C sfunction & RTLAB?

Question

I wrote a C sfunction and used it as a simulink block.It was OK.the answer is OK with the error about 10^-14 (in matlab enviroment)... Now I want to connect to RT_LAB, to run my model real time. To execute a simulink model in RT_LAB, we have 3 steps: 1. compile the model. 2. load it. 3.execute the model. So if you have an invalid model U should have errors in 1st step (as I had before completing my model), my model compiles with no errors now, I load it, but when I execute it, the answers are not valid (there is no error but not a valid answer either). When RT_LAB compiles a model, it translates the model to C codes. I think maybe there is something wrong with this part, generating wrong C code. Is any one here who used RT_LAB? How can I find the problem?

Answer 1

How much precision does your answer require to be a valid answer? If you're doing floating point math, the precision of each step in your algorithm can be off by a small amount and if those errors accumulate you may get a slightly wrong answer.

A common method of debugging is to print out or manually evaluate the numerical value, along with the expected value of your algorithm at each step. You'll be able to very efficiently discover precisely what step is going wrong using this method, or if you'd like to schedule my help I'd be happy to take a look at it.