Modelica.blocks.sources.pulse wrong simulation results

[Bruno Scaglioni]

In some cases the Modelica.blocks.sources.pulse model has wrong output, the square wave stops evolving after two periods and the output keeps stuck to zero.
See the output "pulse.y" of attached model, the signal should be a square vawe until the end of simulation

[Bruno Scaglioni]

Simple model with wrong behaviour

[Christian Schubert]

It seems to be connected to event handling. Internally the pulse relies on the following when statement

when integer((time - startTime)/period) > pre(count) then
  count = pre(count)+1;
  T_start = time;
end when;

According to -lv LOG_EVENTS the when condition becomes true but never changes back to false. Thus the when is not fired again.
However, the when condition should become false immediately afterwards, since the variable count is modified inside the when equation.

[Adrian Pop]

This might also be connected with #1959 where it said that "Yesterday the sawtooth block worked, today it runs amok".

[Willi Braun]

Yes, #1959 and #1960 belongs together. Both relies on same when equation as Christian has pointed out, which generates an events only one time.

[Willi Braun]

and both models stopped working with MSL3.2.1, because the models changed from a sample when equation to the one above.

[Willi Braun]

fixed in r14197.