NRF51 time synchronize sensors

[pascal defesche]

Hello,

I have some questions regarding bluetooth low energy and the NRF51 series. [list=1]1) I have to synchronise maximum 17 sensors, so i will maximum need 4 masters: 3 piconets connected towards one master. Is this possible with ble? In my opinion i see 2 possibility's to synchronise the sensors: let the host advertise, if sensornodes receive the advertisement packet start a timer. Or each connection interval let the host write a timestamp(with acknowledge) and let the sensors send their data(1 packet/ 7.5ms interval). [/list]

[list=2]2) The NRF51 series can handle 6 packets transmitting or receiving, so is it possible to do one normal write (so with acknowledge) to get in the same interval the ack packet back or will this always be the next interval? For example i took the picture from the nordic tour presentations in my attachments. At the second interval one packet transmits from master to slave and immediately after this the slave transmits one packet (can this packet be the acknowledge?). Or should i take in account the next interval i can only send 5 packet transmissions and 1 reponse(the acknowledge)? [/list]

Regards, Pascal

[Ole Morten]

It's much cleaner to keep unrelated questions separate, instead of messing around like this. However, please take a look at the S120 SoftDevice specification, which I think covers what you need, specifically in chapter 7: https://www.nordicsemi.com/eng/content/download/38961/658501/file/S120_SoftDevice_Specification_v0.5.pdf

[pascal defesche]

Is there any guarantee that the timing from application towards link layer is always the same in each message transmission(Ts) or message reception(Trt) at the S120 and S110 softdevice? Is there any minimum and maximum timing specification for the s110 and s120 protocol? The Ts and Trt i've mentioned in my attachement is this the timing parameters in chapter 8.2 in the S110 and S120 softdevice specification?

Regards, Pascal

[Ole Morten]

This time will not be constant, and there isn't really a strict upper bound. Depending on when you actually queue your data, compared to when the next connection event is, the time delay will vary. Also, if there is already data queued for the first event, your data could in theory be delayed until the next event.

[pascal defesche]

Again thanks for your fast response Ole!

I do understand there will be none packets lost and the link layer retries each nACK message, in my attachment shown as Ta.I'm not trying to find the time needed to access the radio for one or more packets(Ta) but i'm trying to find the time needed to send a packet from Microcontroller(ARM M0) towards the S110/S120 device, will this always be the same, or what will be the max ΔT: for example sending 1 packet(Tmin) or 6 packets(Tmax)?

I've found some possible methods for time synchronization in BLE, but the only unknow timing parameters are Ts and Trt which are S110, S120 dependent. Ts: sending a packet from mcu to s120 and Trt: s110 to mcu.

Regards, Pascal

[pascal defesche]

I've added another picture to explain my comment above a bit more. First setting the Tndist for setting data in the S110 buffer(Ts), after data has been put in the buffer the oldest element of the buffer will be sent over radio first(Ta). This picture shows when the device is in connection. Will this be the same in advertising mode? So will the advertisement been queued in the buffer?