Problem specifying model for a partly nested ANOVA design

[#1]

My study design is mixed-model with one fixed between-plot fixed factor (A), a random plot factor (B), nested in A (B[A]), and two crossed within-plot fixed factors (C and D).  I'm able to enter a full factorial model that includes the 3 fixed factors and all their interactions, using either the Least Squares or Mixed-Model Personality.  I'm also able to add the random factor (B[A]), but things go awry when I try to add the interactions between this random factor and the 2 fixed factors and their interaction (C x D).  For example, when I cross B[A] with factor C, the program generates the factor B x C[A}, which is incorrect.  It should be C x B[A].  The same thing happens when I try to cross D or C x D with B[A]. I thought that maybe the program would automatically generate these interactions in the background and use them as denominators for the F-ratios of the fixed factors and their interactions.  However, only factor A has the correct denominator (B[A]); the other fixed factors and their interactions use the residual MS as their denominator, which is incorrect.

I'm a novice JMP user and must be missing some trick for adding the random x fixed interactions to the model.  Any help with this would be much appreciated!