drivers/at86rf215: Add driver for the AT86RF215 dual-band IEEE 802.15.4-2015 radio #12128

benpicco · 2019-08-30T14:42:05Z

Contribution description

This adds a driver for the SPI based AT86RF215 transceiver.
The chip supports the IEEE 802.15.4-2015 and IEEE 802.15.4g-2012 standard.

This driver supports two versions of the chip:

AT86RF215: sub-GHz & 2.4 GHz radio & baseband
AT86RF215M: sub-GHz radio & baseband only

The AT86RF215 behaves like two chips in one package. There is a radio fronted and a baseband controller for both bands and they operate independently. They share the SPI interface and IRQ pin, it is the necessary to check the interrupt status registers to find out which of the two radios caused the interrupt. Each interface is an independent netdev_ieee802154.

Both radios support the following PHY modes:

MR-FSK
MR-OFDM
MR-O-QPKS
O-QPSK (legacy)

All modulations are implemented and configurable through ifconfig.
The driver defaults to O-QPSK (legacy) at 250 kb/s on both bands for compatibility with existing 802.15.4 devices.

Each radio has a separate RX and TX framebuffer that can hold 2047 bytes each. The device is able to transmit 2047 byte frames according to IEEE 802.15.4g-2012 when operating in non-legacy mode.
The larger frame size is automatically used by RIOT when switching to such mode.

What works:

normal 6LoWPAN operation
dual-band operation
software based automatic re-transmission
switching modulation at run-time with ifconfig set page
sleep mode & wakeup (ifconfig set state sleep/ifconfig set state idle)
sending > 127 byte frames in non-legacy mode
reduced power consumption

Whhat's missing:

~~proper interface to configure modulation~~
~~driver should handle ping6 -i 0 as graceful as at86rf2xx - currently all packets but one are dropped. (-i >= 10 ms works fine)~~
displayed / configured TX dBm do not match reality

Testing procedure

An example configuration is provided for connecting the ATREB215-XPRO module to EXT3 on the same54-xpro evaluation board.

Configuration for EXT1 on samr21-xpro

#define AT86RF215_PARAM_SPI        SPI_DEV(1)
#define AT86RF215_PARAM_CS         GPIO_PIN(PB, 3)
#define AT86RF215_PARAM_INT        GPIO_PIN(PA, 22)
#define AT86RF215_PARAM_RESET      GPIO_PIN(PA, 18)

Configuration for esp32-wroom-32 with a bunch of wires

#define AT86RF215_PARAM_SPI        SPI_DEV(0)
#define AT86RF215_PARAM_SPI_CLK    SPI_CLK_1MHZ // loose wires
#define AT86RF215_PARAM_CS         GPIO16
#define AT86RF215_PARAM_INT        GPIO34
#define AT86RF215_PARAM_RESET      GPIO17

In examples/gnrc_networking add the following to the Makefile:

USEMODULE += at86rf215
GNRC_NETIF_NUMOF := 2

You should now be able to communicate with other 802.15.4 radios.
(Tested with samr21-xpro (at86rf2xx) and mrf24j40 as communication partners)

OpenMote-B

The OpenMote-B board is supported by RIOT and contains an at86rf215 chip.
I do not have that hardware and could not find proper documentation, but I provided a configuration anyway based on the OpenWSN-fw code. This is entirely untested.
@MrKevinWeiss kindly provided me with documentation, this revealed that defines regarding the antenna switch were wrongly named. Sub-GHz is always connected to the at86rf215 whereas 2.4 GHz can be switched between cc2538 and at86rf215.

With this PR, at86rf215 will always be used for both sub-GHz and 2.4 GHz.

TODO

Channel Pages / Modulation selection

The device provides 3 different PHY modes that all come with different options.
IEEE 802.15.4-2012g ^{Amendment 3: Physical Layer (PHY) Specifications for LowData-Rate, Wireless, Smart Metering Utility Networks} puts those modes to channel pages 9 & 10, but introduces 'sub-fields' to select the modulation & mode. The standard does not provide a nice numerical mapping though (or I did not find it?) and I honestly have no idea how to map this to the channel page used in Riot.

I now added options to ifconfig to configure the PHY modes and modulation options individually.

Since they depend on the modulation, the values for getting/setting txpower are also somewhat bogus right now.

Issues/PRs references

#12082 allows for a small optimization where we can route the interrupt to the sibling device that last called _send(). This may save us the wake-up of the other thread in the common case.

MrKevinWeiss · 2019-09-09T11:14:47Z

I can try to help with the testing on the openmote. I really don't deal that much with radios but can do my best!

MrKevinWeiss · 2019-09-09T11:15:35Z

I know the sub-GHz was wanted for quite some time. It would be nice to support it.

benpicco · 2019-09-09T18:49:55Z

rebased to the changes made in #12171

@MrKevinWeiss Is there a data sheet for the openmote-b somewhere, so I can already prepare a setup for you to test?

MrKevinWeiss · 2019-09-10T10:50:59Z

The best I can do is email you the schematics... I don't know if they are public though so if you email me [email protected] I can forward them.

benpicco · 2019-09-24T14:48:46Z

Rebased to include #12295 so the build on AVR will not fail due to the added case in _print_netopt.

tinstructor · 2019-09-27T08:59:25Z

Let's say I have 2 OpenMote B nodes and I'd want to run a quick test, then how would you suggest I do that? @benpicco

benpicco · 2019-09-27T09:02:22Z

@tinstructor make -C examples/gnrc_networking BOARD=openmote-b GNRC_NETIF_NUMOF=2 -j flash should do.

Then when you run ifconfig you should see two interfaces.

If you ping the other node using the link-local address you have to make sure to specify the interface over which you want to reach it, so e.g ping6 fe80::f874:8230:d3e:1b8f%8 if you want to ping fe80::f874:8230:d3e:1b8f using the 2.4 GHz interface.

You can use ifconfig <id> phy_mode [FSK|OFDM|O-QPSK] to select different modulations.
ifconfig has also gained some options to further configure the individual modulations.

Of course you have to have the same settings on both nodes for them to be able to communicate 😉

tinstructor · 2019-09-27T10:07:42Z

@benpicco I don't seem to be able to get a response and I'm sure both nodes have the same config. However, is it possible that I first need to set the TX-Power and radio states? Also how can I see which interface is 2.4GHz/Sub-GHz respectively?

benpicco · 2019-09-27T10:14:27Z

You shouldn't need to configure anything, they default to legacy O-QPSK on both interfaces for compatibility with 802.15.4-2003.

Can you post your ifconfig output?

Also, try ping6 ff02::1%7 and ping6 ff02::1%8 for a boardcast ping on the sub-GHz (7) and 2.4 GHz (8) interface.

The internal oscillator can be tuned by adding up to 16*0.3pF to the RC circuit. To do so connect a frequency counter to the chip's clock output (and set AT86RF215_USE_CLOCK_OUTPUT to 1), then tune the value until the measured frequency is closest to 26 MHz. This should improve the radio characteristics of the transceiver.

If the device is in deep sleep it will not answer to reads of the PN register. We have to reset it before.

benpicco · 2020-03-09T18:29:37Z

#13598 should fix the issues on the openmote-b.

roberthartung · 2020-03-10T09:37:58Z

@benpicco Regarding #12128 (comment) - Fragmentation should not be an issue of the radio driver. If higher layers need fragmentation, this should be handled on higher layers, not here! There should be a higher level queue for this purpose.

benpicco · 2020-03-10T09:42:42Z

Fragmentation should not be an issue of the radio driver. If higher layers need fragmentation, this should be handled on higher layers, not here!

The driver does not handle fragmentation. It will just ensure that send() blocks while a transmission is still ongoing. This is the behavior seen on other radio drivers in RIOT as well.

There should be a higher level queue for this purpose.

#11263 would implement this.

we want to keep the diff small.

When the radio is in DEEP SLEEP, reading the state register will return 0. This makes the reset routine abort early even though the device is connected and still waking up. Instead, just add a timeout to the polling of the WAKEUP IRQ bit. In practise it has shown that the loop will be taken two times on 'normal' reset and four times when the radio was in DEEP SLEEP. Polling 255 times sure does not hurt. To reproduce the issue, put both interfaces into SLEEP mode, then reboot the module: ifconfig 7 set state sleep ifconfig 8 set state sleep reboot

fjmolinas · 2020-06-19T09:30:21Z

@benpicco are there futures in this PR still missing in master?

benpicco · 2020-06-19T09:49:35Z

are there futures in this PR still missing in master?

MR-FSK - I have a WIP branch for that
- KConfig integration
- ACK / CCA timeouts need to be calculated with the proper formula
Reduced Power Consumption - will cut it out of this branch after MR-FSK as this requires some additional handling in the MR-FSK case
Battery Monitor - I want to move it away from a callback based solution to posing a message on a message bus

TODO but not implemented anywhere:

TX dBm values are still bogus. Not sure if having large look-up tables is worth it just for a nice ifconfig output
small optimisation: currently the radio switches RX -> TXPREP -> RX for CCA. This is not needed - it can just start CCA in state RX
we need a layer on top that makes sure the frequency / modulation options comply with local regulations. Right now the user has to ensure the configuration is legal, we should have a way to just be able to set a config for EU/US/… to spare users the burden of meticulously going to the standards to figure out what combination they can use in their region.

benpicco · 2020-09-08T14:20:41Z

All features from this PR have now been split off to separate PRs or have been merged.

benpicco added Area: network Area: Networking Type: new feature The issue requests / The PR implemements a new feature for RIOT Discussion: RFC The issue/PR is used as a discussion starting point about the item of the issue/PR Area: drivers Area: Device drivers labels Aug 30, 2019

benpicco requested a review from jia200x August 30, 2019 16:37

benpicco force-pushed the at86rf215 branch 2 times, most recently from 369990b to 791f334 Compare September 5, 2019 01:03

benpicco mentioned this pull request Sep 5, 2019

shell_commands: gnrc_netif: add support for getting/setting PHY modes & options #12171

Closed

benpicco force-pushed the at86rf215 branch from 791f334 to 962b34a Compare September 5, 2019 02:02

benpicco requested a review from MrKevinWeiss September 5, 2019 02:04

benpicco force-pushed the at86rf215 branch from 962b34a to e069b1d Compare September 9, 2019 18:47

benpicco force-pushed the at86rf215 branch 3 times, most recently from dbf0718 to d477617 Compare September 16, 2019 23:19

benpicco force-pushed the at86rf215 branch from a98beef to 5ae2cd5 Compare September 24, 2019 13:32

benpicco marked this pull request as ready for review September 24, 2019 13:32

benpicco added CI: ready for build If set, CI server will compile all applications for all available boards for the labeled PR and removed Discussion: RFC The issue/PR is used as a discussion starting point about the item of the issue/PR labels Sep 24, 2019

benpicco force-pushed the at86rf215 branch 2 times, most recently from dfebf8c to 209b2ac Compare September 24, 2019 14:47

benpicco force-pushed the at86rf215 branch from 21362ec to 707c8e2 Compare September 24, 2019 16:18

benpicco added 5 commits March 9, 2020 19:28

at86rf215: don't read part number before reset

2f3218e

If the device is in deep sleep it will not answer to reads of the PN register. We have to reset it before.

at86rf215: don't change state while the device is busy TXing

0fa8179

at86rf215: adapt to auto_init move

92b78e0

at86rf215: fix OFDM channel spacing

edd1832

benpicco force-pushed the at86rf215 branch from f78fde3 to edd1832 Compare March 9, 2020 18:29

benpicco and others added 11 commits March 11, 2020 18:50

at86rf215: first round of review comments

d33bd4a

at86rf215: fix module handling

a7f0442

at86rf215: properly handle at86rf215m

29a59b8

at86rf215: clean up get_random()

2b149ac

at86rf215: fix default energy detect threshold

bae75e8

at86rf215: add pre-condition to get_random()

4654766

at86rf215: separate vendor file from custom additions

cfc22a1

at86rf215: add defines for default parameters

27c8b29

at86rf215: more review comments

4185472

at86rf215: backport changes from mainline driver

b289b4e

we want to keep the diff small.

benpicco mentioned this pull request Apr 20, 2020

drivers/at86rf215: implement MR-O-QPSK #13912

Merged

benpicco mentioned this pull request May 4, 2020

drivers/at86rf215: implement MR-OFDM #14010

Merged

benpicco added the State: demonstrator State: This PR (loosely) demonstrates a concept and is not necessarily meant to be merged. label Jun 19, 2020

This was referenced Sep 6, 2020

drivers/at86rf215: implement MR-FSK #14959

Merged

drivers/at86rf215: implement Reduced Power Consumption #14960

Merged

drivers/at86rf215: implement battery monitor, add power bus #14973

Merged

benpicco closed this Sep 8, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

drivers/at86rf215: Add driver for the AT86RF215 dual-band IEEE 802.15.4-2015 radio #12128

drivers/at86rf215: Add driver for the AT86RF215 dual-band IEEE 802.15.4-2015 radio #12128

benpicco commented Aug 30, 2019 •

edited

Loading

MrKevinWeiss commented Sep 9, 2019

MrKevinWeiss commented Sep 9, 2019

benpicco commented Sep 9, 2019

MrKevinWeiss commented Sep 10, 2019

benpicco commented Sep 24, 2019

tinstructor commented Sep 27, 2019

benpicco commented Sep 27, 2019 •

edited

Loading

tinstructor commented Sep 27, 2019 •

edited

Loading

benpicco commented Sep 27, 2019

benpicco commented Mar 9, 2020

roberthartung commented Mar 10, 2020

benpicco commented Mar 10, 2020

fjmolinas commented Jun 19, 2020

benpicco commented Jun 19, 2020 •

edited

Loading

benpicco commented Sep 8, 2020

drivers/at86rf215: Add driver for the AT86RF215 dual-band IEEE 802.15.4-2015 radio #12128

drivers/at86rf215: Add driver for the AT86RF215 dual-band IEEE 802.15.4-2015 radio #12128

Conversation

benpicco commented Aug 30, 2019 • edited Loading

Contribution description

Testing procedure

OpenMote-B

TODO

Channel Pages / Modulation selection

Issues/PRs references

MrKevinWeiss commented Sep 9, 2019

MrKevinWeiss commented Sep 9, 2019

benpicco commented Sep 9, 2019

MrKevinWeiss commented Sep 10, 2019

benpicco commented Sep 24, 2019

tinstructor commented Sep 27, 2019

benpicco commented Sep 27, 2019 • edited Loading

tinstructor commented Sep 27, 2019 • edited Loading

benpicco commented Sep 27, 2019

benpicco commented Mar 9, 2020

roberthartung commented Mar 10, 2020

benpicco commented Mar 10, 2020

fjmolinas commented Jun 19, 2020

benpicco commented Jun 19, 2020 • edited Loading

benpicco commented Sep 8, 2020

benpicco commented Aug 30, 2019 •

edited

Loading

benpicco commented Sep 27, 2019 •

edited

Loading

tinstructor commented Sep 27, 2019 •

edited

Loading

benpicco commented Jun 19, 2020 •

edited

Loading