Hi Lukas,
Could you please be more specific? what do you want to do? What do you mean by warm-up?
The time from a command being written to the GENFSK command register to the time when bits can be received by the PHY?
Or do they mean the mode change and startup time overall?
Mode change time might have some opportunities for improvement as it is a software-driven time. The time from writing an RX_START_NOW command into SEQCMD field of the GENFSK->XCVR_CTRL register is a hardware determined the time for any type of RX and should be around 105usec.
Internally, you could look at ANTGENRX and GENSTATE DTEST pages on external pins to see the ANT/GENFSK RX and state machine signals. You can look at RXWARMUP page to see the complete receive warmup time from LDO regulator startup to rx_init signal (almost the end of warmup).
The TSM timing for RX warmup is pretty much fixed by the hardware requirements and not open to changes. You should look at the software layers that are being wrapped around the GENFSK to see how much overhead they are adding. Set a GPIO when the app layer asks for RX, then another GPIO when the RX_START_NOW sequence command is written into the register and finally observe one of the DTEST signals showing the end of RX warmup. This will give you a complete picture.
Best Regards,
Mario