nRF9160 Feather - Suggestions for next version

zpm1066 · 2023-05-17T02:51:26+00:00

Hi Jared, Any chance you can share what new changes you have in mind for the next iteration of nRF9160 Feather? Although I do understand that board space i...

AchimKraus

Will it be equipped with the new nRF9161 variant?

Built-in programmer/debugger/console access

Does this refer to the USB serial?

Control of RP2040 power for advanced users

I’m not common to the RP2040 and a short look at the spec didn’t help me too much. For me, the nRF9160/61/zephyr is a platform, which is robust and OK to develop software for. The quiescent power consumption of the feather v5 is simply the “best in class” and I’m not sure, what the consequences of a RP2040 will be on that aspect. And building systems on two MCU isn’t too easy. I don’t know, if there is so many software for the RP2040 that’s worth to consider that.
And if I would consider to use a RP2040, I guess it will be a SparkFun RP2040 mikroBUS Development Board with an LTE IoT 4 Click / nRF9160 and cut of the LED on the Mikro-Click.

All GPIOs are dedicated now (no sharing with other features)

I guess this is related to the feather’s GPIO, not the SOC, right?

Single antenna operation

OK. (I mainly do not use GPS, so it’s no difference for me).

RGB led

That’s something, I will enjoy.

Maybe, you consider also to connect some GPIO to the COEX interface? Some internal GPIO not used by the feather. the idea is to detect “radio activity” even when the app is not aware of that (HPPLMN search e.g.).

zpm1066

AchimKraus
Re RP2040 - If I’m not mistaken, the RP2040 is intended to be used as a debug probe running standard CMSIS-DAP debug and UART interfaces over USB. Probably a modified version of the open-source Raspberry Pi DebugProbe firmware. Currently lacks RTT support but debug probe works well on the Picos.

The RP2040 is a pretty inexpensive and a powerful SoC with programmable I/O (PIO) that can execute code independently of the Cortex-MO+. The PIO opens the door to a whole host of applications that manipulate IO and transfer data, essentially it can be used to emulate different peripherals and communication protocols (e.g. CAN) independent of the RP2040’s ARM core. The RP2040 is pretty well supported by Zephyr RTOS.

Re RGB led - A pretty useful addition. I hope the RGB led is individually GPIO addressable and isn’t a WS2812 led.

jaredwolff

zpm1066 Re RP2040 - If I’m not mistaken, the RP2040 is intended to be used as a debug probe running standard CMSIS-DAP debug and UART interfaces over USB. Probably a modified version of the open-source Raspberry Pi DebugProbe firmware. Currently lacks RTT support but debug probe works well on the Picos.

Correct! It will either use the pico probe firmware or probe-rs. More testing required though.

zpm1066 Re RGB led - A pretty useful addition. I hope the RGB led is individually GPIO addressable and isn’t a WS2812 led.

It’s controlled by the LED drivers on the nPM1300. Each one is individually controlled via I2C

AchimKraus

To be frank: I also don’t want to “waste my time” with replacing the segger jlink ;-).

zpm1066

AchimKraus
I can understand where you’re coming from. I use the Segger J-link & RTT all the time with nRF/Pico/STM32 devices with the SWD 10 pin connectors. If there’s no space for a Min SWD connector, an onboard RP2040 hardware debug probe is preferable to a Tag Connect.

AchimKraus

zpm1066 If there’s no space for a Min SWD connector, an onboard RP2040 hardware debug probe is preferable to a Tag Connect.

If the RP2040 is removed, there will be space ;-).

I guess the question is more, will the new feather be a “complete DK with debugger”. I’m not sure, if that will be the right way. The new nRF9161-DK comes at an even lower price than the nRF9160-DK. Both comes with a well working software stack.

My interest in the feather is to have a device close to a “productive device” (at least a good prototype), not a “development kit”. I’m only one user, if more feel well with a DK, then I guess it will go that way. If others are also don’t consider a DK as too useful, then I would recommend not to go that way.

What’s about the other additional functions? Extra flash? Accelerometer?

jaredwolff

@AchimKraus you’re more of a power user. Many of my customers have complained about the fact that they have to buy a proprietary cable and programmer. You can still program via the SWD pins on the bottom of the board but that would require a programming fixture.

AchimKraus What’s about the other additional functions? Extra flash? Accelerometer?

Same flash as before (32Mbit) and same accelerometer. Trying to keep feature parity with the nRF9160. I did remove the RTC though since the nRF916x can maintain the time (as long as it’s not rebooted of course!)

zpm1066

jaredwolff Thanks. A BME280 at half the cost of a BME680 may be worth a consideration.

I use the BME688 in Thingy:53 and Adafruit BME688 breakout for prototyping; and BME688/nRF52840Dongles in a Thread network. In addition to BME680 features, the BME688 has a gas scanner function for ML. BME688 costs ~$7 per 100, about the same price as a BME680; and BME280 is ~$4 per 100. Personally, I’d pay the extra and go for the BME688.

AchimKraus I guess the question is more, will the new feather be a “complete DK with debugger”. I’m not sure, if that will be the right way. The new nRF9161-DK comes at an even lower price than the nRF9160-DK. Both comes with a well working software stack.

I didn’t realize that nRF9161-DK is considerably lower in price than a nRF9160-DK. A quick check on mouser.com shows nRF9161-DK ~$99 vs. nRF9160-DK $155.

It’ll be interesting how Nordic will price the upcoming Thingy 91:X (nRF9151, nPM1300, nRF5340, nRF7002, nPM6001, and same sensors as Thingy:53) and nRF9131-EK (nRF9131, nPM1300, Segger J-Link OB Programmer/debugger). Both of these offerings will be excellent for prototyping.

I would use the nRF9161 Feather boards as deployment devices with easier development accessibility, either a Mini SWD 10-pin connector or the proposed design.

jaredwolff

Alternatively, I can change the 3pin SWD connector to connect to the nRF9161 instead making the RP2040 only updatable over USB bootloader.

zpm1066

jaredwolff Yes, adding a SWD connection to the nRF9161 sounds good.

I know it’ll increase the cost but how about adding a BME680? It opens the door to a quite a few other use cases for customers, including embedded machine learning.

I guess, one of the reason for the RP2040 selection was the availability of an open-source DebugProbe firmware vs. licensing Segger J-Link OB firmware running on nRF52840 or nRF5340. Plus the open-source option allows future enhancements.

jaredwolff

zpm1066 I know it’ll increase the cost but how about adding a BME680? It opens the door to a quite a few other use cases for customers, including embedded machine learning.

This has something other customers have asked for in the past. I’d consider it but it may push the cost way to far to make the project sustainable.

AchimKraus

In my experience of the past years:

you will get troubles developing a cellular application, if you not only run demos.
in trouble you need the Nordic Forum and the first question back is, “Please provide a modem trace”
it’s also easier to solve trouble, if that could be reproduced with Nordic equipment

All that, and the new price of the nRF9161-DK, ends up in my conclusion, that a developer will have such a nRF9161-DK. Those who only use demos and slightly modify them will be happy with the USB bootloader.

The Thingy is nice, I also use a couple of them. But once you need your own sensor, or other batteries or antenna, then the nRF9160 feather is still the best in class.

zpm1066 I would use the nRF9161 Feather boards as deployment devices with easier development accessibility, either a Mini SWD 10-pin connector or the proposed design.

That is exactly also my usage. For example the Mobile BeeHive Scale is built with an nRF9160 feather, a Thingy:91 is missing expandability and a DK is just to large in size.

zpm1066

AchimKraus The Thingy is nice, I also use a couple of them. But once you need your own sensor, or other batteries or antenna, then the nRF9160 feather is still the best in class.

In the few Thingy:91X perspective photos shared from MWC Barcelona 2024 and the recent Embedded World exhibitions, it looks like Thingy:91X may be supporting an EB (Expansion Board) like Thingy:53. If so, then we may get I2C, SPI and several GPIOS for sensors and peripherals. It definitely has an I2C STEMMA connector, plus holes for ‘charge led’ & ‘led2’ above the USB-C and STEMMA connectors. As yet, I don’t see an “edge_connector” defined in the thingy91x devicetree in NCS v2.6.0/nrf/boards/arm. Thingy:91X release was suppose to be in Q1 but looks like it’ll be in Q2, so perhaps work in progress.

zpm1066

AchimKraus That is exactly also my usage. For example the Mobile BeeHive Scale is built with an nRF9160 feather, a Thingy:91 is missing expandability and a DK is just to large in size.

For LTE, I’ve been using Particle Boron Feather boards with nRF52840 / UBlox SARA modem, though these are not low power like the nRF91xx. I’ve been pretty satisfied with their Borons. BTW - Their EOL Xenons (nRF52840) make excellent Zephyr dev boards. A few online places still have old stock.

I haven’t come across any nRF9160 boards in a Feather format that satisfy my requirements in terms of a SWD 10-pin J-Link connector for ease programming/debugging. I was planning on designing my own nRF91xx Feather but thought I’ll check what Jared’s plans are going forward for the next iteration of his board. Why re-invent! Hence, the reason for starting off this topic last year.

However, Jared’s proposed design for a nRF91xx Feather checks off several asks - nPM1300, SWD Programmer/Debugger, and eSim. So, I definitely hope to use these nRF91xx Feathers for future deployments and development.

zpm1066

AchimKraus Thingy:91 is missing expandability and a DK is just to large in size.

Recently Nordic Semi has been making a few changes to the Thingy91:X devicetree files, e.g. added support for the expansion board and other misc changes. I don’t see an “edge_conenctor” defined in the "“thingy91x_nrf9151” devicetree yet like in nRF5340. Somewhat slow progress at the moment, perhaps there are delays in nRF9151 availability.

I expect the Thingy91:X Expansion Board will be the same spec as that in the Thingy:53 (nRF5340) or an enhanced version with backward compatibility. The nRF7002-EB WiFi 6 ( I use several with Thingy:53 and nRF52840 boards) should work with the Thingy:91X. With I2C (EB and I2C STEMMA), SPI and several GPIOs, the Thingy:91X should be pretty expandable for sensors and peripherals.

Originally, both the Thingy91:X and nRF9131-EK were planned for Q1 release. Hopefully we’ll see a Q3 release.

zpm1066

jaredwolff
Is there any reason why you wouldn’t want to use the nRF9151 over nRF9161 other than perhaps pin compatibility with nRF9160?

From the specs, nRF9151 looks to be identical to a nRF9161 with the addition Power Class 5 20dBm and , of course, a smaller packaging. What advantages does nRF9161 provide over nRF9151?

jaredwolff

zpm1066 Is there any reason why you wouldn’t want to use the nRF9151 over nRF9161 other than perhaps pin compatibility with nRF9160?

They are actually not exactly pin compatible and does require a footprint change.

From the specs, nRF9151 looks to be identical to a nRF9161 with the addition Power Class 5 20dBm and , of course, a smaller packaging. What advantages does nRF9161 provide over nRF9151?

It’s available now vs waiting a while for the nRF9151. I do have an idea for the nRF9151 since it allows for more room. There’s still time to switch but I’m limited by when Nordic can deliver.

AchimKraus

Thingy:91X

If it gets extendable, that will be nice. Exchanging the battery and antenna has also some benefit, so at least I will continue to use feathers for prototypes and the thingy:91 mainly as “network discover and tester”.

AchimKraus

I’m only interested in real low power devices. Leave the other stuff for the MQTT folks ;-)
I guess Jared remembers, that before I ordered the first feathers I asked him in a private e-mail, if the feather really goes down to 10µA (because the Thingy:91 before v1.6 was more about 35-40µA and only, if you disable the 3V in deep sleep).

A “low power device” (beside of a lot of marketing plurr, the nRF9160 is the only kid) accompanied by a “low power protocol” (CoAP/DTLS 1.2 CID) enables to build battery solutions and makes such devices much more independent.
E.g. the Mobile BeeHive Scale runs from 3xAA 2000mAh (LSD) more than a year in the “wild” (in theory even 2 years ;-) ).

zpm1066

jaredwolff It’s available now vs waiting a while for the nRF9151. I do have an idea for the nRF9151 since it allows for more room. There’s still time to switch but I’m limited by when Nordic can deliver.

I suspect nRF9151 availability may be closer to Q3. Probably similar for the nRF54xx as well.

Btw - Have you considered doing a Feather board with nRF7002 with either nRF54L15 or nRF54H20?

AchimKraus

In the meantime I got it, that the RP2040 will be used as USB bridge, and though on the board, maybe for swd. For the USB bridge it would be great to have two channels, that makes getting trace easier. I guess, there are then drivers for such an USB bridge, for Ubuntu and Windows? Mac?
About the swd: The thingy:91 has a switch to select the target. Maybe that’s also an idea.

« Previous Page Next Page »