you being adamant that you can't make good threads that way ;-)
That seems like horsesxxt to me
Yes I was.
yeah sounds useless
I liked the 360 ppr we used at mazak
I would think at least 100? ppr
if the spindle speed is absolutely constant (even as load changes) it could work
jmkasunich: it never is, at low speed
Right. It would have to be a large mass spindle.
a single index per rev would be fine for single-pass threads (which are rare, AFAIK)
10HP induction motor at 1800 RPM, geared down (with back gears) to 100 RPM would do it
Yes it would but would we want to advertize that ability with emc?
my worry is this: this project will be hard enough - I don't want to fight with inadequate hardware
SWP: I disagree - the problem isn't multi-pass indexing (single pulse per rev takes care of that) its drunken threads
cradek: right, the single ppr thing is a distraction
Right. I remember arguing that the spindle drive needed to be a full servo.
wish I had real data about the sherline closed-loop speed control
cradek: you don't have a dividing head or rotary table do you?
the spindle drive shouldn't need to be servo
it may have something suitable built-in
jmkasunich: yes, I have a cnc rotary axis, but no driver for it currently
but its vee-belted to the spindle
yes, would need to change that, but that's easy
if you have rotary, you could machine teeth into the OD of your existing pulley
either for a belt, or to be sensed directly
SWPadnos, That is exactly what a lot of folk said. They got pretty quiet after the guy from Giddings and Lewis said yes it would have to be.
heh - I may have to be quiet, then ;)
ray, swp: let's please not go there now ;-)
Nah. It really depends upon how tightly we can electronically couple the axis to the spindle.
jmkasunich, it's necessary to discuss whether emc threading will require a servo'ed spindle
that's the whole point of this, afaic
if it will, then it's a lot different in software
if it does, that rules out the entire hobby class of machine
SWP: threading as I envision it does not (but does require a decent ppr encoder)
I'm in favor of starting with a wild or nearly so spindle,
in that case, I'm not interested in working on it
and see that we can get the cutting axis to match it.
I have an induction motor on my spindle, so I'm in the same boat
I still don't think this is going to be too difficult, aside from hardware issues.
I'd think that the encoder has to be able to provide a reasonably accurate spindle speed indication every servo cycle
start with at decent resolution sensor, get that working, then see if lower resolution sensors can be made to work
Yep. I'm probably the only one with real servo spindles laying around.
SWP: not speed, position
actually, not at the servo rate, but at the expected spindle response rate (to applied / removed loads)
jmkasunich: I could machine those teeth on my manual mill if they're not an odd profile
cradek, There are belts with half round teeth.
your pulley isn't steel is it?
steel can be sensed with magnetic sensors
no such luck
and the tooth geometry isn't finicky
could go buy a $100 crankshaft position sensor
belt tooth geometry is simpler
... if I had $100
or a $29 encoder ;)
I might even have some magnetic sensors here too
(in a box somewhere)
seems like that won't help though
insert ferrous dots into alum pulley and sense speed and rpms( not accurate for posn tho)
so say I can rig up an encoder - what counts it?
can I hook it right up to the parport?
if the count freq is less than about 20KHz, the HAL encoder counter block could do it
higher than that and you'd need a board with an encoder input
I think I have several parport bits left
thats one thing that concerns me about the encoders I have here
I think they are 1000 ppr
I think I can make a fairly inexpensive encoder counter that operates serially or something
I dunno if that is 1000 counts, or 1000 cycles (which is 4000 counts)
so that means I have to be under 20 rps
either way, software encoder counting tops out pretty quick
but something like 360 counts per rev would be nicer
if I have magnetic sensors, could you machine a steel ring that presses onto the pulley, then machine teeth in it?
say 1/4" wide, 1/4" thick, machine teeth 1/16" deep and wide
what radius variation is necessary to make it work?
don't we need an index pulse still?
if the OD is 3", circumfirence is about 9" and you have 9x8 = 72 slots
it's probably best to have an index
yes, that can be as simple as a screw head that a third sensor sees once per rev
72 slots = 288 counts per rev
fyi the driven pulley is 1.5 OD
so the OD of the ring wouldn't be more than about 2"
until I find the sensors and see how big of a tooth they need its moot I guess
if they need a tooth that is 1/4" wide and deep, forget it
about how much would it be for me to just buy an appropriate encoder?
the problem is still mounting it
there are a number of encoders on ebay as well
the thing is so small there's not much extra room anywhere
most of the usdigital's expect a 1/4" stub shaft (or at least, some small diameter shaft)
or they have a shaft
I have some of those here - 1/4" shaft and ball-bearings
what's the hub diameter on the spindle pulley(s)?
but you're right back to how to mount and drive it
the spindle pulley fits over the entire spindle, probably 3/4" or more
use the spindle like a lathe, and bore a hole into the hub or axle, then put a pin in it
and the drawbar sticks out the back of the spindle
hmmm - drawbar - damn
ID of spindle pulley is .625
the other option is to mount the encoder on the motor, and hope for the best
I actually thought about replacing the drawbar with a stubshaft
but there is also a clearance issue above the spindle
encoder on motor will never work
unless you change the main drive to a toothbelt
that would be best
but under light loads, it may be good enough
even under light loads, you don't get exactly 1:1 with regular belts
there's always creep
of course, you'd need different configs depending on which ratio you're on
it might be 1.001 to 1
this gets a little esoteric, but...
speed sensor on motor, index sensor (1ppr) on spindle
was ray suggesting a separate encoder shaft, belt driven from the spindle?
thats what we did for the mazak
the belt can be tiny cause there's no load to speak of
I think the mazak used 1/4" wide belt, but 1/8" would be plenty
that still seems like the best general idea
... if I can manage to mill the existing front pulley
or take a commercial toothbelt pully, bore most of it out leaving only a ring with teeth, and press it onto the main pully
the front pulley looks like a flat cylinder to me (with an inside-out toothed belt on it)
(or fasten it some other way)
there would be a tooth over the setscrew, but that wouldn't matter
SWPadnos: it is a cylinder, but the belt is meant to be that way
SWPadnos: the belt is a V on the back
I noticed the V-groove pulley on the motor
I actually made a toothbelt pulley once, but it was for a bigger belt (0.200" pitch IIRC)
I still have the cutter with the tooth profile
the trouble with belt-driving an encoder is the side load
but I think you want something smaller
SWP: the ones I have have ball bearings, and toothbelts don't need much tension at all
bearing above and below belt pulley on encoder... no side load
I was thinking of what vibration would do to a not-very-tight belt
SWPadnos: remember this will be very low speed
this isn't a very demanding belt application
large diameter and short span
SWPadnos: I often run 20krpm but would disconnect the encoder (right?)
that's why a belt is going to be best
it must be 1:1, so both pulleys will be about 1.5" OD, center distance might only be 2"
Those automationdirect encoders will handle some side load.
at least some of them.
or a 2 pulley spindle with a hollow shaft encoder.
jmkasunich: why 1:1?
it doesn't need to be 1:1 (except for index)
even that can be done with a trivial HAL module
do these encoders have the index pulse?
notice how easily you guys say trivial when thinking of hal
most are available with index
mine? I think so, but haven't tested them
automation direct, I dunno
but you can't add it later ;)
rayh, the module I'm thinking of would be a very simple module. using it or explaining its use may not be ;)
swp: how exactlly would you deal with indexing?
have a component that has a "ratio" parameter, and only set the index output true every "ratio" index pulses
my parport pins in use are 2-9,1,14 which I think means I have >= 3 inputs free
ray: one of the fundamental concepts of HAL is to divide the problem until the inidividual tasks are trivial, then write trivial code to do the individual tasks that make up a non-trivial whole
having another output for eventual spindle reverse would be good
then put it together with complex config files ;)
and then check to see if the derrived index pulse matches the real every ratio times around.
SWP: software gearing ( electronic earing ) yes
the derived pulse would be output as a divided input pulse, so they would always match up
SWP: I think I see - so you'd still require an N:1 ratio?
not if the param is a float
hmmm - yes ;)
yes N:1 so get a fine encoder to get 'more float like'
No I'm thinking of really knowing where in the rotation of the real spindle we are.
a "rational" ratio
yes, the drive ratio between spindle and encoder
any gear ratio is rational
all it does is change the "scale" - counts per rev
as long as you don't power down or reset between passes you'd be OK
jmkasunich: nah, if my pulleys are 1" and pi", the ratio isn't rational
if they got teeth they are rational
* cradek waves his hands
can't machine pi teeth
sorry I forgot we were talking about teeth
teeth are for eating pi
the index ratio does need to be an integer, or you aren't basing the output on the input any more
not float, big int ( think Forth), the encoder has pulses, is digital, not analog ( else you got leftover fractions )
mmmmm - pie
now I want some pie
I could go for some warm chocolate pecan pie
pie are snot square, pie are round, cornbread are square
with a scoop of 'niller ice cream
that would do it
now I'm hungry
heh. maybe I'll just eaqt that eclair, and see of my wife notices it's gone
and I gotta go back out in the garage, and I forgot what I came down here to fetch
you'll remember when you get back out there
I remembered - part drawingh
* jmkasunich takes his box of t-nuts, studs, washers, nuts, clamps, and misc crap outside
heh - gotta know where to cut ;)
well crap, I still don't know what to do
Got room on the spindle top to add a plastic pulley?
looks like there is .125 of height available
I've not seen timing belts smaller than .25
and maybe 1.7 dia before something hits something else
The diameter isn't critical at all. Just have to match to the encoder pulley.
I think sdp-si has 1/8 width belts
I don't think I can add .25 of height anywhere
yep - they do
there you go.
mcmaster only has 1/4 and up.
is the picture on http://timeguy.com/cradek/cnc
at all valid?
staggerlytom: pretty close, front pulley is still the same, back pulley is much bigger now
it looks like you can crak the motor plate up, openin all the vert room you want
on those wing nut stand offs
yes but I can't raise the front pulley
the spindle is only so long and the drawbar tightens against it
How about the alum below the belt?
yes, is the belt to pulley still very small?
rayh: I don't understand what you're asking
there's another V-groove down below, right?
on the motor pulley
both pulleys have two grooves
the bottom gives a 1:1 ratio, the top set about 3:1
I don't see a groove on the spindle pulley
well, one, but it's above the belt in the photo
there are two grooves, in that pic the belt is in the bottom one
the top one is small
in that space under the bottom groove, I could possibly mill teeth
ok. maybe it's just behind the belt, and not deep enough for the sides to show with the belt in the groove
right, the belt fills the groove
cradek: tried gimp to edit your foto.. ng
it looks like theres room below where the belt is now, to ad another belt for encoder
cradek: an alternative to milling teeth down there is to machine away some alum so you have a step
then bore a stock pully until its a slip or press fit over the step
the problem with cutting that away is the set screw that you don't see in the photo
never mind ;-)
I was thinking if I could mill the teeth, they would go right over that hole
trick is the tooth profile
I have a cutter I modified to cut tooth-belt teeth
but I think its for either 0.2 or 0.25" pitch belts
and I doubt it would fit your mill
its a horizontal mill cutter
I can only do .25 dia tools
1" bore, 3" OD, gotta run it at about 150RPM or so
about 24 teeth
you'd be coming at it 90 degrees from that
(cutter axis pointing at the centerline of the pulley)
I know sherline sells cnc lathes - do they have spindle encoders at all?
don't think so
the belts with curved teeth have a minimum width of 12mm
curved teeth could be cut with a ball end mill I think
I could do that but .5" is way too thick
the trapezoidal teeth come as narrow as 1/8"
pitch is 0.080
tiny little thangs
another problem - depending on the belt's DP my pulley's diameter may be incompatible
that can be fixed, you have a lathe don't you?
maximum diameter change is belt pitch / pi I think
.080 pitch means teeth are 0.040 wide
can't cut the whole cylinder, because of the groove, but I could make a step if there isn't a lot of change required
the step would even act as a flange on one side
if the teeth are only 0.040 wide, that takes a mighty tiny endmill
I'd be very temted to try to do it with a 0.040 thick slitting saw
yeah I have some 03125 but I'd hate to try to cut Al with them
uh, I'd just break them
oh, you mean 0.03125 end mills
I thought you meant 0.03125 slitting saws
are you done making your parts already?
no, thawing fingers again
ends are milled, sides are milled
need to make a cleanup cut on the faces, then I'm done outside
but still several more hours of inside work
boring, and marking a scale on the (dang, wish I had CNC for that!!!)
sounds like you need some heat out there
and then making plungers to fit in the bores
you don't have cnc?
whats that they say about the shoemakers kids?
well come on over, I have cnc and 60 degree pointy tools
bring your encoders :-)
if you were less than an hour away I'd take you up on that
as it is, I can do it manually in an hour or two
(I don't mill them, I scribe the lines and stamp the numbers)
I'd mill the numbers made with truetype-tracer - it would be fun
I know it would be fun
maybe I should mail them to you, and you could mail them back ;-)
except I have to ship by thursday evening
yeah probably no good then...
mcmaster has 0.032 wide slitting saws, 1" OD, 3/8" hole, for $4.60
seems like you could make a trapezoid with three very careful passes
1/16" thick 1-1/4dia 1/2 hole is $7.77
it would probably deflect on the later passes
you'd want to cut the "middle" last
maybe the two outside passes would be enough
don't know how critical the shape is
depending on how bad I wanted to do this, I'd get the 1/16" one, set up my dremel on the lathe, put the cutter in the chuck, and with the cutter spinnind I'd modify the 0.062 wide cutter to cut a 0.040 wide trapezoid
starting with a pre-made $10 pulley seems more tempting
yes it does
(the dremel trick is how I made the cutter when I did it before - started with a 3" diameter cutter about 3/16 or 1/4 wide
took a long time to grind to shape
sounds like not much fun
but I was motivated - I needed a 6" plus pulley for my mill's power feed, and a purchased one was either $100+ or unobtainium, I don't remember which
the original pulley was for silent chain, but worn beyond usability - so I turned off the teeth and recut it for toothbelt
bought the motor end pulley, that was only about 1.5" dia and $10-15
it's fun to fix stuff like that
I can't help but think we're missing some obvious easy way on my mill problem
wish I had an appropriate junkbox to dig through - usually that's what I do, but I'm woefully inequipped
how about taking a stock pully and using the drawbar to hold it to the top of the spindle?
remove when not threading
there's virtually no height available above the spindle
where's that pic again?
I can't raise the motor any
I need to take a new one
just a minute
another way to make a tooth profile cutter:
turn the tooth profile on a piece of 1/4" drill rod
so it would be about 0.15 on either side of the tooth, 0.250 center of tooth
what's inside the housing?
the cut flutes and harden
presumably nothing that spins ;)
then cut flutes and harden
like this: http://www.metalworking.com/DropBox/_2001_retired_files/Worm06.jpg
the cutter on the left has made 6-8 40 tooth wormgears from bronze and is still sharp
that's 5/8 OD, much bigger and heavier than needed here
check out the rest of the pics in that series - I'm kinda proud of that project
pic 5 is another pic of the tooling
you'd only need the middle one
pic 8 is it in action
[21:55:22] <cradek> http://timeguy.com/cradek-files/emc/spindle.jpg
I'm just looking at #8
I can't see if the gear is tilted (relative to the cutter plane)
this is what it looks like today
SWP, it is
pic 7 shows the tilt better
cradek: and the motor plate comes out how far?
Ah, OK. you let the gear spin with the worm cutter during a later pass
to the spindle centerline? beyond?
first pass is gashing, second pass is hobbing
slightly over the bolt head at this pulley setting
whats that under the bolt head? a washer?
the top of the spindle
the fillet you see is the spindle
for threading you'd be using the lower drive belt groove...
that's ~ 1:1 with the motor
so the "easiest" thing to do would be to mill a toothed pulley profile in the bottom
so, bore a stock pulley to slip over the top step of the pully and rest on the top of the bottom pulley
that would interfere with the belt in the top position, wouldn't it?
put a washer under the head of the bolt to hold the pulley down
you can't see it but the two pulleys overlap somewhat
if you're doing high speed work, take it off
oops, so much for that idea
so the stock pulley would have to be < 1.5 dia
can you remove the spindle pulley easily?
no, it could be fine if it was small enough
SWPadnos: not very, it's a tight fit
(not on a regular basis, for machining)
ah, in this pic I can see the setscrew
oh yes, sure I could take it off
is the lower part of the spindle pulley needed for anything else?
or could you turn that down a lot?
how much space between the bottom of the pulley and the casting?
it's just there for the setscrew.
from the pulley to the top of that clip that is part of the spindle
I think the most attractive idea is machining teeth down there using a homemade cutter
can you just put a spur gear on the outside of the toothed belt that's already there?
I suppose that could slip though
the belt could slip wrt the pulley
oh, that reminds me of an off the wall idea I had last night as I was falling assleed
you can see the pulley could be mounted a tad higher
assleed - good one ;)
get a verys small but wide tooth belt
make a gear out of it
turn it inside out, make the pulley a tight slip fit
make a pulley out of it actually
yeah - pulley
but it would still obstruct the setscrew hole
also, I think turning the belt inside out would fsck up the pitch
the setscrew isn't accessed often, is it?
the profile would have to be trapezoidal, and symmetric
more of a gear profile than a tooth belt
no, the setscrew is not used often
but I think turning a belt inside-out would badly screw up the DP
the standard toothbelt profile is a trap, and is symmetric (or pretty near)
I think I was going to say that, but in many more words :)
although belts more commenly use regular pitch, not DP
same issue tho
trying to find the correct one would be a bear, since none of the specs would be what you're looking for
number of teeth tells all
well that and pitch
thickness of tension member or backing would also be needed
[22:07:59] <cradek> http://timeguy.com/cradek-files/emc/spindle-side.jpg
I'd order a belt that is close then machine the pulley bit by bit to fit
ok, heres another idea
the side pic shows some clearance above the bolt
bore a 1/8" or 3/16" hole in the bolthead, drill one flat of the head for a setscrew
run a 1/8" shaft up to the top of the motor, and mount the encoder up there
shaft is flexible enough to tolerate minor misalignment
a flexy shaft
remove it before you do 20,000 rpm milling
if I'm using the 3-jaw, the drawbar would have to be removed though
is the top of the motor amenable to mounting anything
with zip ties, sure
(so not really)
could you put an empty toolholder in the spindle to give the drawbar something to thread into?
umm I would have to cut up a toolholder, but yes I could
just make a threaded shaft for the encoder - a second drawbar, as it were
if the drawbar isn't gonna be there anyway
you still need the tool "blank" to thread into
take a stock gear, bore the lower side of it to fit over the top of the spindle
countersink the top
and use a long flathead screw instead of the drawbar
the gear thickness could be the bolthead thickness + the spinde "nub" height plus a little, and still clear the motor mount plate
make a dummy toolholder, the "drawbar" thread doesn't have to match the original drawbar, and the "toolholder" doesn't have to be steel or even particularly accurate
depends on what you can mount on top of the motor
I take it theres a setscrew in the bottom part of the motor pulley too?
not much unless I just use foam tape or something
yes same thing
depending on how deep in the hole the setscrew is, I'd be tempted to make another groove in the bottom of the spindle pulley, and turn a smaller step (as small as the setscrew allows) in the motor pulley
gives you a slower speed (more torque)
how long is the setscrew (and are shorter ones available)?
can't do that without making a bigger one on the spindle
or getting a different belt
the motor can't move forward or back more than a tiny bit
you'd have to move the motor back
right, I wasn't thinking
either move the motor (not practical) or get a shorter belt
it can't go back more than about .2 before it hits the stepper
side issue anyway
bridgeports are so much ... roomier on top ;)
the top of the motor looks pretty flat
it is quite flat except for in the center
piece of 1/8" alum sheet, drill a hole to clear the bump in the middle
and some foam tape
a hole for the encoder
let it hang over on the sides, run 6/32 threaded rods down to the motor mount plate with nuts to hold it on
what hookup is required on your encoders?
mechanical you mean?
1/4 shaft, about 1/2" long
I'm imagining something that is like a screen-door spring (?) with 1/4" ID
it would just press over the encoder shaft
for the shaft itself?
thats too flexible
could put a solid core in it maybe
1/8 drill rod I tell you!
yeah, good idea
setscrew in the drawbar head
simple lathe-made coupler on the encoder
with two setscrews
only annoyance is you have to make a coupler to connect to the encode
do the us digital encoders (the ones without bearings) come in 1/8" shaft size?
mount a single 1/8" ID ball bearing on that alum plate at the top, mount the encoder electronics module to the plate, add the codewheel to the shaft, done
would the whole shaft pull out the top then to remove it?
the whole works has to come off easily
[22:27:55] <jmkasunich> http://www.usdigital.com/products/e2/
[22:28:37] <jmkasunich> http://www.usdigital.com/products/e2/e2assem.shtml
10 mins maybe?
remove cover, electronics module, then either remove disk, or pull shaft and disk together up and out
I'd be worried about handling that disk all the time
maybe I could come up with a shaft that would come out more easily
disk is mylar
I could just remove the whole assembly from the top of the spindle motor
use a wingnut somehow
maybe use two plates
one attached permanently to the motor
the other, smaller, sits on top and is held on with a couple screws
this sounds promising
I can make all the parts here
drill two holes, tap one in the drawbar
two pieces of flat alum sheet
and some small drillrod
oh, I see a cheaper encoder
do I still need a bearing?
you need a bearing at the top
you could buy an encoder with a bearing in it
then just make a shaft coupler to extend the shaft
I would be more comfortable with that
then to store it, I could unhook the shaft without taking the encoder apart
[22:35:07] <jmkasunich> http://www.usdigital.com/products/s1s2/
the ball bearing version is good to 10,000 rpm
cool, ttl outputs
if you want the cheapest choice, either use my encoders (assuming they work) and make a 1/4" to 1/8" coupler
or use this encoder and use a bearing:
[22:37:28] <jmkasunich> http://www.usdigital.com/products/e4p/e4passem.shtml
encoder is $19
I probably have a 1/8" ID bearing around here you can have
this one is very interesting, it looks like the electronics only go on the one side of the disk
so you could remove the disk and shaft together
just loosen the setscrew in the drawbar head and lift the whole thing off
hmm this S1/S2 is 1/4" shaft too
looks like the $19 encoder has no index pulse
wow, $10 each in high volumn
I should get back to work
thanks for your help
I think we have this nailed
make a 1/4 to 1/8 spacer, if my encoder works you're home free, otherwise buy the shafted encoder
these are 1.5" OD, with a 3/4" section that has the bearings in it
there are four threaded holes on the face
I could even make those parts right away
so a 3/4" hole in the alum plate and two or four clearance holes for screws is all you need to mount it
that sounds great as long as the resolution isn't too high for software counting
they have a $40 shafted encoder
I shouldn't worry about $10-20 difference, I should get the surest bet
you'll have to look closer at pricing anyway, the ball bearings might be an adder
when you have your work done, let's put together an order
the sleeve bearing versions are only good to 100RPM
yeah I need the ball bearing then
found the pricing
ball bearings are a $6 option
ok no problem
S4 series is the cheapest shafted ones
$46.95 for 1 with ball bearings
too bad there's no 1/8 shaft
looks like there ius
S4-ppr-B8 is probably the PN you want
B = ball bearings
8 = 1/8 shaft
I'll try to figure out how to order after dinner
need the cable too, <$10
the S4 series doesn't have an index pulse
well, you can read the website while I make chips
if price is a major driver the kit encoders win by a lot
I'll be back in an hour or so
if convenience matters, then the shafted ones win
ok, I'll figure it out
cheap - fast - good, pick two
dfgdfgdfgdfgdfg internet is up/down like crazy today