Gear Hobbing: Servo or Stepper motor - Page 3
Close
Login to Your Account
Page 3 of 9 FirstFirst 12345 ... LastLast
Results 41 to 60 of 166
  1. #41
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,612
    Post Thanks / Like
    Likes (Given)
    2114
    Likes (Received)
    3893

    Default

    A stepper of any kind would be a turd compared to an appropriately sized brushless servo motor.

    A Sankyo roller drive is a nice piece of gear - putting a stepper on it would be a sin. Note that if you put a gear reducer on the input shaft of the Sankyo, that you will need to consider the backlash in that gear reducer divided by the ratio of the Sankyo unit when considering accuracy - generally a low backlash unit would have +/- (1-4) arc minutes of backlash for a typical "low backlash" unit. OR . . . add a secondary encoder to the Sankyo unit and close the position loop with the secondary encoder while closing the velocity loop and managing commutation of the motor with the encoder on the motor.

    When picking a motor, pick the one with feedback that is compatible with the Delta Tau unit and with the lowest back EMF characteristics possible so that you have a high torque / amp ratio and minimal reduction required for any primary gear education. If you can get the primary gear eduction down to under 4:1, you may consider a timing belt pass instead of an input gear reducer to achieve the overall ratio you are seeking. Best to have an overall ratio that is a optimally a binary weighted encoder count resolution per degree of motion.

  2. #42
    Join Date
    Aug 2006
    Location
    Kitchener, on canada
    Posts
    870
    Post Thanks / Like
    Likes (Given)
    15
    Likes (Received)
    193

    Default

    If you need accurate gears in small quantities would you not be better off getting someone to cut them on a wire EDM machine. I have the same problem as you (need small, accurate, non-standard spur gears, don't want to spend much money or wait too long). EDM does not allow for plastic gears, those I have to sub out.

    I would be happy to have a chat offline about gearing, I am looking for a consultant who can help me with design through to procurement of gearing.

    [email protected]

  3. #43
    Join Date
    Jun 2002
    Country
    CANADA
    State/Province
    British Columbia
    Posts
    2,470
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    1821

    Default

    Hi SBAER:
    Yeah but then you first need to find someone, then you need to wait for them, and then you need to pay them. (or you need to buy a wire)
    Our OP wants to eliminate that nuisance factor and roll his own so he can get what he wants when he wants it and not dangle on somebody else's schedule.

    Cheers

    Marcus
    Implant Mechanix • Design & Innovation > HOME
    Vancouver Wire EDM -- Wire EDM Machining

  4. #44
    Join Date
    Feb 2017
    Country
    UNITED STATES
    State/Province
    California
    Posts
    350
    Post Thanks / Like
    Likes (Given)
    6
    Likes (Received)
    89

    Default

    Quote Originally Posted by motion guru View Post
    A stepper of any kind would be a turd compared to an appropriately sized brushless servo motor.

    A Sankyo roller drive is a nice piece of gear - putting a stepper on it would be a sin. Note that if you put a gear reducer on the input shaft of the Sankyo, that you will need to consider the backlash in that gear reducer divided by the ratio of the Sankyo unit when considering accuracy - generally a low backlash unit would have +/- (1-4) arc minutes of backlash for a typical "low backlash" unit. OR . . . add a secondary encoder to the Sankyo unit and close the position loop with the secondary encoder while closing the velocity loop and managing commutation of the motor with the encoder on the motor.

    When picking a motor, pick the one with feedback that is compatible with the Delta Tau unit and with the lowest back EMF characteristics possible so that you have a high torque / amp ratio and minimal reduction required for any primary gear education. If you can get the primary gear eduction down to under 4:1, you may consider a timing belt pass instead of an input gear reducer to achieve the overall ratio you are seeking. Best to have an overall ratio that is a optimally a binary weighted encoder count resolution per degree of motion.
    I like the dual motor/rotary encoder idea a lot, think i'll go in that direction....it's consistent with my idea of getting the encoders taking direct measurements of the cutter and the work piece actual speeds.

    My milling machine is limited to max of 2000 RPM, so for a 40 tooth gear the rotary needs to go at 50 RPM. That will be the challenge for the rotary/server motor, getting the speeds this slow while still maintaining required torque.

  5. #45
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,612
    Post Thanks / Like
    Likes (Given)
    2114
    Likes (Received)
    3893

    Default

    The Sankyos come with a 50:1 or 60:1 ratio . . . so with a 50 rpm output speed, you will need to spin the input shaft from 2500 to 3000 rpm. I would go with the 60:1 ratio and a 3000 rpm rated motor which will cover your speed range from zero to max without difficulty. The Delta Tau amplifier can easily do field weakening (back EMF suppression) if you ever need to go faster.

  6. #46
    Join Date
    Jan 2010
    Location
    Canada
    Posts
    321
    Post Thanks / Like
    Likes (Given)
    61
    Likes (Received)
    164

    Default

    Quote Originally Posted by drcoelho View Post
    My milling machine is limited to max of 2000 RPM, so for a 40 tooth gear the rotary needs to go at 50 RPM. That will be the challenge for the rotary/server motor, getting the speeds this slow while still maintaining required torque.
    Slow speed, that's why I would go with the stepper motor, Nema 34 with 1600Oz would work great.

  7. #47
    Join Date
    Feb 2017
    Country
    UNITED STATES
    State/Province
    California
    Posts
    350
    Post Thanks / Like
    Likes (Given)
    6
    Likes (Received)
    89

    Default

    OK, so this is all starting to crystallize for me, thinking the following:

    - robust sub-plate mounted on Deckel universal table allowing for helix angle setting
    - Sankyo RCD rotary with 60:1 RPM reduction and support mounted on sub-plate, suitable for holding a mandrel to which the work piece is mounted, probably go with a 5C collet on the rotary
    - high resolution encoder on the Sankyo RCD rotary with feedback loop for rotary position
    - 3000 RPM servo motor driving the Sankyo rotary, with high resolution encoder with feedback loop for motor velocity
    - gear hobbing cutter mounted on the Deckel horizontal spindle arbor
    - high resolution encoder mounted directly on the horizontal spindle arbor
    - Delta Tau controller/amplifier, picks up speed from horizontal spindle arbor, maintains matching speed (ratio determined by specific gear tooth count) on rotary....need to factor in both motor velocity and rotary position in programming
    - Deckel z axis used for cut depth control
    - Deckel x axis used for axial feed of hob across work piece
    - use of single thread hobs for higher quality gear cutting

  8. #48
    Join Date
    Aug 2006
    Location
    New Zealand
    Posts
    1,232
    Post Thanks / Like
    Likes (Given)
    70
    Likes (Received)
    246

    Default

    Hi
    For 50rpm, I would use a servo motor and harmonic drive.
    Harmonic drives have zero backlash combined with high torque ratings. The load is spread across many teeth.

    A geared gearbox will never have zero backlash. A belt drive requires a lot of tension to eliminate compliance. Harmonic drives are compact, simple and sealed.


    Dazz

  9. #49
    Join Date
    Jul 2006
    Location
    Hillsboro, New Hampshire
    Posts
    6,760
    Post Thanks / Like
    Likes (Given)
    1602
    Likes (Received)
    4719

    Default

    Will a 5C collet have adequate concentricity and angular (nutation) accuracy for your uses? Might want a fixturing method that allows you to dial-in TIR and straightness.

    On driving the rotary - one rule of thumb I've read is to match the driving servo motor's moment of inertia (MoI) to the driven MoI as a way to have adequate performance. As with any RoT this suggestion has limitations, and if you need a further gear reduction over the inherent 60:1 of the rotary than I agree with dazz that a harmonic drive would be a good choice.

  10. #50
    Join Date
    Feb 2017
    Country
    UNITED STATES
    State/Province
    California
    Posts
    350
    Post Thanks / Like
    Likes (Given)
    6
    Likes (Received)
    89

    Default

    Quote Originally Posted by Milland View Post
    Will a 5C collet have adequate concentricity and angular (nutation) accuracy for your uses? Might want a fixturing method that allows you to dial-in TIR and straightness.

    On driving the rotary - one rule of thumb I've read is to match the driving servo motor's moment of inertia (MoI) to the driven MoI as a way to have adequate performance. As with any RoT this suggestion has limitations, and if you need a further gear reduction over the inherent 60:1 of the rotary than I agree with dazz that a harmonic drive would be a good choice.
    Alternatively, I guess I could put a 4-jaw chuck on the rotary...

    I don't think I'll need more than the 60:1 reduction. I can control the speed of the horizontal spindle and bring it down as needed.

  11. #51
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,612
    Post Thanks / Like
    Likes (Given)
    2114
    Likes (Received)
    3893

    Default

    A harmonic drive has a nonlinear position output for a linear input due to the epicycloidal nature of the teeth climbing over one another as the flex gear is deformed by the off-center bearing drive. If additional gear eduction were required, this would not be a good choice if the gear blank needed rotation during the cutting process as it would result in a defective tooth form with finish patterns associated with the sinusoidal velocity ripple.

    Study the RCD roller drive and you will see this is more than adequate for stiffness, rotational accuracy, and zero backlash.

    At 60:1, inertia matching will not be an issue at all.

  12. #52
    Join Date
    Jul 2006
    Location
    Hillsboro, New Hampshire
    Posts
    6,760
    Post Thanks / Like
    Likes (Given)
    1602
    Likes (Received)
    4719

    Default

    Hi MG, thanks for the clarification on the epicycloidal errors of harmonic drives. I was aware of them for "Cyclo-type" reducers, didn't think harmonic drives would have them to any significant extent. I'd bet that the errors are so small that, especially after going through a further 60-1 gear reduction, they'd be lost in the noise for this application, but it's good to know.

    And certainly, that same gear reduction changes the matching of servo to driven mechanism. Do you have any general approach to servomotor selection? What do you use for first-cut decisions?

  13. #53
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,612
    Post Thanks / Like
    Likes (Given)
    2114
    Likes (Received)
    3893

    Default

    Quote Originally Posted by Milland View Post
    Do you have any general approach to servomotor selection? What do you use for first-cut decisions?
    This is a much bigger question then can be treated in a post. Lots of considerations generally speaking, the hobber is benign from an environmental standpoint. It is also a relatively low torque application and likely more concerning would be break away torque of the preloaded roller cam actuator (Sankyo).

    In this app I would look for compatibility of feedback options with the controller, then back EMF match for peak expected velocity, then maximum acceleration requirements for operation and E-Stop and finally expected max operating torque requirement summed with max accel torque using published actuator / gearbox efficiency plus some margin.

  14. Likes Milland liked this post
  15. #54
    Join Date
    Feb 2006
    Location
    so cal, usa
    Posts
    566
    Post Thanks / Like
    Likes (Given)
    193
    Likes (Received)
    135

    Default

    Electronic gearing is done every day, successfully. Think rigid tapping on a VMC.The spindle motor is usually not a servo motor, but does drive an encoder. You can rigid tap with very small taps, any "lagging" in the electronic gearing would produce lots of broken taps. Just an example of electronic gearing.

  16. #55
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    7,409
    Post Thanks / Like
    Likes (Given)
    365
    Likes (Received)
    6206

    Default

    Quote Originally Posted by drcoelho View Post
    .....
    - high resolution encoder on the Sankyo RCD rotary with feedback loop for rotary position
    Be aware that a high resolution encoder mounted on the slow output side will have cyclical errors due to mounting tolerances.
    If the track has any runout (and it will) there will be fast and slow areas.
    A Micro_E encoder can give you crazy high counts. They will not be accurate divisions.
    A larger body encoder helps reduce this error a lot so bigger is better. There is also another method of using two encoders and some fancy math, electronics or both.
    In high precision work the encoder is mapped in it's current mounted position to help with this problem.
    Cutting or grinding gears or even putting a nice .005 radius on a insert is not like rigid taping on a VMC.
    The latter is falling off a log dead easy. The former is a whole new and much deeper rabbit hole.
    Once inside this hole the devil in the details shows up.

    You have a lot of advice here , I wish you luck and it is doable.
    Be prepared for times in such a project that will almost drive you mad when everything was done "right" but the output is not.
    If this system runs correctly first time out you are a much better man than I.
    Bob

  17. #56
    Join Date
    Aug 2006
    Location
    New Zealand
    Posts
    1,232
    Post Thanks / Like
    Likes (Given)
    70
    Likes (Received)
    246

    Default

    Quote Originally Posted by motion guru View Post
    A harmonic drive has a nonlinear position output for a linear input due to the epicycloidal nature of the teeth climbing over one another as the flex gear is deformed by the off-center bearing drive. If additional gear eduction were required, this would not be a good choice if the gear blank needed rotation during the cutting process as it would result in a defective tooth form with finish patterns associated with the sinusoidal velocity ripple.
    I am curious. I am familiar with sinusoidal ripple, but not related to harmonic drives. Do you have any data or a reference to support this view?

    Dazz

  18. #57
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    7,409
    Post Thanks / Like
    Likes (Given)
    365
    Likes (Received)
    6206

    Default

    Dazz,
    As with encoders mounting error or runout of the generator is usually the biggest problem. You have to be very careful with this if designing from scratch.
    You get speedup/slowdown of the output.
    some more info here, it's a free download.
    https://ntrs.nasa.gov/search.jsp?R=1991001530


  19. Likes DanielG, motion guru, memphisjed liked this post
  20. #58
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,612
    Post Thanks / Like
    Likes (Given)
    2114
    Likes (Received)
    3893

    Default

    Quote Originally Posted by CarbideBob View Post
    Dazz,
    As with encoders mounting error or runout of the generator is usually the biggest problem. You have to be very careful with this if designing from scratch.
    You get speedup/slowdown of the output.
    some more info here, it's a free download.
    https://ntrs.nasa.gov/search.jsp?R=1991001530

    Bob - that article is the bomb! Nice find!!!

    And as previously mentioned - there is the encoder concentric mounting error that you need to account for as primary and this is the primary error of concern with your application because you do not need another gear reducer given the Sankyo 60:1 ratio out of the box.

  21. #59
    Join Date
    Aug 2006
    Location
    New Zealand
    Posts
    1,232
    Post Thanks / Like
    Likes (Given)
    70
    Likes (Received)
    246

    Default

    Quote Originally Posted by CarbideBob View Post
    Dazz,
    As with encoders mounting error or runout of the generator is usually the biggest problem. You have to be very careful with this if designing from scratch.
    You get speedup/slowdown of the output.
    some more info here, it's a free download.
    https://ntrs.nasa.gov/search.jsp?R=1991001530

    Hi
    I like it. Data dispels myths.
    Although this paper is 28 years old, the physics will still be the same.

    Hopefully manufacturers would have made some improvements over that time. I could not find any modern OEM error plots/data.

    I note the error 0.008 pk-pk = 0.004 pk error after millions of rotations.
    That would be a 1 bit error on a 90k rotary encoder.

    For many applications that error may be too much. For many applications, it may be OK.

    Dazz

  22. #60
    Join Date
    Feb 2017
    Country
    UNITED STATES
    State/Province
    California
    Posts
    350
    Post Thanks / Like
    Likes (Given)
    6
    Likes (Received)
    89

    Default

    As originally envisioned for this project, my goal is to equal the quality of an "old school" gear hobbing machine, not too worried about hitting state of the art CNC specs. I'd be curious what accuracy the older mechanical machines were able to hit for spur gears?


Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •