Print accuracy and (under)extrusion

[conny_g 251]

Hi there,

I am currently printing a series of lego base plates for my kids and ran into an interesting observation:

 

The "pimples" need to very accurate 4.8mm for the lego bricks to stick well but not too much. It's sensitive to 0,1mm deviation, if they are 4.7xmm the bricks are too loose, if they are > 4.85mm they are too tight (and will rip off the pimples when removing them) or won't fit on at all (approaching 4,9mm).

I spent some time to find the sweet spot for two UM2+ with Bondtech extruders and 3D Solex Race block + nozzles in terms of extrusion calibration, temperature and measurements of the .STL so that the base plates are usable in a controllable way. 

But still - exact same settings, conditions, filament, nozzle etc. - the "left" printer produces them more loose (4,75mm) than the other, the "right" one where the bricks sit very (almost too) tight, with 4,80-4,85mm of pimple diameter.

In the last test print of a 2x4 pimple version of the base plate (takes only 10mins to print) I tried to simply tighten the feeder pressure screw of the Bondtech tighter by about one turn.
And interestingly the print got better, instead of 4,75mm pimples I received 4,8mm.
In one of the tests before I modified the e-Steps from 510steps/mm (calibrated 100mm feed) to 520 (which is +2%) and it also got better and produced 4,8+mm pimples.

In another test earlier I found that also +5 C more temperature made the accuracy better, but the print quality worse (slight stringing).

So it seems if you want to print measurements like these as exact as possible it's even feeder tension that will influence it.
The extrusion obviously varies very quickly by 2+% if you change feeder tension or temperature.

I admit it's a rare use case to need 0,1 and less of accuracy, but an interesting learning.

Which way would you go, increase feeder tension (I see the danger of grinding doing so) or would you increase the eSteps/mm by these 2% to compensate for this minimal "slippage" of the feeder?
I think the increasing the eSteps is the better way.
 

Edited February 6, 2019 by conny_g

[yellowshark 153]

Hi @conny_g thanks for that, good work you are doing there J. There is some work I do which has a 0.05 tolerance so it is a subject close to my heart. Interesting point on the feed tension, that is something I had never considered. Interesting also on the 5 degree temp increase, assuming that the previous temp. was pretty optimum,  as I have always worked on the basis of slow and cool for accuracy (and consistency).

My process now when accuracy is paramount, having played with this for several years, is

a)      Always use the same filament – i.e. manufacturer and colour

b)      Always use the same slicer and slicer settings

c)      Get as close as I can to my normal ambient temp. i.e +/- 1.0 c

 

d)      And then use e-steps to dial in the accuracy if needed

 

e)      Check every few months and immediately before the 50 micron job, with the reel that will be used which will either be new or hardly used, so there is no impact by the wind on the reel.

 

And now I will need to focus my mind on the feed tension and see what I think. Excluding the latest  models, as I do not know much about them, I know that the feed system on my printer is better than on the Ultimaker’s but I still set the tension by feel, i.e. if the filament surface feels right then it is right! Maybe I should get more scientific on that one 🙂

At the end of the day my view is get everything optimum and consistent and then fine tune with e-steps if need be.

[conny_g 251]

Thanks for your feedback, that confirms and extends what I have learnt about accuracy the last days.
I agree that it seems best to give it the final perfection by adjusting the e-steps.

Just trying that to tune that a little bit with every new piece. In the last iteration the difference between the lego pimples was already much less, having increased the e-steps by 2%.

[geert_2 557]

Changing feeder tension changes the depth of the indents, and thus the effective diameter of the feeder wheel. And it changes the partial slipping of the filament (=filament speed not exactly matching the wheel, due to the indents being stretched by the resistance in the feeding traject).

 

So I can see that both variables (depth of indents, stretching of indents) have an influence. And they will change with material stiffness too. The wheel will bite deeper into a softer material, and it will stretch more.

 

Additionally, speed and temperature are going to play a big role too, because they affect flow-rate, and mechanical accuracy (overshooting, ringing).

 

So I guess there will allways be quite some trial and error for this sort of jobs, especially if you are going to print it in different colors and materials. Which usually is the first requirement of kids, of course.   :-)

 

On the Lego-website I once read that the development of "simple" bricks requires quite a lot of precision engineering, to get a good fit.

 

[yellowshark 153]

Ok so the fact that I never touch my tension, unless I am occasionally maintaining the knurled brass wheel, is actually a good thing! Maybe the point is set the tension to a an appropriate tension so it is not slipping nor gouging, then leave it alone. I touch mine probably once a year.

 

Oh and in case I inferred otherwise, I rarely touch my e-steps; only when the accuracy goes outside my limits, which is very rare. Then I will check everything else and only go to the e-steps as a last resort; playing with the firmware always raises my blood pressure😡