[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
Valen: I think scales sometimes have a periodic index pulse to be used with a home switch
thats what I was hoping to find out
does it have wires for it but you don't get a signal on them?
I don't know if these scales have that
I guess I don't understand what you're asking me
yeah, but i could easily be missing it
you seemed to know what you were on about talking about the home switches ;->
Valen, cradek - i look in google about servo motor, it move only 90 degrees
yes I do :-)
but I don't understand your question - can you say more?
I was just wondering what the best way of looking to see if an index pulse/home pulse exists on these scales
Mire's lathe has a very special homing setup - much more complex than normal
I've never seen fine home switch close in either X or Z. They could be failed but so far *nothing* has permanently failed on this HNC. I believe it will come to life...
sounds like one of my cars
Mire: how are you monitoring that signal?
I was using Halmeter; I will try Halscope again now that I'm beginning to understand it.
I pink please post identifying info for you motor...
you should be able to see it with meter. turn the screw slowly - it should be active for part of each full revolution
if i take servo like this http://img.villagephotos.com/p/2002-12/76747/HitecServo.jpg
is make it easy?
i-pink: that isn't the kind of servo motor cnc people use. same word, but completely different type of motor
i-pink: that is what model airplane builders mean when they say "servo" but it is not for cnc
I thought it might be a tiny blip... if it's that easy to see i may have failed pieces after all...
has wrong feedback method
Mire: no, it's a prox that activates when the screw is oriented a certain way - not a little blip
If i do, is there a workaround?
Mire: you could do homing without index, but that sucks, your diameters won't be reproducible
cradek, i take only electronic from them and amplifier the output for my motor
it can help me?
if both are not working now (but it homed before!) you probably just have wiring wrong or something
could leave the brain on all the time, resolvers are povered by pc and won't forget...
Mire: sure, but yuck
right now I just need this thing so I can show it off!
Mire: then disable index homing for now
cradek, is look like my motor
i-pink: good, maybe that information will help you then.
but without this http://www.truetex.com/servomod5.jpg
what this part make?
i-pink: if you would read the text, you would see that it talks about adding that encoder
that is the quadrature encoder
cradek, yes... i understand this is the encoder, but what the encoder send???
notice this is an 1800 rpm DC motor. You said yours is apparently a 0.3 rpm DC motor which is not the same at all.
he has an output gearbox by the sound of it
the output of an encoder is quadrature, possibly with index signal. Many encoders seem to operate off a +5V supply and generate ttl-compatible outputs, but an encoder datasheet will always give specifics.
i-pink: encoder sends A and B quadrature(code for exact angle) and Z index pulse, once per rev. I had to add one, it is trivial.
ok, my question is, i have 3 wire,( + ) ( - ) and pulse, the EMC2 can work with this???
you already asked that and jepler already explained why that doesn't work
i-pink: with that arrangement you cannot accurately track the axis position
i want to work without angles only full turn
I had similar situation. the encoder on my spindle was wired to only pulse. I replaced it with one from another motor. now I'm in business.
You need A and B fields for direction if nothing else...
please think how emc is supposed to know where your motor is and which way is rotating
look one turn is 1/40 from one millimeter
i work on plastics
They are cheap or free. mine had a hollow shaft and fit right existing shaft, and all the instructions from our emc gurus made sense after...
it doesn't matter how big or small the distance one pulse represents
if you are content to move only 3/40 mm per minute you must be very patient
recall my argument about something pushing against the axis while it is supposed to be stopped
shaft encoder. seek one. I found mine pressfit onto the shaft of another small motor...
if the motor is right at the edge of a pulse, then you can push it "onto" the pulse with a small movement
it doesn't matter if they're 1/40mm apart, you can still under some conditions push it a very small distance onto a pulse
and 1/40mm is a very short distance anyway
i can live with this distance
but how i connect id to the emc?
i look at the stepconf
plonk = ?
emc is not designed to work your motor unless you add an encoder
it means that people are trying to explain why things won't work the way you want, but you keep asking the same questions anyway
i see in the stepconf only options for stepper motor
that's why it's called "step"conf
it's meant for step and direction systems
i dont understand how to get the pulses from the motor
servo is not step!!
yes, I know that
I have some step/dir servos :)
a servo has an encoder for feedback
stepconf for steppers. PID tuning for servos. fundamentally different, all make CNCs.
ok if i make something with 2 wire
archivist: they have encoders :)
here's a hint: servos are trickier...
servos get two wires for power, and a few(3?,6?) for f/b depending on encoder.
up, down, and feedback from the encoder
you do not have an encoder
at best, you have a tachometer
like those used on computer fans
your motor probably can't be reversed then, or there would be no reference for the tach signal
it must be ground, tach, power, which means that it can only go in one direction
it send pulse every one turn
turn of what the motor or output shaft
ok. do you want the motor to move both forwards and backwards?
the EMC2 only need to count the pulses
emc *cannot* infer direction from commanded pulse, as described for decelleration above...
i-pink, you're asking about CNC use of a motor that clearly can't be used for any sort of "normal" CNC application
no no no
if you have something specific you're trying to do, it might work for your application
answer the last two questions from us
Maybe for tracking stars while the earth spins?
you need to see how the tach feedback works before you think about getting EMC2 to run the motor
for example, if the EMC want to move 10 millimeters, and 1 pulse = 2 millimeters, the EMC need to send up or down, and count 5 pulses
your motor probably can not be reversed. it probably can only run in one direction
servo *does not equal* stepper motor...
anyone work with optimised hexapod stages?
SWPadnos onnn is DC motor
the tach output must have a ground reference. if you swap the + and - output to the motor (to reverse it), then you also change the tach reference voltage
SWPadnos, the motor work on fix voltage
is very slow motor
cradek: I'll be replacing thos Z axis balls after all. lash is around .002 and I can induce oscillations in that axis with a hexwrench in the screw.
my encoder is very simple, is black circle with hole and photo diode
so the tach is not part of the motor?
you said earlier that you have 3 wires, but you actually may have 4
SWPadnos, yes, is not part of him
i add it
add a shaft encoder instead.
i add it
does the same stuff plus angle, therefore implies direction.
what do you mean?
do you see that on decelleration your direction is opposite your command?
This is the reason index pulse is not enough.
is fix voltag and fix speed
There may be a way, but shaft encoder is the better way. I fretted about the expense until I looked at what I already had. You may have one also.
You are thinking steps still.
closed loop servo system. simple awesome better. and more complex.
is black circle with hole and optocapler
led and photo diode
[17:32:41] <i-pink> http://www.speedyshare.com/901488824.html
this is the motor
an old mouse optical encoder although will do the job is a bit of a pain to sort the mountings
ok, but in the mouse i have 2 optical encoder
2 for the price of one ,, you need the guts of it not the whole mouse
[17:35:10] <i-pink> http://www.convict.lu/Jeunes/images/angle_sensor2/encoder_mouse.gif
this is ok? ^
i have it
now i have 4 wire
[17:36:46] <skunkworks_> http://members.shaw.ca/swstuff/mouse.html
(+) (-) for the motor and 2 from the encoder mouse
skunkworks_ AchiestDragon hooooooo i need only one sensor??
i need 2 phototransistors? or one?
you strip the mouse and remove the parts for just one sensor ,, you need to mount the phototransistor and led onto a pcb to hold it in its new position and one encoder on your shaft
ok this is what i have now, but with one hole .
I'm the one w a question now. Didn't u build yourself one like this? no quadrature sigs... only pulses. how does this change the situation?
yes! i build it
from led and photodiode
the photo detectors in a mouse has 2 detectors offset in the same device package
i need 2 detectors?
or one is ok to me?
so will give you the quadrature if used with an encoder with the same spacing of the slits
what is quadrature?
I once let the magic out of an encoder to see how it works. 3 photodiodes, one pulses as you describe, two make quadrature sigs w diffraction gratings for accuracy.
what is quadrature?
quadrature is Cos and Sin, added together resolve to an angle
the angle not important to me
[17:45:52] <skunkworks_> http://www.electronicsam.com/images/KandT/servostart/Fastermouse.png
quadrature means 90 degrees apart. don't worry if you don't get it, it is old math w old words
^ what the signal looks like
phase diference between the two so the signal would go from 00 to 01 to 10 to 00 in one difection and 00 10 01 00 in the other so you get the difection
the resolution is 1/40 mm
[17:46:26] <SWPadnos> http://en.wikipedia.org/wiki/Rotary_encoder
very clever! i'm sure there are other ways...
to deduce dir
ok but resolution of 1/160 mm is overkill to me
Mire: yea there is but most need a more complex arrangment to get dir
as long as the resolution is greater than the acuracy you want then more than that is not realy a problem
mmmm i am confused
what to do?
what are you driving this motor with?
its only when the max rpm of the shaft you are trying to postion makes the pulse rate of the encoder too high to read when you get a problem
hehe too high to read
hehe speed of the motor is 3 RPM
3 RPM in the max speed
from what source come it
not a problem 27000 rpm may be
no no is 3 RPM
is 24V and i give him 12V
in 24V i get 9 RPM
then the resolution is not going to be an issue other than needing to be more than the accuracy you want at those sort of speeds
wow, i looked a the pic, not what i expected. where is this motor from?
cuzz if it really is for tracking stars while the earth spins, you may want to save it for a different project.
this is motor from very old CNC
does it give rpms on nameplate? 9 seems a trifle slow.
has a gearbox
is with gear
is very very strong motor
do you have the rest of the machine? you want the servoamps, they will be made for it.
no :( i get only the motor
There *must* have been a source of positional feedback for it to orient on, maybe it read the linear scales.
look for technical info on the machine you got it from. If it is this unusual there may be good clues there...
i dont know about the machine
the machine is dead before 20 years..
dont know type, manufacturer, lathe vs mill vs something else? did you pull it yourself?
even to know the mechanism it ran may be a clew.
my grandfather give him to me
say, pink, where in the world are you?
the text on the motor is very blurry
Mire i am from israel
is very good place we have action all the time
suckramento killafornia is my home. dull by comparison
Mire, what to do with the motor
have you blind-typed the numbers into a search engine? anything on it.
have had funny luck that way
without regard for meaning. FCC IDs work good for random circuitboards, for instance.
i see on it logo in big
pink, go forth and learn. you don't know who's time you are taking, here. We will help you, only if you do your due diligence. Trust me, I'm brand new too. Notice I'm the only one talking? I sympathise w u.
I occasioned to glance at the names on the EMC2 board of directors the other day. I want to personally thank you guys for your efforts, since I know most of you are on here right now!
i-pink: hai hai.
i-pink how old r u? Do you have shop experience?
linux whiz? or machinist?
(who else wd do this to themselves?)
i work with linux!
Mire: but she doesn't know a lot of programming.
I haven't needed any for this, yet. This software is a moving target, tho. You may need to know how to compile it.
dont try to compile it..
make dual boot
download the ISO of 8.04 EMC2
you have skype?
i-pink, how, where, who etc ?
the them abute the problem with the motor
that you have a DC engine that gives a pulse every spin ?
you want to know how to connect it to emc right ?
need shaft encoder in place of pulsegen. then just follow examples.
archivist_attic, AchiestDragon, SWPadnos
i-pink, and you do not want a big resolution of speed because it is already 1/40 ?
I had similar situation. Simple fix.
Mire, what are the examples you are talking about (I'm trying to translate it to her)
i-pink: Skype is Evil.
i-pink: get Jabber with Jingle or something.
i-pink: or Asterisk.
Mire, i-pink already have encoders but she does not know how to configure it with emc
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-10-05.txt
ik_5, the log of tonight irc
i-pink: this is for telescope tracking correct ?
AchiestDragon, no, is from very old mill machin
hmm, in my case I was interfacing to Mesa servocontroller card.
seem to think at end of day by the time you built the servo controlers and sorted out the encoder it would be cheaper to use a stepper
I agree. You have 2 working stepper axes, make a third.
a proper optical encoder alone would cost you the same as a stepper if not more
how exactly emc works with the encoders ? what does it require ? what will it return etc ?
let me find some links
i don't know emc at all, just try to help i-pink, is it on the web site ?
[19:01:24] <AchiestDragon> http://cgi.ebay.co.uk/Japan-Through-Hole-Encoder-for-Cnc-SERVO-Motor-2500PR_W0QQitemZ320426193761QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item4a9ae3b361&_trksid=p3286.c0.m14
the motor would need an encoder like on that link or something to do the same ,, price on that is about average
[19:02:37] <AchiestDragon> http://www.motioncontrolproducts.co.uk/index.php?cPath=1028&osCsid=b622a56b80ad60cf75298ac225fa2633
she wishes to use pulses rather then servo usage
rather than a stepper driver you would use a driver like the above link
yikes i didn't know the cost. hooray for dumpster diving.
well one of those to suit the motor powe
where as another stepper is going to be less than the optical encoder needed for the motor she has
that motor would be better used for eather a bigger machine with another 2 for the x and y or another project all together unfortunalty
cradek: You are a star
threading work better?
Let me get the imagebin urls
Before : http://imagebin.ca/view/XLxWqSm.html
After : http://imagebin.ca/view/RYa7Qqpm.html
was that the patch
does look fairly straight now
Indeed, and perfectly consistent too, which is possibly more imortant
is this still your homemade encoder?
The wiggles were different every time
Yes, still the paper one.
(Which seems to perform better than Steve's real one :-)
thanks for reporting back. I want to try it on my lathe before I commit it (and especially before I put it on the 2.3 branch) but this is very encouraging
well Steve's got other problems...
Yes, no good fixing cheapskate encoders and breaking real ones.
looki want to make something like this http://www.members.shaw.ca/swstuff/vna-enc.jpg
andypugh: did you try it with and without interpolated position? I kind of suspect it will work ok either way
when i build my own electronics, what exactly will emc send me for "foreword" and "backward" signals (for the hardware) ?
I didn't try non-interpolated. I can if you want
ik_5: whatever you want: analog velocity, pwm, pdm, step/dir, step fwd/step rev, etc
i-pink: Is that a mouse encoder?
andypugh: I'd be happy to know if it still works ok - if so that means it's very tolerant.
The problem with that is that the hole through the middle will have to be vanishingly small, and you really want to be direct-coupling a spindle encoder
that looks like a pot spindle with a mouse disk on the back
yeah that might not wear too well...
After the patch that cradek put in I am reasonably happy to recommend wrapping some laser-printed paper round the spindle tube and using reflective optos.
fit directly to motor spindle
cradek: I get a twitch on my F-Error signal when I let Halscope roll, and it is audible. is this normal or fixable?
Motor spindle is fine with toothed belt, probably a mistake with V-belt drive :-)
Mire: do you get the realtime-delay error sometimes?
OK, let me check with raw position.
no, i got a little Atom motherboard and it's the fastest i've tested. do I understand your question correctly?
andy theres a more accurate way to do that optical encoder if you are willing to spend some time reading the datasheets ,,, the new mice use basicaly a 16*16 pixle ccd element focused on a small aria
they detect the change in position by how far the image has moved ,, give the same optical disk it should be posible to increase the resolution by up to 16* or more
stepgan or stepgun
plus it would remove any printing errors laser printing the encoder disk
AchiestDragon: that would be cool - wonder how fast that technology can work
theres a program on the net someware that lets you use one of those mice as a scanner but would need to google for it
maybe not as fast as you could get with an optical encoder ,, but should be fast enough for a 200ipm machine
but do they have A and B outputs
archivist you can create that from the data it provides or even get a dir and step pulse from it if that is better
hm, they may not have the property that the reported distance for a certain path is always exactly the same
encoders have this property and it is essential for cnc
not at all essential for a mouse
as a simple test just rip appart a mouse with that sensor ,, knurl the shaft you want to sence so that it has a textured surface , and position the sensor close to the shaft
cradek: where is the best place to look for a +-12v supply for these servoamps? it occurs to me that this noisy sucker needs to be ruled out as a cause before I assume that it can be tuned out.
AchiestDragon: if the sensor "knows" the knurl is regular, it would work - if not and it's just a picture flying by, I bet it doesn't work reliably
Mire: can you make a plot of what you're seeing?
Mire: (why not use a PC supply)
the spacing of the knurls is not reliveant as the encoder is detecting the distance it has moved not the spacing of it
AchiestDragon: I know - I think that is the potential problem
ie it takes an image of it , then another and the output is the distance that it has moved
if every full turn of the knurl doesn't report the exact same distance (in both directions even!) it will not work for cnc
yaknow, i considered a pc supply. do they have -12v? if they do the Wikipedia page needs attention.
a regular pattern would reduce the usable speed, since it would be difficult to tell if the image had moved -15 or +1 pixels
Mire: sure they do, rs232 ports need it
-12 +3 -3 but they -12 usually has less current tan +12
i thought so. sheesh, the right answer just keeps on being simpler than it looks.
at least one of the negative supplies, and possibly both of them, are not required for ATX2.0 power supplies
[19:32:31] <cradek> http://skylab.org/~chugga/mpegbox/pictures/z_mbconatxmain.gif
the knurl is needed not for the pich of its spacings its to give it a texture contrast , it does not need to be regular just not totaly smoth with no texture like a ground shaft
earlier ATX supplies, and all AT supplies, have -5 and -12
oh really? wild
I know one was removed, it could have been -5
[19:33:22] <L84Supper> http://en.wikipedia.org/wiki/Power_supply_unit_(computer)#Wiring_diagrams
i need -100mA if i remember rite. sounds like another problem solved.
-5 was dropped in ATX
SWPadnos: the speed would have a max based on the sample rate it used ,, but you can move a optical mouse accross the desk at a hell of a rate and it keeps up with it
Anyone know of a plastic component fab that can spit out prototypes of simple mechanical systems cheap?
cheap is relative
SWPadnos: I wouldn't be surprised if they lost the need for -5 which was probably just for those very old eproms that needed split supplies
I can use Puma Kinematiks from EMC for my oun robot? (no puma) six axes similar in movement. or need news?
i just need a set of ball rods and joints, plus a linear slide for it
similar in movement, probably yes
I tried using a PC supply and it was a waste of time.
SWPadnos: see 2708 (pain in neck): http://www.qsl.net/yo5ofh/data_sheets/mem-prom.htm
optical mouse http://www.contrib.andrew.cmu.edu/~ttrutna/16-264/Vision_Project/
Bud different measurements
The one I had, even with power resistors to keep the required load on the critical lines, would still randomly shutdown occasionally.
andypugh: it's true - I had to put a big resistor on the +5 line. they're touchy that way.
Of course, if you have one, you can try it
I needed resistors on +5, +12 and +3.3 and it still didn't always stay on
[19:40:03] <AchiestDragon> http://spritesmods.com/?art=mouseeye
andypugh: Did you enable POWER_GOOD ?
spherspline : SLA's, 3D printed or machined parts?
the one I used is way too old to have +3.3... that's probably the difference.
I used an AT type PS (the kind with the actual physical power switch)
andypugh: Did you enable PWR_GOOD ?
No, wait. I didn't. Power_good is an output isn't it?
It was about this time last year, it's a bit hazy. I bought it new and had the manual, and did all it said.
Input to the PS actually.
Mabe that was the problem.
Just short the GRN wire (PWR_GOOD) to any BLK (GND)
That's not the same as the wakeup line, is it?
SWPadnos: the optical mouse would give you the distance image moved and is not based on the intervles of say knul peeks in the image it scanns ,, think that explains it better than my last attempt
L84Supper just need some parts to finish design of a hexapod style rig
small, about 10cm dia
if i have dc motor + photo diode (work like encoder), type 2 in stepgen help me???
cradek: 4mm thread with non-interpolated encoder position http://imagebin.ca/view/4Oi9bP.html
Not quite as good
I was just pointing out that something regular like graph paper makes it difficult to tell if the displacement was a lot in one direction or a little in the opposite direction
knurls are similar to graph paper in that regard, since they're fairly regular
spherspline : it all depends on the material you need, "plastic" is pretty vague
the final will be cast or milled metal, small rods
None of the RP solutions are exactly cheap.
As an example a part 2"x2"x3" will cost me £300 to have made at work, and that is cost price, no profit.
true but it just needs a uneaven surface that gives a nice contrasty image on the sensor ,, so it has some texture to compare against
denim is the best surface to run such a mouse accross
Yes, another "not exactly cheap" solution
although regular optiacly it gives a nice contrast where as say glossy paper would have no contrast and would seem totaly white to it
paper has some texture but depending how fine the fibers are
emachineshop prices were what set me on the "build your own CNC machine" in the first place...
andypugh: Fine $25 USd and I'll get a 3yo to make it for you out of silly putty
last time i looked at abs injection molding costs the mould itself was arround the £3000 mark for the cnc'ing of it
I bought a gloss white computer table from Ikea. It was totally mouse-repellent
Whoah! hang on there jymm, emachineshop is a really good facility, but Spheresplne requested cheap.
cheapest plastic prototyping solution would be http://www.craft-products.com/polyform-plastic.asp
and a dremel
And I am not the one wanting parts madking.
anyone know where that place is where the money just falls out of the sky?
Cheapest is probably conventional machining from solid, if the design suits that.
Otherwise it is not going to be cheap. That is all I was saying.
L84Supper: near a active vlocano ,, theres dimonds form in some of the stuff it chucks out
look here http://linuxcnc.org/docs/2.2/html/man/man9/stepgen.9.html
in type 2, what is phase-A and phase-B
hmm now to find that volcanic sieve
A and B are to 50% square waves 90 degrees out of phase. They tell you speed and direction.
If A goes high then B you know you are going one way, B then A the other.
anyway got to crack on and try to sort out the stepper config of my machine now i got it finished
Depending on your application you might only need one channel + index
andypugh, how it look like?
[19:59:41] <andypugh> http://en.wikipedia.org/wiki/Quadrature_encoder
Look at the section on "incemental rotary encoder"
you mean to this Incremental rotary encoder
Ah, sorry. I am currently obsessed with encoders. I missed that you were talking about stepgen
That's an interesting manual page actually, I hadn't realised that EMC can do the full bipolar step generation thing. In principle you could just use EMC and a few big power transistors to drive a stepper directly.
andypugh: I think a lot of folks have done that (heck even I have)
I possibly would have if I had thought about it.
you end up with terrible big power resistors to get much performance
Then if you want more power and volts and current, you just upgrade the power stage
late 70s technology
Ah, yes. Software chopper generation isn't going to happen is it?
I am almost relieved, as I have already ordered my replacement drivers
Did you see the non-interpolated trace?
Not quite as regular, but probably still perfectly usable.
yes - but it looked the same to me - seemed straight
did it sound different?
No, both were totally silent due to me having a shortage of stepper drivers....
Spida_ is now known as Spida
The interpolated one gave line-on-line, non-interpolated was giving a 1 or 2 pixel dither
SWPadnos: 2TB = $170
SWPadnos: What is this? I don't know AMD for crap... http://newspaperads.mercurynews.com/ROP/ads.aspx?advid=32664&adid=8403725&subid=27029261&type=
no idea, I don't see the actual ad
SWPadnos: turn off AdBlock
ok. sempron = shit
it's like celeron
SWPadnos: You'r enot familure with RackAble Systems are you ?
not specifically, no
ok, now SGI
SWPadnos: Darn, There is two RJ45 connectors on the Remote Manangement Module, just nfc why TWO
so you can plug in a phone, or multiple units in a chain?
no idea. and access to docs requires a support contract.
Talking about rack systems, my EMC machine is a second hand 2U server from eBay. It does make it easy to build an all-in-one controller box.
Dual Xeons. Far from cutting-edge but does the job.
You need to be a little careful with rack-mount servers, they don't have much call for a parallel port.
This one works, and is what I have, http://cgi.ebay.co.uk/Geotrace-Cluster-By-Rack-Saver-Server_W0QQitemZ150377049223QQcmdZViewItemQQptZUK_Computing_Networking_SM?hash=item23032bac87&_trksid=p3286.c0.m14#ht_612wt_1167
Hi, is there any way to have the remaining cutting time displayed on the Axis screen.
not at the moment
it mightbe possible to add that function, but I'm not sure how involved it is
Total runtime is available
Ok, that is a pity, good to be able to tell the wife when I will be finished ;-)
yep I found that on the properties of the file page
it's a little bit difficult if you want to take into account things like feed override setting, tolerance mode (which can affect the speed around corners), acceleration, etc.
turkshead is now known as dr_head
get wife used to delays
Oddsocks: Tell her "When she can hear again" + 15 minutes
* archivist is single, likes to get people in trouble
Get her her own CNC and double throughput
make her a cnc knitting machine
I will not let her see this as I will end up being banished to the garage
I was watching a friend do Cross-Stitch and couldn't help thinking it looks like a job for a machine.
ok the cnc mill is in there but not much room to sleep
more cnc time, win
andypugh: called a sowing machine
the sewing machiens today are actually pretty cool
yay got machine working now sweet
SWPadnos: as EMC is going to workout all the moves should it not be to difficult to get it to do a quick look ahead
that requires a complete look ahead of the file and a machine does not actually need to do that to start working
does it not do this to generate the figure in the file properties ?
axis reads the file to show the path
is the total running time a guesstimate then
wouldn't it be enough to just calculate total time - elapsed time
would probably be accurate enough for most needs
I think that is all I am after.
guess it depends how the total running time is calculated, if it takes in speed changes
and blending of different types
but it also depends how acurate it needs to be
great discussion on computer power supply, guys. I was digging thru my garage learning about the need for resistors firsthand. In short, neither specimen turned on for testing.
I'm looking at a 20 pin connector. If anyone has hacked out the answer I'd love to know it.
total time - elapsed time isn't much more useful than looking at the clock and adding the estimated time in your head
the estimate isn't intended to be all that accurate either
Mire: it will come on if you ground "power on" which I think is green??
assuming you have an ATX PS there
You need to earth the swith-on wire and then it will turn on, probably only briefly
SWPadnos: that's what i wondered
and you'll probably have to put load from +5 to gnd
[21:30:39] <andypugh> http://www.instructables.com/id/ATX--%3E-Lab-Bench-Power-Supply-Conversion/
way to really answer the heck out of a question!
I thought you only had to short a pin to make an atx power to go?
* oPless shrugs and returns to his model
Yes, but they need a load to regulate, and if they can't regulate they shit donw.
(err, shut down)
switcher ps. this is why my lites flicker on plugin then all is quiet.
they have only one switching regulator which makes all the output voltages and they shut down if any of the output voltages go too high or too low
How large can I make the load resistor(s)? It is a clear efficiency tradeoff...
so if you put a load only on 12V, it will try to put out more power which will make the unloaded 5V line go too high and it shuts down
The spec will state a minimum current
depends on switcher but 20-30% load on main supply
ratings show 400mA@-12v, 8Amps@+12v. Does this assymetry have implications?
i need +-12V@150mA
yea some tend to oscialte when theres no load on them ,, usualy a couple of old hdd drives would be enough to stop it doing that
Mire: you only need 12V 150mA?
there asimetric anyway ,, the -12v is only needed for the rs232 drivers and some odd expantion , but the +12v would be driving disk drives , cd roms , flopies etc
i'd try putting about 1W load on 5V and 12V
as long as you have someitn in place to maintain the minium load it should be fine for +-12v at 150ma
then it will probably supply small currents well on all the output voltages
you only need to have the min load on the 12v and 5v outputs the rest hold fine
Have you looked at Dc-DC convertors?
[21:44:06] <andypugh> http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=3114758
Though, if you have a PC PSU to use, you might as well
i do, but thanks for the tip. You should see what I'm replacing. the idea that a little black monolith can replace it is quite a joke.
The future happened while we weren't looking
yah, then I got a job learning the past from old geezers and now I'm caught between worlds...
Mire when you go?
plenty of old geezers in here
would +-10v do you ?
* jymm kicks archivist's walker out from under him!
donno. why do you ask?
Mire, i make now a 2 bit encoder 00 01 10 11
[21:51:25] <AchiestDragon> http://www.whipy.demon.co.uk/gw1cto-trx-sch.pdf
sheet 20 gives +-10v at 100ma from a single 12v source ,, uprate the caps c325,c326 to 47uf and the 4 on the outputs to 22uf should do for 150ma
if you remove u33 and the 2 resistors by it , and feed 12v to where u33's output is then it will give you +-12v rather than 10v
Big GE servoamps. +-12v is what the terminals say
i-pink: 00 01 11 10
yea because a 11 to 00 change is going to give a 10 or 01 for an instant anyway
is 0 1 2 3
simple binary is bad for encoders
0=00 1=01 2=10 3=11
Or Wiki for Gray Code
you should have 1 transition only see gray code
as you count around, you have a place where both bits change
change order to 0,1,3,2 and only one changes per transition
so in any direction the change on the output is only ever one bit at once ,, when you get to 3 the change for the next step you would assume it to be 0 but because there is a delay or small error in the sensor positions it would drop one bit before the other giveing you a 01 or 10 before the 00 meaning an error
[22:03:42] <i-pink> http://en.wikipedia.org/wiki/File:Encoder_disc_(3-Bit_binary).svg
is 0 1 2 3
you want gray code. is different
If you read on it says why that is a bad plan
i make this
[22:05:23] <i-pink> http://en.wikipedia.org/wiki/File:Encoder_disc_(3-Bit_binary).svg
[22:05:40] <i-pink> http://www.speedyshare.com/886910352.html
it is not a binary count you need binary is not sutable as more than one bit can change between each step and if more than one bit changes they DO NOT change at the same time there is an slight error
and the pic i send is ok?
pink is from israel and speaks good but not perfect english.
[22:07:55] <frallzor> http://www.mechmate.com/
anyone in here made one of these? :)
Unifilarpole yes to me?
you have made one? =)
you are making one?
planning at least
asking around for lasercutting offers
is what it should be like
Not is a machine for laser o for plasma
thats a 3 bit not 2
is for mill
but the parts needed need cutting ;)
thats what im asking around for =)
i pay arround $200 for laser parts
where did you get them?
In a laser company :)
you should have a local company
my nearest is about 10 miles
but the emc use only A and B, i need to ignoring from the external circle
I do, but local allways tend to be the pricey alternative
but ive sent a req. for pricing so hopefully ill know soon
my local was cheap enough but does have a minimum order value
Hang on. You are making a 2-bit encoder because emc has only 2 inputs?
laser should be pretty cheap id assume comparing to other ways of cutting
I think you might want to rethink.
You would normally use a few hundred pulses per rev on one channel, and one pulse per rev on the other
a 2-bit absolute encoder is probably a bit pointless
its a motor for Z not a spindle
Still. Is the plan to have speed and direction only?
frallzor, you can post your comments in the forum
i am =)
you got a log to look in? allways fun
The forum is fairly low-traffic, the mailing list might work better, though that is a bit more emc-focussed
he was talking about the machmateforum I'd assume
Should I shut up? :-)
no keep talking, allways fun :P
So, 2 bit encoder. SImplest would be to laser-print something, though that 3-bit wiki one won't work properly.
Do you have suitable software? Inkscape would do the trick nicely (you can type in object sizes directly, which is how I got my encoder to wrap exactly round the spindle.
wibbly wobbly clunky machine finished and working ,, other than needs dremel mounting on it http://www.whipy.demon.co.uk/dscf2342.jpg
It looks a bit under-exposed to me :-)
yea batteries a bit low on cam
now i have it working can replace those stepper drivers with bipolar microstepping ones
i-pink: What is your native language?
Hmm, I was going to suggest reading http://he.wikipedia.org/wiki/סרוו
But is looks a bit empty
It's the hebrew version of http://en.wikipedia.org/wiki/Servo_motor
but i not make servo
i make encoder
If you want to use your motor with an encoder you cold try reading this
[22:34:26] <andypugh> http://he.wikipedia.org/wiki/תורת_הבקרה
But it is possibly a bit more theoretical than you need
An encoder is part of a servo system.
I assume you intend doing something with the information from the encoder? Once you do that, you have a servo
i make now new encoder with gray code not binary code
00 01 11 10 =)
i-pink, and how will emc control it
archivist, i think with type 2 on stepgen
Stepgen is output, not input.
wrong, its a dc motor you have
i am confuse?
type 2 is work?
I admit I don't understand how pink's setup is meant to work... apparently she needs to control only the number of rotations of the motor. Encoder was to establish direction, she had just a hole and photodiode before.
type 2 is for steppers
dc motor and 2 bit gray code encoder
Stepgen type 2 will drive a particular type of motor. But you are making an encoder, which will have to provide inputs to an encoder function
You probably want a PID controller to drive the motor.
And use the encoder so that the controller knows where the motor is.
[22:49:29] <i-pink> http://linuxcnc.org/docs/2.2/html/man/man9/stepgen.9.html
and see type 2
Yes, that will drive a stepper motor. The phase A and Phase B are motor windings, not encoder inputs.
andypugh: what is needed to convert PID controller output to drive this dc motor?
Probably PWM+direction pins to an external PWM-voltage convertor (which is just a capacitor and resistor at its simplest)
you need a simple pwm amplifier..
I see you have a breadboard to build it on.
then an amp and an encoder on the motor (not the gearbox out)
[22:53:20] <skunkworks> http://emergent.unpy.net/projects/01142347802
The bit I am struggling with is Gray-code to software encoder counter. I am not sure that EMC supports that
That's a relief after we have told i-Pink to make one
Which HAL module is it in?
the software endocer takes greycode.
i am almost finish the greycode endocer
It only seems to mention quadrature in the manual page
greycode is quadrature
[23:00:00] <skunkworks> http://en.wikipedia.org/wiki/Gray_code
if i dont learn slower soon my head will balloon like a Martian.
Yes. Of course it is, when you have two channels. Allow me to slap my forehead in a comedy manner
wait - I might be wrong - sorry - long day.
No, you are entirely correct.
not quite true, two bit only has 4 values
i finish it!
and a high count encoder is many lines
The point is that a 2-bit gray code encoder is also a 2-channel quadrature encoder. Connect A and B to the pins of an Encoder in HAL, and the encoder position count will ramp up and down correctly
Hello all. The big picture is setting up another CNC machine using EMC2. I'm using a similar setup to my first EMC2 machine, but this one uses a charge-pump to enable the motor power supplies. The problem is that if I tell stepconf that I have an estop switch (which I do :), the charge-pump signal doesn't happen. If I say no estop (that is, set that parallel port pin to unused), then the charge pump works.
invert the pin
I think my problem is that iocontrol.0.user-enable-out is always false. I can't find exactly what internal conditions set this to false.
Do you have emc working i-pink?
[23:05:50] <i-pink> http://www.speedyshare.com/321672552.html
Ought to work.
Sorry--I need to type faster. EMC seems to read the estop fine. iocontrol.0.emc-enable-in, estop-ext, and pin10-not (the estop input) are all true. But the critical (I think) estop-out, charge-pump, and so on are false.
Seems to track back to iocontrol.0.user-enable-out being false.
To find out you can use stepconf to create a set of config files, and tell it that A and Z are on the same parallel port pin, and that B is on another.
You should then be able to watch the output of encoder.0.position in the hal-meter (in Axis that is Machine-Show Hal Meter)
It should increase when you rotate the encoder over the detectors one way, and decrease when it rotates the other way
(I am simplifying here, and skipping a few weeks of screaming and cursing)
is very very hard!
That will just be enough to check the encoder. To set up the motor to be controlled you will need a servo setup in EMC, not a stepper one
As for being hard, yes. People spend whole apprenticeships and degrees learning this stuff :-)
ok. and the emc support servo?
but i have only 0 90 180 270 degrees
But I think you will need to setup your own HAL configuration files.
Can she fake something that will stop motor on a given rotation? she says that
just remember you are doing this for FUN !!! ;-)
's all she needs
Yes, but the encoder can cheat a bit.
Yes, you can probably use stepgen to produce up-down signals to drive the motor one way or another, and just choose up and down according to whether the encoder position is bigger or smaller than required
In a sense, she wants to use it is a stepper, but with 360 degree steps
Can I point out that you really, really, want to be putting opto-isolators between the parallel port and the rest of the electronics
How fast does the motor spin, and is there a gearbox?
not 360 steps
full turn per step
You will have 90-degree steps
Actually, you might find you have better than that with the interpolated encoder function
archivist--the estop is already inverted, and seems to read fine. If I activate the estop, estop-ext and iocontrol.0.emc-enable-in go false.
if the EMC send very short pulse the motor cant rotate 90 degree it move only few degree
First things first, get your motor, encoder disc, sensors and interface set up and connected to the parallel port, and make sure you can see the channels in halscope or halmeter.
Yes, but then the encoder won't move far enough, and emc will send another pulse
The encoder "closes the loop" telling emc when it has moved the motor far enough.
it will pulse until encoder changes.
You don't really want to be using a stepper drive to control a dc motor, though. That PWM servo amplifier was a much better soultion.
Yes, assuming you have EMC working, there are software voltmeters and oscilloscopes that will let you watch what all the signals are doing.
how i can build PWM servo amplifier at home????
You have a prototype board, I saw it in your first picture
And skunkworks posted a circuit diagram
[23:23:33] <Mire> http://emergent.unpy.net/projects/01142347802
there it is again.
You might need to get him to explain some of it.
how come that ubunto wifi drivers for my laptop dont work but the suse ones do
you mean that?
[23:25:37] <i-pink> http://media.unpythonic.net/emergent-files/projects/01148303608/l298-servo-sch.png
it did connect when i first set ubuntto /emc2 up as it managed a online update ,, but it now refuses to connect
just keeps requesting the pass key
That's the one. It only looks to contain one regulator and a few capacitors and diodes
i have L298N and the diodes i am not shore about the 3.3K resistors
what is C1 - C4? and what is the lm340lz???
what are some considerations to keep in mind when building your first mini-mill? how hard would it be to convert this machine's wood parts to metal? http://buildyourcnc.com/default.aspx
Halscope; this is a halscope trace I took earlier today, showing my encoder pulses (at the bottom), the encoder position (the rising lines) and the motor position (the wobbly line). Halscope is going to be really useful to you for working out what is happening: http://imagebin.ca/view/XLxWqSm.htm
akafubu, you might want to look at some of the 80/20 plans that are out there
That's not a mini-mill. Check Youtube for Blacktoe, it's huge
blacktoe is huge
for mdf anyway
LM340 seems to be a regulator.
what does the 80/20 stand for?
brand of metal strut
and what about diy 5-axis projects, i haven't seen any of those
80% Meat ratio to 20% fat
start with a 3 axis
okay, if i can find a how-to somewhere
i-pink: You can look up what components are (and get the data sheets) at http://il.farnell.com/
cnczone is where to find info about machines
's PWM controller is only 2 chips. neato
1 min let me sort out my network problems
although lost scroll back in the process
hey, Adragn, you posted PWM comtroller schematic, rite?
although have one someware
very elegant. still need amps after that, right?
or will this drive motors direct?
Here is the one I use.
[23:47:01] <andypugh> http://imagebin.ca/view/75aik0k.html
It does more (and less) than you want, but does opto-isolate the p-port
pwm as in pwm speed controled sutable for up to 15A and 1.5 to 48V motors http://www.whipy.demon.co.uk/h-bridge-sch.pdf
with electronic break control on there also
shows it as 3.3v logic but will work at 5v without changes
It produces 2 isolated voltages from 1 PWM signal, then swaps lines around using relays. You could use one of the PWMs and keep the relays for reversing. I had run out of pins to the 2 relays do off-lathe forwards-lathe backwards-mill forwards depending on the combintion
Yeah, use his :-)
That's exactly what you want
Though I am not spotting motor-minus
the output fets will need heatsinks and need to be mounted with insulated thermal pads and insulated screws
sorry, I am being daft.
The motor connects in the middle (M1 and M2) +Vmotor is motor power supply.
the sensor is not work,
i use yellow led :(
i did an update to that schematic without the break and without all those diodes ,, probabay better with that as its tested
What is the sensor?
but the file is on another machine will take me 10 mins or so to get it booted up
the phototrnsistor with led
You might be better with an all-in-one sensor,
is all the time 39-40 K
normally they are infra-red with a filter so that ambient light isn't a problem
i have ultra white led
Do you have a part number for the sensor?
I am using these, and they work fine
However, are you just measuring the resistance of the sensor?
I forgot the link
[23:57:49] <andypugh> http://il.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=4550931
Are you measuring the sensor resistance with a multimeter?
i put black shrink on the phototrnsistor
OK, the multimter test voltage is probably too small. does the miltimeter have a diode test setting?
Try it with that, and you should see something change