TLE4926C-HTNE6747 Datasheet, PDF (5/27 Pages) Infineon Technologies AG – Dynamic Differential Hall Effect Sensor

TLE4926C-HTN E6747
stage for maximum because of falling edge and high output stage for minimum
because of rising edge). This behaviour continues until the first valid minimum and
maximum values are found. With this pair of values there is sufficient information for
getting a quite accurate new calibration result, so that the output can switch with the
result given by the internal comparator. The average of the minimum and maximum
value gives a representation of the offset. More precisely, the minimum and
maximum value (8 bit values respectively) is summed up, the result is subtracted by
256 (=100H), this result is shifted to the correct position taking into account the
current setting of the PGA, and finally this value is added to the current offset value.
The whole procedure can be repeated for many times and converges to an offset
value which compensates for the signal offset. In other words, if the minimum and the
maximum have equal magnitude, their sum will be 100H (80H is the mid-value) and
after subtraction of 100H a correction value of zero will appear.
The shifter, which multiplies the sum off minimum a maximum in corresponding to the
PGA position, calculates the offset- update. In PGA = 3 no shift is applied and the
sum is added (or subtracted) directly from the offset. In PGA = 2 the sum is divided
by 2, in PGA = 4 the sum is multiplied by 2 and so on. But the so calculated update is
not applied every time; in fact there is a nonlinear filter that avoids small offset-
correction in order to improve jitter.
Continuous calibration
Once the device has finished its first calibration it enters a continuous calibration
mode. Basically this means that after each edge transition going out of the circuit a
new offset value can be adjusted. To avoid a offset- jumping due to a unregular
wheel or noise there is implemented a fincal state and a update filter. The algorithm
enter in the fincal state if the difference between the maximum and minimum is less
then 8 Lsb, and the finecal state will be left if the difference is more then 16 Lsb.
Below the 8 Lsb value the offset is not changed, between 16 Lsb and 8 Lsb only 1
Lsb steps are done, and over the 16 Lsb threshold value full adjustment is possible.
The update- filter lets perform the calculated update- step only if the last and the
current update are over the 8 Lsb threshold an if the update- directions are the same.
This avoids unwanted offset- updates due to long notches or teeth (long notches
generate higher amplitudes). The offset calculation unit is protected against overrun
errors so it will clip the values at zero and full scale (3FFH). A set of rules apply to the
calibration process which regulate under which condition and to what amount an
offset calibration is done.
Mathematical relation between max, min, PGA and offset:
Offset(mT)=Offset(Lsb) * Fullscale/1023(Lsb) – Fullscale/2;
120mT = Full-scale of the Offsetdac with 1023 Bit;
Max(mT)=Offset(mT) + (Max(Lsb) – 128)*2^(PGA –3) * Fullscale/1023(Lsb);
Min(mT)= Offset(mT) - (128 - Min(Lsb))*2^(PGA –3) * Fullscale/1023(Lsb);
Data Sheet
Page 5 of 27