User Tools

Site Tools


Performance Evaluation

For the graphs shown here, the background subtraction time window is equal to the graph interval, so the lowest point on each sensor response is always zero. The Y axis (labeled capacitance) is the increase in capacitance above this minimum value. This is a change (motion) detection mode, where only increases in the recent past have any effect on the output.

Currently no object detection function is implemented, however reasonably good results could be obtained by using a simple threshold of 300 attofarads.

Sensor noise:

This graph shows the sensor noise under good conditions:

Sensor noise floor

The noise floor is about 10 attofarads peak-to-peak, which corresponds to a current change of about 4 picoamperes at the antenna. When more interference is present, the noise floor can increase by over an order of magnitude.

In practice, the sensitivity is also severely limited by the need to detect a very small relative change in the background capacitance. The noise floor is about 1/20,000 of the typical no-object capacitance, so very small drifts in the background measurement can swamp the tiny changes caused by small or distant objects.

The electronics were designed for high stability, and the residual intrinsic electronic drifts are very well controlled by normalizing the response according to the reference channel amplitude, so the measured drifts are mostly due to actual changes in the capacitance caused by small mechanical motions of conducting objects very close to the sensor. This motion can be caused by thermal expansion or flexing of the mounting structure.

Person detection:

For detecting an adult, the reliable range is from contact up to about 1 meter. The noise limited detection range is about 1.6 meters. This graph shows the response of the three sensor array to a person walking in a along the length of the array at 1 meter, 0.7 meters and 0.35 meters:

Response walking at different ranges

We can clearly see the direction of the motion by the relative timing of the peaks of response from each sensor, and can tell the approximate position of the person at any given time by the relative strengths of the sensor responses.

Respiration monitor:

To demonstrate the potential for other uses of the sensor, I used it to measure chest motion due to breathing. The capacitive measurement principle has been used in products to detect respiration, typically by means of a large mat placed on the bed. Due to the high sensitivity of this sensor, it can be placed some distance way. This graph shows the response when the sensor is 30 cm away from the chest:

Breathing detection

My electronics page | Up | Prev | email

You could leave a comment if you were logged in.
wiki/user/ram/electro/capsense/performance.txt · Last modified: 2010/05/31 12:57 by ram