Analog sampling not guaranteed at fast IR values?

I have a router ZNet 2.5 accepting a 1 Hz square wave off a function generator into its A/D port. The router is configured to take periodic samples every X ms, and transmits it to a coordinator connected to a Mac OS X via the UART. On the Mac, I am able to plot the data transmission from the router’s A/D port.

What I noticed is that 1 Hz square wave is reproduced accurately if the IR rate X is greater than 100 ms. If however, I ramp down the IR to 10 ms, I get a square wave with an apparent higher frequency (i.e., > 1 Hz). I suspect that when the IR rate is too fast, the Rx buffer on the coordinator becomes overflowed and Tx packets from the router are lost.

Because the API mode of the new Znet 2.5 is too simple, I’m not able to dig deeper into the module’s operation to see if this is the case.

Does anyone have similar experiences and/or suggestions?

Thanks in advance,

There are a few things to understand when working with both ZigBee and ADC sampleing. First, ZigBee isn’t terribly fast, even in unicast compared to some other protocols at least. Sending a unicast message may require several hundred ms to accomplish.

Second, that when sampling, the IR parameter allows only one sample per packet at the specified interval. When IR was greater than 100ms, a sample was taken once every 100ms. At this rate, one packet (sample) was sent every 100ms, which was apparently slow enough for the ZigBee network to be able to process before the next sample was sent.

You were trying to sample once every 10ms, generating a new message each time that the network needed to be able to route. This was simply too fast to process through the network. Buffers overfilled, messages were lost, and your suspicion was correct.