Dear MSP team,
My customer is running some final evaluations and they asked me about the noise immunity, as they were applying a magnetic field to the device and saw variations on the readings (using a magnet near the ESI inductive sensor).
I recommended double checking the calibration through the AFE2:
LC Sensor Rotation Detection With MSP430 Extended Scan Interface:
4.6 Optimal DAC Voltage Level Setting:
The DAC in the ESI generates a reference voltage for the comparator. To eliminate or reduce the oscillation when the voltage of input signal is close to voltage reference of DAC, two DAC registers "ESIDA1R0" and "ESIDAC1R1" are used to provide two reference voltage for analog hysteresis. This is especially important when the sensor is on the border line of metal and nonmetal part of the rotor disc. The issue is how large should it be for the separation of these two levels. If the separation is small, oscillation may frequently occur. If it is too large, it increases the voltage difference required between the maximum and minimum of sensor signals and, in turn, reduces the detection distance.
Noise level of signal is then a significant data for the optimal setting of analog hysteresis. With a closer look to the sensor signal (see Figure 11) there is a small noise when the sensor is located onto either the metal or nonmetal part of the disc. This noise can be measured wherever the signal level is. As noise amplitude is within the "Noise Level", seldom exceeding it, the optimal hysteresis setting can then be set as the mid-level between the maximum and minimum of signal plus and minus the noise level.
The noise level is measured when the rotor disc is not turning. The function "Find_Noise_level()" is responsible for this measurement, in which "FindDAC_Fast_Range()" is the algorithm to search for the signal level. This function is to measure the variation of signal for 0.5 second with the sensors staying at the same position over the disc. The noise level is then the difference between the maximum and minimum of the measured data.
Do you have some inputs on this or do you have some other ideas on how to solve it?
Thanks and best regards,
Juan.
