CC1310: Time Synchronisation question / example availability?

Hi,

yes, there will be a "rfSynchronizedPacketTx/Rx" example in the upcoming SDK release. The SDK release is supposed to happen next week during Embedded World. The example is not based on EasyLink, but raw proprietary RF commands. I have the felling that you know already pretty well what is required, so I don't see the need for publishing it here.

Craig E said:

Secondly, I want to check that I'm calculating something correctly... If I understand the Radio Timer is used for this sort of synchronisation, which is clocked from the 32.768KHz crystal.

The radio timer (RAT) is clocked from the HF oscillator which runs at 48 Mhz and has a tick rate of 4 ticks per µs. The RF driver disables the RF core (and the RAT) whenever possible, but syncs the RAT to the RTC during power on/off. This makes it possible to keep the RAT "virtually" counting and schedule a RF command in the future. During standby, the (virtual) RAT drift depends on the 32.768 KHz crystal of course.

Craig E said:

If the crystal on the Rx and Tx node are the same 20ppm device, does that mean that the maximum drift will be 0.00002 * 2 = 40us per second between last synchronisation? 

Yes, correct. But depending on how long the RF core stays in active mode during that time, the accuracy of the HF crystal applies as well. 

Craig E said:

If I want to send a packet every 15 minutes, the maximum drift would therefore be 15 * 60 * 40us = 36ms. This means the Rx side would have to start listening at 36ms before the calculated Tx time until 36ms after so as to not miss the packet?

Under the assumption that this is the only thing what both sides do, the calculation is correct. It would then be even possible to optimize because you can assume that the drift changes only slowly over time. You may need to experiment a bit with the time window and a safety margin.