Blocks.Sources.{Sine, Cosine} do not behave as described

[AHaumer]

The parameter description says (defined in Blocks.Interfaces.SignalSource):
parameter SI.Time startTime=0 "Output y = offset for time < startTime";

Yes the output is equal to offset for time before starTime,
but this parameter causes an additional phase shift:

The implementation is wrong:

I used a frequency of 10 Hz and a startTime of 0.025 s (i.e. a quarter of the period).
The sine actually is shifted to a minus cosine, and the cosine is shifted to a minus sine.

I know that it is not easy to change that behaviour (an addtional correction by parameter phase is needed - very confusing)
but to correct the misleading parameter description, and offer additional blocks with the correct implementation:
y = offset + (if time < startTime then 0 else amplitude*sin(2*pi*f*time+ phase));
instead of
y = offset + (if time < startTime then 0 else amplitude*sin(2*pi*f*(time - startTime) + phase));

Bear in mind that this effects Electrial.Analog.Sources and Electrical.Polyphase.Sources!

[HansOlsson]

This has been the case at least since MSL 1.3.1 from Dec. 13, 1999 (despite a number of other changes in the block), so it seems simplest to update the parameter-description - and the original idea was to use startTime to shift when the model acticates.

I'm not sure if an additional variant is needed; as it wasn't detected until now.

[AHaumer]

Ok updating the parameter description is dangerous since the partial Blocks.Interfaces.SignalSource is used in various other sources.
Maybe it is the case since MSL 1.3.1 but it's a fact that the description is misleading and I have to fiddle around to get a sine (which is zero at time = 0), but gets active at startTime and before startTime I get offset.

And for threephase voltages it is even harder, because we have to use phase for the 120° phase shift between the three voltages.

[HansOlsson]

Ok updating the parameter description is dangerous since the partial Blocks.Interfaces.SignalSource is used in various other sources.

Yes, but as far as I can see all of the derived classes:

• Modelica.Blocks.Sources.ContinuousClock
• Modelica.Blocks.Sources.Cosine
• Modelica.Blocks.Sources.Exponentials
• Modelica.Blocks.Sources.ExpSine
• Modelica.Electrical.Analog.Sources.LightningImpulse
• Modelica.Blocks.Sources.Pulse
• Modelica.Blocks.Sources.Ramp
• Modelica.Blocks.Sources.SawTooth
• Modelica.Blocks.Sources.Sinc
• Modelica.Blocks.Sources.Sine
• Modelica.Blocks.Sources.Step (well, it doesn't matter as its a step)
• Modelica.Blocks.Sources.TimeTable (as default - but you can set shiftTime differently)
• Modelica.Blocks.Sources.Trapezoid

use similar formulas, so it's "just" the parameter description that has been lagging behind for decades.

[AHaumer]

No it's not just the parameter description.
For other sources (e.g. Step) the description is 100% correct.
That's not the case for Sine and Cosine which get an additional (unexpected) phaseshift.
Have a look at Electrical.Polyphase.Sources.SineVoltage (which contains m Electrical.Analog.Sources,SineVoltages):
phase is used to achieve the correct shift between the voltages.
Then try to start each of the voltages at the maximum, try startTime={(k-1)/m/f[k]}
Large transformers are started that way to minimize the inrush current.
SineVoltage1 is the original from MSL, SineVoltage2 is my corrected version;

At least there has to be a hint that startTime causes an additional phaseshift.
Although the user should start thinking what does that mean it is not obvious,

[beutlich]

This issue reminds me of #1771 where users came up with the very same confusion for Modelica.Blocks.Sources.CombiTimeTable. The solution was to introduce parameter shiftTime in a backwards-compatible manner.