The Beeb Body Build course 136
Touchless Temperature Testing
By Mike Cook

Always on the lookout for a project I was scanning the sensor section of a component catalogue last month when I came across a pyroelectric sensor. These are normally used to make passive Infra read movement sensors, the sort that you make into burglar alarms. However, I thought I could put it to a little more serious use. One of the experiments that my first year students perform is that of the verification of Newton's laws of cooling. This states that the rate of loss of heat is proportional to the excess temperature. The students basically plot the temperature of various objects as they cool. Now as any physicist will tell you, when ever you make a measurement on a system you alter that system. As a simple example, suppose you put a thermometer in a liquid to measure it's temperature, then the thermometer will be heated up and thus lower the temperature of the system. In fact this inability to take any measurement without disturbing the system is a fundamental law of physics.
The trick of performing any experiment is to make the effects of the measuring instruments, on the system, as small as possible. Hopefully reducing it to a negligible amount. Well, in the cooling experiments if you could measure the temperature without any contact at all then the disturbance to the system would be negligible. Now the temperature of an object is defined as the amount of radiation it will emit. There is a curve, known as the black body curve, which describes the distribution of energy in a perfectly radiating body. The peak of this curve is used to define the temperature so if you can measure the wavelength of this peak then you can tell the objects temperature. This is almost impossible to do without quite expensive equipment, however if you measure the amount of radiation at any one frequency you will find that this varies quite predictably with temperature. Therefore I thought that the pyroelectric might be useful in determining, if not the actual temperature, the change in temperature.
The E100SV1 sensor uses a ceramic ferroelectric material which produces an electrical change due to change in polarisation intensity, in the Infra red region of the spectrum. It also has incorporated into it a FET (field effect transistor) which buffers the very high impedance of the sensor. The fact that it only responds to changes means that it responds, not to absolute temperature, but to changes of temperature. This is what suites it for use as a movement detector. What happens if you place a hot object in front of it is that the sensor gives out a voltage which then falls back, this is shown in Figure I. You might think this is fine for a cooling experiment but in fact the changes in temperature would be much too slow to record. Well I am sure you are thinking that there must be a solution to this or else the article is suddenly going to stop, which it has not. The solution is to use what is known as a chopper, that is constantly switch between the sensor looking at the object we want to measure and some reference object. In that way we will get a response from the sensor that is proportional to the difference between out object and our reference.
This sounds all quite grand but in practice it turns out to be remarkably simple. As our reference we simply use a piece of cardboard placed in front of the sensor. For preference this should be painted mat black but I have found it not too critical. What we do is take a reading with the cardboard in front of the sensor and then with it removed. Now we can't do this by hand for two reasons, first of all it would be terribly boring but second the movement of our hand in the sensors view would affect the result. However as we have a computer on hand we can use this to automate the entire process.
What we need in the way of a computer interface is a digital output to control our cardboard and an A/D (analogue to digital) converter to measure the reading. This sounds an ideal job for the I2C interface but any other combination with these capabilities will do. The A/D need not be very fast as the response time of the sensor is very slow. In fact in electronic terms it is almost static, taking about two seconds to stabilise. I arranged the cardboard to be controlled by one of the outputs by driving a solenoid connected to a pivot arrangement. Although there is limited movement on the solenoid, if you make the leaver long enough you can get enough movement to cover the active area of the sensor, the general arrangement is shown in Figure II.
Now the output from the sensor is very small so we need some form of amplifier to get it into range. Unfortunately the sensor will not work at all at 5 volts and so it will need to be powered by a separate 9 volt battery. This has the danger that you might feed a 9 volt signal into the A/D which would certainly not do it any good, therefore we have to prevent this from happening. This is done by connecting the output of the amplifier via a diode to the 5 volt reference signal of the A/D. Therefore if the voltage should exceed 5 volts the diode will conduct and limit any further voltage rise. The electronic circuit is shown in Figure III, and all the parts are available from Maplins, the shop not the holiday camp.
Notice the variable resistor, this is used to provide an adjustment in gain of the second amplifier and so allow you to change the sensitivity of the temperature reading. I used a ten turn helical potentiometer for this so that I could get very fine and repeatable control over the sensitivity. The circuit is rather unusual in that it combines a conventional twin stage amplifier with a low pass filter and a bit of integration. The filter smooths out the rapidly changing thermal noise and the integrator collects the small changes in the sensor's output to give a much more stable signal. The third stage is a simple inverting amplifier so that an increase in temperature presented to the sensor results in an initial positive voltage. Finally the solenoid is driven directly by a power FET, and as the coil requires a current of about 250mA needs an external 12 volt supply.
The only job remaining is the software. First of all we need to make sure that the sensor is covered and the system has settled down. Then the sensor is uncovered and readings taken from the A/D, we must take several measurements and choose the one that represents the positive going peak of the sensors output. Once this peak has been detected we can drop the cardboard in front of the sensor again.
In practice the circuit is quite sensitive but I must warn that it is an experimental circuit, that means it can be quite fiddley to set up. The distance of the sensor from the object to be measured is important as the further away they are the lower the readings. I found that I could get a reading from my hand at a distance of about a meter, a soldering iron at this distance sent it off the scale. It would be useful to perform a calibration of the system so that values could be placed on the readings, remember you are measuring the difference between the temperature of the cardboard and that of the object. In practice it illustrated the difference in cooling between a shinny black sphere and a sooted one, next I am going to try it to illustrate latent heat. This is where you measure the temperature of something going through a phase change, that is changing from liquid to solid. Here you see a flattening of the curve as the substance looses heat but does not change temperature whitest solidifying. Other uses spring to mind like checking the lagging efficiency of you hot water tank, but that will have to wait until another time. See you next month.