For example, system modeler is probably never going to be as accurate as a dedicated multi-physics FEA but it will get reasonable ballpark answers very quickly for complex multi-physics system-level problems (which is a big advantage)

There are a few techniques that can be applied with System Modeler in order to get fast solutions to problems where FEA is usually used. For example, when modeling flexible elements. A flexible bar can be simulated splitting it into small segments and approximating the deformation. It's a tradeoff, with FEA you can get a more accurate result, but with System Modeler you can get good enough solution in a reasonable time. @Vedat Senol has made a few applications like this.

In the case of electric circuits it is not always necessary to perform simulations with the most accurate semiconductor models. For example, when developing circuits with operational amplifiers, one can use a detailed model of the op-amp to analyze extreme cases. But in most situations, for the work I do, a behavioral model of the op-amp is more than enough. System Modeler makes very easy to create these behavioral models and perform very fast simulations. For example, in the course I run a simulation of a full synthesizer and render a few seconds of audio in a reasonable time. Using a SPICE-like simulator would take much longer.

Is there any strategy for automatically taking outputs from software like Ansys, Comsol, LTSpice or other simulation programs and integrate the outputs as part of a system-level multi-physics simulation that System Modeler could model?

It depends on what the other simulators allow. System Modeler can produce FMUs, under the FMI standard https://fmi-standard.org/. This allows models created in System Modeler to be exchanged with other simulators therefore making sending an receiving data much simpler. Is up to the other simulators to adopt such standard in order to make interoperability easier.

In the case of FEA, Mathematica provides tools to perform FEA analysis and it is possible to combine with System Modeler. Again @Vedat Senol has made a few examples.

Another application where I'd love to see what System Modeler can do is as a tool something like LangChain where LLM outputs can be chained together to action complex tasks.

I haven't seen anything combining System Modeler with LLMs. What we do have is running neural networks built in Mathematica running with System Modeler using the ONNX format.

That's a very good question. I quickly created a model comparing 3 different cases: ideal (no thermal interaction), thermally isolated from the ambient and exposed to the ambient. Here is the model and results:

The model can be found here if you want to play with the parameters: https://amoeba.wolfram.com/index.php/s/TWQFSjnT9WTyxSj

Also, there are multiple examples in the Wolfram Demonstrations project that mention phasors. You can download, run, and peruse the source code for each of these little projects.

Hi Ankit;

In calculating both the amplitude and phase angle in electrical circuits, that contain capacitors and inductors, there is a technique called Phasors - which greatly simplifies the calculations. The phasor calculations are easily performed using my TI-89 calculator but cannot find a similar method using Mathematica. However, I would much prefer to use Mathematica. Does Mathematica have something similar to Phasors or is there a recommended method to calculating amplitudes and phase angles in Mathematica?

Thanks,

Mitch Sandlin

I have checked the questions you mention and it seems that the answer key is incorrect, so the options we took as correct are wrong. I will update the quiz to reflect the correct answers.

Wolfram U prefers that students not discuss quiz answers in the DSG for a course. You should e-mail wolfram-u@wolfram.com or you can put info in the feedback form that pops up at the end of each lecture.

FWIW, I'll also be providing them feedback to the quizzes.

Hi Leonardo,

I can see how it can be a real advantage to take a more simplistic approach. For example, system modeler is probably never going to be as accurate as a dedicated multi-physics FEA but it will get reasonable ballpark answers very quickly for complex multi-physics system-level problems (which is a big advantage). Accurate simulation for electronics can be very complex, System Modeler seems to be taking a more high-level simpler approach which is nice.

Is there any strategy for automatically taking outputs from software like Ansys, Comsol, LTSpice or other simulation programs and integrate the outputs as part of a system-level multi-physics simulation that System Modeler could model?

This would help solve scaling issues with the packages I mentioned above; they are so computationally intensive that doing large systems are not really feasible. It would be interesting to see some System Modeler examples doing this.

Another application where I'd love to see what System Modeler can do is as a tool something like LangChain where LLM outputs can be chained together to action complex tasks.

Apologies, my question maybe a little off-base, just a few general thoughts/questions.

The capacitor model in the Electrical and the Spice library looks the same to me. Unfortunately, we do not support it yet. For this course, you can create all the models using the Modelica.Electrical library.

class Modelica.Electrical.Analog.Basic.Capacitor
  Real v(quantity = "ElectricPotential", unit = "V", start = 0.0) "Voltage drop of the two pins (= p.v - n.v)";
  Real p.v(quantity = "ElectricPotential", unit = "V") "Potential at the pin";
  Real p.i(quantity = "ElectricCurrent", unit = "A") "Current flowing into the pin";
  Real n.v(quantity = "ElectricPotential", unit = "V") "Potential at the pin";
  Real n.i(quantity = "ElectricCurrent", unit = "A") "Current flowing into the pin";
  Real i(quantity = "ElectricCurrent", unit = "A") "Current flowing from pin p to pin n";
  parameter Real C(quantity = "Capacitance", unit = "F", min = 0.0, start = 1.0) = 1.0 "Capacitance";
equation
  n.i = 0.0;
  p.i = 0.0;
  v = p.v - n.v;
  0.0 = n.i + p.i;
  i = p.i;
  i = C * der(v);
end Modelica.Electrical.Analog.Basic.Capacitor;

class Spice3.Basic.C_Capacitor
  Real v(quantity = "ElectricPotential", unit = "V") "Voltage drop of the two pins (= p.v - n.v)";
  Real p.v(quantity = "ElectricPotential", unit = "V") "Potential at the pin";
  Real p.i(quantity = "ElectricCurrent", unit = "A") "Current flowing into the pin";
  Real n.v(quantity = "ElectricPotential", unit = "V") "Potential at the pin";
  Real n.i(quantity = "ElectricCurrent", unit = "A") "Current flowing into the pin";
  Real i(quantity = "ElectricCurrent", unit = "A") "Current flowing from pin p to pin n";
  parameter Real C(quantity = "Capacitance", unit = "F", min = -1.7976931348623157e+308, start = 0.0) = 0.0 "Capacitance";
  final constant Real WSMServices.Machine.Real_MAX = 1.7976931348623157e+308 "Maximum finite Real number.";
  parameter Real IC(quantity = "ElectricPotential", unit = "V") = 0.0 "Initial value of voltage";
  final constant Real ModelicaServices.Machine.inf = 1.7976931348623157e+308 "Biggest Real number such that inf and -inf are representable on the machine";
  evaluated parameter Boolean UIC = false "Use initial conditions: true, if initial condition is used";
  protected Real vinternal(quantity = "ElectricPotential", unit = "V") "Capacitor voltage";
  final constant Real Modelica.Constants.inf = 1.7976931348623157e+308 "Biggest Real number such that inf and -inf are representable on the machine";
equation
  n.i = 0.0;
  p.i = 0.0;
  v = p.v - n.v;
  0.0 = n.i + p.i;
  i = p.i;
  vinternal = p.v - n.v;
  i = C * der(vinternal);
end Spice3.Basic.C_Capacitor;

In today's class, somebody asked:

why are frequencies represented with imaginary numbers?

I highly recommend ChatGPT for this kind of question. That exact question yields a comprhensive answer in the AI. Since ChatGPT 4o is available freely for anyone who requests an account; I highly recommend using that tool. OpenAI even provides an app for MacOS to help keep things uncluttered on your desktop.

In class today, our instructors were discussing the variety of sinusoidal behaviors associated with electronics and signal processing. I believe they were searching for a particular word. There is a pertinent list of such terms available online in the Electromagnetic Terms section of the Wikipedia entry for Oliver Heaviside. The terms are admittance, elastance, conductance, electret, impedance, inductance, permeability, permittance, and permittivity. Two other terms were coined by Heaviside's peers around the same time: susceptance and reactance. These terms were coined from 1885 to 1894. Some are widely used today in physics and engineering; others have withered into obscurity. Some terms deal with real values; others deal with imaginary values. The Wikipedia article contains links defining each of these terms.

Anyone who thoroughly understands these eleven concepts is a signal processing wizard.

Hello Dr Ruiz,

In yesterday's lecture 11, towards the end, we went from the simple 2 loop 4 node resistor circuit to what System Modeler did with it, which was very complex. You went by that really fast. We should dwell on that. What all can you do with that model? That seemed to put a finger on the strength of System Modeler. In particular the associated thermal modeling. Can you model resistors whose values change with temperature? Can you model thermal runaway? Can you show the effect of putting a fan on the circuit, or a thermal fin, to prevent thermal runaway? A simple circuit like that seems to be a perfect pedagogical tool for looking at complex issues like that. I've seen circuits fail because even though that had low dissipation, there was no way for the heat to get out. A beginner's mistake, but one that is easy to make because thermal analysis is not traditionally included in circuit analysis.