I guess I really didn’t expand on the idea of sensing in the original goals of the project, but that’s what this project is really about. Measuring, recording, displaying, and trending a large number properties inside and outside the home is a large part of this project. I’m not sure of the exact number, but let’s just say when I’m finished, it should be more than a hundred and less than a thousand. Since a significant amount of these properties will be in the form of temperature, we need to determine a reliable, somewhat simple, and cost effective solution to measure ambient and surface temperatures.
Having worked with temperature measurement in the past, there really are three ways to cheaply and accurately measure ambient and surface temperatures;
1) Integrated Circuit (IC) sensor
2) Resistive Thermal Device (RTD) – a type of thermal resistor
3) Thermistor – another type of thermal resistor
Of course, there are other ways to measure temperature such as thermocouples and fiber optics, but I really see these three as my main choices.
Let’s list out the types and some pros and cons of each:
| Sensor Type |
Integrated Circuit (IC) |
Resistive Thermal Device (RTD) |
Thermistor |
| Advantages |
InexpensiveLinear output relative
to temperature
Large output change
for small temperature change |
Linear output relative to temperatureWide temperature range |
Large output change for small temperature rangeFast Response
Inexpensive |
| Disadvantages |
Small operating temperature range – added complexity when dealing with lower temperaturesMay be subject to self-heating – affecting accuracy |
Small change in output with change in temperatureSlow response
Relatively expensive |
Not linear across temperature range |
Three parameters that are important to sensors are accuracy, repeatability, and sensitivity. Many times it’s a trade-off between these parameters, complexity, and cost to get the best type of sensor for the application. Let me cover these three parameters briefly;
1) Accuracy - This tells us exactly how close the sensor’s output is. Here is a definition I had:
Accuracy of a displayed value is characterized as an uncertainty of a measurement display representing the actual value being measured. It is expressed in terms of how far off any given reading could be from the true value, given in terms of a fixed value, a percent of reading, a fixed value plus a percent of reading or a percent of the instrument’s full scale value.
2) Repeatability – A measure of how closely the sensor will measure the same value every time. Here is a more formal definition:
The repeatability of an instrument or sensor is a measure of its random accuracy. In general, the more accurate a sensor or instrument reading is, the more repeatable it will be.
3) Sensitivity – This indicates how much the sensor’s output changes when the measured quantity changes. Sensors that measure very small changes must have very high sensitivities.
Here is a good article on various temperature sensors.
Although I have used both an
IC type as well as
RTDs, I would like to explore the thermistor as my primary temperature sensor based on the data above. Since this data will eventually be logged into a computer we could apply the
Steinhart-Hart equation to hopefully remove some of the non-linearity of thermistors. Since I have some
IC sensors (
National Semiconductor LM34) that have worked quite well, I’ll probably use them also. The other nice thing about thermistors is their high sensitivity. This helps in measuring very small, relative changes and that is really what I’m after. I’m not as concerned with absolute measurements as long as it’s repeatable.
I suppose I could buy a solution from a Home Automation vendor, but I’m really after a high-volume, repeatable, and simple (maybe relatively) solution that keeps me in my limited budget. After some initial trials (more on them later), I hope to find a good solution for a large number of temperature measurements.
Next, I’ll have to determine what kind of circuitry I’ll use to convert the property changes in these devices to a usable signal (think transducer and sensor) and then some method (analog to digital conversion) to get them into a computer. That sounds like at LEAST two more topics for future articles.
