Calibrating flow vs. material diameter, stiffness, etc.

[UltiSpangler 2]

I'm trying to calibrate flow rates for different materials on my S7.  In particular, PolyMide PA6-CF and PA6-GF seem to generate lines which are too wide, and so the print itself ends up ~0.15mm wider than it should be.  I can correct with several different settings in Cura (Flow, Horizontal Expansion, etc.), but which setting is the right one depends on whether I'm seeing the wrong volume of extruded plastic (affects infill as well as outer walls) vs. the extruded plastic being the right volume but just spread out in the wrong shape (infill would still add up to 100%, but outer walls might be too wide or too narrow).

 

Material diameter

One factor is material diameter, since flow volume is proportional to filament cross sectional area.  Most of my filaments are 2.84-2.86mm, where I'd expect them to be.  Hatchbox ABS is actually 3.00mm.  I have one spool of PolyMax PLA which is 2.81mm.  Easy enough to correct for; just set flow rate = (2.85/measured_dia)^2, so flow=103% to compensate for 2.81mm, and 90.2% to compensate for 3.00mm.  Not sure how fast filament width changes during filament manufacture, but it's easy enough to measure at the start of a job.

 

Feed distance vs. stiffness?

Another factor seems to be related to material stiffness.  I tested manually feeding material by pressing the "down" button for the material on the S7 front panel 100 times and seeing how far the filament moves.  (It should be possible to write a job file to move the filament exactly what the S7 considers to be 100mm, which would be easier on my tapping finger.)  

 

MatterHackers Build PLA moves 103mm.  PA6-CF, which is stiffer, moves 106mm.  UltiMaker TPU moves 92mm.  To reduce the effect of resistance in the rest of the system, I'm using a ~20cm piece of filament, and measuring from the top of the feeder up the Bowden tube in the back.  All of those filaments measure 2.85mm wide.  So I'm wondering if the difference is that the knurled feeder digs into the soft material (reducing its effective radius to the inside of the teeth) while riding on top of hard materials (effective radius is the outside of the teeth).

 

Print weight actual vs. expected?

I can calculate actual material density using a microgram scale and measured filament volume (Pi r^2 h for a given filament sample's radius and length).  Cura and Digital Factory show the expected filament length.  If I weigh a printed part, I can calculate back to the actual filament length used.  (Using Cura's filament length is better than the weight.  Displayed weight is based on the density and width from the material profile, not the actual filament reel loaded, and isn't displayed to as many significant figures as filament length.)  This *should* give me the same numbers as the feed distance deltas.  Though there could be additional variance if there's slippage in the feeder during retractions (TPU, I'm looking at you...)  That's next on my list this evening.

 

Other factors affecting XY dimensions?

Still to figure out:

• Material shrinkage as it cools (depending on how fast, could shrink the whole print, or just add stress as warm expanded plastic prints on top of already-cooled-and-shrunk lower layers)
• Nozzle actual dimensions (Pretty sure the nozzles are still in good shape; a thin walls test with PLA measures out fine.)
• Material foaming?  Fibers squishing out?  (I've dried the filaments in a PrintDry, so pretty sure it's not steam foaming.)

 

Any ideas?