On the KW45 product, there is a way to enable the 32kHz clock without using a crystal externally. Indeed, a FRO32K can be used instead. this article proposes to show you at a glance how to activate it and which performances to expect in comparison to a 32kHz crystal.
This Crystal-Less mode allows to reduce the cost of the system, without compromising the 32 kHz clock accuracy thanks to a software calibration mechanism called SFC standing for Smart Frequency Calibration. One other advantage of the FRO32K is the shorter start up time, including the calibration. The FRO32K clock is calibrated against the 32 MHz RF oscillator through the Signal Frequency Analyzer (SFA) module of KW45.
Software enablement:
The Crystal-less feature is available since the SDK version 2.12.7 (MR4) , all measurements in this document are done with softwares based on this version of SDK.
To enable the Crystal-Less mode, simply define the compilation flag gBoardUseFro32k_d to 1 in board_platform.h or in app_preinclude.h.
In this mode, the SFC module measures and recalibrates the FRO32K output frequency when necessary. This typically happens at a Power On Reset, or when the temperature changes, or periodically when the NBU is running.
By using this mode, higher power consumption is expected. The FRO32K consumes more power than the XTAL32K in low power mode (around 350nA), and the NBU wakes up earlier while FRO32K is used, which also entails a higher power consumption.
FRO32K vs Xtal32K performances:
For these measurements, we used an early FRO32K delivered feature but, even if it is still in experimental phase, the results below will already give you some information.
- Clock accuracy at room temperature:
In steady state, the output frequency of the FRO32K is even more stable than that of the XTAL32K thanks to the SFC module. The clock frequency accuracy of the XTAL32K is a bit better than the FRO32K, however, both are within the permitted accuracy range and are compliant with the Bluetooth Low Energy specification.
- Clock accuracy after recalibration (triggered by a temperature variation):
This test proved that the FRO32K provided a source clock that is within the target accuracy range even during a temperature variation.
- Throughput test at room temperature: Throughput measurements are performed using two different clock sources to verify if there is any connection lost due to the potential clock drift entailed by using the FRO32K as a clock source.
The BLE_Shell demo application is used for the throughput measurement. (refer to KW45-EVK Software Development Kit).
The DUT is programmed with software using either the XTAL32K or the FRO32K as the source clock. After the communication establishment, the bit rate measurement is triggered manually, and the result is displayed on the prompt window.
Results: Two clock configurations show identical performance, which proves that the 32 kHz crystal-less mode presents no disconnection and no performance degradation.
- Throughput test over a temperature variation: it is the same test set up as above but within a 60 °C temperature variation. The results are identical to previous ones. No disconnection or performance degradation is detected.
Conclusion
Various tests and measurements proved that the FRO32K can be used as the 32 kHz clock source instead of the XTAL32K, with the help of the SFC module. It is capable of providing an accurate and stable 32 kHz clock source that satisfies the requirements of connectivity standards.
However, please note that this feature is still in experimental phase, tests are still ongoing to ensure that the feature is robust in any circumstances. Customers who want to enable this feature in production must validate this solution according to their own use cases.
For more detailed information, a draft version of the application note is attached to this article but an updated version will be available on NXP.com website when a new SDK is released.
