When you added more complexity, did that include sheet metal parts and additional loads? The errors sound like they are telling you that some connection/fixture/load is improperly attached to a shell entity. The error for the frequency calculation has nothing to do with the bonding condition; it has to do with predescribed loads.
1.4 million Degrees of Freedom is very high. For reference, the cut-off between the more accurate Sparse Solver and the Iterative solver is 100,000 DoF. Running such a complex mechanism all at once may be unfeasible. I would suggest running parts or sub-assemblies individually and recording their results on boundary conditions, such as where they are attached to another part, then inputting those results as base excitations in another study.
Thank you for replying I really appreciate you taking the time to answer. The error popped up when I added some solid parts to my assembly.
All the errors went away when I removed the remote masses from the frequency calculation and I achieved the solutions I was looking for. I did not consider the remote masses as external loads but apparently they are.
Since the frequency analysis is only used as a check on my system it served its purpose. I am unsure on why the remote masses affected it since early in the process they seem to solve fine.
Now I have moved on to using the model in my linear random vibration analysis.
Thank you for your suggestion on breaking down the system I understand how it would work on a strictly vibration analysis, but unsure if it would work on the random vibration analysis.
It would be unrealistic to do it for a random vibration with a reasonable degree of accuracy. You might considering calling your VAR; they could suggest methods to cut down on the total number of features/elements. Perhaps you could also consider outsourcing it if the entire size of the assembly is too large for SolidWorks to handle.