Connection Stability, Packet loss, etc...

[Marius Heil]

Hello,

I am having a lot of trouble with connection timeouts. With my current settings (10ms connection Interval and 6 seconds connection supervision timeout) I would not expect any timeout to happen. The RSSIs between the connected beacons are withing good range. I am using many beacons in a Scatternet formation. Is there an API available to get information about connection quality? Dropped packets, CRC errors, etc...

Connection losses seem to get worse when advertising and scanning at the same time. This is not what I would expect because connections should clearly have a much higher priority. I would really like to get some insight on this.

Thank you, Marius

[Kenneth]

There is no API to get information about connection quality no. The multilink scheduling priority for S130 should be described in the softdevice specification, e.g.: infocenter.nordicsemi.com/.../multilink_scheduling.html

Typically I recommend to set the scan window to minimum, and adjust scan interval to fit the current consumption requirements of the application. The reason I recommend a short scan window, is that it will have minimum effect on other softdevice functionality you might want to use in the application. For instance connections will be blocked during a scan window, and a long scan window can delay execution of flash commands and timeslot api.

[Marius Heil]

Mhh, that's too bad that there is no API to get access to these measures. It could certainly help a lot. I wouldn't have guessed that connections could really time out while advertising or scanning. I thought that connections had a very high priority. I guess that takes my to trial and error for configuring the settings.

[Kenneth]

As long as scan interval is shorter than the connection interval, then scanning will have no impact on communication. Exception might be if you have multiple connections at the same time.

[Marius Heil]

Yes, I do have multiple connections. You do have tested Scatternets internally I would guess, any chance to get access to some of these measurements and results?

[Kenneth]

Sorry no, we haven't really done any proper investigation or measurement on this. We have solved this slightly differently using a rebroadcasting mesh: github.com/.../nRF51-ble-bcast-mesh

[Marius Heil]

Yes, I know about that one, but that would mean you have done no proper measurements on nodes that have multiple connections (as Central and Peripheral) with the S130?