When simulating MSM photodetectors under illumination I face the problem of slow (sometime very slow) convergence. Anybody knows how to get around this problem?
It is not clear if the convergence rate (the rate by which the residual error of the equations is decreasing) is slow or the CPU time of the simulation is large. The later could be due to the large number of discrete unknowns.
It is the first that I am talking about, the residual (or the correction to the variables) to the equation decrease very slowly. Even sometime it shows certain oscillation.