MCU clock drift and radio frequency drift - are they the same?

Question

Let's assume that I have two sensor nodes, one with crystal oscillator running at 24 MHz frequency, the other at 24 MHz + 10 ppm frequency.

To my understanding, in system-on-chip (for example, Texas instruments CC2650) based sensor nodes this single high-frequency crystal (MHz range) powers both the MCU and the radio (where GHz range is needed). I mean, the "local oscillator" component in RF diagrams that generates the 2.4 GHz sine wave is calibrated by using the MHz oscillator as the source (through PLL).

Assume that on both nodes the radio is configured to use the 802.15.4 channel 11, which has 2405 MHz center frequency. Is it the case that one node will communicate using 2405 MHz and the other 2405 MHz + 10 ppm?

Answer 1

Is it the case that one node will communicate using 2405 MHz and the other 2405 MHz + 10 ppm?

Yes, one will transmit at 2405 MHz and the other will use 2405.02405 MHz plus there will be jitter caused by the PLL (in both systems) and this may be in the region of +/- 100 ppm at a variable frequency in the hundreds of Hz to low kHz range (with some randomness too). This dwarfs the static 10 ppm error.

When it comes to receiving, it is likely that the "misaligned" receiver will lock-onto the precise transmission frequency irrespective of its own slightly misaligned local clock. This can also be done using PLL techniques.

On more complex transmissions where the actual centre frequency can be missing (such as in phase modulation or other suppressed carrier types), a special PLL (called a Costas loop) is employed.

Answer 2

Yes, the frequencies will be off by 10 PPM. However, consider that is only ±24 kHz. That's way less than the spectrum width of a WiFi channel, so isn't going to matter.

Such frequency mismatches are built into WiFi receivers. They either don't care about the small mismatch, or they tweak themselves slightly based on the actual received frequency. Think about how expensive such equipment would be if it couldn't be based on even 10 PPM crystals.