Posted on February 22, 2013 at 12:36Hi,
If we take an oversight of the STM32 MCUs documentation, u will see that the coarse calibration was first introduced by STM32F1 MCUs family. STM32F2 MCUs’ RTC is also built-in with coarse calibration. But for recent products wave (STM32F3 and STM32F4), it is noticed that a new calibration technic is introduced : the Smooth Calibration.
Just, if we follow the chronological evolution of the RTC peripheral for STM32 MCUs, it is clear that the smooth calibration is the replacement of the Coarse technic. I think, that the Coarse calibration is kept only for compatibility reasons.
The Smooth Calibration can be used to:
- Either adjust the offset drift on LSE oscillation due to Crystal loading mismatch (static correction; it is done normally one time for each PCB)
- Or to dynamically compensate the temperature effect on the oscillation drift (Crystal oscillation frequency depends on temperature)
To measure the RTC drift, dedicated equipment are needed. But, basically, the principle of calibration is simple: just we need to calculate the time difference between two RTC updates.
Now, if you refer to the reference Manual, you find that ST people said: in order to get the right RTC deviation you need to measure the time between much more than one RTC updated : much more than one second:
They specify 8 seconds, 16 seconds and 32 seconds windows. For me of course I will adjust my RTC using the 32 seconds because it gives that better resolution as they said. But for someone how is producing millions of units, each second means money.
Now, how to measure the RTC one seconds updates? Because if you remember we need to count 32 RTC second updates and measure the time interval needed for that. The deviation will be simply the time elapsed divided by 32. And if you multiply by 10Exp6 you will get the drift expressed in PPM.
Hopefully, on STM32F3 RTC we have an 1-Hz output which can be directed externally to the MCU through the AFO_CALIB also called RTC_Calib.
Calibration seems to be transparent to the RTC operation. In other words calibration is done on the fly. If not, I think dynamic calibration has no sense. Isn’t ?
For calibration demonstration soft, as far as I know ST doesn’t provide a one (I checked on ST web site). From my side, personally, I didn’t need to develop such soft. So, sorry man I think you need to try to do it by yourself.
May be you can share with us later on your experience with RTC calibration.
That’s all !!
