CC2640R2F: Connection interval changes without request

Hi Jerome,

I don't believe this is expected behavior. Can you share annotated/commented Ellisys lags that marks where the issue is seen? Also can you try the latest SDK version for the CC2640R2 which is the 5.30 SDK?

Best Regards,

Jan

Thank you Jan for the help. 

On our product #2, with multi_role as master, we compared BLE stack 1.40.0.45 vs 5.30, I am not including the data here as to not confuse, but essentially 5.30 performed worst than 1.40 in those regards.

The following is for #1 simple_peripheral (as slave).

- The capture here shows a BLE5 Coded PHY connection, but we have seen the same misbehaviour with BLE4 legacy.
- Issue is the highlighted packet where it switches to 40ms interval, can also be seen in the Timing view. (14:29:07.824 965 900) 
- Magenta packets are identical, as far as we understand those are "repeated" by the BLE stack and the changes in the data is not sent during that time. the repeated data always starts with an unexpected switch to 40ms interval. 



here we see an ATT Read was successful and the data is correct, so it's not "entirely stuck". 



in the end it just stops the repeated duplicates and continues sending the correct data



We may provide the btt file through private channel if needed.

Best regards,
Jerome

Hi Jerome,

Can you clarify what you mean by saying that 5.30 performed worst? Do you mean that it took even longer for the connection to recover to its original connection interval in 5.30 than in your SDK?

Best Regards,

Jan

Hi Jan,

The following will only concern product #2, multirole as master:
As we can see on the captures here bellow, there is a param update request for 450ms, but the master unexpectedly goes on to use 1.35sec. This was stack 1.40.
With stack 5.30 (no screenshot), we also get that wrong interval, but worst as it increased to 2.3 sec! Our slave is able to hold the connection at 1.35sec but at 2.3sec that was too difficult, thus we could never upgrade our SDK version and we are forced to keep using stack 1.40 on that product.

Regards,
Jerome 

Hi Jerome,

Thank you for the clarification. This is indeed unexpected. Could you provide the ellisys files themselves? If you would prefer to send them via private message, then you can do so through the E2E inbox. I have sent you a friend request to facilitate this. I would like to fully rule out any custom application code that may be causing this. Are you able to observe this behavior in an unmodified example?

Best Regards,

Jan