Welcome to the SeeMeCNC Community

Just like the subject implies, it would be great to integrate Automatic scaling algorithms based on a filament's known shrinking values to get a more precise printing/engineering match. Currently I am waiting for my printer to be shipped then built.. I have no experience printing and don't know if this kind of feature exists already but just looking in Matter Control and Repetier-Host I see no Material/Filament Auto Tuning /Offset Scale-Model-To-Filament Shrinkage Values.

From many videos and reading through forums I hear that one of the reasons why people don't often print big or can't effectively print big is because the printed object becomes less accurate.. I feel that some auto model scaling could help this.

Shrinkage itself isn't a big deal. What it results in can be, shrinkage of hot plastic deposited on cooler plastic can cause warping of the model which gets worse as models get bigger. You cannot counter this with software.
The other people people blame on shrinkage is holes coming out undersized, but IMO the real issue here is the filament cutting corners and it tends to vary from printer to printer, so consistently correcting in software is tricky.

Polygonhell wrote:Shrinkage itself isn't a big deal. What it results in can be, shrinkage of hot plastic deposited on cooler plastic can cause warping of the model which gets worse as models get bigger. You cannot counter this with software.
The other people people blame on shrinkage is holes coming out undersized, but IMO the real issue here is the filament cutting corners and it tends to vary from printer to printer, so consistently correcting in software is tricky.

Hi Polygonhell I just finished watching a video stating the exact same thing about hot and cool layers causing warping. However some other valuable info on filament specs are posted on Taulman's website ... here is an example http://www.taulman3d.com/bridge-nylon.html Some reports I've listened to stated that warping wasn't as big of an issue when using the hotbed but when printing large the size throughout don't quite match.

I think both filament shrinkage and controlling layer temps could help solve the printing accuracy .. I understand the hotbed helps and many of the bigger Industrial printers have the printing area enclosed to keep the filament cooling slower and dust/particles out.

I understand the hole size issue as being largely about using INscribed polygons as the approximation for circles.
It is acceptable once you know about it and make the required allowances in the design, i.e. it is little worse than the usual need for tolerance allowances.

I am getting within 0.5% on 10 cm parts, which is certainly not machine shop quality of even mid 20th century, but good enough for what I do (hobby tinkering).

10 cm seems close to the limit of what I can get ABS to stick and STAY STUCK.
I haven't done the arithmetic, but I acknowledge that ~220C ABS cooling to ~90C will shrink with enough force to either break the ABS/glass bond or the glass itself at some length.

Once that happens at the corners or edges the part curls up - that ain't a scaling issue anymore (-:

What to do, what to do...
Plastics that shrink a lot LESS would help, but given that the plastics industry doesn't have to worry about this I doubt that we will see any progress on that front.

Maybe that is the REAL problem, we are using materials that have been developed to work very well in "all at once" production, it is probably an advantage that parts shrink quickly and release easily from molds, etc.
Our "additive process" takes TOO MUCH TIME !
I don't see that changing by any HUGE factor in the near/mid term.

Where I think "software" might be able to help with this is in allowing more flexibility of the part during thermal expansion/contraction.
I am not convinced that infill ALWAYS needs to be bonded SO firmly to the outer shell of many parts, or that the infill needs to bond SO firmly to previous layers of itself.
I think there may be a lot of cases where it could "be there" just behind the surface and "be available" to support the surface if/when needed.
Some parts could benefit from more flexibility - SOME prototypes that don't quite fit might be a bit more useful (to the development process) if they were more flexible.
It could be a "feature" (-:

Support material is already kinda/sorta like this, though it is probably optimized for weight support and minimum material usage, i.e. ONLY in the vertical direction and UNDER overhangs.

The software in our heads ?
Yeah, maybe more thought devoted to the design of the part would help a lot MORE than mere code hacks, but that becomes skill dependent and we all HATE that. Right ?

Arduino due is out. And the price has gone down very significantly (15 usd for the mega equivalent board) maybe the enhanced processor could make better calculations and make better circles.

teoman wrote:Arduino due is out. And the price has gone down very significantly (15 usd for the mega equivalent board) maybe the enhanced processor could make better calculations and make better circles.

The polygon approximations of circles are made by the CAD (and slicer) programs, just as curved surfaces are approximated as a mesh of triangles - they are smoothed out well enough by the melt.
If you use polygons with too many segments a different problem pops up, lots of little straight lines with stop/pause/start in between.
For what I do octagons are a good enough choice, the filament takes a bit of a short cut around the apex corners anyway, so they are "blunted" octagons with enough diameter where it matters.