S32K3 MBDT | HV BMS | Temperature readings with higher number of CMU boards

[narsimh_kamath]

[posting on behalf of Team Raftar]

We’re currently using a K358BMU connected to 8 peripheral boards, each equipped with the MC33774 chipset. The setup is configured and running the MBDT model for all BMS functionality checks.

Each peripheral board has 8 analog input channels, and we’ve properly configured the TD handler indexes in MBDT.

The issue we’re facing is as follows:

When we set the number of boards to till 6 in both S32CT and MBDT, all 8 analog inputs from each board are read correctly.

However, when we increase the number of boards to 8, only the first 4 analog input values per board are displayed in FreeMASTER — meaning only the Primary Measurements are showing up, while the rest are missing.

Has anyone encountered this issue before or knows what might cause the secondary measurements to not appear when scaling up to 8 boards? Any insights or suggestions would be greatly appreciated.

[mariuslucianand]

Hello @narsimh_kamath,

This is a tricky one, and we would need more details regarding the implementation to be able to propose a solution for this.

The first thing which come into my mind is if everything works for 6 AFEs but not for 8, this might suggest a timing issue. What is the Simulink step size vs. the time it takes to execute all the conversions for the 8 devices and to communicate the values via TPL? What can happen here is that the measurements and transfer time might be longer than the step time and some values lost. How is the data acquisition procedure implemented? Did the team used the global conversion and then in the same step window read the values of all the 8 AFEs? Or they have continued to use a round-robin approach as in the enabling example we provide in our toolbox?

Also, depending on the approach, the Transaction Descriptor buffer might overload. Was the TD buffer sized increased or resized to be able to store all the requested data from the AFEs?

It is curious that only the first 4 AFEs readings are correct per each device, when it is increased to 8 AFEs. I think this depends on the actual implementation for the reading sequence. Is there every device handled separately (start conversion, waiting for the conversions to be finalized and then reading the results) per each device. Of course the external configuration tools project and the initializaion sequence must be double checked here, but maybe the data acquisition approach should be optimized.

Secondly, I am thinking rather or not it is an issue with the FreeMASTER. We have noticed that in the FreeMASTER we add a way higher amount of variables to be monitored, some of the data is not displayed. To elimintate this, you can create a new watched sublock and add there only the 8 analog values from the 8th BCC and test only with that.

Please describe the reading technique used, so that we can have a better understanding of the implementation to try understanding where is the issue coming from.

Regards,
Marius