Nrf52832 periphral disconnects timeout when BLE/ANT+ scan

[Linxi]

Dear Nordic:
My use scenario may be a little complex. I try to describe it clearly.
A device developed on NRF52832/S332/SDK13, acts as a combination of Bluetooth central, peripheral, ANT+ display.
What i expect is: when it starts up, it tries to connect a target Ble sensor with known peer address, it also starts Ble scan and ant+ search, and sends the realtime results to PC by Nordic Nus notification.
Without connection behavior, it works well. But when I added the connection operation today, the peripheral role (nus) quickly disconnects. The reason shows 0x08 BLE_HCI_CONNECTION_TIMEOUT.
By the way, i use extended packet in nus service which support MTU 247. I checked all the parameters configured for BLE initialization, and found nothing to solve my problem.

Does the environment i am using support my scenario? It's a prototype of a bike computer. Please give me some advise to check my issue or how to accomplish the scenario with better solution.
Thanks very much.

[Kenneth]

Hi,

Check out this application note on ANT channel search:
https://www.thisisant.com/resources/an11-ant-channel-search-and-background-scanning-channel/

Likely you need to disable high priority search, in addition the BLE link should use slave latency (for instance >10) to allow some blocking by the ANT channels.

Best regards,
Kenneth

[Linxi]

Hi Kenneth:

I printed the connection parameters when peripheral receives BLE_GAP_EVT_CONNECTED. And it shows superversion timeout is 500 with Android and 75 with IOS. I think it's the key that causes disconnection when i use IPhone. Is there any way wo update the superversion timeout to a bigger value？It's expected to be MSEC_TO_UNITS(4000, UNIT_10_MS) that i defined.

[Kenneth]

Typically it is the conn_params_init() that can initiate and handle this for you. For instance in ble_conn_params_init() you specify a FIRST_CONN_PARAMS_UPDATE_DELAY delay, which is the delay from connection event until it will try to negotiate new connection parameters (MIN_CONN_INTERVAL, MAX_CONN_INTERVAL, SLAVE_LATENCY and CONN_SUP_TIMEOUT). A supervisor timeout of only 75ms sounds very short yes. 

My suggestion may be that you stop ANT channel search when you receive the BLE_GAP_EVT_CONNECTED event, and wait until BLE_GAP_EVT_CONN_PARAM_UPDATE event to check the updated connection parameters in &p_ble_evt->evt.gap_evt.params.conn_param_update.conn_params, for instance that the parameters use slave latency and that supervisor timeout is more than for instance 500ms before you start ANT channel search again.

Best regards,
Kenneth

[Linxi]

Hi Kenneth:

It seems very hard for me to fix it. I struggled this week. In my use scenario, i need ANT+ display, central and periphral keep active. This means ANT+ search for tracking, Ble scan,ANT+ background search may work together.I am not sure how much you know about bike computer. It should keep tracking sensors that user assigned.It also need to be connected by App to receive and response command.I am going to be crazy.Because I often receive disconnetion timeout event.Could you give me some advise?

[Kenneth]

I think looking at my previous suggestions will help, they may not help individually, but combined they should make a considerable difference. There is one thing I have not asked about, what LFCLK source and tolerance have you used when init the softdevice stack?

Best regards,
Kenneth

[Linxi]

#define NRF_CLOCK_LFCLKSRC                                                     \

  {                                                                            \

    .source = NRF_CLOCK_LF_SRC_XTAL, .rc_ctiv = 0, .rc_temp_ctiv = 0,          \

    .xtal_accuracy = NRF_CLOCK_LF_XTAL_ACCURACY_20_PPM                         \

  }

[Linxi]

I will double check my code and procedure. Actually, I just wanna know something about this situation. Is there any official instructions that Nordic provides to guide developers?

[Kenneth]

For testing I suggest to try NRF_CLOCK_LF_XTAL_ACCURACY_250_PPM.

Best regards,
Kenneth