ANT Forum

We are trying to run three similar tx-only master channels from one nRF51422 and noticed that channel contention fails when the channel periods differ by only one clock count, i.e. 8192, 8193, 8194. As the tx windows approach each other, they collide as expected, but they never slide past each other and get stuck. When channel periods differ by 2 clock counts, the tx windows successfully slide by each other.

This is also the case with ANTware and a 24AP2U-based USB stick.

Thoughts? We're okay with having the channel periods differ by 2 clock counts, but are seeking reassurance that there are not any underlying issues that will affect us when devices are in the field.

Thanks!      

Is there a particular reason you configured your channels as TX only? This is not a recommended configuration, as it disables the coexistence handling mechanism of ANT, resulting in situations like what you are seeing.
     

Yup, battery lifetime. The product in question is spec'd at a 5 year battery lifetime.      

Tx only configurations are really not recommended, specially if you will be expecting to have multiple devices coexisting.. If you absolutely need this feature because of the battery life, the clock counts must be set at least 10 apart for transmissions to slide across each other in a timely manner, limiting the duration of the outages.      

Thanks! Will do.      

Can you confirm that the bug is limited to channel period offsets of 1.      

Just to clarify, this is not a bug, this is the expected behaviour when using transmit only channels.

As mentioned before, transmit only channels are NOT recommended for scenarios where multiple master devices will be coexisting in the same RF space. If you decide to use transmit only channels with different channel periods, make sure they are offset by at least 10 counts. But the recommended configuration would be to use bidirectional channels and a single channel period.      

Semantics.

We don't anticipate that our customers will overlay cells, but acknowledge that it's inevitable. So we've added enough variation in channel periods across a network that channel collisions will be unlikely and short-lived.

     

Just want to make sure you understand our issue, though.

On one device, with 3 channels, the channels are having problems with coexistence and sharing of the radio resources. When the channels overlap in time, and the channel counts differ by 1, it seems that the TX windows get stuck and never slide by each other. When we increase the channel period difference to 2 or more, the issue seems to disappear. This seems to indicate that there's a bug in the calculation of the wake-up events, or similar.

In the 2+ device situation, in which there are multiple tx-only master devices, it is expected that they will collide occasionally, or perhaps for a long time, as there's no channel arbitration mechanism or CSMA.

On the single device, though, one would assume that there is 'channel arbitration', regardless of whether or not the RX windows are enabled as the ANT stack knows when it plans on opening TX windows for the 1-8 channels.