This has been a really interesting print for me, and a very frustrating one; Thingiverse 36661 Key Chain Clip. I must have printed this about 20 times, often stopping half way due to it messing up. The full size print is too large for a keychain so I’ve been printing it at 75%, scaling in Slic3r. I tried all sorts of setting options, extrusion multiplier, filament width, speed etc. The one on the left was the best I got, the wall was weak, had holes, it seemed to be overextruding, different temps didn’t help. I was getting frustrated when I decided to try changing the scale, to 80%. Look at the difference just an extra 5% made, pictured right.
So this is an example of how scaling a model can produce mathematically bad gcode for your printer. I spent way too long thinking it was my print settings messing up.
I’d love to see the gcode for those! The only thing that makes sense to me would be that your scaled version is matching the extrusion width of your nozzle…
@John_Driggers and I’d be happy to send it to you. I’m printing it again right now. Once it is done I’ll see if I can PM it to you.
Never thought this could happen, Thanks for sharing!
I was getting bad prints until I made the Layer height exactly correspond to my Z axis steps, so 0.2 gave me a number of steps that had a fraction over eg 237.45 whereas 0.25 gave me an integer result eg 220 (figures are only an example).
I could see the same thing applying to the X and Y, may be you hit something like that?
Isn’t that a cheap printer problem? Spent some money one of these laser and liquid printers on kick starter like the FORM1.
It is a home built Prusa Mendel. So, yes 
@Geoff_Drake
Well, a Prusa may be cheap to build, but can produce stunning prints. Z-Axis banding can be caused by using layer heights that result in your z-motors having to hold on a microstep rather than a full step - so you solution works a treat.
@Matthias_G Resin printers have their own challenges, not the least of which is that the material itself is not very strong!
I did one cheap reprap as a school project with my students, buying a proffesional one could have given us better prints, but never so much fun and learning! 
It was fun building mine, but at over 2 years old it is showing its age. I am in the process of building a Hadron. Then I have plans to upgrade the Prusa.
@Geoff_Drake those rounding errors only apply to the z axis, you don’t stipulate the horizontal resolution in the same way. @Richard_Blackman looks like things are still getting too hot at the top there.
Did you check to see what the minimum temperature was you could reliably extrude at?
Amazing!
That’s pretty impressive. I wonder how many of the little flaws I’ve dealt with over the last few years could have been explained by this effect.
Yes me too!
It definitely look like over-extrusion – lots of “shove outs”.
You should run the gcode through a path visualizer and see if the slicing for both sizes have similar construction of the walls – that is, the shells and the fill patterns look similar.
My guess is that by scaling up from 75% to 80%, you’ve increased the size of your model by about 8% – and, if those walls are only a few lines wide (and from the looks of it, it looks like it could be as little as 5 or so line widths), an increase of 8% isn’t nearly enough to add another line-width worth of fill (or perimeter) - so the resulting gap can absorb the over-extrusion.
I dunno for sure, but I suspect the resulting plowing actually creates air pockets from the nozzle riding the bumps, causing poor layer adhesion where the fills criss-cross, or where the perimeters don’t align well along the layers.
A lot of interesting comments here. Just to be clear, both these prints used exactly the same settings, LH, filament width, temp etc. The only change was the scaling percentage. Most prints are probably done at 100% so this wouldn’t apply. The lesson here is that if you are scaling your print and it comes out crappy try adjusting the scale a little.
I’m happy with the 2nd print, I now have some working clips, so on to the next project.
Is there a way to scale a model and still have it print right? Or do you essentially have to redesign the model at the smaller scale? Does this mess happen when you scale a part to make it larger?
Good question, haven’t tried scaling up but I would doubt it would be such an issue.
@Sean_Ong if its not a printer issue (I still think it is) , its an issue with slic3r I suspect. There’s no reason why you shouldn’t be able to scale an stl and still have it print nicely. I’ve not heard of this before or experienced it myself when scaling models in Slic3r for that matter.
@Tim_Rastall @Richard_Blackman thanks. I was worried for a second. I’m considering getting a 3d printer and I’m just looking around at what issues exist.