MSP-EXP430FR2311: High current draw, longer then expected after LPM3/4

George Karavaev

Part Number: MSP-EXP430FR2311
Other Parts Discussed in Thread: ENERGYTRACE

Hello All!

Would really appreciate your help in this. I've been very confused for about 4 days on this problem now. Re-read all documents and no luck.

I have following requirements:

*Simple IoT sensor application.

*Wake up MSPFR2311 every ~5s, communicate with sensor for ~5ms through I2C, back to sleep.

*Every 1hr transmit through SPI to wireless chip for ~50 ms and back to sleep.

*Battery life is critical, which is why I am trying to use LPM4 to sleep for ~5s (timing not critical in slightest).

However, the issue I am seeing is that the CPU is awake for far too long. I am very very confused as I have timed with GPIOs all available code and it runs much shorter then the power draw. My ISR runs for ~30 uS and code runs at max for ~50ms. However current measurements with EnergyTrace reports that my activemode current is way too long (and variable too!) It's sometimes as long as a full second somehow.

I narrowed my program to do the minimum possible that still repros this on at least 2 boards.

#include "io430.h"

int main( void )
{
  // Stop watchdog timer to prevent time out reset
  WDTCTL = WDTPW + WDTHOLD;
  //set pins to low power mode.
 P1OUT = 0x00;
 P2OUT = 0x00;
 P1DIR = 0xFF;
 P2DIR = 0xFF;
 //unlock pins
 PM5CTL0 &= ~LOCKLPM5;
 //set RTC to trigger ~8 seconds
 RTCMOD = 10000UL;
 RTCCTL = RTCSS_3 | RTCSR | RTCPS_1 | RTCIE;
  

 

 while(1)
 {
   //pretend to do work for 5000 / 2MHz = 2.5 mS
   unsigned volatile long wait_var = 5000;
   while(wait_var>1)
     wait_var--;
   
   __bis_SR_register(LPM4_bits | GIE);
 }
  return 0;
}

#pragma vector=RTC_VECTOR
__interrupt void RTC_ISR(void)
{
  (void)RTCIV; //clear interrupt
  __bic_SR_register_on_exit(LPM4_bits);  //return to main program
  return;
}

Here is the energy trace showing mysterious long delays everywhere. How could this?

over 5 years ago

0 Winter Yu over 5 years ago

TI__Genius 12315 points

Hi George,

I will try to repeat your issue with your code. Get back to you later.

0 George Karavaev over 5 years ago in reply to Winter Yu

Intellectual 315 points

Thank you! I will be waiting with held breath.

I suspect it's something to do with FLL. I added some code in ISR to do FLL lock (fll settle + load param) and it made time between ISR and main code 50 ms.….

0 George Karavaev over 5 years ago in reply to George Karavaev

Intellectual 315 points

Did you manage to reproduce? If not I can submit more code that shows the issue.

0 Johnson He over 5 years ago in reply to George Karavaev

TI__Mastermind 28416 points

Hi George,

I'm trying to repeat your issue with CCS9.0.1, but not encounter this issue, could you tell me your compiler tools version?

The result show this :

0 George Karavaev over 5 years ago in reply to Johnson He

Intellectual 315 points

Hello Johnson!

Your power looks very high. (9 mW?) is there anything else connected to the board?

Also, it's still looks long. If you zoom in, do you get almost half a second of "awake" times?

However, mine still looks different from yours. I tried using CCS and this is what I got. The code is exactly the same as IAR, but #include msp430 instead if iot430. I have:

CCS 8.3.0.00009

MSP430 Compiler tools 18.1.5

TI MSP430 tool-chain 9.0.1

0 George Karavaev over 5 years ago in reply to George Karavaev

Intellectual 315 points

Here is my exact out file I am using. I had to 7zip it to let system handle it. Run -> load should give the same performance.

On my EXP board I have only 4 jumpers populated -- GND, 3V3, SBWTDOP. SBWTCK. All others (UART, 5v, light sensor, LED) are not populated.

broken.7z

0 Johnson He over 5 years ago in reply to George Karavaev

TI__Mastermind 28416 points

Hi George,

My power have around 9mw because the board have other connection, Now my test result are similar to yours after disconnect all connections in Eneraytrace.

Then I do a experiment as following:

1. Test your code run time, around 86ms, show this:

2.A series resistor is connected in 3.3V power, then use oscilloscope to measure the voltage across the resistance, we would found there have a little bit high voltage around 87ms every 10 seconds, show this:

3.So that the result show active mode current is not too long such as Eneraytrace, I think the time should subject to the actual experiment.

If my post helped solve your issue, please click on the This resolved my issue button.

0 George Karavaev over 5 years ago in reply to Johnson He

Intellectual 315 points

Does that mean that we can't rely on EnergyTrace with <100 ms timing?

0 George Karavaev over 5 years ago in reply to George Karavaev

Intellectual 315 points

I think I got it figured out.

This post helped a little bit:

The graph is not accurate below certain power levels. However, energy is still correct. The EnergyTrace essentially does power counting and if spike is too fast it'll still count it correctly, but it'll display the timing as longer but power as lower (and still get the same result).

e2e.ti.com/.../357227

**Attention** This is a public forum

MSP low-power microcontrollers

MSP low-power microcontroller forum

MSP-EXP430FR2311: High current draw, longer then expected after LPM3/4