BLE disconnect every 30sec depending of the PC since Windows update

[Sebastien]

Our BLE HID mouse devices has work well during several months and since some windows update, on some PC, the BLE is disconencted every 30 sec.

We received a BLE_GAP_EVT_DISCONNECTED from the host.

Then we re advertize and reconnect in loop.

Only at the first pairing, we have no disconnection:

We are using SDK 15.2.

No issue with Nrf Connect on an Android Smartphone 

Regards,

  Sebastien

[Turbo J]

The BLE_GAP_EVT_DISCONNECTED event has an additional status code. Which one do you get?

[Sebastien]

0> <info> app: BLE_GAP_EVT_DISCONNECTED, Reason 34 (Decimal)

[Kenneth]

Hi,

Reason 34 = 0x22 = BLE_HCI_STATUS_CODE_LMP_RESPONSE_TIMEOUT

Seems to be related to an  procedure has timed-out. I have seen in the past that some peer may not handle the case where MTU exchange/DLE update is initiated on BLE_GAP_EVT_CONNECTED event. Using a delay of for instance 100ms can workaround in such case, e.g. in on_connected_evt() in nrf_ble_gatt.c add a delay like this:

        nrf_delay_ms(100); // Add a delay
        err_code = sd_ble_gattc_exchange_mtu_request(conn_handle, p_link->att_mtu_desired);

Though, it can also be something completely different, e.g. that the peer is initiating a BLE procedure that is not handled by the mouse (e.g. an event is not replied to). E.g.  case BLE_GAP_EVT_PHY_UPDATE_REQUEST event is not handled, and adding this may help:

        case BLE_GAP_EVT_PHY_UPDATE_REQUEST:
        {
            NRF_LOG_DEBUG("PHY update request.");
            ble_gap_phys_t const phys =
            {
                .rx_phys = BLE_GAP_PHY_AUTO,
                .tx_phys = BLE_GAP_PHY_AUTO,
            };
            err_code = sd_ble_gap_phy_update(p_ble_evt->evt.gap_evt.conn_handle, &phys);
            APP_ERROR_CHECK(err_code);
        } break;

But to be 100% sure we need an on-air sniffer log.

Btw, what softdevice and version are you using here?

Best regards,
Kenneth

[Sebastien]

Hi,

I will try what you proposed.

We are using s140_nrf52_6.1.0_softdevice.

Regards,

Sebastien

[Sebastien]

the first 100ms delay seems to fix the issue.

Thanks a lot.

100 ms is a lot, do you think this delay could be reduce ?

[Kenneth]

Shorter than 100ms can likely be used since it is a mouse, which typically use a connection interval of 7.5ms, you only need to make sure a few connection interval have elapsed. So try with 50ms.

Best regards,
Keneth