Hello everyone,
I am Pierpaolo Dini, an assistant professor at the University of Pisa. We made an investment, buying 10 s32k144 q100 evaluation boards, for a laboratory course on model-based design in the field of control and monitoring algorithms, but we facing difficulites in using the toolbox.
I have no longer been able to use the PIL mode with the MBD toolbox for s32k1xx evaluation boards, for models that are more complex than the example of switching on the LED via blocxco GPIO, with simulation input managed with the toggle button. It's really the only example where it allows me to do that.
In particular I noticed that anything I add to that example starts giving me the following error:
"Cannot perform a processor-in-the-loop (PIL) simulation for "xxx".
The configuration parameters for this model do not support the following connectivity configurations: IBIS_AMI, Texas Instruments Code Composer Studio (IDE), Texas Instruments Code Composer Studio (TCP/IP), Texas Instruments Code Composer Studio (Serial), s32k_pil PIL, s32k_as_pil PIL
To fix this error, update the configuration parameters or create a supported connectivity configuration. See Configure a SIL or PIL Simulation, Create a Connectivity Configuration for a Target, and Processor-in-the-Loop (PIL) Simulation in your product help. "
I really tried them all: uninstall and reinstall the toolbox, both version 4.3 (latest available) and 4.2 and 4.1. I tried with older versions of matlab, now I'm using 2020b, but I get the exact same error with versions 2022b and 2023b.
The error is quite frustrating since I have never had problems with the s32k144, and through the PIL approach we were able to test even complex algorithms such as MPC and KF. Now even the simple PI for a first order system causes error.
It makes me think there is a problem with the toolbox and not with the procedure I use which has always been the same and I have never had any problems.
I report some screens below:
1) in the configuration block the card is recognized without problems
2) in the solver configuration I used fixed step and a numerical integration time which for this type of control scheme should not cause problems when it comes to computational complexity.
3) via Simulink Test APP, the model for test harness is created, without any errors, and the control loop with the discretized first order system is set "outside". I point out that the dynamics of the system are quite "relaxed" so I doubt it is something related to computational complexity.
(l'ho chiamato pidPIL_Harness2 invece che 1 per mia convenienza, non è un errore...)
4) with SIL/PIL APP I set processor-in-the-loop in the settings, and the preliminary simulations are completed without errors (because in fact the model contains no errors...)
I hope tech support is able to resolve such a problem.
Hello @PierpaoloDini,
Thank you for posting on our MBDT Community!
From your message I understand that you are no longer able to run complex models in PIL mode on any of the MATLAB versions newer that R2022b. Was the toolbox and PIL mode working for you in a previous version than R2022b? Is this starting to happen when installing another toolbox/mechanism for code generation other than/from NXP?
First of all, the settings from your screenshot of the MBDT_S32K1xx_Config_Information look good. I hope that the same block and settings are also placed in the referenced model.
Placing the configuration code in the model is the way of setting the hardware target for the referenced model as well. I know that only the application is supposed to be there but this is how the MBDT for S32K1 works.
What I have noticed from your model is that the pidPIL, does not have black corners, and there is no PIL written between parenthesis and as you say, the Test Harness sets the model to be executed in PIL mode. This is why I would try first that the System under test is Model blocks in SIL/PIL mode and select the system which is to be executed in PIL mode, just like in the screenshot attached below:
For this purpose, I have used the MathWorks example Battery State of Charge Estimation in Simulink Using Deep Learning Network and added the coder target to its settings, in order to be executed on the S32K144 EVB, similar with what you have, and this model ran for me successfully on the R2023b. Please find attached this model below.
To run this example on your end, please run first the BatterySOCSimulinkEstimation_ini.m script. Open the BatterySOCSimulinkEstimation_NXP.slx file. Open the main config model, set the proper COM port in the settings, and then, under the MCU tab press the "Propagate configuration" button. Back in SIL/PIL press the Run SIL/PIL button.
In your MATLAB, I would try also to run the mbd_s32k_path script. This will add some required files into path and maybe this is also related to your issue.
If the problem persist for your model, could you please share with us one of your model on which you encounter issues? If it is something confidential, you could also send it in private.
I have sent the archive with the example to you via a private e-mail.
Hope this helps,
Marius
Now i Have MATALB 2024a and versione 4.3 of MBDT, so no issure must be linked with the version i use.
Thinking on what you sayd about bolded black corners, as a proof of that is not related on how you define the test harness at the first step, now i explicit the PIL mode, and the block have the form you think is correct.
Exactly the same error appear again (again!!.).
I noticed also i cannot run anymore neither the simple example of taggle led switching...
I also try to run the example you send me by e-mail and...same error!
I have no words...
