Notifications
Clear all

Tired and indignant  

Page 7 / 10
  RSS
jonathon.b
(@jonathon-b)
Estimable Member
Re: Tired and indignant

Brigandier,

there's still more commits and merges happening now. seems like its started moving again, can you let me know what your test prints come out like please after you compile it etc.

Publié : 28/03/2018 9:04 pm
Brigandier
(@brigandier)
Reputable Member
Re: Tired and indignant


Do you think its the stepper bumping it up to 256 that's causing the problems? I was thinking about trying it with interpolation disabled on the E axis only.

reading online it does say it shouldn't cause any problems but id try anything to get it sorted right now.

Pure speculation on my part, but I suspect it's due to the stepper having increased resistance the closer it is to a full step. Consider this to be one step:

  • Start of step, very hard to move from, lots of resistance. If tuning is off, the step may not start in time and "jump" ahead when it finally breaks over,

  • Middle of step, likely the easiest area to move,

  • End of step, lots of pull, needs good tuning to keep the stepper from snapping to that full step spot.

  • Brigandier,

    there's still more commits and merges happening now. seems like its started moving again, can you let me know what your test prints come out like please after you compile it etc.

    Yep I am still a few hours out from being able to. Usually when I get home their waves of commits have stopped. 🙂

    My MK3 Parts: [Bowden] [New Shoes] [TPU Micro Springs]

    Publié : 28/03/2018 9:22 pm
    Kwaad2
    (@kwaad2)
    Honorable Member
    Re: Tired and indignant

    Whoa Whoa Whoa Whoa... Wait.

    It was set at 64 microsteps? There's the problem.

    I was discussing at MRRF2018 with someone that more steps are not always better.

    EG: 1.8° motor has a LOT more torque than a 0.9° motor. A 1.8° motor, with 8 microsteps, with have similar torque to a 0.9° motor with 4 microsteps.

    This is no where near accurate, but basically

    Let's say 200 steps per revolution = 1ft/lb of torque. (totally hypothetical and not a real number, but just follow along)

    200 = 1
    400 = 0.5 (2 microsteps)
    800 = 0.25 (4 microsteps)
    1,600 = 0.125 (8 microsteps)
    3,200 = 0.0625 (16 microsteps)
    6,400 = 0.03125 (32 microsteps)
    12,800 = 0.015625 (64 microsteps)
    25,600 = 0.0078125 (128 microsteps)
    51,200 = 0.00390625 (256 microsteps)

    The issue we see here, is when we start microstepping, ESPECIALLY on the extruder which has to work VERY hard to extrude the filament, the more microsteps we get, the "weaker" the motor gets! I think the extruder on the Mk3 is a disaster personally, BECAUSE IT IS DIRECT DRIVE.

    If there was a simple 10:1 gear reduction, the filament would move 1/10th as much per step. This would improve your ability to print in two ways.
    1. 32 microsteps (6400 steps per revolution) would extrude LESS filament than running at 256 microsteps without gear reduction. Meaning you can keep the microsteps bigger for MOAR POWAH!!!
    2. Because it will be 10x "easier" to extrude each microstep, you can now run at 256 microsteps, instead of 32 microsteps. So effectivly, we can change the number of steps per bondtech revolution from...

    3,200 to 512,000, with SIMILAR torque at the bondtech gear.

    I'm sure my math is all kinds of jacked up. Please ignore that, but just understand the concept.
    More steps = less power
    Skipped steps = problems
    More power = less skipped steps.

    Until proven otherwise, I don't care what Jo says. Gear reduction > Direct Drive.

    Hi, I'm Sean. I used to work on CNC machines.
    I try to not make mistakes, but the decision is YOURS.
    Please feel free to donate to my filament/maintance fund.

    Publié : 28/03/2018 9:25 pm
    john.n13
    (@john-n13)
    Estimable Member
    Re: Tired and indignant



    Like these pictures?

    Are you going to tell us how you achieved those results?


    I’m not to sure what you mean, I didn’t make the prints so crap it’s was the mk3 that did that lol

    Sorry not to be clear - I wasn't commenting on the quality. I'm lost on "who's testing what" - so what FW version, what slicer, any gcode mods?

    I'm just starting with my MK3, calibrated and tested as best I can determine. Now I want to optimize the print to the degree possible and determine "what I have" from a quality point of view.

    Best looking test cubes I have so far are default S3D - less X and Y axis ringing and moire but slightly more corner "bulging" and definitely more unevenness than default slicr3 - on FW 3.1.3-45. So, I want to see what options I have to test improvements and I'm asking this question to you to determine if I'm at a good place to move forward or already way behind on a firmware version, for instance. Your corners didn't seem to have any significant bulging.

    I'm still researching communicating with the installed firmware and other related topics. Not ready yet to say I can't find the information I think that I need - but I'm guessing I'll be back for confirmation at least.

    Publié : 28/03/2018 9:42 pm
    jonathon.b
    (@jonathon-b)
    Estimable Member
    Re: Tired and indignant



    Are you going to tell us how you achieved those results?


    I’m not to sure what you mean, I didn’t make the prints so crap it’s was the mk3 that did that lol

    Sorry not to be clear - I wasn't commenting on the quality. I'm lost on "who's testing what" - so what FW version, what slicer, any gcode mods?

    I'm just starting with my MK3, calibrated and tested as best I can determine. Now I want to optimize the print to the degree possible and determine "what I have" from a quality point of view.

    Best looking test cubes I have so far are default S3D - less X and Y axis ringing and moire but slightly more corner "bulging" and definitely more unevenness than default slicr3 - on FW 3.1.3-45. So, I want to see what options I have to test improvements and I'm asking this question to you to determine if I'm at a good place to move forward or already way behind on a firmware version, for instance. Your corners didn't seem to have any significant bulging.

    I'm still researching communicating with the installed firmware and other related topics. Not ready yet to say I can't find the information I think that I need - but I'm guessing I'll be back for confirmation at least.

    no problem, so i'm on 3.1.3. Slic3r with mostly default profiles. calibrated extrusion. all newest parts, belts at 280. what are your straight wall prints like?

    Just to be clear the problem is mostly visible on straight walls but its happening all the time.

    its good to all share information

    Publié : 28/03/2018 9:49 pm
    john.n13
    (@john-n13)
    Estimable Member
    Re: Tired and indignant


    no problem, so i'm on 3.1.3. Slic3r with mostly default profiles. calibrated extrusion. all newest parts, belts at 280. what are your straight wall prints like?

    Just to be clear the problem is mostly visible on straight walls but its happening all the time.

    its good to all share information

    Thanks for the comeback. I'm going to print some larger straight wall prints this evening - singled thickness - and see what I can learn. I plan to post some pics for informed comment.
    BTW, belts here at 283 and 289. I had paid no attention to them yet based on my visual examination of their behavior and the assembly test suggestion except to look at the their tracking and make sure they are not rubbing the pulley flanges.

    Publié : 28/03/2018 10:03 pm
    fulcrum
    (@fulcrum)
    Trusted Member
    Re: Tired and indignant


    EG: 1.8° motor has a LOT more torque than a 0.9° motor. A 1.8° motor, with 8 microsteps, with have similar torque to a 0.9° motor with 4 microsteps.

    According to this article, the step degree of the motor does not have anything to do with the torque of the microstep. The formula seems to be T(inc) = T(holding) * sin(90/microsteps). Now, the question is whether the driver bumps everything to 256 microsteps or it honers the setting in the firmware. If the former than the definition in the firmware have absolutely no effect on the torque.

    Publié : 28/03/2018 10:12 pm
    ben.c5
    (@ben-c5)
    Eminent Member
    Re: Tired and indignant

    the OP first pic is exactly what happen to me when my filament is not dried. I eventually buy a food dehydrator and designed ring sections to stack 4 rolls on spindles and feed directly out the side.

    you can test by baking a roll or a few meter of filament for a few hours at temperature less than material "glass transition" temp, then print and see any changes

    Publié : 28/03/2018 10:13 pm
    Kwaad2
    (@kwaad2)
    Honorable Member
    Re: Tired and indignant


    According to this article, the step degree of the motor does not have anything to do with the torque of the microstep. The formula seems to be T(inc) = T(holding) * sin(90/microsteps). Now, the question is whether the driver bumps everything to 256 microsteps or it honers the setting in the firmware. If the former than the definition in the firmware have absolutely no effect on the torque.

    What my point is, the amount of torque goes down vastly with smaller steps/microsteps. Without a gear reduction, it already takes a LOT of torque to feed the filament, ESPECIALLY if you are running your filament "cool". The faster you print, the harder you push the printer, the more "torque" you need to spin that extruder. Yes, I'm looking at you *looks at default infill settings*. The biggest issue I have bumped into is whatever microstepping they are using, is not enough to print with 0.05mm layer height, with small nozzles. I'm looking at ordering a 0.15mm nozzle, and honestly, I'm not sure if I will be able to print with a 0.1mm layer height, without... every... step... being HIGHLY visible! (This is just becuase I'm crazy, and do stupid things though...)

    My thought is though.
    What if the stepper motor is "missing" microsteps, because there is not enough torque. (it's not really possible to tell if a stepper motor successfully took a microstep. You just kind of have to "assume" it worked. I tend to print slow, but when I ramp the speeds up, I start getting weird issues like this... (which increased feed rate, would decrease the temp in the extruder, thus making it harder to extrude)

    Hi, I'm Sean. I used to work on CNC machines.
    I try to not make mistakes, but the decision is YOURS.
    Please feel free to donate to my filament/maintance fund.

    Publié : 29/03/2018 12:05 am
    Brigandier
    (@brigandier)
    Reputable Member
    Re: Tired and indignant

    Now, the question is whether the driver bumps everything to 256 microsteps or it honers the setting in the firmware.

    No question about this. In Configuration_prusa.h:

    #define TMC2130_USTEPS_XY 16 // microstep resolution for XY axes
    #define TMC2130_USTEPS_Z 16 // microstep resolution for Z axis
    #define TMC2130_USTEPS_E 32 // microstep resolution for E axis
    #define TMC2130_INTPOL_XY 1 // extrapolate 256 for XY axes
    #define TMC2130_INTPOL_Z 1 // extrapolate 256 for Z axis
    #define TMC2130_INTPOL_E 1 // extrapolate 256 for E axis

    The lines directly below the step values define if the 256 interpolation is enabled or not.

    Edit: Also got the latest 3.2.0 Alpha compiled. They've enabled some linearity correction by default now, but it only affects the moire issue and not the erratic extrusion widths we're seeing. Feels bad man.

    My MK3 Parts: [Bowden] [New Shoes] [TPU Micro Springs]

    Publié : 29/03/2018 4:56 am
    jonathon.b
    (@jonathon-b)
    Estimable Member
    Re: Tired and indignant

    I think it’s going to be down to us to disable some of the features to make it more normal, there not going to go back on any of the “selling” points of the mk3 I very much doubt. It’s very over complicated for what it needs to be especially the extruder, sometimes there’s no need to reinvent the wheel especially when the mk2 is such a great printer.

    I’ve completed the steps support suggested this morning and so far it’s not much different, I’ve took loads of pictures so I’ll send them a replay and see what else they can come up with but I can’t imagine it will be much.

    They seem to just want to blame the assembly which is always going to be a get out clause I feel unless I had got a pre built then it would be a different story.

    Publié : 29/03/2018 11:08 am
    bo.j
     bo.j
    (@bo-j)
    Active Member
    Re: Tired and indignant


    EG: 1.8° motor has a LOT more torque than a 0.9° motor. A 1.8° motor, with 8 microsteps, with have similar torque to a 0.9° motor with 4 microsteps.

    Maybe you are onto something here sean.h8. From all I have seen so far, it wouldn't surprise me at all if the quality problem turns out to be some kind of extrusion issue. It may be possible to test if steps is skipped by printing the same object twice, and then see if the Bondtech drive gear revolution count (or begin-end angle difference) is the same (marking the drive gear with at little scratch or dot of paint, should make this easier).

    Publié : 29/03/2018 2:27 pm
    david.s86
    (@david-s86)
    Eminent Member
    Re: Tired and indignant


    Now, the question is whether the driver bumps everything to 256 microsteps or it honers the setting in the firmware.

    No question about this. In Configuration_prusa.h:

    #define TMC2130_USTEPS_XY 16 // microstep resolution for XY axes
    #define TMC2130_USTEPS_Z 16 // microstep resolution for Z axis
    #define TMC2130_USTEPS_E 32 // microstep resolution for E axis
    #define TMC2130_INTPOL_XY 1 // extrapolate 256 for XY axes
    #define TMC2130_INTPOL_Z 1 // extrapolate 256 for Z axis
    #define TMC2130_INTPOL_E 1 // extrapolate 256 for E axis

    The lines directly below the step values define if the 256 interpolation is enabled or not.

    Edit: Also got the latest 3.2.0 Alpha compiled. They've enabled some linearity correction by default now, but it only affects the moire issue and not the erratic extrusion widths we're seeing. Feels bad man.

    Would it be possible to turn off the extrapolation on E, or would that bugger up all the code? Maybe that would be a way to check to see if that was the issue?

    I'm a coding noob, so...I have no idea, myself.

    Publié : 29/03/2018 3:40 pm
    Brigandier
    (@brigandier)
    Reputable Member
    Re: Tired and indignant



    Now, the question is whether the driver bumps everything to 256 microsteps or it honers the setting in the firmware.

    No question about this. In Configuration_prusa.h:

    #define TMC2130_USTEPS_XY 16 // microstep resolution for XY axes
    #define TMC2130_USTEPS_Z 16 // microstep resolution for Z axis
    #define TMC2130_USTEPS_E 32 // microstep resolution for E axis
    #define TMC2130_INTPOL_XY 1 // extrapolate 256 for XY axes
    #define TMC2130_INTPOL_Z 1 // extrapolate 256 for Z axis
    #define TMC2130_INTPOL_E 1 // extrapolate 256 for E axis

    The lines directly below the step values define if the 256 interpolation is enabled or not.

    Edit: Also got the latest 3.2.0 Alpha compiled. They've enabled some linearity correction by default now, but it only affects the moire issue and not the erratic extrusion widths we're seeing. Feels bad man.

    Would it be possible to turn off the extrapolation on E, or would that bugger up all the code? Maybe that would be a way to check to see if that was the issue?

    I'm a coding noob, so...I have no idea, myself.

    Seems it would be as easy as setting TMC2130_INTPOL_E to 0 above; however, I don't see this constant getting referenced anywhere else in the firmware, and in most cases it looks like tmc2130 "intpol" settings are instead hardcoded. Maybe they added this as a placeholder until they could code in some #ifdef statements to actually check its value?

    When I get home, I will start going around the code and disabling the intpol hardcoded values and see what kind of results I get.

    My MK3 Parts: [Bowden] [New Shoes] [TPU Micro Springs]

    Publié : 29/03/2018 4:02 pm
    david.s86
    (@david-s86)
    Eminent Member
    Re: Tired and indignant



    No question about this. In Configuration_prusa.h:

    #define TMC2130_USTEPS_XY 16 // microstep resolution for XY axes
    #define TMC2130_USTEPS_Z 16 // microstep resolution for Z axis
    #define TMC2130_USTEPS_E 32 // microstep resolution for E axis
    #define TMC2130_INTPOL_XY 1 // extrapolate 256 for XY axes
    #define TMC2130_INTPOL_Z 1 // extrapolate 256 for Z axis
    #define TMC2130_INTPOL_E 1 // extrapolate 256 for E axis

    The lines directly below the step values define if the 256 interpolation is enabled or not.

    Edit: Also got the latest 3.2.0 Alpha compiled. They've enabled some linearity correction by default now, but it only affects the moire issue and not the erratic extrusion widths we're seeing. Feels bad man.

    Would it be possible to turn off the extrapolation on E, or would that bugger up all the code? Maybe that would be a way to check to see if that was the issue?

    I'm a coding noob, so...I have no idea, myself.

    Seems it would be as easy as setting TMC2130_INTPOL_E to 0 above; however, I don't see this constant getting referenced anywhere else in the firmware, and in most cases it looks like tmc2130 "intpol" settings are instead hardcoded. Maybe they added this as a placeholder until they could code in some #ifdef statements to actually check its value?

    When I get home, I will start going around the code and disabling the intpol hardcoded values and see what kind of results I get.

    Definitely keep us posted. It would be wonderful to fix this issue and make it so my extruder could actually put down a flat wall. ><

    Publié : 29/03/2018 4:15 pm
    Brigandier
    (@brigandier)
    Reputable Member
    Re: Tired and indignant



    Would it be possible to turn off the extrapolation on E, or would that bugger up all the code? Maybe that would be a way to check to see if that was the issue?

    I'm a coding noob, so...I have no idea, myself.

    Seems it would be as easy as setting TMC2130_INTPOL_E to 0 above; however, I don't see this constant getting referenced anywhere else in the firmware, and in most cases it looks like tmc2130 "intpol" settings are instead hardcoded. Maybe they added this as a placeholder until they could code in some #ifdef statements to actually check its value?

    When I get home, I will start going around the code and disabling the intpol hardcoded values and see what kind of results I get.

    Definitely keep us posted. It would be wonderful to fix this issue and make it so my extruder could actually put down a flat wall. ><

    Sad to report, setting the hard coded "intpol" settings from 1 to 0 made zero difference on the inconsistent extrusion issue. I can confirm the MK3 sounds just like any other printer with this set to 0 though, so I know it was disabled (even the extruder stepper was louder). Mark that one off the list.

    My MK3 Parts: [Bowden] [New Shoes] [TPU Micro Springs]

    Publié : 30/03/2018 3:50 am
    jonathon.b
    (@jonathon-b)
    Estimable Member
    Re: Tired and indignant



    Seems it would be as easy as setting TMC2130_INTPOL_E to 0 above; however, I don't see this constant getting referenced anywhere else in the firmware, and in most cases it looks like tmc2130 "intpol" settings are instead hardcoded. Maybe they added this as a placeholder until they could code in some #ifdef statements to actually check its value?

    When I get home, I will start going around the code and disabling the intpol hardcoded values and see what kind of results I get.

    Definitely keep us posted. It would be wonderful to fix this issue and make it so my extruder could actually put down a flat wall. ><

    Sad to report, setting the hard coded "intpol" settings from 1 to 0 made zero difference on the inconsistent extrusion issue. I can confirm the MK3 sounds just like any other printer with this set to 0 though, so I know it was disabled (even the extruder stepper was louder). Mark that one off the list.

    Gutted to hear that, was worth a shot anyway! Did altering that effect the extruder estep value or anything?

    When did you compile it? I see more commits we’re done yesterday...

    Publié : 30/03/2018 7:22 am
    reid.b
    (@reid-b)
    Reputable Member
    Re: Tired and indignant

    Are you all sure you don't have some other issues, like slicer settings, etc? I have tuned my S3D settings and my walls are excellent. I am running the tagged 3.1.3 but with LA enabled (I compiled this version myself). It's been working like a champ on PLA, PET, and TPU. No way am I ready to go to 3.2 alpha, that is a mess right now IMHO.

    Please don't assume the problem is entirely in firmware. With a good build and a tuned slicer, this thing is putting out excellent prints.

    Publié : 30/03/2018 7:38 am
    jonathon.b
    (@jonathon-b)
    Estimable Member
    Re: Tired and indignant


    Are you all sure you don't have some other issues, like slicer settings, etc? I have tuned my S3D settings and my walls are excellent. I am running the tagged 3.1.3 but with LA enabled (I compiled this version myself). It's been working like a champ on PLA, PET, and TPU. No way am I ready to go to 3.2 alpha, that is a mess right now IMHO.

    Please don't assume the problem is entirely in firmware. With a good build and a tuned slicer, this thing is putting out excellent prints.

    I’m pleased you’ve got it sorted pretty well, would you be up for slicing a 1 wall, 0 infill cube and posting the gcode for PLA so I can compare it with yours? Just as a little test?

    Publié : 30/03/2018 11:00 am
    Brigandier
    (@brigandier)
    Reputable Member
    Re: Tired and indignant


    Are you all sure you don't have some other issues, like slicer settings, etc? I have tuned my S3D settings and my walls are excellent. I am running the tagged 3.1.3 but with LA enabled (I compiled this version myself). It's been working like a champ on PLA, PET, and TPU. No way am I ready to go to 3.2 alpha, that is a mess right now IMHO.

    Please don't assume the problem is entirely in firmware. With a good build and a tuned slicer, this thing is putting out excellent prints.

    For our reference, can you try this with a single wall cube with zero infill and post some pics plus the gcode? Make sure to have light shining down the face of the cube to highlight the varied layers if they are there. This will let us see the difference in your version plus let us try it on our MK3s and compare. I will get my copy of S3D reinstalled and see if I get different results on my own as well.

    My MK3 Parts: [Bowden] [New Shoes] [TPU Micro Springs]

    Publié : 30/03/2018 2:30 pm
    Page 7 / 10
    Partager :