Connect with us

Homebrew NC Mill

Discussion in 'Electronic Design' started by Rich Grise, Jan 18, 2004.

Scroll to continue with content
  1. Rich Grise

    Rich Grise Guest

    My office opens onto a machine shop. I was just now chatting up the
    machinist, and idly musing on the workings of the transport mech or
    whatever it's called - it's got motors to drive the bed in X and Y.
    This is a Bridgeport end mill, with electric motors to slew the bed.
    And I was wondering about turning it with a stepper motor - with
    the gear ratio, it looks like you could get about .0005" per step.
    "Hmmm ..." I mused to myself, and the machinist (Jimmy) was talking
    about closed loop and open loop and the accuracy of servos because
    they use the sine and cosine, and that a stepper motor can miss a
    step, and so the feedback comes up. "How does the controller know
    if you've missed a step?" And I start thinking about position
    sensing, and asking about where's the best place to put a pot,
    and he says, "Well, you could just use the readout." I slaps
    meself in the haid - "Boy! Am I Stupid!" "What? For not thinking
    of that right away?" "No. For slapping myself in the head like
    that. That really hurt!" And I went, That's trivial. I've written
    stepper motor controllers in Z80. You can get a driver for a few
    bucks. And I have not only THREE computers (I inherited the one
    that they replaced in the front office), but an M68ICS05P in-circuit
    simulator/development system in my office.

    I'm gonna _BEG_ the boss to let me do that. (slap together a
    68HC705P6A and a stepper motor driver, hack into his Bridgeport,
    and have NC on a beer budget!)

    Wish me luck!

    If I get the project, I promise to make frequent progress reports,
    so everybody think good thoughts! :)

    Cheers!
    Rich
     
  2. Garrett Mace

    Garrett Mace Guest


    Could be simple, could be tough. These mill retrofits vary. If you've got
    the motors and drivers lying around already, half the battle (and 75% of the
    expense) is over.

    But then you need software to run the thing. TurboCNC
    (http://www.dakeng.com) will let you read G-Code files and generate all the
    pulses you need with only a Grandma-grade computer. No need to mess with
    HC705's or anything.
     
  3. Mark J.

    Mark J. Guest

    In news:JzrOb.92137$-kc.rr.com (Garrett Mace):

    Technically, this isn't "Homebrew NC", it's "Workbrew NC". :")

    The nicest (cheap, fast, simple) driver board I've seen is:
    http://www.hobbycnc.com/driverboards.htm

    One of these days I'm gonna get one or three of those, and build a zippy
    little PCB driller. And NC the lathe, and NC the mill... one of these days.
    Maybe. :)
     
  4. The other reason to use FB is to avoid backlash errors and screw
    inaccuracy.. it's also better if the mill uses ballscrews which get
    rid of the first, but retrofitting them costs $$.
    When/if you get it, consider surplus servos as well. They have shaft
    encoders and you can even make them mimic steppers if you so desire.
    Good luck!

    Best regards,
    Spehro Pefhany
     
  5. I've come across web sites describing DC pm motors used for this, rather
    than steppers, used. Steppers are more cost-effective for lower powers,
    if your interested in maximizing torque-speed-$$$

    It is possible to sense back EMF on the stepper to figure out if your
    missing steps, but analog methods take tweaking and time.

    And I start thinking about position
    Yep, I wondered about hacking my own linear encoder from a laser-printed
    mylar strip, or magnetic tape. But for my (PCB drill) app, I didn't need
    to worry.

    (I inherited the one

    I love PICs. Far superior to the archaic Motorola chips.

    Post this to rec.crafts.metalworking for some good advice! There's also
    a spam'n yahoo group mailing list for CNC too.

    Don't know about that, I think I tried it & didn't like it. You need an
    RTOS for milling; Windows can't give you precise time-control for
    coordinated curves. I'd checked out EMC for Linux.
    Checkout my web page for my PCB drill bot. Can hardly do it simpler, but
    perhaps cheaper if you use acme rather than leadscrew & make your own
    anti-backlash nuts.

    --
    Scott

    **********************************

    DIY Piezo-Gyro, PCB Drill Bot & More Soon!

    http://home.comcast.net/~scottxs/

    **********************************
     
  6. Rick

    Rick Guest

    join cad_cam_edm_dro on yahoo groups. they have done everything you want to
    do and will give you much advice.
     
  7. Rick

    Rick Guest

    If you can get servos with quadrature output encoders (two channels of
    square waves 90 degrees out of phase) then you can use Gecko G320 drives
    which are about $100 and allow you to use step/dir from a micro or parallel
    port. Get an old PII running linux and download the free EMC software to
    drive your motors using industry standard G-code.

    If you think the palm sized Gecko drives are too small, consider that I am
    using them right now to move the x,y, and z axis on a 7000 lb Shizuoka
    bedmill. I can do rapids of 180 IPM. I probably could go faster, but the
    mill starts dancing around on the floor!

    Rick
     
  8. Rich Grise

    Rich Grise Guest

    Bad form to reply to oneself? Oh well, there's such an
    abundance of great replies, I couldn't pick a weiner. ;-)

    The "which motor" problem is solved. While talking with
    Jimmy about how to hack into Joe's $1200 readout,
    http://www.acu-rite.com/index.cfm?PageID=2399B4AB-2CF2-11D5-BC6E00A0CC271CB6
    he says "Hey, let's use mine!"

    It turns out that he's got a drive for _his_ old Bridgeport,
    well, at least the X. It's got what's obviously a universal
    motor - both the armature and field windings are coming
    out on the connector; I'll have to do a little digging
    to see if it's supposed to be series or shunt. I don't
    want to try to hack into his control circuit - it looks
    like just a pot and switch maybe an SCR or resistor or
    two, but it's covered in about 1/8 thick conformal of
    some kind.

    So I'm thinking, like wow! The best of both worlds. If it
    turns out to be a 115V motor, I can drive it with a MOC3030
    and a triac - and I have a whole 1/60 second to decide whether
    to energize a particular cycle, so it might give a reasonable
    approximation of a stepper motor. Or I could use phase control,
    making it more like an analog servo motor. AC? DC? Who Cares?
    Voltage? Duty Cycle!

    And he says he can get a position transducer, and kept talking
    about the sine and cosine. Well, a very little digging shows
    that the transducer, above, uses "quadrature TTL," which is
    tantalizingly little to go on.

    But I told him, "If you can give me a signal that tells
    me where the bed is to a ten-thousandth of an inch, I can
    write the software (and build the driver - one 3030 and
    one triac, big whoop) that will put it there."

    Another thing - how do we get the data from the computer
    to the machine? Serial port, of course. What kind of
    software? Well, lessee - how about list a polyline and
    paste it to COM1?

    I could certainly write a parser for

    " at point X= 0.0000 Y= 0.0000 Z= 0.0000\x0d\x0a".

    Thanks for the PIC idea, whoever it was, but I already have
    the Motorola development, and already speak HC11, so O5
    shouldn't be much of a stretch. And I _love_ that timer!

    Oh, and I just inherited a broken motherboard and perfectly
    good power supply, and there's a floppy lying there -
    with $30.00 worth of RAM (maybe less) I could run a DOS system
    and bitbang the whole thing in MSC 5. :) (well, it's still
    got the com port and the LPT port, heh, heh ...) I once hacked
    into Telix or Procomm with some debugger and lifted their
    interrupt handler so I could write a comm program of my own,
    which I called "dt," for "dumb terminal."

    Ah, the good old days brought back to life!

    Cheers!
    Rich
     
  9. In Rich Grise typed:
    Why hack into the readout? I think you want to just hack into the cable
    and tap into the 2 quadrature signals for each axis. If it has a
    connector you could do it with a breakout box.

    Little to go on? Obviously it's a pair of signals from a relative shaft
    encoder. You know, they're 90 deg out of phase, like a sine and cosine.
     
  10. Smiley

    Smiley Guest

    www.cnczone.com is another great resource for both the pre-made and home-brew.
     
  11. Some scales have glass grids, these align to give an light sensor a signal,
    two sets, one alined 1/2 step in front of the other alow for direction to be
    taken into account.

    If you contact some scale makers, they will be happy to tell you what their
    scales output is.
    O.K. the mill got there..... pity about the cutter :-(

    Speed ( feed rate ) control is very important.


    --
    Jonathan

    Barnes's theorem; for every foolproof device
    there is a fool greater than the proof.

    To reply remove AT
     
  12. Oppie

    Oppie Guest

    This magazine often has ads for home shop automation.
    http://www.homeshopmachinist.net/

    If you haven't seen the magazine but are interested in mill/lathe work, get
    a copy. There are some impressive projects in it.

    Back the the original question - Any serious machine uses some type of
    feedback. We have 5 CNC mills now and they all use AC servomotors with a
    variety of encoder devices. The most impressive is how a 500 pound machine
    head (whole machine is about 9 tons) about to tap a 2.5mm diameter threaded
    hole plunges at warp speed to within a mm of the workpiece, spins up and
    feeds while tapping. As much as I know about control systems, I am impressed
    every time I watch the machines.

    Oppie

    For correct response address, remove -nospam-
    =========
     
  13. Garrett Mace

    Garrett Mace Guest

    Oh, and I just inherited a broken motherboard and perfectly

    The DOS bit-banging you speak of has already been fully implemented in
    TurboCNC, at http://www.dakeng.com. I probably forgot to mention in my
    eariler post that it is free to use, and $20 for the source code. A new
    version is on the verge of coming out, too, think it's in fourth beta now.

    Also, since you seem to be in the mood for oddball hacks, I've heard of at
    least one case where a guy used automobile alternators as stepper motors in
    his CNC router.
     
  14. spindle RPM ~= CS* 4/D
    CS = cutting speed in surface feet/minute
    D = diameter of tool

    FEED rate = RPM * FEED per TOOTH * # of TEETH


    Eg. aluminum with an HSS (High-speed steel cutter), 300 FPM, 5 thou
    per tooth. Using a 1/2" diameter 4-flute end mill,

    Spindle RPM = 2400
    FEED rate = 48 inches/minute

    The above is for cutting (G01/G02/G03).

    CNC machines also do G00 "RAPIDS", which can be as fast as the machine
    can move (and often don't follow a straight line between the two
    points). They are not supposed to be cutting when they do this.

    Best regards,
    Spehro Pefhany
     
  15. Rigid tapping or with a tapping head?

    Best regards,
    Spehro Pefhany
     
  16. Bill Garber

    Bill Garber Guest

    : This magazine often has ads for home shop automation.
    : http://www.homeshopmachinist.net/
    :
    : If you haven't seen the magazine but are interested in
    mill/lathe work, get
    : a copy. There are some impressive projects in it.
    :
    : Back the the original question - Any serious machine uses some
    type of
    : feedback. We have 5 CNC mills now and they all use AC
    servomotors with a
    : variety of encoder devices. The most impressive is how a 500
    pound machine
    : head (whole machine is about 9 tons) about to tap a 2.5mm
    diameter threaded
    : hole plunges at warp speed to within a mm of the workpiece,
    spins up and
    : feeds while tapping. As much as I know about control systems, I
    am impressed
    : every time I watch the machines.

    Run them for 20 years as I did. You will cease
    to be impressed after about 5 years. ;-)

    Bill @ GarberStreet Enterprizez };-)
    Web Site - http://garberstreet.netfirms.com
    Email -
    Remove - SPAM and X to contact me
     
  17. Wild Bill

    Wild Bill Guest

    Quadrature TTL is a common standard output for rotary and linear encoders,
    which is two squarewave signals which are 90 degrees out of phase. You can
    count pulses, but a better approach is to read (and keep track of) all the
    transitions.
    There are also sinewave differential encoders, and some manufacturers make
    modules for converting the sinewaves to TTL.

    I dunno about the practicality of approaching .0001" accuracy though. In
    linear encoders, that's about 400 to $500 per foot in length.

    For the axis drive unit, you might be able to adapt a rotary encoder to the
    leadscrew, but leadscrew backlash will be a factor. Because the motor isn't
    a stepper, there might not be good resolution for repeatable accuracy. This
    wouldn't be bad for some operations, but it wouldn't work well for
    predictable precise positioning.

    WB
    .....................
     
  18. Oppie

    Oppie Guest


    No tapping head. Just a tiny tap in the spindle of this huge machine.

    Then there is also single point thread cutting where a single pointed
    cutting tool is spinning fast in the spindle. It decends slowly as the XY
    table moves in a circle. We use this for cutting larger diameter threads on
    occasion where it isn't practical to buy a tap.

    Machines are all RS-232 but connect into an eternet to RS-232 converter so
    the files can be downloaded from the office server. parts are designed in
    AutoCad, edited with AutoCAM tooling path editor and downloaded to the CNC
    mill.
    Ain't technology wonderful?
     
  19. Garrett Mace

    Garrett Mace Guest


    I think the cool way to do it now is threadmilling. Bore a hole out to the
    root diameter, insert a smaller tool with a single row of teeth at the right
    pitch, move tool to the edge of the hole and make a one helical rotation
    around the inside of the hole. Don't need to sync or even turn off the
    spindle, makes accurate threads and you can use one tool to replace a number
    of taps, and do non-standard diameters.
     
  20. I read in sci.electronics.design that Garrett Mace <>
    That's 'thread chasing'. It's getting on for 200 years old.
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-