584,854 active members*
4,254 visitors online*
Register for free
Login
Page 9 of 11 7891011
Results 161 to 180 of 202
  1. #161
    Join Date
    Jan 2005
    Posts
    15362

    Re: Jerk Control in machines

    Quote Originally Posted by peteeng View Post
    Hi MT - I agree that particular video was not good. But if you look at the Tinyg closer its algorithms are very impressive and can be implemented in other systems as they are open source. The TinyG approach is what I expect best practice open loop motion to be. To go further, closed loop approaches need to be used and these are appearing in 3D printing systems at the cost level that I operate at. The big machine players have proprietary open and closed systems that do amazing things in terms of motion control but they are outside the scope of my machines. Milli my first mill will use servos which is a new world for me.... Peter

    In a slight tangent there are a couple of old threads in which flywheels and eccentrics were added to steppers to improve motion (by way of mechanical damping I expect). I can't find if anyone concluded that this was purely a flywheel effect and the eccentrics (small free weights in cavities in the flywheel) where not needed or the eccentrics were functional? Anyone comment on that? Peter
    Yes, I remember the weighted wheels that were attached to the stepper motors, it's like everything people try to make things work better, but there are tradeoffs so what appears to work well, may not be as good as it seems, or it would have stuck and been implemented to the stepper motor, or someone would have been manufacturing them as an add on.

    There are servos that do not work well, so when you go that route careful what you chose.

    Any quality Ac Servo Drive is closing the loop, this does not have anything to do with the control, the control only has to send a signal for it to move, Servo Drives are very sophisticated.
    Mactec54

  2. #162
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi All - Starting the think about the next build and this is what I shall be using as the controller. This is the first time I have seen centripetal accel discussed. Peter

    https://www.youtube.com/watch?v=d68U80HeKc4

  3. #163
    Join Date
    Jan 2005
    Posts
    15362

    Re: Jerk Control in machines

    Quote Originally Posted by peteeng View Post
    Hi All - Starting the think about the next build and this is what I shall be using as the controller. This is the first time I have seen centripetal accel discussed. Peter
    Thats a big improvement for a setup like this but does not come close to adaptive jerk control.
    Mactec54

  4. #164
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi Martec - Please explain what you mean by adaptive jerk control? By nature all motion control in machines is "adaptive" there has to be some look ahead done and some change (adaption) of motion (path, velocity or accel) to make it smoother yet maintain the set path tolerance. Peter

    image summary
    ideas - describes in words what is happening in many controllers
    adaptive - shows a typical calculation flow chart
    jerk limit - shows the screen for Buildbotics path jerk limit setting
    centripetal accel - shows Buildbotics screen for setting the centripetal accel which they call junction accel. Not sure why they call it that, I have asked Doug about that, waiting for a comment...

    Buildbotics do comment that centripetal accel can be dominate and override the path accel (tangential accel). As centripetal (radial) accel is V(squared)/r the velocity can dominate or as R is the divisor if it is small it can create large accels. The vector sum of the tangent and radial accels has to be calculated and limited to provide smooth motion ie limit inertial forces. This is done by smoothing the path within the tolerance to "control" the accels and jerk.... its not simple math!!

    Some advanced systems don't just look ahead and adjust once they run forward and back through the path smoothing several times. This maybe what you are calling adaptive? Peter

    ARggghhhh!! Buildbotics have got back to me and they are out of stock due to the microprocessor shortage and don't know when the next run will be available! bummer!! I'll be patient and look at the 3D printer controllers a bit more...

  5. #165
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi All- I think the 3D printer controllers have gone past the router/mill controllers with input shaping ie jerk control via accel feedback loop.

    https://www.youtube.com/watch?v=er7q-CJL1lc

    I'll have to try to get the one tree octopus to run as a router controller... Peter

  6. #166
    Join Date
    Jan 2005
    Posts
    15362

    Re: Jerk Control in machines

    Quote Originally Posted by peteeng View Post
    Hi Martec - Please explain what you mean by adaptive jerk control? By nature all motion control in machines is "adaptive" there has to be some look ahead done and some change (adaption) of motion (path, velocity or accel) to make it smoother yet maintain the set path tolerance. Peter

    image summary
    ideas - describes in words what is happening in many controllers
    adaptive - shows a typical calculation flow chart
    jerk limit - shows the screen for Buildbotics path jerk limit setting
    centripetal accel - shows Buildbotics screen for setting the centripetal accel which they call junction accel. Not sure why they call it that, I have asked Doug about that, waiting for a comment...

    Buildbotics do comment that centripetal accel can be dominate and override the path accel (tangential accel). As centripetal (radial) accel is V(squared)/r the velocity can dominate or as R is the divisor if it is small it can create large accels. The vector sum of the tangent and radial accels has to be calculated and limited to provide smooth motion ie limit inertial forces. This is done by smoothing the path within the tolerance to "control" the accels and jerk.... its not simple math!!

    Some advanced systems don't just look ahead and adjust once they run forward and back through the path smoothing several times. This maybe what you are calling adaptive? Peter

    ARggghhhh!! Buildbotics have got back to me and they are out of stock due to the microprocessor shortage and don't know when the next run will be available! bummer!! I'll be patient and look at the 3D printer controllers a bit more...
    Note their demo was cutting air, no changing load, they are only setting controlling motor parameters, (lack of microprocessors)

    Controlling the changing inertia acceleration / deceleration, with every move, is what adaptive control is, it is superior to anything that you have posted.


    To simplify tuning while still allowing highest versatility and control over servo behavior, we have developed Adaptive Tuning technology, meaning relative to the inertia load, the servo adapts a wide margin of permissible parameters that will ensure stable control. This allows to easily select the settings for smoothest motion, and fastest response.

    The margin of stability being adaptive is critical for many applications. For example, in machine tool, the dynamic load inertia can drastically change. The load inertia ratio is different at acceleration/deceleration, constant traverse and workpiece processing. The same is true for robotic applications. By maintaining a wide region of stability, the DMM adaptive tuning provides seamless performance transition between these events. Even with the same parameter settings, as the load inertia ratio change, the servo is still well within the stable region.

    Second generation tuning algorithm further improves smoothing with wider frequency range leading to wider domain of inertia load capability.
    Mactec54

  7. #167

    Re: Jerk Control in machines

    To modify acceleration curve from trapezioidal to s-curve on single Axis is not a good thing.
    Better to apply S-Curve to interpolated TCP Point (result of all-axis acceleration quantum quantized to what an axis have to move).

  8. #168
    Join Date
    Jun 2015
    Posts
    4131

    Re: Jerk Control in machines

    hy pete, after insistong soooooo much on splines, if you wish, please take a look at how different interpolation methods can satisfy same machining error

    a perfomant cnc can output different interpolation methods, balancing accuracy vs speed / kindly

    www.youtube.com/watch?v=FlieT66N9OM
    Ladyhawke - My Delirium, https://www.youtube.com/watch?v=X_bFO1SNRZg

  9. #169
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi Mactec - Then the Klipper/Bigtreetech is adaptive as it changes the velocity & accel on the go via feedback from the accelerometer. The DMM maybe superior but at a high cost point. As I have said before I build machines at a Maker/DIY level. High performance servo systems with position/vel/accel feedback feed forward are outside the scope of my environment. I expect that the accelerometer system can do as well as the DMM system as the algorithms and basic physics are the same. It may even be better as it includes the whole mechanical system response vs the complex feedback/calculation requirements of the servo type systems. Adaptive, input shaping are words to describe using the second differential of position to create a smooth path to minimise jerk (or jounce, depends were you were brought up) I'll keep looking. Peter

    Hi DK - The video explains what the motion controller is doing. If the path strategy is about minimising curvature (ie max second differential of position) within the "track" ie the machining tolerance then this would be the smoothest least jerk path possible. The trick however is that many paths have transitions and if you max the curvature around one curve this may not max the curvature around the next curve. That's why different race car drivers have slightly different race lines. Their car and personality are slightly different, same as machines. So optimising transitions is a bit of a fuzzy area. The maths is straightforward as the racetrack example shows, the implementation can be a bit tricky. Peter

  10. #170
    Join Date
    Jan 2005
    Posts
    15362

    Re: Jerk Control in machines

    Quote Originally Posted by peteeng View Post
    Hi Mactec - Then the Klipper/Bigtreetech is adaptive as it changes the velocity & accel on the go via feedback from the accelerometer. The DMM maybe superior but at a high cost point. As I have said before I build machines at a Maker/DIY level. High performance servo systems with position/vel/accel feedback feed forward are outside the scope of my environment. I expect that the accelerometer system can do as well as the DMM system as the algorithms and basic physics are the same. It may even be better as it includes the whole mechanical system response vs the complex feedback/calculation requirements of the servo type systems. Adaptive, input shaping are words to describe using the second differential of position to create a smooth path to minimise jerk (or jounce, depends were you were brought up) I'll keep looking. Peter
    No not at all, unless it is using a closed loop system between the motor and the Drive, it can't really do much of anything, I'm sure they can make the control execute smoother code, which would help some, the only way is to control the inertia acceleration / deceleration, while it is running the program, the DMM Adaptive control was developed for Heart / lung machines, used for severely injured lungs, it is the only true Adaptive control available in a servo drive unless licensed to another manufacture, there are hundreds of hobby users using DMM servo systems so not out of reach for hobby users today, a lot don't get the full benefit of a servo system because the machines are not built that well.
    Mactec54

  11. #171
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi Mactec - OK I have done some more reading and digging. The input shaping functions more like a damper on a car wheel (commonly called a shock absorber) it initially uses an accelerometer to test and calculate the spectral frequency response of the structure. Once this spectral response is known a suitable damping factor can be built into the motor signal, same as you can get different damping ratio motion dampers. You can also get dynamic dampers that have adjustable damping ratios. I suspect that's what's happening here with the algorithms . In Klipper there are a few damping strategies that can be selected. So its not an active control or an adaptive control. But it is a valid strategy to improve smoothness/decrease jerk same as a automotive wheel motion damper provides.

    see convolving - https://en.wikipedia.org/wiki/Convolution

    'I assumed that since the accelerometer was mounted permanently on the spindle in the articles I looked at it was actively involved in the motion control. Since then I have found it is only used when recalibration of the system is needed...

    https://all3dp.com/2/klipper-input-s...ply-explained/

    The search goes on. I shall use the bigtree/klipper combo on the printer I'm building to get some experience with this approach.... I have found a local supplier that seems to have good knowledge/ support. So I'm setting up a question list for them at the moment. Considering the board cost $125AUD and the 8A 60V ( up to eight drivers per board) drivers cost $40AUD each seems its good value compared to mill/router type controllers etc. Peter

  12. #172
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi All - I have been speaking to the local 3D printer place and they have given me ideas. The duet3d drivers have input shaping and have a good CNC "ecology" as they say. D3D can use Marlin or Klipper. More digging to do. Peter

    Duet 3D

  13. #173
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    more info on shapers - Peter

  14. #174
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi All - Looks like Marlin has had s-curve accel in development , but I think shaping is better. Sorting some accelerometers at the moment. Peter

    Search · s-curve · GitHub

  15. #175
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi All - Ard found a good video on damping in another thread from Rexroth. Here's a clip for the video showing the feedback loop. The feedback goes into a PID filter and the command signal gets modified depending on the settings. The feedback can be acceleration or position. It can also be sensorless and this would use an input shaper type function. peter

    https://youtu.be/eikPy2vJKyI

  16. #176
    Join Date
    Jun 2015
    Posts
    4131

    Re: Jerk Control in machines

    as high cnc's have allready their loops, rexroth senzors are add-ons to such machines ( for complementary tasks ), but more likely they are used for less automatized systems

    feedback can be acceleration or position
    or sound, or pressure ... whatever type of senzor there is



    jerk control = jerk + control = motion change + control = planer + mcs

    for hobby, is a great jump from linear to s acceleration, and this makes one focus only on start and end accelerations

    a high end machine modulates acceleration based on curvature, and may use not only linear and s, multi stage deceleration, so it's focus is not only start&end, but continuously; this are functions for die-molds



    as stated from begining, the answer is data from feedback loop for example, for that 3d pritner of yours, if you could map the head accel data with position data, you would be able to fine tune it / kindly
    Ladyhawke - My Delirium, https://www.youtube.com/watch?v=X_bFO1SNRZg

  17. #177
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi DK - Indeed that's what I shall do. But you don't have to map accel with position only correct velocity at known vibration zones. Peter

  18. #178
    Join Date
    Jun 2015
    Posts
    4131

    Re: Jerk Control in machines

    hy pete, i was not reffering to changing acceleration, thus not to what to do with recorded data, but, for example, for that 3d printer, as to record data from accelerometer senzor, togheter with axis position, because only accelerometer data alone is hard to interpret - is needed to relate it to something

    the loop is reactive upon reading, regardless of what kind of loop is, or what process/machine is there

    reading data involves sampling averages, that form paterns, and tuning the loop, involves a reaction upon each patern ( there is no clue for the machine that it is actually vibrating, but it can detect an abnormal patern; is it vibration ? who knows )

    such functions are not open to client, being more machine builder specifics, but it can allow some partial control over it

    is possible to record data from systems without loops, but is not only recording, but also processing, because, like this, a result is created and initial data is deleted ( or large portions of it ), otherwise, with those high frequencies, one would end up with an enormous amount of data

    unfortunately, i don't know what such tools are there at hobby system

    Indeed that's what I shall do
    i don't know what you are after so far, it seems to me that going with S curve capable controllers is a nice step

    if the controoler does not allow too much acces, you can use some senzors over your construction and 'check' how is it performing ?!



    your entire idea with 'jerk control' can be put in a larger area, like stability, or how much to push a system until it deteriorates

    define a method to control, to quantify the 'thing' you are after; within buget

    for example :
    ... there are cases where senzors are used to detect bending, so to avoid cnc machine wear, simply because the closed loop can not detect such a case
    ... or bridge deformation under stress ... or helicopter blades twist ( rexroth - siemens )
    ... or to detect sound, checking versus a baseline, for wear detection

    i have seen that you focused a lot on jerk, but for expensive cnc's, smoothnes is much more than jerk and splines, so to say it; you can define smoothness as long as loop deviation is tolerated

    somehow the discusion is a mix, balanced between hobby and profesional systems

    if i was to decide on control for building a machine with reasonable costs, i would contact all such guys that are into this, so to bring my self to-date fast with what is there; i don't know, maybe i am wrong, but my intuittion says that s curve capable hobby controlers is not a new thing / kindly
    Ladyhawke - My Delirium, https://www.youtube.com/watch?v=X_bFO1SNRZg

  19. #179
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines

    Hi DK - The 3D printer industry have gone well past s-curve velocity profiles and use input shaping to improve machine vibration using electronic damping. For my level of technology and economy input shaping is the one to pick for CNC machines. It provides nearly all of the performance of full active control in a simple implementation. This thread and investigation has covered a lot of territory and there's not much more to investigate. My overall comment is that the Maker level CNC tech is far behind the 3D printer tech. Things like Klipper and Marlin are very attractive and the printer hardware is cheaper and better developed then the CNC boards.

    The next generation of CNC builders will use Arduinos, 3D printer boards and Klipper etc as they will have been exposed to all of that at school and its straight forward to use and cheaper then CNC stuff. There are several 3D printer retail shops within 2hrs of me, there are no retail desktop router shops anywhere. So the printer market is well developed and will continue to develop as it has the money and customer support to do this. So printers now have input shaping and in the future will have full active control of motion standard. Mach4 does not make a motion controller so they will lose momentum , their machine motion control is decades out of date. UCCNC has the advantage of making their own motion controllers. These need to be integrated and updated to input shaping and then they would be hobby market leaders. There are several people working on CNC style GUIs for Klipper and Marlin , once this happens they are the pick as then there will be no major front end difference but there will be a large performance difference.

    So anyone starting to design a machine now I suggest they look at Big Tree Tech control boards and Duet3D boards they offer better tech and better $$$ for the people interested in this sort of thing... Peter

  20. #180
    Join Date
    Jul 2018
    Posts
    6253

    Re: Jerk Control in machines


Page 9 of 11 7891011

Similar Threads

  1. Reversing direction causes motors to jerk
    By Despontene in forum Gecko Drives
    Replies: 20
    Last Post: 01-10-2020, 01:49 AM
  2. Reversing direction causes motors to jerk
    By Despontene in forum DIY CNC Router Table Machines
    Replies: 2
    Last Post: 12-29-2019, 08:43 AM
  3. cnc jerk controller
    By rcmaniak in forum Linear and Rotary Motion
    Replies: 1
    Last Post: 06-21-2018, 12:30 AM
  4. I+j arc jerk issue driving me nuts
    By Jon.N.CNC in forum UCCNC Control Software
    Replies: 31
    Last Post: 06-14-2016, 12:03 AM
  5. A few questions : jerk, offsets, hotkeys .
    By Abarmot in forum Dynomotion/Kflop/Kanalog
    Replies: 7
    Last Post: 03-13-2013, 04:51 PM

Tags for this Thread

Posting Permissions

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