Power Profiler Kit II measurement error

[MartinZ]

I made some tests with PPK-II in source meter mode by connecting a load resistor in series with a Fluke 8846A ampere meter to the PPK output. A also hooked a GW Instek GDM-8351 voltmeter in parallel to the PPK output.

Here are the results:

The PPK is out of spec around 500uA. From time to time, while changing the output voltage up and down, those red values suddenly got much better, in accordance with the +/-10% specified max. error. But I couldn't figure out what exactly makes the PPK suddenly jump in/out of spec. However, most of the time I observed the large errors from the table above.

Given the described behavior, I suspect a FW error in range changing logic. Are you aware of this? Can this be fixed in a future FW upgrade?

[Kenneth]

Hello,

First thought: It may look like a problem when PPK II change measurement range (which happens around 500uA), we will try to look into it.

Kenneth

[Audun]

Hi,

thanks for the detailed results and setup!

As you mention, it could indeed be an issue with the measurement ranges. We will investigate internally with the same setup as you have used and see if we can make adjustments to improve this behavior.

As a side note, the updated PPK2 firmware has a secondary COM port used for command line access. To determine if it is actually an issue with the ranges, you can open this COM port and execute the "sampling metrics" shell command after sampling has stopped. The command line interface is non-responsive while sampling, to avoid interfering with the primary function of PPK2.

See this screenshot as a reference:

If the "samples dropped" metric is high, it indicates that there is an issue with the measurement ranges.

Finally, you can use the shell command "sampling discard_switch_samples 0" to never discard any samples. I would be very interested to know if this improves the behavior or not.

Best regards,

Audun

[Audun]

Hi again MartinZ ,

I do have a new version of the firmware you can try if you want. I don't expect this to drastically improve accuracy, but hopefully it will be a bit more predictable. Some of the measurement range detection timings have been tightened, and the VDD output accuracy might be a bit improved.

You can program the new firmware using the nRF Connect for Desktop Programmer app. Power Profiler will complain that there is a version mismatch, but you can just click "No" in the prompt.

During testing on my PPK2, I observe that measurement range 3 (0.5 mA - 5 mA) shows an offset of about 100 µA. For a load in the 500 µA range, the accuracy depends on whether PPK2 uses range 2 or range 3. Note that the range selection is done purely in hardware, and cannot be directly controlled from firmware.

I can compensate by changing the gain for this particular range under "Advanced configuration":

It might be possible to improve the calculation done on the Power Profiler side, but I have no timeline for when we might look into this.

Best regards,

Audun

pca63100_ppk2_0.0.0_5bdd04e.hex

[MartinZ]

EDIT: This of course replies to Audun's 1st post but even if I click Reply, it gets placed at the bottom of the thread.

Hi,

I tried what you propose. Here is the setup:

- 6k8 load resistor

- start at U_set=2500mV, sample for a few seconds, current is correct

- switch to U_set=2600mV, sample for a few seconds, current is correct

- switch to U_set=2700mV, sample for a few seconds, current is >20% too high

- stop the sampling

- execute "sampling metrics":

shell:~$ > sampling metrics
sampling metrics
Metrics:
  Samples in session=2741521
  Samples dropped=3 (+1)
  Blocks dropped=114 (+39)

Immediately after that, I typed in the command "sampling discard_switch_samples 0" and repeated everything from above. The results were practically the same. Sampling metrics output:

shell:~$ > sampling metrics
sampling metrics
Metrics:
  Samples in session=3622159
  Samples dropped=3 (+0)
  Blocks dropped=135 (+21)

Please let me know if this gives you any insight into what might be happening.

[MartinZ]

Hi,

Thanks for your feedback. Before trying the new FW, I'd like to know how to safely revert back to official FW after this test.

Second thought, I find it hard to imagine that this is a gain problem in any particular range because I get correct results both in the 50uA-500uA range well below 500uA as well as at the 500uA-5mA range well above 500uA.

I saw in the schematic that the range switching is indeed fully automatic by the HW. I checked the state of some test points at different output voltages:

	U_set=2600mV R_load=6k8 I_meas=380uA	U_set=2700mV R_load=6k8 I_meas=478uA	U_set=2700mV R_load=3k3 I_meas=870uA	U_set=2700mV R_load=1k0 I_meas=2.75mA
TP30	3.3V	3.3V	3.3V	3.3V
TP31	0V	3.3V	3.3V	3.3V
TP32	0V	0V	0V	0V
TP33	0V	0V	0V	0V
TP3	1.56V	0.22V	0.38V	1.13V
TP21	60mV	60mV	60mV	60mV

The problematic behaviour is thus at the lower end of the 500uA-5mA range, and it doesn't seem to be caused by the offset dictated by POT 3 (TP21).

If it's not the FW, could it be some non-linearities in the FETs through which the measured current passes (Q2, Q16, Q5, Q7)?

Don't get me wrong, I'm well aware that the PPK-II is not a laboratory instrument but more of an indicator. I'm just curious about what's going on, but also pointing out that the results may be outside the stated specification. (Please don't go the way of some car manufacturers, which is to adjust the specification when they can't fix it).