Does KmotionCNC simulator support trajectory planner? - Page 2


Page 2 of 2 FirstFirst 12
Results 13 to 19 of 19

Thread: Does KmotionCNC simulator support trajectory planner?

  1. #13
    Registered
    Join Date
    May 2012
    Location
    canada
    Posts
    331
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Hi Tom,

    Quote Originally Posted by TomKerekes View Post
    Well I would think 0.02 inches would be significant errors.
    Yes perhaps even a smaller value would be fine.

    Yes only the IIR Filter was applied (besides subsampling and recombination)
    Ok great, i wont worry too much about the run length average and 3 point filters.

    Yes I was thinking to control the amount of error would be most desirable and intuitive for the Operator. I was also thinking a constant feedrate for the entire path, so assuming ideal feedrates with no trajectory planning, distance and time are proportional. But I now see your point. I find this all very ironic on how we work so hard to assure the machine follows the exact path specified in a fully optimized manner with exactly controlled errors, but now you are sort of saying errors can be uncontrolled like machines with poor trajectory planning work.
    Well i guess there is a trade off between accuracy and speed. And in many cases a little filtering like this is desirable because it will allow things to run faster without making a whole lot of difference to the finished part. And of course we will be able to specify a tau value to decide exactly how much accuracy we need. There may also be times we want to avoid using this and follow the exact path. There is actually a G61.1 mode available on the industrial machines which is a point to point mode and this guarantees that the machine slows down enough to meet each point before moving to the next, basically disabling any smoothing. Our Fanuc and Mazak mills had this at my last shop, but by default this mode is disabled as it runs slower. I guess this g61.1 mode is similar to how Kmotioncnc is working now.

    So do you think we will be able to have the amount of smoothing proportional to feedrate? Makes sense to me to smooth more while roughing fast, and seems to be the industry standard. I'll download those files and play around with them. Thanks!

    Mark



  2. #14
    Gold Member TomKerekes's Avatar
    Join Date
    May 2006
    Location
    USA
    Posts
    2958
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Hi Mark,

    And of course we will be able to specify a tau value to decide exactly how much accuracy we need.
    Well except the accuracy will vary depending on feed rate. I was thinking of the Users that expect high accuracy at all speeds.

    There may also be times we want to avoid using this and follow the exact path. There is actually a G61.1 mode available on the industrial machines which is a point to point mode and this guarantees that the machine slows down enough to meet each point before moving to the next, basically disabling any smoothing.
    It is physically impossible to go exactly through each point without coming to a complete stop at each point. So this mode is usually totally unusable unless the Job is extremely simple (ie square).

    I guess this g61.1 mode is similar to how Kmotioncnc is working now.
    Well not really. In fact I almost take offense at this Many systems have only two modes. "Exact Stop" as described above - which is mostly unusable. Or "No Stops" which goes at feedrate through everything regardless of what errors are created - which is often also unusable. KMotionCNC allows you to specify the allowed error and plans the optimal motion that always meets the allowed error while honoring all acceleration and velocity limits of all the axes. Additional improvements could still be made, but please don't lump us in with those other systems.

    So do you think we will be able to have the amount of smoothing proportional to feedrate? Makes sense to me to smooth more while roughing fast, and seems to be the industry standard.
    That should be possible if it makes sense. Upsample each segment based delta time x Feedrate rather than delta distance. For now you can get the same results by "transforming" each segment portion of GCode separately based on feedrate.

    I'll download those files and play around with them.
    Thanks. I'd appreciate the feedback. I'd love to see a comparison with optimal Trajectory Planner values shown earlier compared with the Low Pass Position Filter.

    btw do you have any idea why the CAD/CAM system generated such bizarre GCode for what you seem to want to be a basically flat surface?



  3. #15
    Registered
    Join Date
    May 2012
    Location
    canada
    Posts
    331
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Quote Originally Posted by TomKerekes View Post
    Well except the accuracy will vary depending on feed rate. I was thinking of the Users that expect high accuracy at all speeds.
    Hi Tom, I think having accuracy vary with feedrate will actually provide greater accuracy overall. If we decide to use a certain LPF value to give good performance when roughing at 150IPM, and a little later were cutting at 30 IPM, it would be very easy for the machine to follow the path much more closely at that speed, so doesnt make sense to be using the same LPF value for all speeds. I think users would decide on an acceptable value to use at high speeds, and whenever your going slower then that, you know your following the original path much more closely.

    It is physically impossible to go exactly through each point without coming to a complete stop at each point. So this mode is usually totally unusable unless the Job is extremely simple (ie square).

    Well not really. In fact I almost take offense at this Many systems have only two modes. "Exact Stop" as described above - which is mostly unusable. Or "No Stops" which goes at feedrate through everything regardless of what errors are created - which is often also unusable. KMotionCNC allows you to specify the allowed error and plans the optimal motion that always meets the allowed error while honoring all acceleration and velocity limits of all the axes. Additional improvements could still be made, but please don't lump us in with those other systems.
    Sorry to offend you Tom. I've only ever tried g61.1 on a Mazak 510c II, and after doing some more reading, it looks like Mazak actually handles g61.1 differently. Mazak calls it "high accuracy mode" so probably a little different then the point-to-point mode as they call it on fanuc. When I tried it on the Mazak there would certainly be no stopping in g61.1 mode for each block, or slight angles or large radius curves, but it would slow down more then normal to avoid corner rounding on sharp corners, and follow the programmed path closer. Which seems to me to be like what we have now in Kmotioncnc. When in default g64 mode it would round corners significantly at very high speeds to help maintain programmed speed.


    That should be possible if it makes sense. Upsample each segment based delta time x Feedrate rather than delta distance. For now you can get the same results by "transforming" each segment portion of GCode separately based on feedrate.

    Thanks. I'd appreciate the feedback. I'd love to see a comparison with optimal Trajectory Planner values shown earlier compared with the Low Pass Position Filter.
    I'm trying to catch up on a few things over the next couple days, but ill try and do a test sometime over the weekend maybe. Ill take a smaller section of the code and cut three parts, and compare run time and part finish. Ill try my current trajectory planner settings as baseline, Ill try one with the LPF and Ill also try one with no LPF but increased tolerances in the trajectory planner like your first video. Any suggestion on values for the last one?


    btw do you have any idea why the CAD/CAM system generated such bizarre GCode for what you seem to want to be a basically flat surface?
    Sometimes you get weird solid models from customers and cam software doesn't do a great job with it. Not really sure what causes it, only happens occasionally.

    One last question. Do you think someday we could have this LPF added directly to the trajectory planner instead of having to use the transform feature? Ideally thats what id like to see. Be nice to have it built into the TP and active all the time, and just a few clicks away if you want to change the value to suit a specific job etc.



  4. #16
    Gold Member TomKerekes's Avatar
    Join Date
    May 2006
    Location
    USA
    Posts
    2958
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Hi Mark,

    Ill try my current trajectory planner settings as baseline, Ill try one with the LPF and Ill also try one with no LPF but increased tolerances in the trajectory planner like your first video. Any suggestion on values for the last one?
    I'd suggest the parameters used in the first video. Basically tolerances of 0.003 or 0.006.

    Do you think someday we could have this LPF added directly to the trajectory planner instead of having to use the transform feature? Ideally thats what id like to see. Be nice to have it built into the TP and active all the time, and just a few clicks away if you want to change the value to suit a specific job etc.
    Yes ideally that would be the goal. It may take a significant effort, make future changes more difficult, introduce bugs, require much testing, etc... So we would really like to establish the real value before committing.



  5. #17
    Registered
    Join Date
    May 2012
    Location
    canada
    Posts
    331
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Hi Tom, I ran a small section (1"x3") of this part today as a test. I ran my existing settings as a baseline (0.0003 co-liner and 0.001 corner rounding), second test was with 0.003 co-liner and 0.003 corner rounding, and third test was back to original settings and a 0.010" low pass filter applied. Results were as follows:

    Test 1 (baseline) - 3 minutes, 59 seconds. Best finish of the 3.

    Test 2 (0.003 co linear and corner rounding) - 3 minutes, 26 seconds. Largest error in surface.

    Test 3 (0.010 Low pass filter) - 3 minutes, 1 second. Some error in surface, small gouge at start of filtering (possibly my fault, see below).

    This was the total run time to cut the top surface only. Other features were machined beforehand. See pic for details on surface finish. Test 1,2,3 are from left to right. But overall the low pass filter seems to be useful. Reduced cycle time considerably, and considering this helps smooth sharp corners, i could probably increase acceleration rates as well since they have been limited to keep things running smoothly. So possibly even more gains to be had if acceleration could be increased. Increasing the co-linear tolerance and corner rounding also made a huge difference to cycle time, but seemed to give slightly more error to the surface then the low pass filter.

    Im fairly pleased with the finish from the LPF considering how bad the g code was to start. Im hoping this gouge wont be a problem in the future. I started the filter just as the tool hit the part, maybe i should have added a couple of lines and started it above part. I might experiment with this later if i get a chance.

    Edit: The gouge was my fault, i started the transform on a line with only the Z value. Adding the X and Y values to the first line and starting above the part fixed the problem

    Attached Thumbnails Attached Thumbnails Does KmotionCNC simulator support trajectory planner?-20181021_184959-jpg  
    Last edited by mmurray70; 10-21-2018 at 08:14 PM.


  6. #18
    Gold Member TomKerekes's Avatar
    Join Date
    May 2006
    Location
    USA
    Posts
    2958
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Hi Mark,

    Thanks! It looks like low pass filtering has value. What Trajectory Planner settings and KLP filter in KFLOP values were used in the 3 cases?

    This type of Low Pass filter is fast and relatively easy to implement as it only involves the previous filtered value and the next point. But it is asymmetrical and has a lag at the end that has to be handled somehow.

    I'm still wondering if the other filters (running average or Gaussian type) could work better.



  7. #19
    Registered
    Join Date
    May 2012
    Location
    canada
    Posts
    331
    Downloads
    0
    Uploads
    0

    Default Re: Does KmotionCNC simulator support trajectory planner?

    Quote Originally Posted by TomKerekes View Post
    Hi Mark,

    Thanks! It looks like low pass filtering has value. What Trajectory Planner settings and KLP filter in KFLOP values were used in the 3 cases?
    No problem Tom, thank you. See attached pic for trajectory planner settings. These settings were used for baseline and the low pass filter tests, colinear and corner rounding were changed to 0.003 each in the second test. Low pass filter in Kflop is set at 0.020 in all three tests.

    This type of Low Pass filter is fast and relatively easy to implement as it only involves the previous filtered value and the next point. But it is asymmetrical and has a lag at the end that has to be handled somehow.

    I'm still wondering if the other filters (running average or Gaussian type) could work better.
    Im not too sure about the differences between different types of filters. A quick google search shows a bunch of data thats a little over my head, but seems like the end result is somewhat similar from all of them, so maybe either could work. Maybe somebody else who understands this better could offer some advice.

    But honestly it seems like whatever your using now in the this transform feature is working very well. Seems to be similar to the smoothing ive noticed over the years on the fanuc and mazak mill I used to run, with the exception of those machines smoothing more proportional to speed as we discussed earlier. If you think this is fast and easy maybe you should just stick with it? Im guessing the lag will be small, maybe a straight move from the last filtered position to the actual position in g code could work? I dont think this will be a big deal as it should be a very small amount and people will be using lead in and out anyway on anything important.

    Attached Thumbnails Attached Thumbnails Does KmotionCNC simulator support trajectory planner?-tp-jpg  


Page 2 of 2 FirstFirst 12

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
  •  


About CNCzone.com

    We are the largest and most active discussion forum for manufacturing industry. The site is 100% free to join and use, so join today!

Follow us on


Our Brands

Does KmotionCNC simulator support trajectory planner?

Does KmotionCNC simulator support trajectory planner?