MSP430FR5994: Low Power Mode current consumption measurement issues

I am checking a prototype of our new main board with a MSP430FR5994. At the time of writing I already (partially) built a number of boards and everything is functionally OK, my focus now is checking power consumption.

There is nothing else on my board now than a MSP4305964 and a reset pull-up resistor, the board is powered by an external power supply. However, even when idling in LPM3 mode I still measure (varying) values between 330 and 380 uA. As I embark quite a lot of code, I finally put some initialization code into my main function, to cut off the rest, without significant results.

So finally I turned to a MSP430FR5994 (same family, same specs) LaunchPad with a minimal program (modified blink.c) which I used to set the device either in LPM4 or LPM3 modes:

#include <msp430.h>

#pragma vector=unused_interrupts

interrupt void Unexpected_ISR()
{
    // Interrupt handler for unexpected interrupts.
    __delay_cycles(16000000);

    WDTCTL = 0xDEAD;
}

void main(void) {

    WDTCTL = WDTPW | WDTHOLD;

    while(1)
        _BIS_SR(LPM3_bits); // Either LPM3 or LPM4 modes
}

Now the issue is the following, believe me this has taken quite a lot of time to find out and it is reproducible in my original project and this minimal code project:

- if I launch this program without a breakpoint in the code on the LaunchPad the consumption is round 350uA, when I disconnect the target (connected in SPBW)  it falls down to something around  220uA, both in LPM3 and LPM4 modes, on my own board these currents are more or less 100uA higher,

- now if I stop my target and set a breakpoint in the interrupt catcher (this is why I added it) and launch again, in LPM4 mode I measure 0.6uA (which I believe is correct) and in LPM3 mode I measure around 50uA (which I believe is still twice the specs) on both my board and the LaunchPad.

Si finally my questions are :

• what settings should I apply to measure the current of my device (fard/soft breakpoints, lower power mode debugging or not asf) ?
• what setting will influence the consumption ? are my findings known issues/features ?
• what could still prevent me from reaching the max. 25uA in LPM3 mode ?

Any help is appreciated,

Marc