Extrusion Calibration?



  • Hi *,

    I have the next strange problem.... 😞 I have the strange feeling that I'm a magnet for them.

    I printed a tolerance test yesterday. I used "https://www.thingiverse.com/thing:2318105" for that. The result was a block of plastic with no moving parts. Tnot even the 0.5 distance. I reduced the extrusion rate to 90% and the "0.5" was loose afterwards.
    So I reduced the extrusion rate in the slicer even further.
    80%=0.35 moving
    70%=0.25 moving
    60% is still printing.

    I do not trust my calibration and anymore. I dialed in the extruder by measuring the millimeter trough a hot nozzle etc. (Original Bondtech on my CR10 v2 pro with the original hotend)

    M92 X80.16 Y80.16 Z400.00 E431.62
    

    I wounder about the 70%! I would have excepted something like 95% or 90% but 30% reduction seams a bit to much to me.

    The XYZ cube looks good to me and I know "https://duet3d.dozuki.com/Guide/Ender+3+Pro+and+Duet+Maestro+Guide+Part+4:+Calibration/40#s165" 😉
    The filament is very close to the 1.75, sometimes 1.74 and sometimes 1.73. But very accurate over 1m. The nozzle is a 0.4 measured and configured in the slicer.

    Any tips where I should start now? I'm bit lost now... 😞

    Cheers, Chriss


  • Moderator

    Are you sure you're using the right thermistor values? Your temperature could be quite a bit off which could be influencing things.

    If you're using a legit bondtech I would try using the steps per mm provided by them to start with. It should be pretty close to true.

    How do the prints look? 30% over extrusion would be pretty obvious.



  • @Phaedrux said in Extrusion Calibration?:

    Are you sure you're using the right thermistor values? Your temperature could be quite a bit off which could be influencing things.

    Well, that is hard to answer here. Not all of the values are hard facts, right? I can not remember where "B" came from, but I think that the parameter are not soooo wrong:

    M308 S1 P"e0temp" Y"thermistor" T100000 B4138 
    M950 H1 C"e0heat" T1 
    M143 H1 S280
    M307 H1 A461.2 C156.1 D3.1 S1.00 V24.0 B0
    

    I'm currently not able to verify the values, my IR-thermometer is on vacation. 😉

    If you're using a legit bondtech I would try using the steps per mm provided by them to start with. It should be pretty close to true.

    Indeed, it is. Bondtech asks to start with 415 I have "431.62" after I measured 100mm 5 times.

    How do the prints look? 30% over extrusion would be pretty obvious.

    That is a other point, the difference between 100% and 60% do not so different as I expected.

    IMG_20201020_064832.jpg

    Cheers, Chriss



  • @Chriss
    The Extrusion Calibration I went with was make a box with 0.8mm walls and keep turning the extruder factor down until you say the walls don't join.
    You can't use the naked eye to see the walls , without I went down to 60% and could not see that the walls wasn't joined.
    With a Magnifying glass I found it was about 85-87% extrusion factor.
    I tried that tolerance test and wasn't a good test of tolerance.



  • @peter247

    The Extrusion Calibration I went with was make a box with 0.8mm walls and keep turning the extruder factor down until you say the walls don't join.
    You can't use the naked eye to see the walls , without I went down to 60% and could not see that the walls wasn't joined.
    With a Magnifying glass I found it was about 85-87% extrusion factor.
    I tried that tolerance test and wasn't a good test of tolerance.

    I got your point but I'm not sure that I'm already at this point to tune in the last 15%.

    You can't use the naked eye to see the walls , without I went down to 60% and could not see that the walls wasn't joined.

    What do you mean with "wasn't joined"? I ask this question myself since I discovered this docuemntation. Does it mean that the two parallel strings do not touch at all? And does a "good touch" mean that they are perfectly joined and look like a perfect flat surface? I would like to see a picture from the expected result. Made with a good macro lens. 😉

    I ordered a new nozzle this morning. I do not trust this guys from CN any more, the new one is from BondTech. I hope that they provide a reliable nozzle. I will perform the "wall calibration drill" when I have the new nozzle. I think that I need to gob back to start.

    Cheers, Chriss



  • @Chriss

    When I did my calibration test , at 60% -70% you could see a gap between the walls and as I increased the extrude factor you could start to see it fully joint into a solid single wall. ( but could not see any difference with the naked eye )

    I think a did step 15 method 1 to calibrate it ?
    This is my test of calibration

    https://www.thingiverse.com/thing:4575774



  • @peter247
    I was not very successfully with method 1, I tried my best and went forward with method 2 than.

    I will wait for the new nozzle now and start over from beginning. I will take some from the process than. I hope that this will make it clear than. (Damn I was not so unhappy with the results till I started this test drill.)

    The tiny motor block is very nice, I will print that for testing from now on. 😄

    Cheers, Chriss



  • @Chriss

    My prints for models wasn't that bad , but printing things like boxes each layer was different you could fell a roughness between all the layers , tried calibrating the e-step for my BMG clone which I found to be 420 e-step .
    The model I give you fused together and didn't work , I made a box in fusion 360 which fitted inside a second box printed at the same time, no way would they fit.
    Printing my boxes over and over again cutting the extrusion factor by 1% each time I found 87% my boxes became smoother and my boxes fitted each other.
    Keep trying , I don't see the point of e-step calibration.
    I also set up pressure advance , but not sure if that helped or made things worse.


  • Moderator

    @Chriss said in Extrusion Calibration?:

    Well, that is hard to answer here. Not all of the values are hard facts, right? I can not remember where "B" came from, but I think that the parameter are not soooo wrong:
    M308 S1 P"e0temp" Y"thermistor" T100000 B4138

    What kind of thermistor is it? If it's a stock creality than that looks pretty good.



  • @Phaedrux It is the stock hotend.
    Since 20 minutes with a Bontech nozzle. I start now with the extrusion tests.... keep your fingers crossed.



  • @peter247 said in Extrusion Calibration?:

    @Chriss

    My prints for models wasn't that bad , but printing things like boxes each layer was different you could fell a roughness between all the layers , tried calibrating the e-step for my BMG clone which I found to be 420 e-step .

    Yes, that is what I see here too. The layers in the gabs are very rough and do not look very right.

    Printing my boxes over and over again cutting the extrusion factor by 1% each time I found 87% my boxes became smoother and my boxes fitted each other.

    87% seems a bit to low to me but I guess that you are right as long as you have results which are OK for you.

    Keep trying , I don't see the point of e-step calibration.

    You bet that I will. My e-steps are okish, I do not see a reason to restart a dail-in process.

    I also set up pressure advance , but not sure if that helped or made things worse.

    I will use pressure advance in the future, but I think that I must be able to make acceptable prints without it. Other people where able to do so in the past too.

    Cheers, Chriss



  • Update:

    I went down till 70% till they lines separated. So I went back to 80% and the lines on the X wall where back melted but not so the wall following the Y axes. 🙁

    I think that I go back to start now and calibrate X and Y first. Blimy! It seems that something is wrong there.
    I did it more than 3 times now, I do not understand what's wrong with that peace of Chinese crap. 😞



  • @peter247 @Phaedrux

    Here is what I did:
    I placed a needle on the hotend to measure the real driven way to calibrate the steps for x and y. I used 250mm for that. So I was able identify a tiny miss configuration

    Old:

    M92 X80.16 Y80.16 Z400.00 E431.62 
    

    New:

    M92 X80.14 Y80.00 Z400.00 E431.62
    

    I printed the cube and followed Step 1, I made two pictures trough a magnifying glass. The first picture is from the right Y wall, the second from the left Y wall. Both with 100% extrusion.

    IMG_20201023_052804.jpg IMG_20201023_052738.jpg

    I have no other explanation than: "wrong steps for X".

    So I printed the XYZ-Cube which came out as
    X: 20.07-20.18mm
    Y: 20.21-20.25mm

    That confirmed my theory, don't you think so?

    Is there a better method to calibrate the steps? Needle on the hotend and a steel lineal and the longest possible travel seems to me as the best possible way.

    Cheers, Chriss


  • Moderator

    Usually it would be best to use the calculated steps for X and Y since we have quite accurate values for the relevant factors (belt teeth pitch, pulley tooth count, motor steps, etc) and modern manufacturing ensures we're within a pretty tight margin of error, so the calculated values are likely to be more accurate than we can reliably measure.

    So I would ask, what do you get for results if you just used 80 steps per mm in X and Y?

    The extrude is a bit different because the hobb diameter is much harder to quantify. Not only is it rare to get a value from the manufacturer but it's hard to measure in practice because it depends on how squished the filament is. That's why we do a measurement calibration and then fine tune with an actual print. Though if you're using a genuine quality extruder (E3D, Bondtech, Dyze, etc) you're likely to get a steps per mm value from the manufacturer, which should be pretty bang on and only need a bit of tuning due to filament differences.

    xy_steps_per_mm = (motor_steps_per_rev ∗ driver_microstep) / (belt_pitch ∗ pulley_number_of_teeth)

    e_steps_per_mm = (motor_steps_per_rev * driver_microstep) * (big_gear_teeth / small_gear_teeth) / (hob_effective_diameter * pi)



  • @Phaedrux said in Extrusion Calibration?:

    Usually it would be best to use the calculated steps for X and Y since we have quite accurate values for the relevant factors (belt teeth pitch, pulley tooth count, motor steps, etc) and modern manufacturing ensures we're within a pretty tight margin of error, so the calculated values are likely to be more accurate than we can reliably measure.

    I will not argue here about error rates and accuracy in consumer products from CN. 😉

    So I would ask, what do you get for results if you just used 80 steps per mm in X and Y?

    I had the same in my mind, well, the result was very much the same. One wall is melted the other has a tiny gab. I will do some more tests with the belt tension.

    The extrude is a bit different because the hobb diameter is much harder to quantify. Not only is it rare to get a value from the manufacturer but it's hard to measure in practice because it depends on how squished the filament is. That's why we do a measurement calibration and then fine tune with an actual print. Though if you're using a genuine quality extruder (E3D, Bondtech, Dyze, etc) you're likely to get a steps per mm value from the manufacturer, which should be pretty bang on and only need a bit of tuning due to filament differences.

    And the Filament is much harder to measure. 🙂 I use this calculator to calculate all of that. (Yes it is in German)
    I think that the steps for the extrudor are not so bad. My current goal is to get a similar surface on all walls first.

    I will do some tests with the belt tension and I will checks the wheels and the v-slots to make sure that everything works smooth here....

    Cheers, Chriss



  • @Phaedrux

    Hi...

    So v-slots clean and everything is smooth. Belt tension is fine.

    M92 X80.00 Y80.00 Z400.00 E431.62
    

    PLA@210°C 100% Extrusion.

    Left Ywall:
    IMG_20201023_082149.jpg

    Right:
    IMG_20201023_082139.jpg

    I can see that the gab between the two lines start appearing at the curve from x to y. It seams to me that there outer wall is a tiny bit longer on the X axis. But this do not make sense to me.

    I will start with a fresh cura profile now. 😞

    Cheers, Chriss



  • @Chriss

    You are the first person who as calibrated their x , y.
    The way to find out if you have X , Y error and not under / over extrusion is make a stepped cube.
    if you make a cube 50mm squared and it is 50.2mm , what is the problem ? , but if you make a second cube at 100mm squared and it is 100.2mm it is not X and Y



  • @peter247

    You are the first person who as calibrated their x , y.

    I'm sure that I'm not the first person. A buddy gave me that "tip". 😉

    if you make a cube 50mm squared and it is 50.2mm , what is the problem ? , but if you make a second cube at 100mm squared and it is 100.2mm it is not X and Y

    I got your point and I fully agree. I was running into the wron direction here.

    I made a 20mm and a 60mm cube.
    20mm: X=20,20 Y=20,20
    60mm: X=60,20 Y=60,20

    I can hear your: "Toult you so" till here to my desk. 😉

    The values for z are missing here, That is a other story and I'm not able to get the cube off the bed at the moment. (3DLAC have it's downsides. 😉 )

    I will do some more extrusion tests later or tomorrow.

    Cheers, Chriss



  • @Chriss said in Extrusion Calibration?:

    That is a other story and I'm not able to get the cube off the bed at the moment. (3DLAC have it's downsides. )

    Yep I use 3DLAC for them times you just can't get it to stick or you really don't want a small part to come loose.
    I try not to use it on PLA unless I have to , but PETG is a must or you will crack the glass.

    What is your bed temperature ? and what is your bed temperature when you are trying to remove your part ?.

    I've found that there is no part trying to remove your part if the bed is above 30c , but if you heat the bed to 60c it will remove it's self or with a slit knock at about 25-28c.
    With PETG and 80c bed you know when that is ready you hear a very loud crack, if you are using something like 3DLAC that is your part removed if not it is your bed glass cracked.



  • @peter247 said in Extrusion Calibration?:

    @Chriss said in Extrusion Calibration?:

    That is a other story and I'm not able to get the cube off the bed at the moment. (3DLAC have it's downsides. )

    Yep I use 3DLAC for them times you just can't get it to stick or you really don't want a small part to come loose.
    I try not to use it on PLA unless I have to , but PETG is a must or you will crack the glass.

    I use the 3DLAC since I switched from PEI to Pertinax. I was not able to get big prints to stick on the bed. That was a bad decision, the next surface will be PEI again.

    What is your bed temperature ? and what is your bed temperature when you are trying to remove your part ?.

    I print PLA with 60°C and usually I wait till it is back to room temperature, well smaller parts get off if the bed is still warm.

    I've found that there is no part trying to remove your part if the bed is above 30c , but if you heat the bed to 60c it will remove it's self or with a slit knock at about 25-28c.

    I which that a slit knock would be enough. I sharpened the front of my spattle to slight better under the object. 😉

    With PETG and 80c bed you know when that is ready you hear a very loud crack, if you are using something like 3DLAC that is your part removed if not it is your bed glass cracked.

    🙂 I wish that this could happen with PLA. 🙂

    Well, I dialed in my z-offset today, for the nozzle change. I had to lift it 0.1mm, so the Elephant foot has gone now and the smaller cube was easy to remove now. I think that this will be the case with bigger object, too.

    I printed a first layer which covered the complete bed. And I saw the ruff first layer appearing at the "max Y - min Y" corner:

    IMG_20201024_175259.jpg

    Is that caused by an uneven bed? I saw that in the mast from time to time on my PEI sheet, too and I was not able to find the reason for it.

    Cheers, Chriss



  • @Chriss

    Does your printer have auto / mash bed levelling and does it have a bltouch or micro switches for Z height ?
    Do you own a pair of digital callipers ?
    How do you levelling your bed ?
    It is hard to get scale on the print how big is it ?
    Looking at your print the first layer looks like it is thicker in parts of the print by the edges.
    To me it looks like it is under extruding .


  • Moderator

    Can you get a better photo?



  • @peter247 said in Extrusion Calibration?:

    Does your printer have auto / mash bed levelling and does it have a bltouch or micro switches for Z height ?

    There is a BLTouch in place.

    Do you own a pair of digital callipers ?

    At least 10, but most of them are analog, I hate the battery changing drill. I used a micrometer screw to measure the thickness: The numbers in circles are the number of the stripe, the other numbers are the thickness at this place. The order of 4 and 5 are wrong. 😞

    IMG_20201025_045020.jpg

    How do you levelling your bed ?

    What do you want to read?
    homez.g:

    M300 S500 P1000		; Beep
    G90
    G1 Y155 X155 F6000 ; go to first probe point
    M400
    G91               ; relative positioning
    
    ; Run 1
    M558 F250          ; Chriss - set the down speed
    G30
    
    ; Run 2
    M558 F60          ; Chriss - set the down speed
    G30 
    M558 F350         ; Change the speed back
    

    The mesh:

    M557 X10:270 Y40:270 P10:10 
    

    Zlevel:

    G30 P0 X0 Y155 Z-99999; probe near a leadscrew, half way along Y 
    G30 P1 X250 Y155 Z-99999 S2 ; 
    

    Hightmap:
    0df481bd-42bd-48c4-aeca-7ac1f39d3e21-image.png

    I used a feeler gauge at every screw in the corners to level the bed mechanically.

    It is hard to get scale on the print how big is it ?

    20x35mm ruffly

    Looking at your print the first layer looks like it is thicker in parts of the print by the edges.
    To me it looks like it is under extruding .

    I think that the the z offset is wrong. The first layer should have 0.20 from the slicer, but it is more in average over the complete bed. That is a bit strange. I will tune in the z-offset again and print the first layer one more time.

    Some more pictures:
    IMG_20201025_044339.jpg

    IMG_20201025_044314.jpg

    Back site:
    IMG_20201025_044417.jpg

    @Phaedrux Are they good enough?

    Cheers, Chriss


  • Moderator

    @Chriss said in Extrusion Calibration?:

    @Phaedrux Are they good enough?

    Yes. Confirms that it looks like the nozzle is too close in those areas.

    The image of your heightmap shows a bit of tilt front to back, raised in the front. I think you could probably level the bed a bit better.



  • @Phaedrux

    Thanks...

    Let me repeat it: The roughness comes from over extrusion because the gab between the nozzle and the bed is to narrow.

    I will re level the bed and will print tiny batches of first layers. Lets see how they look than.

    Cheers, Chriss


Log in to reply