OK, apologetics first, i am not a protein chemist. But I ran this molecule in its entirety (it got truncated by the programs) so I diced it up into many many pieces, running each separately and trying to figure out how the pieces might fit together. This diagram (on the right) is the one I will use to create the octadecamer, the ball and stick diagram on the right is what i think is pretty accurate (and it matches some, but not all of the images I found in publications on SP-A, and it is just a little more accurate than those that make the CRD unit a round ball and the neck region a spring.  It seems that only when I get the structure of the collagen-like portion along with the coiled coil of the neck to i get a structure that sort of fits what is seen with shadow cast TEM images (also published in peer-review journals). So this is what I am using.  feel free to criticize.  ha ha.
Here is a composite of several shadow cast images of SP-A, you can see that there is similarity. The dissimilarity with the ball and stick model might come from the fact that there are two configurations, open and closed, and that the closed configuration (which looks like the shadow cast molecules are in) is related to ion concentration. Almost all models of SP-A say there is an important “kink” in the collagen-like portion, and I am guessing that this is important for the images seen in traditional TEM of the alveolar type II cells which contain putative surfactant protein-A granules.