nRF9160 power supply selection

[tedhavelka]

Hello Devzone,

My name is Ted.  Recently I posted here and received some good help on low power sample code for the nRF9160DK board.  This code helped me prove in real life that it is possible for the nRF9160 draw just single microamps of current, in a proper deep sleep configuration.

I've returned to a custom board design on which I am working, and I adapted that hello_world application for our board.   On this custom hardware my finding is that I can achieve a floor current of about 25uA.

Everything outside the 9160 SiP is turned off, save for a single switching regulator which provides VDD and VDD_GPIO to the SiP.  The SiP application core enters Zephyr's idle thread, and hence the particular processor's deep sleep mode.  The Nordic Power Profiler Kit II in source meter mode gives these measurements.

Our DK board draws ~4.5uA with the low power hello_world code.  Our custom board draws about 25uA.  I suspect the switcher is drawing the extra current, but I cannot turn that off as nRF9160 requires a continuous supply at VDD and VDD_GPIO.

A key difference I understand is that the nRF9160DK measurement set up captures only the current to the nRF9160 SiP.  Measuring current on my custom board I am capturing total board current.  I need to measure total board current, as the overall design is powered by battery.  Given a desired battery life I must bring average total board current down to single microamps, in order to justify using the nRF9160 in our design.

This brings me to my question: does Nordic Semi have reference designs which account for keeping an external regulating part or circuit to very low current consumption?  I know this question may be a long shot.  I know component availability is ever changing and Nordic's estimating power profiler cannot account for arbitrary circuit designs.  Nonetheless there seems to be a catch here.  According to various datasheets and documents, the nRF9160 itself can draw as little as 1.4uA in deep sleep.  In a practical circuit however, there is normally a regulated supply to provide a steady power rail to the SiP.  In a battery based circuit, any regulator circuitry must be able to draw comparably low current, else the super low current draw possible with the nRF9160 does not mean much, when to keep the nRF9160 powered another circuit element must use multiple times the current draw of the SiP.

Is my understanding here accurate?  Would all bets for stable nRF9160 operation be off, if I power the SiP directly from a high impedance battery, one whose discharge curve is relatively flat among different battery type discharge curves?

- Ted

[Kenneth]

Hi Ted,

Not sure if this answers your question directly, but below the https://devzone.nordicsemi.com/power/w/opp/3/online-power-profiler-for-lte you can find the comment:

"The tool does not show peak currents. The different bars in the graph show the average current within a specified event type, e.g. iDRX event. The radio RX/TX will typically give the highest peak currents, but it is only a small part of the event itself, hence the average current will be lower than the peak current. Please refer to the nRF91 Product Specification for expected min/max values for the different current components."

So the actual peak currents are even higher than shown, and they are the highest when the modem is used (highest peak is 435mA):
https://infocenter.nordicsemi.com/topic/ps_nrf9160/_tmp/alta.nRF9160/autodita/CURRENT/parameters.id_current_modem.html 

So, based on this I can answer your question:

Is my understanding here accurate?  Would all bets for stable nRF9160 operation be off, if I power the SiP directly from a high impedance battery, one whose discharge curve is relatively flat among different battery type discharge curves?

You can't make a design with a high impedance battery no, the voltage drop would be way too high. You might get some useful tips from the thingy91 design in terms of design:
https://www.nordicsemi.com/Products/Development-hardware/Nordic-Thingy-91

Best regards,
Kenneth

[tedhavelka]

Hello Kenneth,

Thank you for the fast reply, and patience with a basic hardware question.  I have read Nordic's on-line Power Profiler description and notes on limitations a few times now.  Repetition helps cement understanding.  I understood the absence of full peak current reporting.  Thanks however for the "sanity check" reminder on the obvious point "voltage drop too great running LTE modem and SiP direct from high impedance battery".

Regarding Nordic's Thingy91, my team's hardware lead has designed a custom board for our battery based product idea, and his design follows the patterns of the Thingy91 in many respects.  The boards are not identical but they are similar.

So last summer 2021 I had read much of Thingy91 documentation on-line.  Lots of good information on your site!  One thing I could not find, however, is a schematic of the Thingy91. 

Question (1)  Is Thingy91 schematic not publicly available?

We have on hand one nRF9160DK board, version 1.0.0 as silk-screened on the back of this board.  With a modified "hello_world" app from another ticket I started here in 2022 March we can measure a floor current of 4.5uA used by the nRF9160 SiP.  On a custom board I can reach a floor current of ~22uA.  I don't have test points which permit me to measure exclusively SiP current.  Practically, however, I need to reduce the total board current to single microamps current use, in deep sleep mode, in order to justify use of the nRF9160 in our design.

Question (2) does the nRF9160 SICA B1A version of this SiP no longer have an internal leak current issue when VDD_GPIO is supplied with 1.8 volts rail?

One concern I have is that I find floor current increase when I set just a subset (four GPIOs) to Zephyr's GPIO_DISCONNECTED pin config value in firmware.  I may be opening a separate ticket if I cannot find a root cause for this.  From another recent post 86752 I made one of your team confirms that Zephyr's DISCONNECTED configuration is the highest impedance way to configure a GPIO.  So it is unclear to me how nRF9160 pins would turn on in some default setting that draws less current than when they are explicitly configured to be disconnected.

Final comment regarding my team's low power search efforts:  to reach a floor current of 22uA with nRF9160 in our custom board I needed to disable Zephyr 2.6.0's power management feature, that is, to set CONFIG_PM=n.  This feature increases floor current by about 25uA.  I have some Zephyr source tree code to review, and perhaps this is best in a new ticket.  In any case,

Question (3)  Is it known that Zephyr 2.6.0 run time power management Kconfig feature steadily increases nRF9160 floor current?

Thank you for your help, Kenneth.  You and your fellow engineers make this an excellent support forum!

- Ted

[tedhavelka]

Good evening Kenneth,

Thanks for the further help.  I have just created a new ticket 87158 for nRF9160 low power and Zephyr CONFIG_PM=y questions.

Thank you especially for the pointer to the hardware files.  It must be me, that I've had some trouble locating the schematics files for various Nordic Semi development kits.  I don't know that they are too buried, but I believe developers would benefit from having a more direct link or search result pointing to them.

On a Devzone use note, I don't yet fully know the normal "ticket life" flow, though in previous tickets I have on a few occasions asked Nordic Team to change a given ticket status to "closed" or "answered" once the issue is resolved.

For now I would like to keep this ticket open.  I'm hoping to find answers to these questions.  Once answered I will make a final reply here, pointing to the answers (documentation or other Devzone posts), and request to close this ticket then, if that is alright.

Thank you again for your help Kenneth.  Stay safe,

- Ted