LAUNCHXL-F28379D: Issue Syncro SPI if slave is in CCS debug

Hi Biagio,

Just to confirm I understand the issue correctly- if you boot the part from flash without an emulator connected, the code works as expected. But if you have the emulator connected and you are running from CCS, the data becomes invalid. Is that correct?

Do you have any waveforms of the transactions?

Regards,
Kris

Yes, if I boot the slave from flash (the master is always running from CCS ) the code works as expected. if I have the slave connected and running from CCS, the comunication SPI is wrong.

I haven't waveform of communications, but I can read from the master the measures sent by the slave.

For example, If the tensions I apply to the system with the generators are V[0]=100 V, V[1]=250V, V[2]=400 V, V[3]=600 V, and the duties I send from master to slave are d[0]=0.3, d[1]=0.8,
If i boot slave from flash, from debugging the master and oscilloscope connected to the slave pwm, all is ok, but if slave run from CCS I have the following random situation:
1)the measurements sent to the master are wrong and are V[0]=0; V[1]=100 V, V[2]=250V, V[3]=400 V and the duty are correct
2)the measurements are ok but the duty cycle sent are wrong and are d[0]=0, d[1]=0.3, I see this also from the oscilloscope connected to the slave pwm.

I have to understand the reason for erroneous communication because the DSPs control a H-Bridge structure of 500 kVA and if the microcontrollers fail during operation it is really very dangerous for everything and everyone.

Regards,
Biagio

Biagio,

A few questions come to mind:

How are you syncing the devices when running with the emulator? For example, could your master be sending data before your slave is ready? Remember a device may boot differently when the emulator is connected.

What does the flow look like from CCS? Start master then start slave?

How are you verifying this is an issue with the SPI data as opposed to a PWM frequency issue or something along those lines?

Regards.
Kris

I am sure the slave is ready when the master sends the data, I start (play from CCS) the master and the slave together but the first time the master waits 10 seconds before starting the algorithm and SPI communication. In addition, the slave performs only 4 acquisitions and 6 spi communications (nothing else), but the master performs the following operations: 8 acquisitions, small algorithm, 4 spi communications with the slave (MODULE B) and other 4 communications with another slave (MODULE A), small algorithm, 2 communication spi (MODULE B) and other 2 communications (MODULE A) with another slave , small algorithm.

The problem is not due to the frequency pwm. The PWM frequency is 4 kHz => every 250 micro seconds the routine is repeated,
6 comunication spi = 42 microsec
Algorithm = 120 microsec
8 Acquisition < 15 microsec
Tot= 177 microsec


The only thing that comes to my mind is that somehow the SPI - A module of the master, used to communicate with another slave (this slave is always started by flash and use its module SPI - A) interferes with the SPI - B module of the master used with this slave.

MASTER SPI A <=> SLAVE1 SPI - A (no communication problems)
MASTER SPI B <=> SLAVE2 SPI - A (yes communication problems)