1. Got a question or need help troubleshooting? Post to the troubleshooting forum or Search the forums!

Auto Tuning without damaging Flexplate System

Discussion in 'Mods and Upgrades' started by Lance Weston, Apr 3, 2020.

  1. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    The FlexPlate System loses it's magnetism at temps above 70C. Robo Auto tunes at 80C. I have found that auto tuning the Bed at 50C gives better tuning values than at 80C. I found that auto tuning the extruder at 150C gives better results than at 240C. The Marlin code has only one smoothing constant ( K1 @ 0.95 ) that it uses for both the extruder and the bed.

    I have modified the Marlin code to give separate smoothing constants ( K1bed ) and ( K1xtrdr ) I found that K1bed tuned best at 0.855 and K1xtrdr tuned best at 0.96. I modified the code to intercept the Robo tuning instruction and change the Bed temperature to 50C and the extruder to 150C.

    The lower temp on the Bed allowed me to auto tune some beds that did not tune at the higher temp.

    This link has the code, the modified hex file and the original hex files for the R2, C2 and R2 dual. To load the hex file: put the hex file on a USB drive and plug into the R2. On the LCD display go to files -> USB then load.

    https://drive.google.com/open?id=1mns7X2dDRph8n8hS5YT-Ob3mhnQbt6Z1

    The pictures below show the tests I performed to get to these values. I also found that some effects do not take place until 15 minutes into the print as shown in the pictures.

    It is just that easy to restore to the original code.
     

    Attached Files:

    #1 Lance Weston, Apr 3, 2020
    Last edited: Apr 4, 2020
    mark tomlinson likes this.
  2. tkoco

    tkoco - -.- --- -.-. ---
    Staff Member

    Joined:
    May 7, 2018
    Messages:
    721
    Likes Received:
    273
    I'm missing something here. Autotune? Are you referring to the PID wizard?
     
  3. Lance Weston

    Lance Weston Active Member

    Joined:
    Jun 2, 2018
    Messages:
    664
    Likes Received:
    230
    Yes, I am referring to the PID wizard. The wizard raises the bed to 80C and the extruder to 240C. By moving the wizard to the middle of it's range I get values that give me more constant temperatures. When moving to the Flexplate system, particularly with a glass bed, the mass of the Bed is increased a lot. One of my R2's would not auto tune with the wizard. I began to investigate. Moving the wizard temp down to 50C allowed the PID wizard to complete. Failure of the wizard is not uncommon. The wizard uses a smoothing constant K1. K1 is a value between 0 and 1 for smoothing the sampled data points, where 0 looks at more data and 1 looks at very little. 0 would be used for very stable systems. The Marlin software uses a K1 of 0.95 for both the Bed and Hotend.

    I have modified to code to do a couple of things. Break out K1 as two separate variables K1bed and K1xtrdr. I now can ( tune ) select the number of data points for each independently. As you can see from the pictures the value of K1 ends up to be different between bed and extruder and at 0.93 the wizard could not auto tune the bed. I also modified the PID command code to intercept 80C for the bed and then send 50C and 240C for the extruder and send 150C. These PID ( tuning ) temps give more stable PID constants than the tuning at the higher temps.

    The only down side is that if you issued a manual PID tune "M303 E-1 C10 S80" from the terminal to auto tune the bed it would use 50C instead. You could use 81C or 79C, but 80C can not be had. Same goes for "M303 E0 C10 S240" for the extruder. Use either 239C or 241C, any value but 240C.
     
  4. tkoco

    tkoco - -.- --- -.-. ---
    Staff Member

    Joined:
    May 7, 2018
    Messages:
    721
    Likes Received:
    273
    Thank you for explaining the situation. I was not aware of these issues.
     

Share This Page