sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

ovrebekk — Tue, 19 Mar 2024 13:52:08 GMT

Hi Rbelli

Hung is currently on leave, so I will follow this ticket in his place.

[quote user="rbelli"]The only idea I have is that the problem is related to corruption of packets due to other bluetooth devices. When I first made the product I could look at the sniffer output and see the communication happening in real time. After the problem was observed the sniffer output is an uninterpretable blur.[/quote]

Bluetooth devices generally don't interfere a lot with each other, since they use a narrow frequency band, a low output power, and usually transmit very infrequently. Wi-Fi is a larger cause of concern, since Wi-Fi uses a much wider band, and can transmit at higher output power levels than most Bluetooth devices.

If you have connection/link quality issues its definitely worth taking a look at the hardware, as mentioned by Hung earlier. If the problem is present on your custom hardware but non on standard DK's then it is a sign that the hardware has room for improvement.

[quote user="rbelli"]The project will be transitioned to a mesh network so I will not do any more work on this problem.[/quote]

The best of luck 🙂

Best regards
Torbjørn

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Fri, 15 Mar 2024 20:29:25 GMT

I found that the scanning TX Power level was a separate setting. After increasing this to the maximum I found that the problem decreased. We also have some units that only do scannable advertising and no connections. These work much better but that is due to many more messages being attempted.

One partial solution I found is to reset the device when the connection problem presents itself. This covers up the problem to some extent but doesn't fix it entirely.

The only idea I have is that the problem is related to corruption of packets due to other bluetooth devices. When I first made the product I could look at the sniffer output and see the communication happening in real time. After the problem was observed the sniffer output is an uninterpretable blur.

The project will be transitioned to a mesh network so I will not do any more work on this problem.

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

Hung Bui — Wed, 06 Sep 2023 09:30:56 GMT

Hi Rbelli,
My suggestion is to do the test using 2 development kits. This way we can be sure it's not the hardware or poor radio tuning causing the issue.

Please try testing without the black list, you can do what I suggested, simply put the central to sleep for dozen of seconds before you start connecting again.

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Tue, 05 Sep 2023 20:05:26 GMT

I implemented blacklist functionality in the stock NUS example code. This works correctly.

I got the sniffer to see the connections during failures. Each time I ran the test I got a different result. Usually no connect went out. Sometimes both sides stopped doing anything. Once I saw the connection succeed but the central still locked up after many "Empty PDU" messages.

In your experience, what would be the next thing to look at so I can find out what I've changed that is causing the problem? Is there any other functionality that conflicts with BLE connections?

I am using USB, UART, the watchdog, some app timers, and the bootloader. Only the UART is active in the NUS example.

Thanks

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

Hung Bui — Wed, 16 Aug 2023 13:09:41 GMT

Hi,
You can take a look at this Academy course : https://academy.nordicsemi.com/courses/bluetooth-low-energy-fundamentals/lessons/lesson-6-bluetooth-le-sniffer/

It about nRF Connect but you can still read about the instruction how to use the sniffer.

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Wed, 16 Aug 2023 13:04:12 GMT

There is no filtering the the peripheral.

The Bluetooth toolbar was off. This is the first I knew about that. I will see if I can get any clarity using it.

Thanks

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

Hung Bui — Wed, 16 Aug 2023 07:56:42 GMT

Hi Rbelli,

You may want to click on the Advertiser list and select the peripheral you want to follow. This way you can see the communication between them when the connection is established, not just ADV_IND and CONNECT_IND.

From what I can see the connect request was sent in the bad connection. So we need to check if it's been rejected by the peripheral or not.
Did you implement any black/white list on the peripheral ?

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Tue, 15 Aug 2023 15:58:08 GMT

f7:0b:15:51:5d:88 is the peripheral and 5d:c2:aa:8f:4f:08 is the central device.

This is a good conversation filtered to only show the peripheral:

26400   2023-08-15 11:27:00.635548   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
26401   2023-08-15 11:27:00.637132   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
26427   2023-08-15 11:27:00.946492   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
26434   2023-08-15 11:27:01.046582   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
26435   2023-08-15 11:27:01.047068   5d:c2:aa:8f:4f:08   f7:0b:15:51:5d:88   LE LL   38   SCAN_REQ
26436   2023-08-15 11:27:01.047394   f7:0b:15:51:5d:88   Broadcast   LE LL   50   SCAN_RSP
26437   2023-08-15 11:27:01.048166   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
26455   2023-08-15 11:27:01.259195   c1:24:88:c2:b1:8f   f7:0b:15:51:5d:88   LE LL   60   CONNECT_IND
27512   2023-08-15 11:27:16.303002   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27519   2023-08-15 11:27:16.405359   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27520   2023-08-15 11:27:16.406944   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27521   2023-08-15 11:27:16.408528   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27528   2023-08-15 11:27:16.511828   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27529   2023-08-15 11:27:16.513412   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27530   2023-08-15 11:27:16.513898   5d:c2:aa:8f:4f:08   f7:0b:15:51:5d:88   LE LL   38   SCAN_REQ
27531   2023-08-15 11:27:16.514224   f7:0b:15:51:5d:88   Broadcast   LE LL   50   SCAN_RSP
27537   2023-08-15 11:27:16.617155   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
27542   2023-08-15 11:27:16.718145   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND

Then 90 seconds of ADV_IND, SCAN_REQ, and SCAN_RSP.

Then this is a bad connection:

36762   2023-08-15 11:28:44.927753   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36763   2023-08-15 11:28:44.929337   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36764   2023-08-15 11:28:44.930922   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36774   2023-08-15 11:28:45.034624   c1:24:88:c2:b1:8f   f7:0b:15:51:5d:88   LE LL   38   SCAN_REQ
36775   2023-08-15 11:28:45.035722   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36776   2023-08-15 11:28:45.037306   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36783   2023-08-15 11:28:45.137878   f7:0b:15:51:5d:88   Broadcast   LE LL   61   ADV_IND[Malformed Packet]
36784   2023-08-15 11:28:45.139462   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36785   2023-08-15 11:28:45.141047   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
36786   2023-08-15 11:28:45.141532   c1:24:88:c2:b1:8f   f7:0b:15:51:5d:88   LE LL   60   CONNECT_IND
41369   2023-08-15 11:29:25.211077   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
41370   2023-08-15 11:29:25.212661   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
41371   2023-08-15 11:29:25.214245   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
41399   2023-08-15 11:29:25.423606   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
41400   2023-08-15 11:29:25.425191   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND
41432   2023-08-15 11:29:25.629609   f7:0b:15:51:5d:88   Broadcast   LE LL   58   ADV_IND

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Tue, 15 Aug 2023 13:08:11 GMT

I have other peripherals that connect and they send the disconnect. These work fine. They will start advertising again later and everything works correctly.

[quote userid="2121" url="~/f/nordic-q-a/102792/sd_ble_gap_connect-returns-nrf_success-but-no-events-are-generated/441332"]Can you do a test that after you disconnect from the peripheral, you wait for 20 seconds and then try to connect again to the same advertiser ? [/quote]

That is how the blacklist works. It is currently set to 90sec. I get no events after that.

[quote userid="2121" url="~/f/nordic-q-a/102792/sd_ble_gap_connect-returns-nrf_success-but-no-events-are-generated/441332"]If you don't see any event, could you run the test again, this time connecting to a dummy address . If you do see the BLE_GAP_EVT_TIMEOUT event, then we need to look into why re-connecting to a previously connected advertiser causing the issue. [/quote]

This is what my edit yesterday was about. On the second connection attempt with the malfunctioning device I changed the address before connecting. It timed out as expected.

[quote userid="2121" url="~/f/nordic-q-a/102792/sd_ble_gap_connect-returns-nrf_success-but-no-events-are-generated/441332"]You may want to capture a sniffer trace to see if any connect request is sent. [/quote]

I will do that. I will also try to get this problem to occur with as close to stock NUS pair as I can get.

Thanks

BoB

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

Hung Bui — Tue, 15 Aug 2023 09:55:44 GMT

Hi Rbelli,

Since the backlist is not affect the softdevice I think we can just put it a side to narrow down the issue.
Can you do a test that after you disconnect from the peripheral, you wait for 20 seconds and then try to connect again to the same advertiser ?

When you do that do you receive any BLE_GAP_EVT_TIMEOUT or BLE_GAP_EVT_CONNECTED event ?
If you don't see any event, could you run the test again, this time connecting to a dummy address . If you do see the BLE_GAP_EVT_TIMEOUT event, then we need to look into why re-connecting to a previously connected advertiser causing the issue.

You may want to capture a sniffer trace to see if any connect request is sent.

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

rbelli — Mon, 14 Aug 2023 18:17:44 GMT

I get the same results if I change the peer address before calling sd_ble_gap_connect(). The blacklist is implemented as a modification to nrf_ble_scan.c. The softdevice knows nothing about it.

We have communication on the USB port and the blacklist is still operating so the CPU is not locked up.

Edit: I forced a connect to an invalid address with no blacklisted devices and I got the timeout event.

RE: sd_ble_gap_connect returns NRF_SUCCESS but no events are generated

Hung Bui — Mon, 14 Aug 2023 08:03:59 GMT

Hi Bob,

The issue is pretty strange.
Could you try some simple test to see if you call sd_ble_gap_connect() to connect to a non-exist advertiser do you get the BLE_GAP_EVT_TIMEOUT timeout event ?
It should timeout after the scan timeout period.

My assumption is that the blacklist mechanism is simply implemented on the application, and doesn't have anything to do with the softdevice , correct ?