PowerProfiler 2: How to set up ampere-mode properly?

[danielschultz]

Hi guys,

I'm trying to measure deep-sleep current of my application quite precisely to obtain a reference measurement to evaluate against for further improvements.


The setup is as follows:
I connect a battery in ampere mode. Here, I'm struggling between two ways of doing this (which yields approx. a factor 3-4 of difference):


The PPK2 is connected to my laptop as the only USB device and the laptop-charger is not plugged in.


In setup (1), my deep-sleep currents measure approx. average 384 uA (10ksps) @ 10 sec:

Zooming in, noticing periodic transients:


The frequency of these harmonic transients to occur changes from DUT to DUT, so might be a HW issue (ground bouncing or so) in there.

Nevertheless, if I (everything else being equal), removes the GND from the PPK2 so only the hot wire goes through the PPK2, and then connects GND from battery -> GND on DUT (setup 2), then these harmonic transients (same frequency) are much less excitated and the avg. deep-sleep current falls to ~109 uA:


So this is a significant difference. I have also tried in source mode, but this generally yields a result of ~40 uA below configuration (2).
It seems to be a problem only connected to my DUT, or just a dynamic load. Measuring on a precision resistor yields more or less the same result.

Also, in configuration (2), I get a burden voltage of such significance that my application is not able to wake-up from deep sleep when intended to.

So I guess the primary question is; how do I assure the most precise output of my PPK2?

Can anyone me point me in the right direction? I have found some patterns of which I have a hard time concluding on, so any advice is very much welcome!

Have a nice weekend,

Br Daniel

[danielschultz]

Hey everyone,

I found myself having made a rookie mistake on the measurements. I see that with passive loads, setup (2) is yielding incorrect results. So I guess this method can be discarded. 

I observed there is a ~4 sec of obtaining a correct result when powering the (in this case passive) load. I must've been fooled by this "delay" and concluded upon the result within 4 seconds earlier ...

Q1: Is this behaviour to be expected? I'm curious if someone, with a reference to schematics can explain this behaviour if so.



This is some tests I have done with our 5 PPK2's. With "ampere mode" i am referring to test setup 1 mentioned above:


This yields a difference of ~40 uA from ampere mode to source mode on the same PPK2. Generally source mode is less. Can't figure this out if this is to be expected, but I guess it is expected behaviour when it's seen on all PPK2's

In ampere mode, there is a difference of max. 21 uA (~10%) from PPK2 to PPK2. With passive loads I don't see that much of a variance. Maybe this is the difference from deep-sleep cycle to deep-sleep cycle on the IoT devices. 

Q3: Will switching of the sense resistor be captured in measurements? For which current thresholds will they be switched?

Q4: The incorrect measurements on PPK2 no. (2) is due to the control of the source mode supply is malfunctioning. 3.6V output setting gives 4.15V out. When obtaining these tests, another device suddenly got this error aswell. Have you seen other instances of this failure to occur and what might be causing it?

Hope relevant answers to this post might serve to other trying to obtain reference measurements for further optimizing battery usage. Hopefully getting an answer to be sure to obtain measurements as should be done so one can question the DUT instead of the measurement gear.

Br Daniel

[Kenneth]

Hello,

Only setup #1 will work, you need to have ground between PPK and DUT for them to ensure they have the same ground potential:
https://infocenter.nordicsemi.com/topic/ug_ppk2/UG/ppk/measure_current_ampere_meter.html?cp=11_9_5_2

Normally "spikes" in measurements are caused by the internal regulator is operating in refresh mode, which means it only turn ON for short periods to charge it's decoupling capacitor. This overall reduce current consumption quite a lot.

It's difficult to provide any more details here without actually knowing what hardware and software is doing here, without knowing more of your DUT I would say setup #1 seems to be most correct, both in terms of what they look like and actual current.

Kenneth

[danielschultz]

Hi Kenneth,

Thank you for your answer. I also noticed setup #2 wouldn't work whatsoever- so that was a bummer by me!

DUT is an ESP32 with some peripherals connected to it, everything powered by the same Li-Ion battery. Measuring while the DUT and the peripherals are in deep-sleep.

So the question is if the very light but dynamic load (the DUT) can cause the above mentioned behaviour with transients in the measurements, even if this is not seen with a passive load (470 ohm - 1Mohm)? 

[Kenneth]

Hello again,

It may be that the spikes are caused by the PPK when it switches between two measurement resistors, because of the frequent change in current consumption.

Kenneth

[danielschultz]

Hi again,

I thought about that too - hence my question if someone knew what threshold values are for current ranges to switch the internal shunt resistors.

Either that or the internal regulator's operating in refresh mode. Do you know at what rate the internal cap is charged? Does this depend on DUT current consumption?