Hi,
We're using the nRF54L15 driven by a 5ppm TCXO connected to the LFCLK. Our application measures the timing external pin events; the accuracy of this measurement is critical. We'd like these measurements to be timed by our TCXO.
It seems the TIMER peripheral doesn't support LFCLK/LFXO as a source (from section 8.22 "TIMER runs on the high-frequency clock source (HFCLK)"). So we're using the GRTC with CLKCFG.CLKSEL=0 (LFXO), and have set up a capture channel to measuring the external pin events through GPIOTE events. This process works, but we have accuracy concerns.
Section 8.10 of the datasheet states:
""
[...]. Due to the combination of clock sources, it has a 1 μs resolution and an accuracy
equal to the 32.768 kHz clock.
The main features of GRTC are the following:
- System timer – SYSCOUNTER
- 1 μs resolution
- Runs on a fast 16 MHz clock
- Automatic synchronization of SYSCOUNTER with the internal low frequency timer for ultra-low
power operation
• Internal low frequency timer runs on LFCLK (32768 Hz clock)
• Internal low frequency timer can run while the device is in System OFF mode
"""
Since it claims an accuracy equal to the LFCLK I assume there's a periodic re-synchronization. What we'd like to know in order to estimate accuracy is how often does this re-synchronization occur?
From power profiler measurements I can see otherwise unexplained current spikes every ~0.03 - 0.6 seconds (screenshot attached), which I suspect is caused by this re-synchronization. If this is the case, could you please provide more information on the functioning of this re-synchronization process and how I might be able to confirm this is causing the current spikes?
Note: We keep the GRTC active during the measurement, since section 8.10.2 states: "[...], the the TASKS_CAPTURE[n] is functional only when the SYSCOUNTER is in active state."
