Learning Forum

[rhodes4]

I have been using Maxwell for many years; first as Ansys Maxwell and now as Ansoft Maxwell. One thing I have noticed is that when I compare test cases where I have anayltic solutions, Maxwell is always a little bit low, like it is missing some of the energy. I have tried the things I know about like decreasing the solution %error and making the region larger. The answer always seems to converge on a number a few percent lower than the analytic fomulas. Here is a recent example. I modeled a single turn cylindrical coil. The thickness of the cylinder is .054 inches, the length is .648 inches, and the mean diameter of the cylinder is 2 inches. Maxwell converges fairly quickly on an inducatnce of 62.7nH. I have three different analytic formulas for short, cylindrical coils and they all give 65.4nH. I have seen this same thing with other designs and also when using the electrostatic solver to compute capacitance of known gemoetries like parallel plates and coaxial cable. Maxwell is always a little bit off. Is there some basic setting I have incorrect?

[Timos]

Hi, for the low-frequency range, Maxwell typically provides reliable inductance results. I recommend refining the mesh and considering the use of the skin depth mesh operation for better accuracy. Another approach would be to use Q3D to validate the results as a cross-check.