Unusual behaviour in model outputs recorded by SystemTest - a value for every outport, on every outport.

I have a model in SystemTest with a number of inports and outports.

With each timestep, the values for all the inputs are read in from an Excel spreadsheet.

Normally in my experience, for every timestep, one value for every outport is recorded.

What I am finding for a particular SystemTest is that many values are recorded for every outport, for every timestep. The number of values appears to correspond to the number of outports, so that if there is 5 outports, on every timestep all 5 outports have a 1x5 double each recorded as the result.

It's a discrete fixed step solver of sample time 0.01 - this corresponds to the size of each timestep in the spreadsheet of data.

Any ideas as to how to sort this or what might be causing the behaviour? Many thanks

[James]

The change from a single recorded piece of output to multiple happens just as the library I'm testing is dropped in to the model.

I can't post it, but it contains such things not seen before as a discrete time integrator, and matrix operations like reshape or multiply, and selector blocks.

What is confusing me is why every single output has the same number of recorded pieces of data outputted as the number of outputs in the model. Signals that simply go in and out (with nothing happening to them) change from a single value going in to multiple (but similar) values coming out. The last value seems to be the one I want coming out, so I'd assume that's the same for all of them.

[James]

Fixing (rather than inheriting) the outport sampling rates or sizes does nothing.

[Bart Addis]

Does the model behave this way without SystemTest? In other words, if you connect a scope to the output signals, what do you see?

[James]

Using a scope it displays precisely what's expected.

Answer 1

[Fangjun Jiang]

It looks like the signal associated with each of your Outport has become a vector signal. Click menu Format->Port/Signal Displays->Signal Dimensions and Wide Nonscalar lines to check it out. It is probably inherited from the Inport, which means the data inport from the Excel spreadsheet is not in the right format.

[James]

There could be something in this actually. Eventually I deleted all inports and outports, and redid them. This fixed it.

I can only assume the first outport was connected when I was still configuring inputs and it inherited some settings, that somehow were copied to every other outport when I cloned the outport for the other signals. Why it didn't update to what it should inherit, I don't know, but it's the best that I can understand.

[Fangjun Jiang]

That is a good point. In our organization, we recommended to drag and drop every block from the library, not to duplicate from the model, unless you are 100% sure that is what you want. There are so many parameters now for every block. You don't know what is different from the default setting.