CST - General setup / choosing a proper solver

Dear community,
I'm further looking on help / detailled manuals about CSTs Low Frequency simulations. In the last two month I simulated a rather simple model of an cylinder capacitor (or short coax cable) to determine R, L, C and G as well as 3D-Fields of E and B. Therfore I dealed with ES, LF-FD EQS, LF-FD Fullwave, HF and pRLC solvers. For example the capacitance varies from 0,19% up to 1,1% accuracy in comparison to the analytical solution - within one and the same model!
For my PhD studies I will have to build a power electronic up to 36kV voltage supplied by AC from DC upto 500kHz. The goal is to reach at least 0,3% accuracy in transmission over all frequencies. Thus I want to simulated as much variants as possible to reduce the prototyping. So far, after simulating my simplyfied hands-on-model, I don't belive in CSTs accuracy when simulating model without analytical solution. This is driven by my experience on the variety of accuracies of the different solver as I mentioned above. May further (mathematical and physical) information about the solvers would give my an better understanding.
Against the backround you may give me an advise for choosing the right solver? Or do I have to use different solvers side by side? Also I urgently need documentations of the solvers pyhsical implementation. Thanks in advance! Tobin

CST Studio Suite ​​​​​​​