EXP430(G2553) Problem with Timer1 using ACLK (SOLVED)

[sporkey 0]

Hello,

 

I was trying to setup 2 different PWM channels on my launchpad and noticed some strange behavior with timer1 using ACLK, and so I ran the following test, checking output of TA0CCR0 and TA1CCR0 on P1.1(TA0.0) and P2.0(TA1.0) with my oscilloscope. The only variables are P1SEL/P2SEL and TASSEL. Here's what I noticed:

 

- Timer0 and timer1 both function properly with SMCLK, with different internal divider (and different CCR1s work as expected)

- Timer1 cannot use ACLK at all. (output = low)

- Timer0 with ACLK works properly alone. 

- Timer0 with ACLK doesn't work if port2 are selected for timer1 outputs. If timer1 uses SMCLK, timer1 works, but not timer0. If timer1 uses ACLK, both timer output = low.

 

The outputs that confuse me are marked in RED in the table attached. Can someone please explain to me why timer1 with ACLK behaves that way? Am I overlooking anything in the datasheet?

 

Thanks in advance.

 

 

#include <msp430g2553.h>

 

void clockConfig()

{

BCSCTL1 = CALBC1_1MHZ; // Set DCO

DCOCTL = CALDCO_1MHZ;

}

 

void main(void)

{

WDTCTL = WDTPW + WDTHOLD; // Stop WDT

clockConfig();

 

P1OUT = BIT1;

P1DIR = BIT1;

P1SEL = BIT1; 

 

TACCR0 = 10;

TACCTL0 = OUTMOD_4;

TACTL = TASSEL_x | MC_1;

 

P2OUT = BIT0 ;

P2DIR = BIT0 ;

P2SEL = BIT0;

 

TA1CCR0 = 10;

TA1CCTL0 = OUTMOD_4;

TA1CTL = TASSEL_x | MC_1;

}

 

 

 

 

Hardware:

EXP430G2 Rev 1.5 (MSP430G2553)

Kit crystal installed with 12.5pf caps in C21/C22

oscilloscope

 

Software:

compiled with CCS 6.1.3

programmed with mspdebug

 

 

[roadrunner84 466]

I think your problem is related to the crystal.

The crystal is connected to P2.6 and P2.7, but you overwrite P2SEL and P2DIR with non-default values.

As a result, the crystal is not controlled correctly anymore and ACLK is dead.

To avoid this problem, try to use |= and &=~ as much as possible to set or clear bits.

P1SEL |= BIT1; // set SEL1 bit 1
P1SEL &= ~BIT2; // clear SEL1 bit 2
P1SEL |= (BIT1 | BIT2); // set SEL1 bit 1 and 2
P1SEL &= ~(BIT1 | BIT2); // clear SEL1 bit 1 and 2
P1SEL = (P1SEL | BIT1) & ~BIT2; // set SEL1 bit 1 and clear bit 2

[sporkey 0]

Roadrunner, thanks so much! You are 100% correct. Great tips.

 

Both ports function properly after changing P2SEL = BIT0 to P2SEL |= BIT0.

 

 

To understand the problem a bit more, I tried messing with BIT6 at P2DIR and P2SEL.

 

If I write P2DIR = BIT6; and P2SEL = BIT6;, it affects P1's use of ACLK.

If I do P2DIR |= BIT6; and P2SEL |= BIT6;, P1's ACLK functions properly but P2 goes low. That is different from what I expected reading the datasheet.

 

I thought that BIT6 at P2DIR/P2SEL turns XIN off and TA0.1 output on, but apparently it doesn't affect the crystal or turn it into TA0.1 output. Am I reading the datasheet wrong?

 

 

 

 

[roadrunner84 466]

As I read the datasheet, after P2SEL = BIT6; P2.6 will be a timer output, but only if you write P2DIR = PIN6; too. So your crystal isn't oscillating anymore.

If you write P2SEL |= BIT6; the crystal will be functioning because you set a bit that was already set. But if you add P2DIR |= BIT6; then you switch to an undescribed situation; it looks like you want the pin to behave like XOUT, not XIN.

[sporkey 0]

Thanks Roadrunner! I was deliberately trying to stop the crystal working to understand what got me in trouble earlier. I think I understand everything now.

 

I did a bit more reading, and found the part I overlooked previously in the user