37 SEPTEMBER 2024 WorldWide Drilling Resource® Sensor Common Sense by Britt Storkson Owner, P2FlowLLC Most computer controls have sensors to provide information the computer is supposed to “process” and correctly respond to. Just what is a sensor? It is a device that “senses” physical events and converts them into information, a proportional voltage (also known as a signal), the computer can use. So what does it mean? It means the sensors can “make or break” the computer control unit because it is totally dependent on getting the correct information input from these devices, and if the information isn’t correct or if there isn’t enough information over a given time frame, it’s “game over.” One can have the best computer control hardware and software ever made, but if it’s getting garbage inputs, nothing good is going to happen. The production of sensors for computer controls, while not as big as other parts of the tech sector, is still a big industry and there are thousands of different sensors for many different applications. There are sensors which “sense” light energy, pressure, temperature, movement, magnetism, odors (like smoke or haze), and many other specialized applications. All sensors output a voltage signal which accurately represents what’s going on out in the real world from which the computer microprocessor (or “micro”, the computer “brain”) decides what to do with this information. Let’s take one type of sensor (I’ll use a temperature sensor) to illustrate the basic concept of what sensors do for the computer control system and the various issues which must be respected to develop a usable product using this sensor. This will also illustrate the perils of not selecting or using the proper sensor for a given task. This sensor is called a “thermistor.” We will use +5 volts for this example. Connected to a regulated voltage (meaning it stays constant and doesn’t change), with a 10,000-ohm resistor wired to one lead of this device and the other lead going to ground, the thermistor data sheet tells us there will be a signal of 2.5 volts at 77ºF (25ºC). While this thermistor is nonlinear, it is repeatable and the manufacturer data sheet provides a chart of what the thermistor resistance will be a various temperature levels. This thermistor is a negative temperature coefficient (NTC) type, which means as the temperature gets hotter the resistance gets lower. There are also positive temperature coefficient (PTC) thermistors available as well. For example, copper wire has a positive temperature coefficient, so if copper wiring is going to be used in a very hot area it must be derated (be larger than normally needed) to ensure acceptable voltage losses. When I write the codes the microprocessor (micro) will use to manage a temperature sensor, if I want to maintain a 77ºF temperature level I tell the micro (using coded instructions) to convert this signal into a digital value and then tell it to check to see if it is above or below the 77ºF threshold. If I want to cool something to below 77ºF, I tell the micro to turn on the cooling if temperature rises above this level. If I want to keep something above the 77ºF temperature, I tell the micro to turn on some heat. For this to work correctly, this signal must be sampled many times with time delays in between samples before changing the output status (turning on or off the heating or cooling). Otherwise, the output would rapidly turn on and off (chatter) when it approaches the temperature threshold, which is undesirable. Temperature is a fickle metric because it can change depending on where you measure it. If the sensor is exposed to sunlight it will output a high temperature reading, even if the temperature of the surrounding air is much lower. And temperature variations can be considerable. In a room, the temperature at the ceiling is typically much higher than the temperature at the floor. Even opening or closing a door can radically change the temperature the thermistor sees, causing heating or cooling to turn on when you don’t need it to. One solution is to install many sensors in various places and average the results, but this doesn’t always work well because different thermistors placed in different areas can skew the values - resulting in inaccurate measurements. And no system is 100% accurate. There will always be errors introduced by the thermistor itself, the wiring to the thermistor, as well as the sampling, converting and processing, of this signal. On a side note, we bought a new manufactured home recently with an EcoBee thermostat in it. While it works well, it does have a quirky side. The thermostat faces west and, for a short time, is exposed to the rays of the setting sun coming through our west-facing windows. Even though the temperature is correct at the device, when exposed to the sun’s rays, the cooling turns on and drives the temperature down several degrees when it doesn’t need to. I’m not sure why it happens. Is it because the thermostat uses light intensity to decide when to turn the cooling on or off? Or does light striking the materials on its face alter the temperature readings in some way? In any case, the next question is: can we live with this level of accuracy? If it’s a noncritical application (if the temperature maintained is off by a couple of degrees), the answer is yes as the user probably won’t notice a problem. If the sensor is an object or motion sensor on a self-driving car - that’s a critical application. And it doesn’t make any sense to have noncritical performance in a critical-sensing application because people can get killed. And that, in a nutshell, is why we are ultimately not going to have self-driving cars. Britt Britt Storkson may be contacted via e-mail to michele@worldwidedrillingresource.com
RkJQdWJsaXNoZXIy NDk4Mzk=