Dimensional error with walls of pipes?

I have a MendelMax 1.5 which I built from a kit.

The Marlin calibration that came pre-installed on the Arduino seems pretty good, and when I print test pieces I'm getting around 0.5% error, which seemed to be reasonable.

However if I attempt to print items that are hollow, e.g. a 90mm drain pipe adaptor (for my rainwater collection system), the wall thickness of the pipes comes out 12% too thick :-(

I've checked the extruder calibration, and if anything, its probably extruding too little not too much, so I'm not sure what would be causing this large inaccuracy.

Can anyone advise on whether its possible to fix this in either Slic3r settings or Marlin calibration ?

BTW. I can easily get around this my modifying the model, but this doesn't seem to be the ideal solution and I'll only end up with the same problem when printing something else.

Thanks

Roger

Have you tried printing a thin wall box, and measuring the resulting wall thickness with a caliper? Compare this to what the slicer thinks the extrusion width should be, and see if you are still out by 12%.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

Hi Mike.

What thickness should I use ? Do you mean something like this [www.thingiverse.com]

Does the thickness of the wall need to be some multiple of my nozzle size (which is supposed to be 0.4mm)


I'm also not sure what you mean by "Compare this to what the slicer thinks the extrusion width should be, and see if you are still out by 12%."
Do you mean look at the GCode to see how far the print head is supposed to move ??

Thanks

Roger

Edited 1 time(s). Last edit at 11/01/2013 05:33AM by rogerclark.

How thin is the wall of your gizmo?
How thin a layer are you trying to print?
How small is your extruder?

@uncle_bob

The wall isn't particularly thin, I'm printing one of the standard drail pipe sizes, which is 90mm external diameter with 2.5mm wall thickness, so that the inside of the pipe is 85mm.

But rather than the walls being 2.5mm thick they are around 2.8mm thick, which means my printed pipes won't fit into the real pipes etc.

Extruder is supposed to be 0.4mm
I'm printing at layer hight of 0.3mm

I made the model in OpenSCad and checked the dimensions in Blender. (Sorry I don't have the model file to post at the moment as its not on this computer, I'll try to post the files later)

Thanks

Roger

Ok, your required wall is 2.5 mm. Your extrusion is 0.4 mm (more or less). You can either print 6 passes of 0.4 and get 2.4 or you can print 7 passes and get 2.8. My *guess* is that it's trying to do a solid fill and on a circle there isn't much room for a zig zag.

If you are after 2.5 and it's off 12%, that's only 0.3. My guess is that it simply did the 7th pass. You could look at the geode and see if that's the case.

Uncle_bob,

I think your are right.

I noticed that when it was printing, that it only seemed to be doing perimeters and no infill (as far as I could tell).

I'll try to take a look using the online G Code viewer.

I've looked at my Slic3r settings (which are the default ones from the wizard, I just set the nozzle and filament sizes as well as the temps for ABS)

Perimeters seem to be set to minimum of 3 (I'm not sure if this includes inner and outer perimeters)

But this looks like its probably not a printer issue (more likely a settings problem with Slic3r, or a bug in Slic3r)

Thanks

Roger

I'd take a close look at the code. You may have to pick weather you want an exact outer diameter or an exact inner.

uncle_bob

BTW. The model I uploaded is just a test piece really. And I made a mistake in that the lower portion of the model is actaually too wide it should be 5mm smaller in diameter as a 90mm pipe is supposed to fit inside the top, and the bottom is supposed to fit inside another 90mm pipe.

So for some parts of the model, I need the outside diameter to be correct (the lower part of the model) and for the upper part of the model the inner dimension needs to be correct.

I may try setting the wall thickness to 2.4 instead of 2.5 in which case it would be 6 x nozzle diameter, so perhaps it would then print correctly and the 0.1mm difference is so small it would make no difference to the operation of the pipe.

Nevertheless, I suppose its a good thing to learn that I can't have objects whose size if not a multiple of the nozzle width, well not for both the inner and outer shells, i.e, I presume that Slic3r is favouring the outer size, and not the inner size.

I could try using another slicer, i.e I have Cura installed, but I've found it doesn't work very well for me, because it doesnt seem to be retracting the filament when it does a move (while not printing), so I send up with a spider's web of filamement which slighly messes up some prints.
(Note this could be because I'm using an All Metal hot end)

Thanks again

Roger

If the (what ever it is) has enough cross section that there is room for infill, you can get around the nozzle width issue. Actually that's not true with all infill patterns, but it's true 99% of the time.

rogerclark Wrote:
-------------------------------------------------------
> Hi Mike.
>
> What thickness should I use ? Do you mean
> something like this
> [www.thingiverse.com]
>
> Does the thickness of the wall need to be some
> multiple of my nozzle size (which is supposed to
> be 0.4mm)
>
>
> I'm also not sure what you mean by "Compare this
> to what the slicer thinks the extrusion width
> should be, and see if you are still out by 12%."
> Do you mean look at the GCode to see how far the
> print head is supposed to move ??
>
> Thanks
>
> Roger

Print a box that is one pass wide. You can tell the slicer to do this. The one you linked to will be fine. If you slice it using the spiral vase option, it will only be one pass wide.

I use a 0.35mm nozzle, and a max layer height of 0.25mm, so the slicer usually uses an extrusion width of 0.52mm. It would be odd if you were extruding with a 0.4mm width, as it is desirable to have the extrusion width about twice as wide as the height. The header in your g-code should tell you the actual extrusion width, and after you print the single wall box, measure the wall thickness, it should be very close to the value specified in the g-code, if your extruder is calibrated correctly.

I never calibrated my X, Y, and Z axis, and just used the calculated values based on the belt pitch, pulley teeth, leadscrew pitch, and steps per rev. With the extruder calibrated properly, I find that objects are within +-0.05mm of what they should be, for both inside and outside measurements. I usually make holes a little oversized though, so there is a little play.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

I just finished a series of experiments with these settings in the current version of Slic3r.

If you print with a 0.41 mm width and a 0.25 height for your single layer box, you measure something (say it's 0.51). You put that into Slic3r. Print another box you have a wall that's 0.65 wide. Put that in and the wall gets thicker still....

There's a bit of playing around to get things just right.

Yes, I think that is what you would expect, because you are telling it to print wider each time. If all you are doing is giving it a larger extrusion width, then it will feed more filament to get you to that width. Instead, what you need to do is back off on the amount of filament extruded, so the extrusion width is narrowed, by either changing your steps/mm, or by changing the extrusion multiplier in the filament settings.

If your extrusion multiplier was originally set to one, and you ask for a 0.41mm width, but measure a 0.51mm width, then set your extrusion multiplier to 0.41/0.51=0.80. You could also change the flow parameter to 80% directly on the printer if you have a control panel.

The layer height tells the printer how much to move the Z. The nozzle width interacts with X and Y motion. The extrusion multiplier, nozzle width, layer height, and extruder calibration all drive the amount of extrusion done. Getting them all "just right" can take a few passes. There are secondary issues from temperature settings and print speed. There's to much interaction to get it all at once with info from a single print.

That's not to say don't do this. Quite the opposite. You very much should get it just right. I'm only saying you should expect the process to take a while and set aside some time to get it all done. If you go into it expecting a 15 minute cal and you are done, your results will not be what they could (and should) be.

uncle_bob

Is there a documented calibrartion procedure to determine all of these settings ?

I presumed that Slic3r was able to calculate the amount of filament needed, based on nozzle size, filament size and layer hight; so I didn't change the extrusion multiplier from its default (which I think is 100%).

I did try printing some parts with a 0.2mm instead of 0.3mm layer hight, but found that the base layers were somewhat squashed together, so perhaps my initial Z axis height, is a bit low; however I made it as low as possible to make sure that the ABS and PLA adhered to the bed.

anyway, it looks like more calibration is needed as well as more care when making models

Slic3r does calculate the amount of filament needed based on nozzle size, filament diameter, and layer height, but if your E steps are wrong, it won't be correct. I read somewhere that it was easier to simply adjust the flow parameter on the printer rather than worrying about e-steps, filament width, and extrusion multiplier, but ultimately all of them control the same thing. If you put in the measured filament diameter, you can then adjust either e-steps, or the extrusion multiplier, as either of those adjustments will give you the same result. You don't have to worry about getting both of them exactly right, as a combination that works is all you need.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

Sorry Mike,

I'm not quite sure what you mean.

I've measured the filament input calibration i.e set the extruder to feed 30mm of filament and then measured how far the input filament moved, and its actually a bit low, i.e in Pronterface I set the filament to move 30mm at 30mm per min, (so hopefully no slipping), and the distance moved appeared to be around 27.5mm, (its a bit hard to read how far the bit of masking tape moved, using a micrometer)

So it looks like I need to update my calibration file in Marlin and re-compile / upload.

Do I also need to look at the extruded filament e.g diameter, and possibly the length extruded (as this is far more difficult)?

Or is there something else I need to do?

Thanks

Roger

What I generally did was make sure that a single wall box is printed with the correct width. Then you know that the extrusion width is what was commanded, so the amount of filament extruded is then correct. You can do this by adjusting the extrusion multiplier, or e-steps. I probably made sure that my e-steps were correct, as I don't usually use the flow or extrusion multiplier parameters.

In your case, if you want to adjust the e-steps, you can calculate a new value to try based on the old one. So, new_e_steps = old_e_steps * (desired_distance / distance_actually_moved). Calibrating this over a distance longer than 30mm might be advisable, as it will helf reduce measurement error. I'm not sure how critical length extruded out of the hot end is, as I've never done it. Lot's of calibration info can be found here: Triffid Hunter's Calibration Guide. The other calibration method that I alluded to earlier can be found here: Extrusion Flow Calibration.

I probably have more tweaking to do on my printer at some point, but for now, it at least prints objects that are dimensionally correct, which was my first goal. Considering how many things can go wrong in this whole process, I've always been really impressed with how well these things actually work. Even my first few prints were really good, and other than a little tweaking the printer has always worked fairly well.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

Thanks Mike

I'll calibrate the e steps first.

Like you, I was very impressed with what dimensional accuracy I get from my printer (MendelMax 1.5 built from a kit).

external dimensions are accurate to well less than 0.5% error, its just the wall width thing I need to get sorted out.

Thanks

Roger

I have never seen a full detailed / end to end calibration procedure on one of these printers. There are bits and pieces here and there. The biggest issue seems to be that there are many ways to do the same thing. Do you adjust flow rate or not? Do you set up this first or second. Is the focus calibration or troubleshooting (it's never just one or the other) ? Are you after top speed / best accuracy / both at once ? How good is good enough (it'll always be "good enough for me") ? What tools are available (dial gauge, sub-nanometer uber-gizmo, calipers, ruler ...)?

There are also many different hardware setups. Does your hot end work this way or that way (do you use retraction or not...). Ramps or some other board? Then there's software, Slic3r or something else. Marlin or Sprinter.

The ideal format would be some sort of Wiki. With so many approaches and variables that seems to have crashed and burned each time it's been tried. You almost need a Roger wiki, a Mike wiki, and a Bob wiki to keep things straight.

Uncle_bob,

I think you hit the nail on the head. This is one of the reasons that these types of printers are not really useful for mass consumer consumption... yet. People that are quite technically inclined usually can get them working to their satisfaction, but not everyone will have an easy time of it, and may not have the required troubleshooting skills.

I guess the good thing about the Reprap wiki that exists, is that we could all contribute to it, as others have. Triffid Hunter's calibration guide is pretty good as far a decent calibration guide goes, but I don't know if it is fully end to end.

There are many adjustments that control the ultimate filament flow rate. I think you only really need to adjust one of them to get it working correctly. You can adjust e-steps, or flow, or filament diameter, as all of these are directly related. This was likely done for convenience, and because it makes sense for the math, but it does add some confusion for the users.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

Considering all the money I paid for the free software - I'm very happy with it. In the "real world" you can spend $50K (or worse $50K per year) on this or that software package and get stuff that's *way* worse thought out.

I have a friend with a Stratasys printer. He just has different problems than us, but still has problems. Though I do have to admit is would sure be nice to have dissolvable support, which I'm sure will come someday. I keep trying to talk him into building a reprap, just so he can have access to less expensive supplies, as Stratasys sure charges a lot of money for their cartridges.

I must admit that I have been very impressed with the software and firmware. The firmware in particular has been amazing. Slic3r seems to crash a lot, but usually I can find a setting that works, and given the cost, there is not much to complain about.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

I would have to get into naming names and that likely would wind me up in court, trust me when I say that there are very expensive commercial software packages that make Slic3r look like the most stable piece of software ever created.

----

There are some filaments out there that are dissolvable. I don't remember the chemistry off the top of my head, but they work in combination with ABS.

Yes, I have seen some of the PVA filaments that are dissolvable, but they are much more expensive than the actual filament. There is an article on hackaday that talks about using HIPS as support material, and dissolving it in Limonene. I think it will be a bit before this becomes more mainstream, and at that point, the prices will likely start to come down. As I get older, I find I also get cheaper...

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

It's the Limonene stuff that I was thinking about. My thought was to make a main support from ABS and then just put a buffer coat of the HIPS over it. That way a little HIPS would go a *very* long way. The main support would be a throw away, but it would be nice cheap ABS.

I think that I've heard of people using PLA for support material as well, but I can't remember how they dissolved it, or if it was just removed like normal support, but didn't stick as well, so it left a better surface finish.

---

Mike Anton
[manton.ca]
[laserlight.wikidot.com]

Let's see, how many of poor Roger's threads have I hijacked in the last few days.... I suspect I should be expecting a letter bomb from Oz ....

Hi uncle_bob,

I do seem to have created several threads that people are interested in ;-)

No worries. I'm just sitting on the sidelines observing and learning

Cheers

Roger