MSP432 Interrupt Vector Handler

Jimmy,

look at the "MSP432 Platform Porting Guide" which can be downloaded here. Chapter 4.2.3 describes the interrupt system.

There are two interrupt vectors for the TimerA:

• TAxCCR0 interrupt vector for TAxCCR0 CCIFG
• TAxIV interrupt vector for all other CCIFGs and the TAIFG

You still have to write an interrupt function for CCR0 and/or CCR1,2,3,... The CCR0 function has a dedicated one and it's flag is cleared when entering the ISR. CCR1,2,3,... and TAIFG share one interrupt function. You can use the TAxIV inside this function to determine the highest pending flag. When accessing TAxIV, this highest pending flag is automatically reset. But this is similar to the MSP430 IV working principle.

For the MSP432 you typically use the startup file to place the name of your specific ISR for the module into the table and you then enable the interrupt in the NVIC module. Your ISR can have any name you want, but it has to be declared in the startup file.

Dennis

Thanks for the fast reply. Ok, so if I'm understanding this correctly, NVIC is a vector table which stores interrupts based on their priorities. Would the following line

NVIC_ISER0 = 1 << ((INT_TA0_0 - 16) & 31);

store a TimerA0 interrupt in the table at some position, to get accessed as soon as all the reserved interrupts are ignored?

And for the silliest question: I see interrupt handlers declared as void functions outside main(), but they're never called inside main(), so how would this work? I compile and run the program from the link below on CCS without a problem, but it doesn't seem to work at all, even with the correction suggested one post below:

e2e.ti.com/.../432089

Thanks a great bunch Dennis, I'm going to chew through this now :) Going to report back once I get something working!

You are also welcome to report back if it is not working :)

Okay, I tried out an example I mentioned above and it's working now. I also implemented the frequency counter originally written for MSP430, just needed to change the timer registers and interrupt handlers for two timers. I think I get the general idea now, thanks again!

You're welcome! Have fun!

Okay, so as expected I ran into some problems :D. Basically, when I call the function containing the timers for frequency measurement, MSP432 freezes. It doesn't always happen, but it happens eventually in debug mode. When I compile in release mode, it happens the first time the function is called.

I'm using two timers to measure the frequency of internal or external signal. I initialize the timers with 

 __enable_interrupt();
NVIC_ISER0 = 1 << ((INT_TA0_0 - 16) & 31); //enable TA0_0 interrupt
NVIC_ISER0 = 1 << ((INT_TA1_N - 16) & 31); //enable TA1_1 interrupt


TA0CCR0 = 20000;			// 20000 * 2400 = 48 MHz operation
TA0CCTL0 |= CCIE;
TA0CTL |= TASSEL_2;

TA1CCTL1 |= CCIS_1 + CM_2 + CAP + CCIE;
TA1CTL |= TASSEL_2;

The interrupt handler function declaration:

void TimerA0_ISR (void)
{
	counter++;
	TA0CCTL0 &= ~CCIFG;
}

void TimerA1N_ISR (void)
{
	freq++;
	TA1CCTL1 &= ~CCIFG;
}

The frequency measuring code itself:

void print_frek(void){

	TA0CTL |= MC_1;				//start the timers
	TA1CTL |= MC_1;

	while(counter != 2406);	//fine tune this number for accurate frequency results, around 2400 for 48 MHz DCO


	TA0CTL &= ~MC_0;				// stop timer
	TA1CTL &= ~MC_0;				// stop timer
	result = freq;
	printf("Frequency: %d\n", result);      //for debugging purposes
	counter = 0;			                        // reset counter
	freq = 0;			                                // reset freq. count
	TA0R = 0;				        	// reset timerA0 register
	TA1R = 0;					        // reset timerA1 register

}

For testing purposes I'm using this in a dirty button-push loop like this:

	while(1)
	{
		printf("standby!\n");
		if(~P1IN & BIT1){
		print_frek();
		}
	}

This works for a first dozen or so calls of the print_frek() function, where it correctly displays the internal ACLK frequency, but sooner or later, it freezes. If I'm not using any loops or button pushes, it works indefinitely.  Any ideas? At first I thought It had something to do with DCO frequencies, but the behaviour is the same at every HF setting. The VCORE level is also at 1. Thanks! 

Again, that's the solution! Thanks a lot. I'm surprised it even worked as it was, I just read the descriptions next to the values in "msp432.h" and set the corresponding bits (in the wrong way!).

Glad it works now :)

One more question, does TimerA run independently of CPU, meaning CPU can do other stuff while the timer is counting?

The TAxR registers are sourced from the clock source you specified for them. But they are connected in hardware. With setting the individual bits you simply route the selected clock source to the timer module and by the divider bits, for example, you route the clock signal through some flip-flops (or grab the signal from another flipflop, I don't know the internal construction), but it is all done in hardware.

Example for a clock-divider:

(Just a pic from the net - f/16 is not a divider from the timer A module)

So in short: The timer is sourced without interaction of the CPU. The timer can ping the CPU in form of an interrupt to react on a special event. The CPU can even be in low power mode and the timer is still running, waking the CPU up to do something.

Dennis

Okay, thanks. So in theory, it would be possible to generate pulses with MSP and have a frequency counter to measure frequency of said pulses without interfering with pulse generation too much?

I don't know your pulses and the amount of CPU processing they require, but the timer could, for example, generate output pulses like a PWM or a square wave without any CPU interaction. For a frequency counter, the timer also runs without the CPU, collecting the incoming pulses, but for processing the result when the timer generates an interrupt, the CPU must jump in, of course.

Okay, thanks, that's what I wanted to know :) .