not sure where the error is, but moving my script to 1.2 seems to have gotten a more accurate result.
we are a mass manufacturer (2 million pieces per annum, and at least 400 new models being printed each month) so our general weight category for gold goods is under 3.5gr.. a 10% at that weight is negligible and easy to rectify in cleanup.
We havent really paid attention super accurate weights until this thing started with these bangles and we had a cad weight of 45gr but a part weighing 61 grams... its my job to isolate and fix that discrepancy, so its part rhino volume calculation factor (which i believe is accurate in "simple" geometry, but can induce as much as 5-10% error in more complex parts with internal convolutions)
Our magics install and our rhino report different volumes on the same parts -but- to qualify, the difference is noted AFTER we clean the models in magics fixer. so some degree of difference of weight in nurbs volume, vs the weight of the horrific mesh rhino spits out vs the cleaned mesh from magics. We now use 3dcoat exclusively to voxelise the mesh, then bring back into rhino to volume and the error is marginalised at about 1.5 to 2% on large parts and a negligible amount on smaller models.
next up is the values used in Matrix as "standard" SG. of course the software is written to use the most common SG's but they do make provision for editing the sg to match your actual sg. so an exercise to measure volume on parts cast was started and our own SG's fixed. now output weight is closer than ever.
the last part of this exercise was determining why the weight goes up in the resin.. for both the pic 100 and b9 resin i added roughly 10% to SG and now we are withing total error <2%.
I dont imagine the SG of the b9 resin is innacurate, i ascribe the error to small difference in scaling and bleeding. Again in small parts >4gr this will have gone unnoticed previously.
to speed up the process of calculating these things i wrote a grasshopper script to adjust a master list of YG, 925, brass and resins, and the team are tracking predicted vs actual. once they values on these are locked down we will add SG for white and rose golds, (we have 4 rose golds and 3 white golds... every step a step closer to dialling out errors... as you might imagine a 0.1gr overshoot is nothing when you selling that one ring at retail, but on a delivery of 50 000 pieces it will add up..
script is available in my grasshopper thread in 3dcad, cannot claim ownership as its assembled from bits of scripts scrounged from GH forum. looks like so. I wrote it because the designers/cad operators are too busy to have to leave rhino to use my excel sheet calculator. Script will eventually update to factor cad>resin>cast>vulcanise>cast>cleanup weights. aim is to get under 0.1% error margin. https://www.youtube.com/watch?v=GUJTfoqmJhk
Matrix8.0, Rhino5, 3DCoat4, 5 Axis MinitechMinimill2, Envisiontech Perfactory, B9 1.2.
Be the Shepherd, or be the Wolf... Just never be the Sheep.