Back to Basics: Sensors and Signals (Part 1)
Sensors and signals; where to start? A sensor is a device that responds to an input stimulus and provides an output relative to the input. The stimulus is anything that needs to be measured – from detecting simple key presses that you used to find this article on the Internet to detecting the precision movement of your mouse so you could click on the article’s link. Sensors typically provide an analog output, such as a voltage, current, or resistance, which is then converted by additional electronics or circuitry into a digital value, allowing it to be interpreted and manipulated by a processor or microcontroller.
An entire career could be spent understanding a specific type of sensor. Therefore, part one of this two part series, will cover the basics of the sensors that Fredericks has specialized in for the past 80 years: tilt and vacuum sensors, which come in a variety of forms and utilize active and passive technologies to measure tilt or vacuum.
- What you need to know about passive sensors
Passive sensors respond to an external stimulus generated by another source. Take a thermometer for instance: as the temperature increases in a fluid in-glass thermometer, the fluid in the tube expands (rises) and as the temperature decreases it contracts (falls). Based on the position of the fluid, markings on the tube indicate the temperature. This is a basic example that requires nothing more than human vision to take a rough measurement. Another example is satellite antennas which often utilize a tilt sensor or inclinometer to point in a specific direction and optimize the signal strength from a satellite in space.
- What you need to know about active sensors
Conversely, active sensors provide a stimulus which interacts with the outside world and is then sensed by the sensor. An excellent example of an active sensor is a force balanced servo inclinometer/accelerometer. This device generates a force to match the gravitational force acting on it – which varies as the device moves or tilts, then measures the amount of force used to counteract gravity and converts it to an angle or acceleration. Another more common example of an active sensor is a camera at night. A camera generates a flash; the sensor in the camera captures the reflected light; and finally, converts it to an image.
- Understanding the four key properties of signals
Put simply, a signal is what is generated by a sensor. It contains information about what is being measured, which can then be converted into a usable value for the person or system reading from the sensor. Signals from a sensor can be measured or sampled and such measurements have many characteristics. Among the most important include:
- Resolution: Resolution is the smallest change in input stimulus that can be detected by your sensor. This is most easily described with an analog to digital converter: let’s say you have an ideal linear ±10° (20° range) tilt sensor with a 0 to 5 V DC output, which is being read by a 10 bit ADC (1024 counts) referenced to 5 V DC. This means that your reading has a resolution of 1024/5 = ~205 counts/V and a maximum sensitivity of 5/20 = 0.25 V/° which, when combined, gives your measurement a maximum resolution of ~51 counts/°. This can then be inverted, converting it to degrees, giving the sensor a resolution of 0.02°/count.
- Sensitivity: This is the amount of change in a sensor’s output in response to a change in the input stimulus – put simply, it is the slope or gain of the sensor output. To use the same example as above where you have an ideal linear ±10° (20° range) tilt sensor with an output of 0 to 5 V DC. This means that the sensitivity (slope) of the output is 5/20 = 0.25 V/°. Oversampling is an important technique commonly used to improve resolution and sensitivity. A common example relates to the use of a microcontroller in sampling a sensor. Many low cost microcontrollers contain a 10 bit ADC, which does not have enough resolution (only 1024 bits) for many applications – however, it is possible to take extra samples and average them together to create a higher resolution sample. For any sample, to increase the number of bits from X to Y, a total of 4 Y-Xmust be sampled and averaged. So for a 10 bit ADC, if you wanted to oversample to generate an accurate 16 bit sample, you would need to take 4 16-10 = 46 = 4096 samples and average them together (add them and divide by the total number of samples).
- Precision, Repeatability & Reproducibility: When the same input stimulus to a sensor is sampled many times, an ideal sensor would output the same value for each sample. No sensor is ideal in that each sample will show some variation from the true value. Consider an electrolytic fluid filled tilt sensor: let’s say you take a reading from the sensor and it indicates that the angular position is 0.00°; the sensor then undergoes some dynamic motion and returns to its original position of 0.00°, but this time when we take a reading it outputs an angular position of 0.01° tilt. This is because there is mechanical and physical interaction between the fluid and the measurement electrodes, causing a slight variation in the output signal at the same position. If we did this many times we might find that the output ranges anywhere from -0.01° to 0.01°, making the precision of the tilt sensor ±0.01°
- Accuracy: Accuracy indicates how close a given sample from a sensor will be to the true value of the input stimulus. It combines errors related to resolution, sensitivity, precision, and other characteristics of a sensor’s output. Note, that it is possible for a sensor to be highly precise, while also being very inaccurate due to, for example, significant non-linearity in its output.
Stay tuned for part two of this series, where we will dive further into sensors and signals, including output characteristics, linearity, compensation, and software filtering!
For more information on sensors and signals, contact us today.
If you are interested in guest blogging opportunities, contact Shawn Orr at 215.938.4422.Return to Blog