Tuning and Signal-to-Noise Ratio (SNR)
In practice, the raw counts (digitized capacitance values) vary due to inherent noise in the system. CapSense noise is the pk-to-pk variation in raw counts in the absence of touch. A fine-tuned CapSense system reliably discriminates between the ON and OFF states of the sensor. To achieve good performance, the CapSenes signal must be significantly larger than the CapSense noise. SNR, which is defined as the ratio of CapSense signal to CapSense noise is the most important performance parameter of a CapSense sensor.
In this example, the average level of raw count in the absence of a touch is 5925 counts. When a finger is placed on the sensor, the average raw count increases to 6060 counts, which means the signal is 6060 – 5925 = 135 counts. The minimum value of the raw count in the OFF state is 5912 and the maximum value is 5938 counts. Therefore, the CapSense noise is 5938 – 5912 = 26 counts. This results in an SNR of 135 / 26 = 5.2.
The minimum SNR recommended for a CapSense sensor is 5. This 5:1 ratio comes from best practice threshold settings, which enable enough margin between signal and noise in order to provide reliable ON/OFF operation.
SmartSense is a FW algorithm that automatically sets all CapSense tuning parameters to optimum values. Advantages of SmartSense vs. Manual Tuning are:
Cypress SmartSense technology allows a device to calibrate itself for optimal performance and complete the entire tuning process automatically. This technology will meet the needs of most designs, but in cases where SmartSense does not work or there are specific SNR or power requirements, the CapSense parameters can be adjusted to meet system requirements. This is called manual tuning.
Some advantages of manual tuning, as opposed to SmartSense auto-tuning are:
To find out more in detail, please check the PSoC 4 and PSoC 6 MCU CapSense Design Guide.