I would suggest that "head and nut diameter" is on the outside of the two clamped components, determining where the bolt load is transferred to them; and "separation test diameter" operates where the two components contact each other, in the middle of the joint.
I agree with Jonathan. I don't understand you question even. can you please elaborate what is your expectation..?
Both terms are considered separately.
Fastener Head and Nut Diameter: it is used to specify a diameter for Head and Nut in case of Bolt type of Fastener or diameter of Head only in case of Screw type of Fastener.
Separation Test Diameter: it shows an area, within which normal stresses are checked during analysis(only to ensure stress which is applied through Bolt/Screw is compressive stress), is created through inner diameter=hole diameter & outer diameter=separation test diameter.
generally value is taken twice of a value of hole diameter.
Maybe this helps for now. Search these threads (no pun indented) as there is much more information to help. Look for Roland Jakel's posts.
by DEFAULT :
All the diameter numbers can be changed manually after selecting the references.
I would propose that the fastener diameter you select should be the stress diameter. See published tables.
It is up to you how you change the head/nut and separation diameters. Defaults are reasonable for most purposes.
Sensibly, the fastener diameter cannot be bigger than the hole. The nut/head diameter cannot be smaller than the hole(s), the separation test diameter cannot be smaller than largest hole.
For a bolt, both holes the same diameter:
For a screw:
Separation diameter, what it looks like and where it is :
Spring stiffness that keeps the opposing separation diameters annular regions apart:
Fastener head/nut diameter what it looks like and where it is
If you use contact then there is no longer any purpose for the separation springs and they are not created. The contact provides the stiffness.
Hint. We tend only to use the separation springs when building a model. These are linear and therefore quick but there are many reasons why the results could mislead. For the complete model we always use contact which is non-linear and takes longer.