Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

Not applicable
Could someone please tell me the clock settings which give the best accuracy for the watchdog timer when the system only has a single 12 MHz clock crystal?

I'm trying to configure the window watchdog for an application which has a 100 us main loop. I want the watchdog to trigger if it is serviced more than once in a loop or if it is not serviced for more than two loops. So, that gives a nominal lower bound of 100 us and upper bound of 200 us.

I want to use a clock other than the main clock, so the backup clock looks like a good option. The uncalibrated frequency for this is 35.5Mhz +-25%, which would mean the bounds would have to be much wider to allow for worst-case variations in the clock. The datasheet says that using the high precision reference clock fstdby to calibrate the backup clock results in a +-0.5% accuracy, which is much better.

However, I can't work out if this means when fstdby is generated using the internal slow clock, the external ultra low power clock, or either of these sources. Given that the internal slow clock has a basic accuracy of +-5%, I suspect it means when the external clock is used. As my system will not have the external clock available, I need to know for sure what clock accuracy will be so that I can chose suitable bound values for the watchdog.
1 Reply
Not applicable
I think I may have found part of the answer to this - the data sheet says that the automatic calibration error of +- 0.5% is "Error in addition to the accuracy of the reference clock", so that means the error is basically due to the error in the slow internal clock.

So, the next question is, what is the accuracy of the slow internal clock? I will be running with Vbat at 3.3v, so looking in the data sheet I can see two accuracy figures which appear to apply:

1) +-5% for 2.4 V ≤ VBAT, TA =25°C;
2) +-5% for VBAT = const. TA <0°C or TA >85°C.

I think this is saying the accuracy at 25°C is +-5%, but it can be +-5% worse over the full temperature range (i.e. +-10.25%). Is that the correct way to read the data?