{"id":17375,"date":"2025-02-25T09:22:09","date_gmt":"2025-02-25T14:22:09","guid":{"rendered":"https:\/\/blogs.mathworks.com\/simulink\/?p=17375"},"modified":"2025-02-25T09:22:09","modified_gmt":"2025-02-25T14:22:09","slug":"the-basics-of-a-control-loop-in-simulink","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/simulink\/2025\/02\/25\/the-basics-of-a-control-loop-in-simulink\/","title":{"rendered":"The Basics of a Control Loop in Simulink"},"content":{"rendered":"
\"\"<\/img><\/a><\/div>
This week I received a series of interesting questions on the basics of putting together a simulation of a control loop involving a Simscape continuous plant and a discrete controller. The questions are summarized in this image:<\/span><\/div>
\"\"<\/img><\/div>
Here are my answers to those questions.<\/span><\/div>

Discrete Controller<\/span><\/h2>
I recommend implementing the controller in a separate model file and bringing it in the simulation using the <\/span>model block<\/span><\/a>. The main reason is that you will likely want to generate code using <\/span>Simulink Coder<\/span><\/a> or <\/span>Embedded Coder<\/span><\/a> for the controller at some point. With <\/span>model reference<\/span><\/a>, the controller is a standalone atomic unit and you can define a strong interface to it. This ensures that the controller behaves identically as part of the closed-loop simulation and when you will generate code for it.<\/span><\/div>
\"\"<\/img><\/div>

Continuous Plant<\/span><\/h2>
Here, I would say: <\/span>It depends...<\/span><\/div>
One important aspect to consider is algebraic loops. In this kind of closed-loop simulations, it's typically the plant that breaks potential algebraic loops. If the plant is an atomic unit, there are more chances to run into <\/span>artificial algebraic loops<\/span><\/a>.<\/span><\/div>
Here is how I typically decide:<\/span><\/div>