High speed internal oscillator calibration

AN3101

12/24

DocID16612 Rev 6

2.3.2

HSI calibration with fixed error

The HSI_CalibrateFixedError() function is provided to calibrate the HSI clock with a

maximum allowed frequency error. It is configured by the user as an absolute value given in

Hertz (the first parameter: MaxAllowedError). This function is the same as

HSI_CalibrateMinError() (refer to Section 2.3.1.), but it searches for the frequency that has

an error (in absolute value) less than or equal to MaxAllowedError.

•

If it finds this frequency, it stops searching, configures the HSITRIMR register

according to this frequency and returns SUCCESS, meaning that the calibration

operation has been successfully performed.

•

frequency). It then sets the HSITRIMR register to the default calibration value and

returns ERROR, meaning that the calibration has failed and did not find any frequency

with an error less than or equal to MaxAllowedError.

The frequency measurements start with HSTRIM = HSICALR + 1 (unlike in the

HSI_CalibrateMinError() function where frequency measurements starts from

HSICALR - 12 to end with HSICALR + 8). The HSITRIM value is computed in loops to

find the next value: the HSITRIM value starts from HSICALR + 1, then goes to the next

value to the right: HSICALR + 2, then to the next to the left: HSICALR, then to the

second to the right HSICALR + 3 and so on until it reaches HSICALR + 8, forming a

“spring loop” (as shown in Figure 7).

This algorithm is based on the fact that the probability of finding the frequency that has

the minimum error increases when the HSITRIM bit value tends to HSICALR. This

algorithm is implemented to minimize the time consumed by the calibration process.

Figure 7. “Spring loop”

The second parameter is used to get the frequency (in Hertz) after calibration in the form of

an unsigned 32-bit integer (unit32_t).

The flowchart in Figure 8 gives the algorithm for this function.