Slic3r and w:h wierdness

Prusa, Slic3r 0.7.0, Pronterface, Windows, 1.8 stepper w/Wade's, hobbed bolt, ABS media, Budaschnozzle w/ 0.50mm orifice

I think I followed the extruder calibration procedure correctly, but maybe blew it as I've performed it with the hotend in place and heated, so made lots of little ABS doggy piles. Ended up with an E steps of 630. Measured the filament at an average of 2.88mm diameter.

The only way I've been able to produce really decent quality prints is to set w:h in the Advanced tab of Slic3r to 1.15. Arrived at this by printing 0.5mm calibration box and measuring thickness of walls after cutting off the thick bottom and making adjustments to w:h until I get a w:h ~1.8. Using 0.37mm layer height, by the way.

Prints look marvelous, except when it comes to solid infill layers. Too much ABS is being extruded and the hotend bounces over the previous solid layer with expected results - looks awful and takes several layers to recover.

I assume Slic3r set at automatic w:h (0 entry in the textbox) works for others (and also assume Slic3r sets extrusion rate to produce a w:h of ~1.8), so first question is, why doesn't it work for me? Second question is what's up with my nasty solid infill layers?

Thanks for the help!

When you tried printing the 0.5 wall box how did you do it? Did you take a solid cube and just turn off the solid layers and infill? If you did you should see at the top of the gcode file a line that says:

; single wall width =

This is the width the wall should turn out.

I would set it back to 0 and try printing it as I said and only adjust the "Extrusion multiplier" until you get the correct width. Then take that number and use it to recalculate the E-steps per mm.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Chowderhead Wrote:
-------------------------------------------------------
> The only way I've been able to produce really
> decent quality prints is to set w:h in the
> Advanced tab of Slic3r to 1.15. Arrived at this by
> printing 0.5mm calibration box and measuring
> thickness of walls after cutting off the thick
> bottom and making adjustments to w:h until I get a
> w:h ~1.8. Using 0.37mm layer height, by the way.
>
> Prints look marvelous, except when it comes to
> solid infill layers. Too much ABS is being
> extruded and the hotend bounces over the previous
> solid layer with expected results - looks awful
> and takes several layers to recover.

Your plastic flow is wrong if this is happening. You can't calibrate it by printing thin walls, they are too hard to measure accurately enough with all sorts of other inaccuracies present (Z wobble, deposited filament shape, X and Y axis repeatability etc.)

Printing an object with 100% infill will tell you whether you're getting correct flow - if the hotend starts to draw grooves in the already printed areas, you're definetely extruding too much. Adjust the E-steps value in the firmware and the extrusion multiplier in the slicer to get this right.

After the extrusion volume is correct, you can start worrying about the w/t value. I'm also using Slic3r 0.7.0 and I have mine set manually to 1.8. With a 0.35 nozzle, I've found that this value produces really good quality at 0.20 layers and quicker but still good prints at 0.25 layer height.

Sublime: I used this [www.thingiverse.com] which I found on the calibration page in the RepRap wiki.

ttsalo: I printed several infill boxes (http://www.thingiverse.com/download:17274) from the calibration page, but it says to play with the w:t (w:h is what I called it) which I did to little effect because I felt like I was making too big adjustments. Probably because I "calibrated" extrusion following the lineal filament feedrate method, which now doesn't seem like a good idea at all. Going from one to three dimensions with any accuracy requires knowledge of too many variables (media density, volatile fraction, and I'm sure others) that themselves vary, based upon what I've read about filament.

If slicing is done on a volumetric basis, then calibation needs to be completed on the same basis. The infill box would then seem to be the best alternative, except the engineer in me wants to know what the volume really is, not just whether the top looks pretty. Guess I'll be taking a volumetric cylinder home with me.

Something then needs to be done about the E steps per mm unit in firmware. It's not a meaningful unit if everything else to do with extrusion is based upon volume. This invites people to do the wrong thing, like lineal filament feedrate calibration.

Linear feed length calibration always works for me but I use my own software. It should work for any slicer if they have the right maths.

---

[www.hydraraptor.blogspot.com]

Sorry if I came across wrong, I share my opinions a little too easily, I guess.

I have nothing but respect for what you all have accomplished! This is one of the coolest things I've come across, ever. I hope I can make some sort of contribution.

Thanks for the help, I'm going to put your suggestions to work tonight!

The calibration cube you linked to has two walls all the way around. This means if it is sliced with a normal profile it should turn out twice as thick as the number mentioned in the gcode file under ; single wall width = because it has to print the outside and the inside.

But I, like most other people only calibrate using the measurement of filament being feed into the machine and don't even bother to check the single filament thickness.

Also we are using the volume of plastic with e-steps-per-mm . Since we know the diameter plus the length of filament being feed in we can easily calculate the volume based on that.

You mention testing with the hotend on and I personally think that is a good thing as it allows for a perfect calibration taking into account the resistance of melting the plastic. But if you are testing at some really low speed and then try and print to find it does not work you may just not have your idler bearing tight enough and it is slipping.

Also all that w/t stuff is for calibrating Skeinforge, not Slic3r. I used Skeinforge for a year and have switched to Slic3r two months ago and have never had to adjust the w/t. Slic3r makes really good code, all the calibration is done for you once the e-steps-per-mm are set correct.

Edited 1 time(s). Last edit at 03/11/2012 03:10PM by Sublime.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

w/t isn't something you calibrate in Skeinforge. It is something you specify and Skeinforge extrudes the correct volume and spacing to make it so, assuming your steps per mm is calibrated and your filament diameter is what you have told Skeinforge it is.

---

[www.hydraraptor.blogspot.com]

I understand that w:t isn't a calibration in the slicing program, but its calculation must rely on the volumetric flowrate calibration (approximated by filament diameter and lineal feedrate).

I think Sublime's note regarding the speed at which I calibrated the extruder is what's impacting my print quality. I think I calibrated at too slow a speed (30mm/s as I recall). I'll recalibrate at higher feedrates and see where that goes.

Thanks again for answering my questions!

It's not an approximation, unless you count the minute amount of volatiles that boil off. The volume that goes in is the same as what comes out.

I don't see what effect speed would have unless your extruder is slipping.

Edited 1 time(s). Last edit at 03/11/2012 10:44PM by nophead.

---

[www.hydraraptor.blogspot.com]

For what it's worth:

Ran filament feedrate calibration at 30mm/s, 60mm/s and 90mm/s generating E-steps/mm of 630, 534 and 516 respectively. Reset E-steps to 525 (mean of last two) and produced better prints, but still extruding too much. Printed the 20mm calibration box at 100% infill to verify over-extrusion. Reduced E-steps more to produce a pretty 20mm box and now the solid infill layers look stupendous with Slicer set to automatic w:t.

I'll press my case that the linear feedrate of filament is not the best extruder calibration metric. I'm no polymer materials scientist, but I do know that polymers absorb water and contain variable amounts of volatiles. I suspect that heating and two phase changes results in some restructuring, too. This and other factors are going to lead to a net change in density from one side of the extruder to the other. Tack on mechanical slip, hotend temperature fluctuations, changing printing environment (spring here brings massive swings in humidity) and there's a multiplicity of unaccountable/unmeasurable variables.

Calibrating the extruder based upon extruded volume at a given set of conditions better accounts for these sorts of variables. This is going to be made (already has been?) more clearly apparent as more people experiment with different plastics.

Every other part of the machine is calibrated based upon what the particular part produces (movement) except the extruder, which is presently calibrated by what it consumes.

Off my soap box now, and thank you all sincerely for the assistance. My printer now rocks!

What if we had a station, with a small roller or horizontal disc, where the fillament is extruded and opticaly measured,
for each layer.

---

Random Precision

About the w/t calibration... I think that was something done in the days when Skeinforge used non-volumetric calculations and the E values meant the length of the extruded filament. You would print a thing with a certain arbitrary E-steps-per-mm value and see what kind of extruded width you happened to end up with? Nowadays you just set it to whatever your printing style requires and the volumetric calculation handles the rest.

Chowderhead Wrote:
-------------------------------------------------------
> Every other part of the machine is calibrated
> based upon what the particular part produces
> (movement) except the extruder, which is presently
> calibrated by what it consumes.

That's the only actual metric available to the machine, so what would be the alternative? Measuring the output volume... how? Sure, if you can do that after a print, you can then plug the result in the extrusion multiplier adjustment and get accurate output, but it still applies only to the specific operating conditions for which you measured that in.

Absorbed moisture and other volatiles that boil off must reduce the filament volume slightly but you were getting far too much plastic, so something doesn't add up there.

I have been extruding plastic for 5 years and have used HDPE, PCL, PMMA, several ABS types and a couple of PLA types. Linear feed rate calibration has always worked for me. I have also done volume and density in and out measurements when people started to claim ABS had a different packing density to PLA. [hydraraptor.blogspot.com]

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> Absorbed moisture and other volatiles that boil
> off must reduce the filament volume slightly but
> you were getting far too much plastic, so
> something doesn't add up there.


The point is that I extruded too much plastic because I followed the calibration procedure based upon filament feedrate, hence my opinion that that procedure is insufficient.


> I have been extruding plastic for 5 years and have
> used HDPE, PCL, PMMA, several ABS types and a
> couple of PLA types. Linear feed rate calibration
> has always worked for me. I have also done volume
> and density in and out measurements when people
> started to claim ABS had a different packing
> density to PLA.
> [hydraraptor.blogspot.com]
> ow-rate.html

With all respect and I acknowledge that we're even having this discussion in part due to your work; that was one test in one setting. Manchester, UK looks to have a pretty stable climate whereas in my location we'll exceed your annual T and humidity variation in a single 48 hour period, depending on the season. More to the point, if you based calibration on volumetric feedrate, then it wouldn't matter whether the extruder feeds PLA faster than ABS - it would all be captured in the measurement and you wouldn't have to blodge anything...

WRT how to measure it, I don't know, but I think it's worthy of exploration.

The point is that PLA feeds faster than ABS for the same pinch wheel diameter, so you have to calibrate for both. It isn't a bodge but a correction factor for the effective diameter of the pinch wheel for different plastics. Volume and density are still preserved.

If your plastic absorbed enough moisture to change the volume enough to notice then you would get steam bubbles and lots of ooze as it was extruded. And the extruded volume would be less, not more. People in humid climates have to dry their PLA otherwise it causes oozing problems.

Slipping and atmospheric moisture would both reduce the volume. You haven't come it with a mechanism whereby it can increase in volume.

I don't see how temperature comes into it. The filament is at room temperature when it goes into the extruder and the object cools to room temperature at the end of the build, so it will have the same density.

Also the amount of moisture plastic absorbs and the thermal expansion coefficients are about an order of magnitude less than the amount of error you seeing.

I don't think you can dismiss a calibration method that works for everybody else. I think something else is wrong with your set up or the firmware or slicer.

---

[www.hydraraptor.blogspot.com]