Dimension issues

Hey everybody!

Been printing perfect sculptures and non-technical models, No stringing, everything LOOKS fantastic, but when I come to print parts which are meant to fit together (aka perimeter width calibration block here [reprap.org] ) I find that following the instructions does NOT have any effect on the size of my parts!

I've increased and decreased the perimeter width by almost double with no effect! My parts consistently measure incorrectly and I cannot figure out how to get them to work correctly...

They are neither undersize nor oversize, there is not "too much plastic", because when I print solid blocks they come up looking perfect and are flat, it's just that the outsides are always about .4mm over, this part in particular has a 10mm block which fits into a 10mm gap, but the block comes up at 10.4 (measuring with calipers) and the gap comes up at about 9.6!

Any tips on how to correct this would be VERY appreciated, I have just now tried reducing the flow by 25% to see if I get an acceptable part but any calibration blog says this is "just tricking the printer" and not really the way to do it!

Anyone shed some light on this problem? it's NOT shrinkage, because one of the parts is oversize, it's just like the perimeter width is wrong, but changing it does not help!

Cheers,
JazzyB

What slicing software are you using? Perimeter width will have no effect on final size of the part.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

Hi NewPerfection,
I'm using Skeinforge 50
In the calibration tutorial at reprap.org (which i assumed was a credible source but, may well not be as the later tutorials state things such as "note: need to include information for calibrating this feature") it states that:

"If the two parts do not fit together at all, turn your perimeter width up."
"If the two parts fit together but rattle around (ie are loose) then turn the perimeter width down"

I'm from a CAD background, running milling machines and the like, so I have a fair understanding of geometry and, one would THINK that if you turn up the perimeter width it would output more plastic AND make the threads more "inbound" on the part, meaning the overall effect would (as you say) be the SAME sized part, just with thicker/thinner threads, is that how you believe it works?

And if that is the case, then what parameter in skeinforge SHOULD I change to get these parts to mate properly? it sucks at the moment because I need to make holes drastically oversize for shafts etc to fit properly.

Belts are tight but not too-tight, printer is very smooth, quieter than my epson inkjet that's for sure! it's just making all walls "too fat"...

Thanks in advance for any help ;-)

JazzyB

AND...

Just to clarify...

What I've read in other blogs on calibration confirms this, I have my E-steps/mm calibrated as closely as I can using digital calipers, and I have set the perimeter width to 1.95 times the layer thickness (height)

So with my layer height at .25mm, when printing an object with a SINGLE wall, I measure this wall as being around .48 - .50mm thick... This is what I'd expect, and turning the perimeter width up to 3 should make this single wall .75mm thick - correct?

I'm lost then? There's obviously another parameter here... it's NOT scaled incorrectly, because the male block is too large AND the female block is too small. My perimeter width and e-steps are what I'd consider to be 'near perfect' - at least not enough to make the parts respectively .4mm undersize and oversize... I've checked over pretty much all of my skeinforge settings using the skeinforge manual (oooooh boy that could use some work, that manual is cryptic!) and now I'm just lost?!

Very few people use SF any longer. People use Slic3r, Kisslicer, Cura, SFact for a few. I would recommend using Cura or Kisslicer as they both produce nice parts and are fairly easy to use (Cura is SF50 made faster and easier with only a few settings).

---

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

:-O

Google led me to reading about SFACT all this afternoon - I was a little hesitant to take the plunge still, saying to myself "but the folks at reprappro have their own version of skeinforge with settings modified already for my printer" and althought I updated it to skeinforge 50 (to get "brim" - ah brim... what an awesome idea that was!) I have spent a lot of time learning it (a LOT of time) and really didn't want to have to start all over again... but...

Sounds like I will!

Although I would have thought skeinforge would at least have given me the correct part sizes? given that my thin-wall calibration is spot-on, my next test is either going to be a different slicer (was thinking SFACT, but I'll now look into cura also, thanks sublime ;-) OR - printing the part with absolutely ZERO infill (not even loops) - I'm really thinking it could be the loops which are pushing out my boundaries, because if the thin wall calibration is spot-on, the WHY does it seem every dimension (inside and out) has "fattened" by .4mm? it's not scale, it's not likely a hardware issue, I'm now thinking the outside walls are being "pushed" outward, and I've tried 20% infill and 0% infill but I always have 2 concentric loops just inside the perimeter... perhaps...?

I'll test that theory now and report back, while looking into Cura!

Cheers guys! Much appreciated ;-)

If you are using one of those hollow cubes people have made to calibrate the flow you will find the part has the wrong dimensions because the slicer draws the line right down the center of the thin wall (half the extrusion inside and half outside the line). It is much better to take a model of a solid cube and use 1 perimeter (0 extra in SF), no top or bottom layers and 0 infill. This should give a you a single wall cube with the correct outside dimensions and one perimeter which you would measure and adjust the flow to get the correct thickness using the extrusion multiplier or 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

SUCCESS!!!

Sort of...

What you've just said I understand completely sublime, the one-and-only reason I see for using one of these thin-walled-cubes is (and could be done with a solid cube except you would need to stop the printer before the top infill) to test the ACTUAL width of a thread (to make sure your extrusion is what you have programmed it to be)

So I used the thin walled cube only for this purpose, to measure the thickness of one wall, not the overall cube itself ;-)

BUT - printing the male/female calibration part from my original post using ONLY perimeter, no extra shells (or extra loops, whatever terminology is correct) and no infill I am now getting (aside from a sliiightly warped wall due to some shrinkage) the two parts to measure 10.00 - and are a nice fit! (although with walls that thin, they can easily compress and be a nice fit no matter what)

So I think the main problem is that skeinforge is placing my "extra shells" too close to the perimeter and when the perimeter gets layed down (note, I lay down Loops>Infill>Perimeter, so the loops have a little time to set before the perimeter gets layed down) so perhaps the perimeter gets layed down too close to the loop and "squashes outward" a bit - resulting in a all-round "fatter" part...

I think I need to fiddle with infill or perhaps the infill width-over-thickness.... Or just give up! and try cura - any idea on what setting to change if it's NOT the infill width over thickness?

This might be a shot in the dark but are you sure you have your filament diameter and nozzle diameter numbers set correctly? All the actual lines of gcode that's generated are based upon these pieces of information. For instance if you want a cube that is 10mm wide, and your nozzle diameter is .4mm, it starts the part laying down a bead (circle) that is .4mm in diameter. We then have 9.6mm left for the first line of the perimeter, so it needs to move the nozzle along the y axis exactly to a centerpoint of 9.4mm, so the tangeant edge of the circle stops at exactly 9.6mm to make our 10mm part. This is an exageration mathematically, but it's how the printer has to move in order to make true dimensioned parts.

If we have our nozzle set to .4mm and you actually have a .6mm head. I've drafted up an example of what I'm talking about which you can see here:

mortinus Wrote:
-------------------------------------------------------
> If we have our nozzle set to .4mm and you actually
> have a .6mm head. I've drafted up an example of
> what I'm talking about which you can see here:

if you set the nozzle to 0.4mm, the gcode will tell the printer to extrude enough plastic to fill a 0.4mm wide line. If the nozzle is wider than that, the line will still end up at 0.4mm wide, because that's how much plastic is being spit out.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

Yes Cameron exactly. That is why gcode is so important and I was saying that if you have it set incorrectly (saying you have a nozzle smaller than you actually do, you will end up with the issue I detailed.

@mortinus you do not understand. Like Newperfection said the nozzle diameter will not make the path larger or smaller. The software determines the volume of plastic to make a path the height and width you want and that is how much it puts out. Since the nozzle is X distance from the last layer and the volume is set via E codes your path will end up the path width calculated regardless of nozzle diameter. Now obviously the nozzle diameter will determine the max width at a certain rate but it will not do as you are saying and some how add extra material that was not called for in the E values. Only having your e steps, extrusion multiplier, or filament diameter set wrong can do that.

---

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

There is no setting in skeinforge for a nozzle diameter anyways!

There is layer height and "layer width over height" (which is specified as a ratio)

For my .5 nozzle I'm using a layer height of .25 and a width over height of 2.0 (giving me an oblong shaped 'thread' of .25 height by .5 width)

I really should take the time to go through the gcode, I'm STILL having hassles with this :-/

If it were skeinforge calculating it wrong this whole time and I'm making, say, a 10mm cube, with my .5 thread width (physically measured at .5 and programmed at .5 thus, calibrated correctly regardless of nozzle diameter) I would expect the code to be something like

G1 X0.25 Y0.25
G1 X9.75
G1 Y9.75
G1 X0.25
G1 Y0.25

All agreed?

I'll go check ;-)

*Goes to check*

Well, that's basically it...

so the positions on x and y are correct for what I've programmed... only thing left is E? damn - cos... like i said, a single thread "wall" measures exactly .5 thick which is what I expect, and my E-steps, I've configured using calipers, I'm confident it's accurate to within AT LEAST .5 over 100mm (I'd like to think it's pretty much perfect)

Ummmmm..... - I do not get it?? These parts are .4 (.2 per side) out of spec, giving a total .8 interference fit... that's pretty bad!

Next lot of x/y moves shows:
X0.688 Y9.313 (and so on)
(I should mention my INFILL width is 1.5 over thickness - giving a width of .375

These x/y figures are indeed .25 + (half of perimeter thickness = .25) + (half of infill thickness = .1875) =.6875
Making them correct...

The code's all good :-( which means it's obviously NOT calibrated right!

*Stamps feet*

I'm just gonna have to start again (again) seriously a pain this reprap pro mendel... highly un-recommended at this stage...

BUT!!!

Extruder steps per mm is right!
And the single wall thicknesses are good!??

Even if I recalibrate I'm sure I'll just end up with the same settings, therefore same results?!

Appreciate all the thoughts so far guys, keep throwing ideas at me!

Filament diameter's still consistent enough, it's actually .03 smaller at the moment - so should see the opposite effect if any!

Extrusion multiplier - I'm calibrating for ONLY this material at this stage (natural ABS 1.75mm) and so therefore the extrusion multiplier is 1.0 (seen lots of people recommend .85 but this is only if you would normally be printing with PLA, and I am using almost exclusively ABS, E-steps was calibrated using this material)

E-steps/mm has been calibrated using a piece of tape around filament, feeding 100mm and checking with calipers, repeating 3 times, exact results, haven't tightened/loosened extruder idler since...

Oh snap!

Wait - I'm using my .5 nozzle, which has like a .65 free-hanging extrusion diameter...

... to try and print .25 x .5 - that's never gonna work, right?!

Damn! I'd like that resoloution! But I'd MUCH rather dimensional accuracy! (on most parts anyways! I did get this to prototype engineering parts, not to print myself a barbie or geared hearts!)

I'm gonna go down that road... just read another post of yours on here sublime, helping out the "blue PLA" issues for someone, just hit me... argh! why did that not hit me like 2 weeks ago!

I'm almost sure that's got SOMETHING to do with it - it's a physical impossibility for a .65 extrusion to take a .25x.5 shape... oh man, don't I feel like a dope :-P

I'll report back - Cheers everyone ;-)

Its only 0.65 before it is stretched so it is by no means impossible to make it 0.25 by 0.5. If you are trying to extrude too small for your nozzle you get effects due to over stretching the filament. Holes shrink more and corners become more rounded, ultimately it starts snapping when bridging sparse infill. Other than those things it doesn't create dimensional inaccuracies.

---

[www.hydraraptor.blogspot.com]

Right...

Thanks for clearing that up Nophead ;-) your blog's been a real help getting started with all this!
I like using .25 also because it divides evenly into most parts - so good news that my nozzle is capable!

I'll still try printing at a lower resolution, don't know too many other things I can try?

You're pretty much fine with layer heights down to ~0.2 with a .5mm nozzle, much below that and you start having issues with spanning sparse infill.
You can certainly print lower layer heights with it, but you start having to print with 100% infill when you get in the 0.1mm range or less.
I usually try and keep the extrusion widths to +/- 20% of the nominal nozzle diameter, and that seems to work pretty well for me.

Ok, lower resolution, not much difference, a little better but problem remains .6 interference fit...

What I did change that gave me a male part of 9.95 and a female of 9.95 (so VERY close, acceptable for now!) was to change the fill order from

Loops>Infill>Perimeter (recommended by someone to give the least shrinkage of perimeters, giving something solid first to "hold the perimeter out")

to

Perimeter>Loops>Infill

This is I think the default? But anyways, the main point is, the two mating parts now fit together with relatively no force (yay!)

So I'll continue down this path for a bit and see how it turns out for other parts. Going to change back to .25 height (cos it's much smoother and to me, more logical when working with metric parts - my last part came up 9.8 high because that's where the final slice was at that resoloution, although the part is drawn at 10.0 high...)

Missus is moving into my bungalow for a few months as we are building a house at the moment and her family just sold their house, so, reprap's gotta go to the factory, meaning I will only be able to use it occasionally, it "disturbs the cats"...

So progress is going to be slow for me for a while but thankfully I'll have a nice big garage or study or spare bedroom JUST for reprapping once the wait is over ;-) (Looking forward to that!)

Thanks for all the info again, what order do you pro's print in??? Out of curiosity... Loops>Infill>Perimeter seemed a very good idea at first but it just wasn't working obviously.

Cheers!
JazzyB

Interesting. I can imagine it coming out a bit bigger when doing loops before the outline because when you do outline first it spreads equally both sides of the nozzle but when there already is a loop on one side it is going to spread more in the other direction. In fact I have noticed myself that holes come out smaller when using two outlines.

I wonder if any slicers take this into account.

---

[www.hydraraptor.blogspot.com]

Yeah, that makes sense... I was thinking along the lines of "thiway the plastic has time to contract a bit after its laid down" but I think you're probably more correct in your thinking... It has nowhere else to go!