Power Profiler 2 reported power increases substantially as input capacitance is increased.

[jstuewe]

I have a NRF9160 design with a TPS63031 regulator providing 3.3v.  When my battery gets weak I am seeing spontaneous reboots, likely due to input voltage droop.

I'm experimenting with adding significant (100-200uf) of input capacitance to try and ride through the peaks but my Power Profiler 2 (PP2) is recording significantly more power draw as I increase input capacitance.

The TI Performing Accurate PFM Mode Efficiency Measurements (Rev. A) application note says that 'large' input cap is needed to correctly measure efficiency in PWF (skip mode) vs PWM mode and suggests adding capacitance will increase measured efficiency by 15%. When I add 47uf, 100uf, or 200uf of ceramic capacitance my PP2 idle current measurements jump orders of magnitude.

Powering my device from the PP2 set to 3.1v., the PP2 measured 1.03 Coulombs during the first minute after power on and 5.1uA while idle with the modem off.
When I added a 47uf MLC cap that jumped to 2.6 Coulombs in the first minute and 215uA idle current.

In another test\configuration I measured 0.96 Coulombs/minute and 5uA idle with no capacitance.
When I added a 2x200 MLC caps that jumped to 3.08 Coulombs/minute and 723uA idle current

If I add capacitance to the 3.3v output of the TPS63031 (VS the input) it doesn't affect the PP2 reported power substantially.

I can't think of an explanation for the drastic power increase reported by the PP2 when input capacitance is added and I don't know if it really changed the power consumption or just what PP2 reported. Can anyone offer an explanation? 

[Kazi Afroza Sultana]

Hello,

We can not see any obvious reason why this should happen, except for the DC leakage currents that some capacitors might have. However still it seems a bit much for a leakage current in the magnitude.

For the nRF9160, the internal DC/DC converter operates in a burst (hysteretic) mode at low current, which periodically charging the supply capacitors. So, when we add large capacitance. it can change the frequency and amplitude of these bursts, however this will not increase the average current. Instead, it may change the profile of current spikes, which would confuse the measurement tool if it is not able to integrate over the burst cycles or if its range switching is affected by the load capacitance.

''nRF9160 contains an internal DC/DC converter operating in a hysteretic mode when the device is in a low power consumption state. In this mode, the internal DC/DC periodically charges supply capacitors. The charging cycle creates spikes in current consumption. During a charging period, current consumption is typically in the mA range, but during a quiet period it can be less than 1 µA.

The performance of the test equipment should be carefully considered when setting up a power consumption measurement system. The most important criteria are good sensitivity for low currents, high dynamic range, and good sample rate. For example, Keysight 6715C power analyzer with auto-ranging power modules and Keithley DMM7510 multimeter are viable options. Nordic Semiconductor's Power Profiler Kit II can also be used for current consumption measurements.''  source: https://docs.nordicsemi.com/bundle/nwp_034/page/WP/nwp_034/pwr_consumption.html 

DC-DC converters are inherently difficult to measure because of the rapid switching currents, and where the PPK2 won't be able to measure the negative currents that might occur.

Can you use a different measurement method with high precision bench power analyzer or oscilloscope to check the result? Then (e.g., a high-precision bench power analyzer or oscilloscope with a current probe) to cross-check the results. if you don't see large differences with other measurement but only with ppk2, we can say the artifacts is due to the measurement method, not related to the actual power consumption of your design.