Troubleshooting help needed Geared Extruder mod

[willefg]

Hi, I have a CTC Dual (Replicator 2 clone) that has been modded heavily over the years. Last year I removed the original MK8 dual extruders and replaced that with a single E3D V6 hotend and standard Bowden extruder. Around then I also started using Octoprint (with GPX plugin for gcode to x3g conversion). I only print using PETG.
My latest change is a 3:1 Bondtech BMG extruder.
I use Simplify3D to send Gcode files to Octoprint and Octoprint controls the print process.

The issue I'm having is that ever since I use a Bowden system, I need to set the Extrusion Multiplier to around 0.80 to get accurately (<1% tolerance) printed parts.
I've used a USB connection directly to the printer and Machine Control in Simply3D to check the extruder motor feed. 100mm filament feed is accurate with the Bondtech BMG 3:1 extruder when the E steps per mm were set at 415, the number Bondtech gives.
With the Extrusion Multiplier set at .80 and the extrusion width set to "Auto" (0.48mm width), prints are exact in size, and the layers bond very well together. It's just that the perimeter shells (I use 3) are somewhat separated and the bottom and top layers don't cover completely. Just as is the case when under extruding.
When I measure the width of a single printed line (on the bed) it's 0.50mm, so that seems accurate as well.

If I decrease the Extruder steps per mm to (0.81 x 415=) 337, I get exactly the same results when I set extrusion multiplier to 1.00.

Can anyone explain to me why my setup needs a 20% deviation to print fine?

[parallyze]

Hi,

Can anyone explain to me why my setup needs a 20% deviation to print fine?

When I measure the width of a single printed line (on the bed) it's 0.50mm, so that seems accurate as well.

You might want to have a look at this thread, especially the difference between s3d/slic3r:
https://3dprinting.stackexchange.com/qu ... tment/1163

If everything looks okay (not overextruded) using an extrusion multiplier of 1.00 and objects always come
out slightly bigger (along X/Y) than expected (30.25 vs 30.00, 60.25 vs 60.00) use "Horizontal Size Compensation" for
fine-tuning.

[willefg]

Thank you for your answer.

Extrusion and print quality is not the issue here. The issue is that I have a 20% deviation in my settings to get to the perfect result. So, the extrusion multiplier is 0.80 instead of 1.00 to get it right and it looks to me like a settings issue or incompatible parameters somewhere in the chain.

After I wrote my message I tried and compared a test design printing it using a Gcode file from Simplify3D through Octoprint with the GPX plugin to convert to x3g for my printer and the same file in x3g directly (over USB) from Simplify3D to the printer. That made no difference, so it rules out a Octoprint/GPX issue. That leaves the printer and S3D.

Regarding extrusion calibration and getting perfect measurements, I've been using following test file and the Excel spreadsheet that is linked there.
https://www.thingiverse.com/thing:1622868

This works great. Easy and fast. You just print the part 6 times in a single machine run, measure the inside and outside sizes, enter these measurements in the spreadsheet together with the extrusion multiplier and extrusion width from you slicer and it will calculate the best extrusion multiplier to get the best result. Worked great for me.

[parallyze]

Extrusion and print quality is not the issue here. The issue is that I have a 20% deviation in my settings to get to the perfect result. So, the extrusion multiplier is 0.80 instead of 1.00 to get it right and it looks to me like a settings issue or incompatible parameters somewhere in the chain.

Uhm...

It's just that the perimeter shells (I use 3) are somewhat separated and the bottom and top layers don't cover completely. Just as is the case when under extruding.
When I measure the width of a single printed line (on the bed) it's 0.50mm, so that seems accurate as well.

If the perimeters don't bond properly you're very likely underextruding. Also the solid layers aren't properly closed, another indicator
for underextrusion. Measuring line width on the bed isn't very accurate, the smallest amount of squish will render results invalid. Print
a box with 3 perimeters @ 0.48mm width. If those walls are <= 1.44mm you're underextruding.

Regarding extrusion calibration and getting perfect measurements, I've been using following test file and the Excel spreadsheet that is linked there.
https://www.thingiverse.com/thing:1622868
This works great. Easy and fast. You just print the part 6 times in a single machine run, measure the inside and outside sizes, enter these measurements in the spreadsheet together with the extrusion multiplier and extrusion width from you slicer and it will calculate the best extrusion multiplier to get the best result. Worked great for me.

Didn't really work that great, did it? I wouldn't recommend anyone to change the extrusion multiplier to compensate for
dimensional inaccuracies, that's on the same level as adjusting x/y steps per mm to fit 30mm test cubes...

Look at the linked thread again. The calculated width of the strands will always have some overlap on both sides - the outsides
of your printed objects will almost always come out a tiny bit bigger than expected if using the correct extrusion amount.
For finetuning use FFF Settings -> Other -> Dimensional Adjustments -> Horizontal Size Compensation.

[parallyze]

Regarding extrusion calibration and getting perfect measurements, I've been using following test file and the Excel spreadsheet that is linked there.
https://www.thingiverse.com/thing:1622868

This works great. Easy and fast. You just print the part 6 times in a single machine run, measure the inside and outside sizes, enter these measurements in the spreadsheet together with the extrusion multiplier and extrusion width from you slicer and it will calculate the best extrusion multiplier to get the best result. Worked great for me.

I just checked the gcode from s3d and slic3r again, confirming what Ryan Carlyle posted in the thread on stackexchange.

Take a 50mm cube, slice it using 0.5mm extrusion width. The resulting gcode from s3d will show this:

Code: Select all

; feature inner perimeter
G1 X75.500 Y50.750 F9000
G1 E0.6000 F2100
G1 X124.000 Y50.750 E2.5205 F1500
G1 X124.000 Y99.250 E2.5205
G1 X75.500 Y99.250 E2.5205
G1 X75.500 Y50.750 E2.5205
; feature outer perimeter
G1 X75.000 Y50.250 F9000
G1 X124.500 Y50.250 E2.5725 F1500
G1 X124.500 Y99.750 E2.5725
G1 X75.000 Y99.750 E2.5725
G1 X75.000 Y50.250 E2.5725

The outer perimeters are spaced apart 49.5mm from each other. The inner perimeters are spaced at
exactly 0.5mm from the outer ones. So if the extruded plastic was in a perfect rectangular shape this
would fit perfectly.

Now let's have a look at one picture from the above thread again:

So the extruded plastic has to be wider than what's defined as extrusion width in s3d. Otherwise the
strands wouldn't bond together, leaving empty spaces. And this extra material will slightly overlap the
objects boundaries on the outermost perimeters.

An extrusion width of 0.5mm in Slic3r will result in this:

Code: Select all

;TYPE:Perimeter
G1 F3600.000
G1 X85.796 Y99.404 E1.12751
G1 X85.796 Y50.796 E2.25502
G1 X134.404 Y50.796 E2.25502
G1 X134.404 Y99.404 E2.25502
G1 X110.160 Y99.404 E1.12473
G1 X110.100 Y99.850 F12000.000
;TYPE:External perimeter
G1 F3000.000
G1 X85.350 Y99.850 E1.14822
G1 X85.350 Y50.350 E2.29644
G1 X134.850 Y50.350 E2.29644
G1 X134.850 Y99.850 E2.29644
G1 X110.160 Y99.850 E1.14543
G1 X109.789 Y99.599 F12000.000

Just as in S3D the outermost perimeters will be spaced 49.5mm from each other. But to get the
proper amount of overlap the inner perimeter is only spaced 0.446mm from the outer one,
not 0.50mm as in s3d.

By decreasing the extrusion multiplier here in s3d one would simply thin out the strands of plastic until
their real width is barely 0.50mm, resulting in accurate outer dimensions of the object but almost completely
missing the required overlap to fill the gaps between the perimeters/strands spaced at 0.50mm. So if the real
width of the extruded plastic was 0.53mm (adding +0.03mm to the 0.50mm extrusion width) you'd end
up with outer perimeters "protruding" about 0.015mm, making the 50mm cube ~50.03mm when printed.

You want to inset the outer perimeters to compensate for this, not lower the extrusion mutliplier.

Using slic3r one could also use the thing/spreadsheet - but only for very minor differences, nowhere near
the 20% you're currently deviating. Because slic3r is using the real extruded width (including the extra material
for overlapping) in it's calculations it will very likely be closer to the objects dimensions when printing using
default profiles not fine tuned using horizontal size compensation.

If your extrusion multiplier was really off by 20% you should've seen massive ridges on solid infill layers
when printing at 1.0/100%...