Electronics

Electronics

EM/Circuit Co-simulation

    • helloycheng
      Subscriber

      Considering two dipole antennas, dipole A and dipole B in free space, the simulation is to be conducted with HFSS transient solver in time domain (Solution type, transient, composite excitation)

      (1) In HFSS 3D project, dipole A is stimulated with a voltage source (Lumped Port excitation), and dipole B is loaded with a 50 Ohm resistor (lumped RLC). Now the question comes up. If I want to replace the 50 Ohm resistor with a subcircuit. How could I define the circuit? Specifically, if I want to include nonlinear elements, such as varactor or PN junction diodes, how could I import the SPICE netlist? Or if I understand correctly, it is currently not available.

      (2) Suppose (1) can be solved, now replace the port excitation with an incident wave, and set both dipole A and dipole B as receiving antennas with 50 Ohm resistors at the feeding points. Is it still possible to perform transient full-wave EM/circuit simulation in HFSS?

      As far as I know, Ansys provide the co-simulation by dynamic link HFSS into circuit simulation. However, the interaction between HFSS and the circuit is based on a spice like equivalent multiport network, which is called a standard EM/circuit co-simulation. This means HFSS project should be launched before the circuit simulation. A true transient EM/circuit co-simulation would be by carrying out both the field simulation and circuit simulation simultaneously. The interactions or data exchanging between field and circuit parts are conduced at each time step. Unfortunately, this is not well documented in the HFSS help documents. Can any one give me some comments or hints about the above two questions?

      Personally speaking, a true transient EM/circuit co-simulation might not be efficient but could be very advanced for nonlinearly loaded RF designs. I am looking forward to the answer. Thanks.

    • AndyJP
      Subscriber
      HFSS itself solves only linear structures. So it is enough to solve the driven-modal or transient, then feed the multiport matrix to spice. Of course, I would love to see nonlinear effects in ferrites, for example, or model a p-n junction, or nonlinear ferroelectric capacitor directly there, but nope. it is not possible.
    • AndyJP
      Subscriber
      P.S. There is no "true transient EM/circuit co-simulation". Any simulation is true inside its model constraints. Model constraints are dictated by the problem and solution complexity and therefore practical feasibility.
      Since ANSYS is a long established company concentrated on business making, rather than on scientific breakthroughs, do not expect much here. Wait for some startup appear in the future, which would bring something new and overthrow the today's monsters.
    • helloycheng
      Subscriber

      Thanks a lot, Andy. Your answers are very helpful.
Viewing 3 reply threads
  • You must be logged in to reply to this topic.