Toolchanging reliability

Discussion in 'Getting Started' started by Andy Cohen, Sep 22, 2019.

  1. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    I am finding that the tool changing has very low reliability. It looks to me as though there is not enough voltage on the X or Y steppers as well as the C stepper. A slight path inhibition and X, Y or the tool lock is out of register.
    Every time I try to go from T0 over to T3 it's a crap shoot. I wonder if it's the homing method?
     
  2. Amr

    Amr Well-Known Member

    Joined:
    Jun 2, 2019
    Messages:
    130
    Likes Received:
    30
    There must be some thing wrong with your setup, I have been using the tool changer for almost a month now, with the exception of a few hickups due to me being eager to start using it and not patient enough to go through the setup guide. other than that all is great so far with overnight prints and consistent tool changes.
     
  3. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,546
    Likes Received:
    481
    You must arrange for their to be no pathing issues. I also had issues with T0, T2 reliability before I realised that the Duet will move from absolute to tool coordinates. Make sure when you dock to pull back to at least 150 so that when the system moves to absolute coords again you don't go somewhere that will make the next move impossible.
     
  4. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    It's 50/50. It doesn't matter which dock. Slowing it down has made zero difference. Sending the head in to on Y 243 did nothing. The amp settings to the motors are way higher then I have on my other Duet machines so it is not that. Only thing I can think of is the senseless homeing.
     
  5. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    There are no physical path issues. When it travels every now and then it goes out of register just a tiny bit. It goes into the dock and every now and then it will not go in all the way. When leaving the tool in the dock every now and then the lock does not turn all the way back sometimes pulling the tool out or causing a crash into the next tool. There's nothing inhibiting the T lock inside the tool. and yes it is lubed. Each dock was "adjusted" per the guide. It is simply not working reliably. If I run a file which picks up a tool, parks the tool and then picks up the same tool over and over, it works fine for that tool. Move to another tool and it will miss. I will be installing limit switches next.
     
  6. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    I persevered and maybe I found the problem... I sure hope so... I did per Spoon Unit's suggestion and moved the T bar using the move control on the PanelDue so that it was in the ideal location in X with the T bar right in the hole and checked what was used in the sys files. Whoops. X was off by .5.
    Then to fix it took changes in both the tPost and the tFree files... in EVERY line (I had missed the last lines on both files).
    It fixed an issue I had at T3, now I will do it for the other 3 tools and see if reliability improves.
    By the way... Thanks to Spoon Unit and Amr for their posts.

    I was #86 on the list. Where are all the others???
     
  7. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    Nope.
    I found a some very small discrepancies in Y which I changed, but the result was the same. When it travels from T0 to T3 it is not accurate enough. VERY discouraging.
     
  8. Amr

    Amr Well-Known Member

    Joined:
    Jun 2, 2019
    Messages:
    130
    Likes Received:
    30
    Can you explain what happens when it fails to pick up the tool, and if this is only to T3 or is this on all tools including T0?
     
  9. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,546
    Likes Received:
    481
    I really need to mount the final tool. I've been putting it off. Until then I can't comment on whether mine will do the same.

    Other things to consider
    • Are you running with enough ampage? I suspect you are as I've found that 1.2A is enough for no lost steps on X, Y moves.
    • Are you experiencing something similar to @Thomas Kamsker who had loose grub screws somewhere in the X, Y motion system.
     
    #9 Spoon Unit, Sep 23, 2019
    Last edited: Sep 23, 2019
  10. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    To test I use the console and randomly insert T0, etc making the toolhead go from one to another. I do it one tool at a time and check on each pickup by hand to make sure that the tool is captured firmly. I have each dock's position carefully dialed into the config files and have verified them repeatedly. I've been able to improve reliability but I find that as I increase the sample size I still get a miss. What's a miss? A fail to release. I stop and I remove the tool by hand by shifting it slowly so as not to move anything. Invariably it's ALWAYS been the T lock which did not return to the complete horizontal position. Why wouldn't it rotate all the way? Not sure. Guessing... it could be the C stepper does not have the rotational torque, it could be that the head did not travel accurately enough, it could be that missed steps in X or Y that slowly increase the error until it's out of tolerance.

    One thing for sure, When I home in X I hear the steppers slip 2 to 3 steps when the carriage crashes on the left side. It's NOT .5 to 1 step per the Duet docs. It's 2 or 3. That CAN'T be good for X axis accuracy and being off a little in X is all that's needed to miss the position at a dock.

    On happier note... I did test prints on each tool and the results were optimal!
     
  11. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    The C stepper is at the default 400 mA. I guess I could try increasing it to see what happens!
    Checked the gantry before I first powered up. It's good. I get sweet print quality, just like my CloneR1 and Craftbot. Gotta love CoreXY!
     
  12. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    BTW... there's quite a bit more TC build activity on FB (unfortunately). Seems quite a few others are seeing the exact same thing and are also blaming the sensorless homing.
     
  13. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,546
    Likes Received:
    481
    I've already added the microswitches for physical stop homing. Just the C left on sensorless, and actually that's not really even sensing. It's just going on forever, then giving up and coming back, which is fine and workable. Definitely a better result than a failed sensorless system which detects a stall too early.

    Are you homing C between tool pickups? If so, why?
     
  14. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    No, I'm not homing between tool pickups. I test by entering T0, T3, T1, T2, over and over randomly. There was 1 time it did not miss after about 20 minutes of it... But when I tried again it missed. So it would seem to me that there is an issue of how reliable as well as accurate the senseless homing is since the only difference between those 2 tests was I had homed all the axes. ANY inconsistency in the homing is problematic for the hard coded dock locations.
    I have my switches, holders and associated cabling ready... now if only it wasn't so hot outside in the garage...

    So here's a question... Going to hard limit stops... Should we remove (or at least rem-out) the line for X and Y Stall Detection?
     
  15. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    Update... Limit stops installed

    Holy CRAP what a diff!!

    100% reliability!
     
  16. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,546
    Likes Received:
    481
    Ahhh. Reliability on positioning for your T3 pickup now? That would make sense right. Sensorless homing only has to move a little bit extra or less to knock things out. Physical stop position is absolute (as long as it can't budge, which I think is fine despite the single screw due to the small inset into the extrusion.

    I removed drops in current and set up for stall detection.
     
  17. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    I wonder if there's value in keeping the detection. Do alerts show in the gcode console? If so it would come in handy.
    I get reliability now forpickup and release on all 4 tools. Finally ready to complete the hot end offsets, polish the start gcode, mount the paneldue and start printing!
    Btw, your post on going in and verifying and resetting the doc locations in the tFree and tPost files should be added to the Commisioning guideline.
     
  18. blarbles

    blarbles Well-Known Member

    Joined:
    Aug 10, 2019
    Messages:
    102
    Likes Received:
    64
    I added limit stops a bit ago to my TC, it is definitely the way to go.
     
  19. Andy Cohen

    Andy Cohen Well-Known Member

    Joined:
    Aug 23, 2019
    Messages:
    217
    Likes Received:
    57
    IMO, E3D would be smart if they simply incorporated the Limit stops into the production design. I am a bit surprised this issue was not worked out during the beta run.
     
  20. dc42

    dc42 Well-Known Member

    Joined:
    Aug 16, 2016
    Messages:
    458
    Likes Received:
    206
    I haven't yet run my tool changer in anger (due to Duet 3 firmware work), so most of the following is speculation. But here goes:

    Stall detection is only accurate to the nearest motor full step, assuming you don't use the M915 F1 parameter (4 full steps if you do). With the long belt lengths used on CoreXY machines, it may be more. If you are lucky, the stall may happen at the same full step every time. If you are unlucky, you may be "on the cusp" and stall detection may occur on one full step or one (or a few) later. It might be temperature sensitive, because belts and aluminium expand at different rates. One full step at 100 steps/mm based on x16 microstepping is 0.16mm.

    This leads me to the view that it's probably a good idea to use an endstop switch for the X axis, because the X placement of the head for tool pickup and release is critical. But it's probably OK to use stall detection for the Y axis, because a 0.16mm or 0.32mm change may not matter when picking up or parking a tool.

    That said, E3D and others are evidently getting reliable tool changing using stall detection homing.

    If you want to use the resume-after-power-fail functionality, you should use endstops switches for both axes, as recommended in the Duet3D wiki. [But resume-after-power-fail only works if the print does not need bed heat to retain adhesion.]
     
    #20 dc42, Oct 4, 2019
    Last edited: Oct 4, 2019

Share This Page