NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Sigurd — Wed, 25 Sep 2024 14:04:05 GMT

[quote user="Anthony W"]So I'm mainly trying to understand if something needs to be done differently specifically for the nRF5340 dual-core platform, since that's where the process is breaking down but it's not clear why that would be the case.[/quote]

The main thing is that configuration of the SoftDevice Bluetooth Controller needs to be done in a child_image, hci_ipc.conf , but you seem to already have that under control. https://docs.nordicsemi.com/bundle/ncs-latest/page/nrf/app_dev/device_guides/nrf53/multi_image_nrf5340.html

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Anthony W — Fri, 30 Aug 2024 18:39:50 GMT

Hi Sigurd, just to clarify, the setup (with multiple NUS clients, I test with four) works perfectly fine for nRF52832 and nRF52840 platforms. Rebuilding for nRF5340 is what fails. So I'm mainly trying to understand if something needs to be done differently specifically for the nRF5340 dual-core platform, since that's where the process is breaking down but it's not clear why that would be the case. The Multi-NUS link shows generically how to support multiple NUS clients, but as indicated, that's already working.

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Sigurd — Thu, 22 Aug 2024 08:20:46 GMT

Hi!

[quote user="Anthony W"]This was extremely helpful.[/quote]

Great!

[quote user="Anthony W"]but maybe the NUS client library does not support multiple clients well in the two core setup, or I need to make some modification. [/quote]

At least some modifications are needed on the central_uart sample, e.g. use of connection context library. See these links:

Enter the Multi-NUS: A Simple Wireless UART Network

https://github.com/NordicMatt/multi-NUS/blob/master/src/main.c

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Anthony W — Mon, 15 Jul 2024 19:49:33 GMT

A different question but maybe a bit easier to answer: what considerations should be made for moving code from the nRF52 series to the dual core nRF53? Are the best practices or common "gotchas" to look out for? To me this seems like a strange bug to have, but maybe the NUS client library does not support multiple clients well in the two core setup, or I need to make some modification. Any advice you have to track down the source of the problem would also be helpful, the BLE software stack is difficult to sift through.

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Anthony W — Fri, 12 Jul 2024 17:05:58 GMT

This was extremely helpful. I don't have a sniffer log because it seems I was able to fix it based on part 2 alone. Previously, I would directly re-enable scanning at the end of the GATT discovery complete callback. Now, anywhere I restart scanning, I add it to the work queue instead, based on what I saw in the peripheral_hr_coded example. This seems to have resolved the GATT issues completely.

However, one odd thing that's happening now is that, at least with multiple sensors connected, it seems to fail to execute all of the BLE callbacks. What I observe happening is something like this example:

1. Device is turned on. No sensors connected.

2. Power on sensor. Device connects, I see full bandwidth (~7KB/s)

3. Power on second sensor. Device connects, one sensor drops bandwidth, the other is close to full (~3KB/s for one, 7KB/s for the other).

4. At this point, powering down either device doesn't result in a disconnect callback. The central hub firmware still thinks they're connected. Powering them on again also doesn't cause a reconnect.

Sometimes, the device with full bandwidth will still see a disconnect event, and I can even reconnect it. The device that has the reduced bandwidth never sees the event. This behavior, by the way, was observed previously on the nRF5340 (not a result of these changes). I do not see it on the nRF52840 so, perhaps like this bug, there might just be extra synchronization required somewhere to account for the fact the BLE host and controller are on different cores.

I attached a trace and a log as an example of this. I connect one sensor, it works fine. I attach another, it connects, but you see a drop in bandwidth on one. I can disconnect the first sensor, which yields no disconnect event, you just see the bandwidth drops to zero. The second sensor I'm able to disconnect and reconnect again. But the second time I power it down, there's also no disconnect event. There seems to be a disconnect somewhere, but I'm not sure where. Any further ideas? I can also open a new ticket if that's helpful, since this seems like a different (but similar) synchronization bug.

Here's the sniffer trace. I only went as far as to figure out how to capture it, so I have no idea if it contains anything useful.

SEGGER J-Link V7.94e - Real time terminal output 
SEGGER J-Link (unknown) V1.0, SN=1050009132 
Process: JLinkExe 
*** Booting nRF Connect SDK v3.5.99-ncs1-1 *** 
[00:00:00.002,441] <inf> posey_platform: Initializing platform... 
[00:00:00.002,441] <inf> posey_platform: Initializing CDC ACM... 
[00:00:00.502,624] <inf> posey_platform: Initializing NUS... 
[00:00:00.502,624] <inf> posey_nus: Initializing BLE and NUS... 
[00:00:00.509,399] <inf> fs_nvs: 8 Sectors of 4096 bytes 
[00:00:00.509,399] <inf> fs_nvs: alloc wra: 0, fd0 
[00:00:00.509,399] <inf> fs_nvs: data wra: 0, 1c 
[00:00:00.525,665] <inf> bt_hci_core: HW Platform: Nordic Semiconductor (0x0002) 
[00:00:00.525,695] <inf> bt_hci_core: HW Variant: nRF53x (0x0003) 
[00:00:00.525,726] <inf> bt_hci_core: Firmware: Standard Bluetooth controller (0x00) Version 54.58864 Build 1214809870 
[00:00:00.526,794] <inf> bt_hci_core: No ID address. App must call settings_load() 
[00:00:00.526,794] <inf> posey_nus: Bluetooth initialized 
[00:00:00.528,045] <inf> bt_0m 
[0m 
[0m 
[00:00:11.576,843] <inf> posey_task_hub: Connected sensors: 0 
[00:00:17.090,728] <inf> posey_nus: NordicNUSDriver::scan_filter_match 
[00:00:17.090,911] <inf> posey_nus: Filters matched. Name: Posey r2405 Rose (slot 0) Address: DE:AD:DC:19:58:5E (random) connect
able: 1 
[00:00:17.092,285] <inf> posey_nus: scan_connecting 
[00:00:17.309,631] <inf> posey_nus: NordicNUSDriver::connected 
[00:00:17.309,814] <inf> posey_nus: Name: Posey r2405 Rose 
[00:00:17.309,844] <inf> posey_nus: Connected to Posey r2405 Rose at DE:AD:DC:19:58:5E (random) 
[00:00:17.309,936] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:00:17.309,936] <inf> posey_nus: NordicNUSDriver::gatt_discover 
[00:00:17.309,967] <inf> posey_nus: Running NUS discovery for Posey r2405 Rose 
[00:00:17.310,028] <inf> posey_nus: 1 of 4 sensors connected. 
[00:00:17.510,803] <inf> posey_nus: MTU exchange done 
[00:00:18.160,186] <inf> posey_nus: LE PHY updated: TX PHY LE 2M, RX PHY LE 2M 
[00:00:18.460,662] <inf> posey_nus: NUS service discovery completed. 
[00:00:18.461,029] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:00:18.461,029] <inf> posey_nus: BLE scanning: Connected to 1 sensors. Active scanning. 
[00:00:21.619,232] <inf> posey_task_hub: Connected sensors: 1 
[00:00:22.321,441] <inf> posey_nus: Connection parameters update request received. 
[00:00:22.321,441] <inf> posey_nus: - Connection interval: [24, 40] 
[00:00:22.321,441] <inf> posey_nus: - Latency: 0 
[00:00:22.321,472] <inf> posey_nus: - Timeout: 42 
[00:00:22.652,618] <inf> posey_nus: Connection parameters updated: 
[00:00:22.652,648] <inf> posey_nus: - Interval: 40; Latency: 0; Timeout: 42 
[00:00:23.570,678] <inf> posey_nus: Slot 3 bandwidth: 36075 B in 5.01s - 7.04 KBps 
[00:00:28.615,447] <inf> posey_nus: Slot 3 bandwidth: 35831 B in 5.05s - 6.94 KBps 
[00:00:31.485,778] <inf> posey_nus: NordicNUSDriver::scan_filter_match 
[00:00:31.485,992] <inf> posey_nus: Filters matched. Name: Posey r2405 Lily (slot 1) Address: D6:81:5C:04:F1:2E (random) connect
able: 1 
[00:00:31.487,243] <inf> posey_nus: scan_connecting 
[00:00:31.660,888] <inf> posey_task_hub: Connected sensors: 1 
[00:00:31.802,459] <inf> posey_nus: NordicNUSDriver::connected 
[00:00:31.802,612] <inf> posey_nus: Name: Posey r2405 Lily 
[00:00:31.802,673] <inf> posey_nus: Connected to Posey r2405 Lily at D6:81:5C:04:F1:2E (random) 
[00:00:31.802,734] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:00:31.802,764] <inf> posey_nus: NordicNUSDriver::gatt_discover 
[00:00:31.802,795] <inf> posey_nus: Running NUS discovery for Posey r2405 Lily 
[00:00:31.802,856] <inf> posey_nus: 2 of 4 sensors connected. 
[00:00:31.968,353] <inf> posey_nus: MTU exchange done 
[00:00:32.718,353] <inf> posey_nus: LE PHY updated: TX PHY LE 2M, RX PHY LE 2M 
[00:00:32.818,389] <inf> posey_nus: NUS service discovery completed. 
[00:00:32.818,756] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:00:32.818,756] <inf> posey_nus: BLE scanning: Connected to 2 sensors. Active scanning. 
[00:00:33.714,324] <inf> posey_nus: Slot 3 bandwidth: 29836 B in 5.10s - 5.71 KBps 
[00:00:36.818,237] <inf> posey_nus: Connection parameters update request received. 
[00:00:36.818,267] <inf> posey_nus: - Connection interval: [24, 40] 
[00:00:36.818,267] <inf> posey_nus: - Latency: 0 
[00:00:36.818,298] <inf> posey_nus: - Timeout: 42 
[00:00:37.073,638] <inf> posey_nus: Connection parameters updated: 
[00:00:37.073,669] <inf> posey_nus: - Interval: 40; Latency: 0; Timeout: 42 
[00:00:37.972,961] <inf> posey_nus: Slot 0 bandwidth: 36075 B in 5.05s - 6.98 KBps 
[00:00:38.765,747] <inf> posey_nus: Slot 3 bandwidth: 19937 B in 5.05s - 3.85 KBps 
[00:00:41.701,507] <inf> posey_task_hub: Connected sensors: 2 
[00:00:42.974,060] <inf> posey_nus: Slot 0 bandwidth: 35587 B in 5.00s - 6.95 KBps 
[00:00:43.810,791] <inf> posey_nus: Slot 3 bandwidth: 19449 B in 5.05s - 3.76 KBps 
[00:00:47.978,118] <inf> posey_nus: Slot 0 bandwidth: 36319 B in 5.00s - 7.09 KBps 
[00:00:48.816,223] <inf> posey_nus: Slot 3 bandwidth: 20042 B in 5.01s - 3.91 KBps 
[00:00:51.744,934] <inf> posey_task_hub: Connected sensors: 2 
[00:00:53.019,989] <inf> posey_nus: Slot 0 bandwidth: 33496 B in 5.04s - 6.49 KBps 
[00:00:58.070,465] <inf> posey_nus: Slot 0 bandwidth: 36458 B in 5.05s - 7.05 KBps 
[00:01:01.786,834] <inf> posey_task_hub: Connected sensors: 2 
[00:01:02.718,292] <inf> posey_nus: NordicNUSDriver::disconnected 
[00:01:02.718,475] <wrn> posey_nus: Disconnected from Posey r2405 Lily at D6:81:5C:04:F1:2E (random) (reason: 8 - Connection tim
eout) 
[00:01:02.718,505] <inf> posey_nus: 1 of 4 sensors connected. 
[00:01:02.718,536] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:01:02.718,994] <inf> posey_nus: BLE scanning: Connected to 1 sensors. Active scanning. 
[00:01:11.829,101] <inf> posey_task_hub: Connected sensors: 1 
[00:01:16.342,895] <inf> posey_nus: NordicNUSDriver::scan_filter_match 
[00:01:16.343,078] <inf> posey_nus: Filters matched. Name: Posey r2405 Lily (slot 1) Address: D6:81:5C:04:F1:2E (random) connect
able: 1 
[00:01:16.344,635] <inf> posey_nus: scan_connecting 
[00:01:16.555,969] <inf> posey_nus: NordicNUSDriver::connected 
[00:01:16.556,121] <inf> posey_nus: Name: Posey r2405 Lily 
[00:01:16.556,182] <inf> posey_nus: Connected to Posey r2405 Lily at D6:81:5C:04:F1:2E (random) 
[00:01:16.556,243] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:01:16.556,274] <inf> posey_nus: NordicNUSDriver::gatt_discover 
[00:01:16.556,304] <inf> posey_nus: Running NUS discovery for Posey r2405 Lily 
[00:01:16.556,365] <inf> posey_nus: 2 of 4 sensors connected. 
[00:01:16.668,365] <inf> posey_nus: MTU exchange done 
[00:01:17.367,614] <inf> posey_nus: LE PHY updated: TX PHY LE 2M, RX PHY LE 2M 
[00:01:17.568,176] <inf> posey_nus: NUS service discovery completed. 
[00:01:17.568,511] <inf> posey_nus: NordicNUSDriver::scan_stop 
[00:01:17.568,542] <inf> posey_nus: BLE scanning: Connected to 2 sensors. Active scanning. 
[00:01:17.723,358] <inf> posey_nus: Slot 0 bandwidth: 31300 B in 19.65s - 1.56 KBps 
[00:01:21.577,575] <inf> posey_nus: Connection parameters update request received. 
[00:01:21.577,575] <inf> posey_nus: - Connection interval: [24, 40] 
[00:01:21.577,575] <inf> posey_nus: - Latency: 0 
[00:01:21.577,606] <inf> posey_nus: - Timeout: 42 
[00:01:21.873,748] <inf> posey_task_hub: Connected sensors: 2 
[00:01:21.901,855] <inf> posey_nus: Connection parameters updated: 
[00:01:21.901,885] <inf> posey_nus: - Interval: 40; Latency: 0; Timeout: 42 
[00:01:22.725,311] <inf> posey_nus: Slot 0 bandwidth: 34716 B in 5.00s - 6.78 KBps 
[00:01:27.871,948] <inf> posey_nus: Slot 0 bandwidth: 36075 B in 5.15s - 6.85 KBps 
[00:01:31.915,252] <inf> posey_task_hub: Connected sensors: 2 
[00:01:32.873,718] <inf> posey_nus: Slot 0 bandwidth: 36702 B in 5.00s - 7.17 KBps 
[00:01:37.919,982] <inf> posey_nus: Slot 0 bandwidth: 35448 B in 5.05s - 6.86 KBps 
[00:01:41.957,733] <inf> posey_task_hub: Connected sensors: 2 
[00:01:51.999,633] <inf> posey_task_hub: Connected sensors: 2 
[00:02:02.041,870] <inf> posey_task_hub: Connected sensors: 2 
[00:02:12.085,052] <inf> posey_task_hub: Connected sensors: 2 
[00:02:22.127,960] <inf> posey_task_hub: Connected sensors: 2 
[00:02:32.171,539] <inf> posey_task_hub: Connected sensors: 2 
[00:02:42.214,172] <inf> posey_task_hub: Connected sensors: 2 
[00:02:52.256,591] <inf> posey_task_hub: Connected sensors: 2 
[00:03:02.298,706] <inf> posey_task_hub: Connected sensors: 2 
[00:03:12.341,674] <inf> posey_task_hub: Connected sensors: 2 
[00:03:22.385,040] <inf> posey_task_hub: Connected sensors: 2 
[00:03:32.427,886] <inf> posey_task_hub: Connected sensors: 2 
[00:03:42.470,825] <inf> posey_task_hub: Connected sensors: 2 
[00:03:52.513,885] <inf> posey_task_hub: Connected sensors: 2 
[00:04:02.556,793] <inf> posey_task_hub: Connected sensors: 2 
[00:04:12.599,945] <inf> posey_task_hub: Connected sensors: 2 
[00:04:22.643,310] <inf> posey_task_hub: Connected sensors: 2 
[00:04:32.685,638] <inf> posey_task_hub: Connected sensors: 2  

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Sigurd — Fri, 12 Jul 2024 06:34:21 GMT

Hi!

[quote user=""]

[00:00:38.954,559] <inf> posey_nus: BLE scanning: Connected to 1 sensors. Active scanning.

[00:00:41.698,547] <inf> posey_task_hub: Connected sensors: 1

[00:00:42.754,364] <inf> posey_nus: NordicNUSDriver::disconnected

[00:00:42.754,547] <wrn> posey_nus: Disconnected from Posey r2405 Lily at D6:81:5C:04:F1:2E (random) (reason: 13 - Unknown)

[/quote]

1) Would you be able to capture a sniffer trace of these disconnects you are seeing? Would be interesting to see what's happening on-air as well. https://www.nordicsemi.com/Products/Development-tools/nrf-sniffer-for-bluetooth-le

2) The problem seem to come after you have connected to 1 sensor, and then start scanning for more for sensors.

What happens if you only connect to 1 sensor, and don't try to connect to any more sensors, is the connection stable then?

If yes, then please check the connection interval you are using for, and that the scanning has a reasonable scan interval/window, that leaves room for the connections as well.

RE: NUS connections on nRF Thingy:53 don't work reliably, compared to the nRF5340-DK

Anthony W — Thu, 11 Jul 2024 18:04:29 GMT

Actually, while the Thingy:53 has more BLE issues, this also doesn't completely work on the nRF5340-DK. Connections happen quickly, but the throughput when connected to multiple sensors is not what it should be. So this must be an nRF5340 related issue instead, something related to how the network and application core are interacting maybe.

While one hack might be just to completely run the code on the network core, I would I guess need to pipe UART data to the application core to then send it over USB. But I'd rather have it work than have to hack something together.

Any ideas would be greatly appreciated!