Keith Gregory wrote:All, Well, you have to be a bit careful in models with point currents. In theory, a coil composed of point currents has an infinite inductance. As you get closer and closer to the point, the field gets higher and higher. Ampere's loop law says that B=mu*i/(2*Pi*r) where r is the distance away from the point, mu is the permeability, and i is the point current. So, if you were integrating energy out to some particular radius, R, away from the wire, you'd be integrating B^2/(2 mu)*2*Pi*r from r=0 to R. Expanding this out a bit, one is integrating mu*i^2/(4*Pi*r). The 1/r type of integral doesn't converge if one of the limits of integration is zero. What this means to finite element results is that you will get reasonable flux density predictions with point currents if you aren't in close proximity to the points (e.g. at points inside the toroid); however, the stored energy (and therefore inductance) that you'd evaluate with point currents is highly dependent on mesh density. For example, I've attached a model that's sort of like a coax cable--a point current in the center of a 1 cm radius circular region with A=0 defined at the boundary. For an increasing mesh size, the stored energy per meter is: 0.1 cm mesh == 3.903e-07 J 0.05 == 4.833e-7 0.025 == 5.354e-7 0.0125 == 6.128e-7 0.00625 == 6.740e-7 0.003125 == 7.391e-7 Dave. -- |
Attachment:
zip00051.zip
Description: Zip compressed data