nRF52832 connection parameters for multilink

[Avadacadabara]

Dear all,

I'm trying to build a system with 1 central to 9 peripherals. All devices are based on nRF52832, SDK16, s132 7.0.1.

I've checked this post to get some clues but still quite confusing while scheduling. 

Here are some questions I have for scheduling and connection parameters:

1. Can peripherals using GAP_EVENT_LENGTH different with each other?
2. CONNECTION_INTERVAL should be unified for all the peripherals and central device? Which eaquals to ∑t_event-Cx + scanWindow + t_ScanReserved ?
3. How to determine central GAP_EVENT_LENGTH? ∑t_event-Cx ?
4. Should I set MIN_CONNECTION_INTERVAL and MAX_CONNECTION to the same value get better performance?

For example, I have Peripheral one with GAP_EVENT LENGTH set to 5 (which is 6.25ms) and other eight peripherals with GAP_EVENT LENGTH set to 2 (which is 2.5ms). Total gap_event length needed is 6.25 + 2.5 * 8 = 26.25ms, thus the central GAP_EVENT_LENGTH should be 22(27.5ms). Considering 10ms scan window and some scan reserved time. I set both CONNECTION_INTERVAL to 40ms for all devices, SCAN_INTERVAL is set to 40ms for central. Is it the correct thinking?

But the peripherals disconnect periodically.

*Besides, I can't set MTU_SIZE to maximum value if I set CENTRAL_LINK_COUNT to 9. It shows .

Following is snippet of compiler message of my cenrtral device code, I've set the RAM exactally the size compiler told me so I guess I've spared as much RAM as possible to softdevice?

Any help is appreciated,

Thanks！

[Sigurd]

Hi,

Can peripherals using GAP_EVENT_LENGTH different with each other?

Yes, you can add different configurations with different GAP event length. Add configurations using the sd_ble_cfg_set() function.

CONNECTION_INTERVAL should be unified for all the peripherals and central device? Which eaquals to ∑t_event-Cx + scanWindow + t_ScanReserved ?

Yes, it's recommend that this is a common factor. It's covered in this chapter here: https://infocenter.nordicsemi.com/topic/sds_s132/SDS/s1xx/multilink_scheduling/suggested_intervals_windows_s132.html

How to determine central GAP_EVENT_LENGTH? ∑t_event-Cx ?

It depends on the connetion interval, and number of links:

https://infocenter.nordicsemi.com/topic/sds_s132/SDS/s1xx/multilink_scheduling/suggested_intervals_windows_s132.html

Should I set MIN_CONNECTION_INTERVAL and MAX_CONNECTION to the same value get better performance?

It's the central that decides the connection interval. The slave can request a different a connection interval, but if you as the central want's to accurately control the connection interval, then set max=min.

Total gap_event length needed is 6.25 + 2.5 * 8 = 26.25ms, thus the central GAP_EVENT_LENGTH should be 22(27.5ms). Considering 10ms scan window and some scan reserved time. I set both CONNECTION_INTERVAL to 40ms for all devices, SCAN_INTERVAL is set to 40ms for central. Is it the correct thinking?

No. 22(27.5)ms is the sum yes, but GAP_EVENT_LENGTH is the time set aside for each connection.

.

I've set the RAM exactally the size compiler told me

Are you sure? If you set DEBUG as a preprocessor symbol define, what does the log shows?

[Avadacadabara]

Sigurd said:

Can peripherals using GAP_EVENT_LENGTH different with each other?

Yes, you can add different configurations with different GAP event length. Add configurations using the sd_ble_cfg_set() function.

I actually want to set fixed GAP_EVENT_LENGTH for a single peripheral device, but the value is different between peripherals. So if I set the value in sdk_config.h, should it work?

Sigurd said:

No. 22(27.5)ms is the sum yes, but GAP_EVENT_LENGTH is the time set aside for each connection.

So if I have 9 connections with different GAP_EVENT_LENGTH, how do I supposed to choose central GAP_EVENT_LENGTH. Fot instance, one 6.25ms peripheral and eight 2.5ms peripherals.

Sigurd said:

Are you sure? If you set DEBUG as a preprocessor symbol define, what does the log shows?

Actually I'm not quite sure about this, I just compile the code on KEIL. Adding RX-data to ZI-data size to calculate minimum RAM size. I'll see what's going on with DEBUG macro.

Thanks.

[Sigurd]

Hi,

When you initializes the SoftDevice, you typically have this function.

/**@brief Function for initializing the BLE stack.
 *
 * @details Initializes the SoftDevice and the BLE event interrupt.
 */
static void ble_stack_init(void)
{
    ret_code_t err_code;

    err_code = nrf_sdh_enable_request();
    APP_ERROR_CHECK(err_code);

    // Configure the BLE stack using the default settings.
    // Fetch the start address of the application RAM.
    uint32_t ram_start = 0;
    err_code = nrf_sdh_ble_default_cfg_set(APP_BLE_CONN_CFG_TAG, &ram_start);
    APP_ERROR_CHECK(err_code);

    // Enable BLE stack.
    err_code = nrf_sdh_ble_enable(&ram_start);
    APP_ERROR_CHECK(err_code);
}

Where nrf_sdh_ble_default_cfg_set is used to set the configuration. Values from sdk_config.h is used, and the tag APP_BLE_CONN_CFG_TAG is used.

You need to add a new configuration / new tag. And when you call sd_ble_gap_connect(), you select what configuration you want to use.

See this Message Sequence Chart: https://infocenter.nordicsemi.com/topic/com.nordic.infocenter.s132.api.v7.3.0/group___b_l_e___c_o_n_n___c_f_g.html

[Avadacadabara]

Hi,

I've checked ble_app_multilink_central demo, the default parameter is shown as follow snippet.

I wonder how could it work since 8 * 6ms = 48ms yet the conntion interval is set in range 7~30ms. Also, scanning interval is 160 * 0.625 = 100ms with 50ms window. Total GAP_EVENT_LENGTH is way larger than connect interval. How does this central distribute its connection and avoid overlapping?

Thanks

[Sigurd]

Hi,

The configuration for the ble_app_multilink_central example does not look optimal. With 8(NRF_SDH_BLE_CENTRAL_LINK_COUNT)*NRF_SDH_BLE_GAP_EVENT_LENGTH > the connection interval, and with scanning as well, you will get scheduling conflicts here. i.e. scanning events and/or some expected connection events will be dropped if this central connects to 8 devices with this configuration. 

[Avadacadabara]

I see, thanks!