TM4C123GE6PMI dying on battery powered product!

Hi Luis,

Yes I've powered it with the same battery and a new battery.  Vcc comes up to 3.4V no problem.  The voltage rail looks fine, but the Tiva part is non-functional.  Does not work and is dead.  It never comes back to life.  I can't program it or communication with it etc....  When it is powered up the Tiva is supposed to blink an LED on my board indicating it is alive.  Normally this works fine.  Upon letting it run off battery until the battery pack dies it appears the Tiva part dies and doesn't come back to life. The battery pack is a 4.8 volt pack and the buck regulates this down to 3.3.  If I replace the Tiva part with a new one and program it everything works fine. 

I thought if the voltage got BELOW 3.15V the chip would just stop working, basically shut off, until the voltage returned to a normal level.  I realize now that this is not a  great design.    I never thought this would kill the Tiva part though.

I've attached a block diagram of my system.  Because I'm charging a 4.8V battery pack from USB I had to use a boost regulator unless there is a simple charging solution chip that can use 5 volts as Vin to charge 4 NiMH cells. 

Block diagram.pdf

Hey Amit,

Great idea!  I put a new battery pack on and checked both VDD and VDDC.  Vdd is 3.3V as expected and VDDC is 1.2V just like normal.  I also checked my oscillator and it is dead.   Unfortunately replacing the oscillator doesn't bring the TIva part back to life :(

When I power up the device and look at the TDO pin I see a square wave with a magnitude of ~4.5V and a period of ~65mS.  I'm not sure where to check the rev.  The exact part number I am using is TM4C123GE6PMI.  Printed on the chip above the part number is 980  YF.  Printed on the chip under the part number is 37C9QQW and underneath that is printed G4.  Does this tell you the rev?

Amit,

Could you elaborate on what you mean by "On the other hand if this is one of many TM4C's that failed then it would be time to change the BOR conditioning of the device."?  This happens on every device because my battery pack eventually drains down to the point where the output voltage of the regulator is dipping from lack of battery power and causing the system (Tiva part) to power up, drawing current which  causes Vcc to dip, resets the chip etc....over and over.  This is at least what I'm thinking.

Could I set brown out detection to detect if Vcc is less than 3.3 nominal to interrupt, shut everything down and not communicate to eeprom.  When the battery is recharged the regulator will output 3.3V plus and the BOR detction will not trip and allow the chip to run the normal code?  Do you see what I'm getting at? 

May I note that much complexity & uncertainty has been added in your "non-standard" attempt to produce an, "orderly MCU shut-down!"   And - very well designed and implemented ICs have long existed - designed to that exact purpose!

Reinventing a known good wheel - via great complexity and some degree of "hope" may not best serve you long-term.

It should not be too hard nor expensive to order your MCU into an orderly shut-down when your primary power supply decays below an appropriate voltage level.   That's a long standard/practiced technique.   The many winding gyrations presented here - unfortunately - not so much...

I agree, but I need a way to make sure the part doesn't lock up if the user doesn't listen :)  Would something like this work:

The first thing I do is enable the BOR0 and BOR1 interrupt setting in the PBORCTL register then the next instruction is to enable their interrupts using the below code:

	HWREG(SYSCTL_PBORCTL) &=~ 0x06;		// Enable BOR to trigger Sys interrupt
	HWREG(SYSCTL_IMC) |= 0x802;			// Enable BOR interrupts

    //
    // Enable Global processor interrupts
    //
    IntMasterEnable();

So this way if there are any power dips that cause resets the cpu will see the voltage is low and get stuck in the while(1) of the fault handler?  If this is the case then it would never reach my EEPROM code.  Does this seem like a possible solution?  Also, is this the correct way to enable BOR reset functionality?

Bravo Amit! Several of our clients have used your TL7700 - and each reported great success.

One client - desperate to "save pennies" implemented a "poor man's version" via a quad, voltage comparator - with 2 free comparator sections "left over."

Rejecting the "known" (MCU/Voltage supervisor) for a far more complex (and incompletely tested/verified) "Multi-Register, sequenced MCU shut-down solution" seems not to satisfy, "Risk - Reward!"