Issue with Spice3 models in NF

[Francesco Casella]

Please check ​Modelica.Electrical.Spice3.Examples.Graetz. The removeSimpleEquations module fails.

I flattened the model with the old and new FE. The equations seem to me equivalent, considering that the connection equations for 3 effort variables can be equivalently stated in the form a = b; b = c; as well as in the form a = b; a = c;

The only other difference I spotted was in the way a connection equation for flow variables is rendered:

D2.n.i = (-D2.icap) - D2.cc.m_dCurrent - D2.igmin;    // Old FE

D2.n.i = (-(D2.cc.m_dCurrent + D2.igmin)) - D2.icap;  // New FE

Of course the unary minus operator and the binary minus operator look the same when pretty printed, but maybe the - in the NF-produced AST is the binary operator, and that causes the removeSimpleEquations module to fail.

There are many other Spice3 models failing in the same place, so I guess this is a commonplace issue, please check.

[Per Östlund]

Fixed in 1ed0e9d. But it seems like the Spice3 models doesn't really like the "put parameter bindings in initial equations and make them fixed" strategy, because it fails during initialization and complains a lot about missing values.

[Francesco Casella]

Replying to perost:

Fixed in 1ed0e9d. But it seems like the Spice3 models doesn't really like the "put parameter bindings in initial equations and make them fixed"

I guess you mean not fixed, right?

[Per Östlund]

Replying to casella:

Replying to perost:

Fixed in 1ed0e9d. But it seems like the Spice3 models doesn't really like the "put parameter bindings in initial equations and make them fixed"

I guess you mean not fixed, right?

That might explain why it's not working so well...

[Francesco Casella]

Ops :)

[Per Östlund]

Replying to casella:

Ops :)

Good thing you caught that, after correcting that slight misstep the Graetz model now simulates without issues.