Hello
I would like to share the following scenario in case someone else may have already experienced something similar or if a possible idea may be contributed to solve this issue.
We are using stack version 2.02.00.31 with our CC2640 chipset on a customized board with the multi role files as the project template. We have paring and encryption enabled using OOB data and we successfully connect as a central with another of our devices acting in peripheral role. When our device is connected in central role, we have our advertising disabled so we can just keep one single connection established at a time.
Once a connection is established and authenticated, we execute a connection parameters update using 2000ms for connection interval and a supervision time-out of 7000ms with a slave latency of 0. Then we measure how long the connection lasts and we are consistently observing that after 17.5 minutes later (average) the connection always breaks due to supervision time-out error.
It gets our attention how consistent this 17 minutes value is. We have measured dozens of repeated connections at the same distance (3 feet between central and peripheral devices) with undisturbed devices overnight.
Then we conducted a test with our peripheral device connecting with another product using a different hardware and chipset acting as a central and we don't have this problem. We have obtained stable connections for more than 3 hours (after several iterations in a row) and the link remains stable. This has allowed us to isolate the condition to our device with the CC2640 chipset acting in central role,
Do you have some suggestions/comments/ideas about how to improve the connections in our CC2640-based product acting in central role? It appears as if we have a synchronization issue of some sort.
