Electronic Thermometer Circuit.
This is what I was after.
Temperature sensor on a lead about 2 meters long.
Accuracy of + – ·2ºC (not confirmed)
Display on a 200 mV digital display. (Multi-meter or panel meter)
There are plenty of digital thermometers with ·1ºC displays but the accuracy is about + – 1ºC and you can not calibrate them. I made this circuit from components that were available at the local electronics hobby shop and it was an educational experience. If you want a simple modern circuit that requires no calibrating I recommend you look at a circuit that uses the LM34 or LM35.
I get a lot of web hits
on this subject.
I only put up this design because I thought it was a neat circuit.
In the circuit above T2 is a fixed current device that sinks approximately 70 Micro amps. P1 is adjusted so that this current does not vary with temperature by balancing the negative temperature coefficient of T1 with the positive coefficient of T2. R3 reduces the effect that adjusting P1 has on the set current.
T4 provides a current proportional to the absolute temperature and is adjusted to equal the fixed current through T2 at 0ºC. Thus at any temperature other than 0Cº a current must flow into or out of the voltage divider formed by R7 and R6 and provide a voltage for the meter as it flows through P3 and R5 .
P3 is adjusted so that at some standard temperature the 200 milli Volt meter reads correctly. Simply put T4 is a Kelvin thermometer and T2 subtracts 273.15 to convert it to a Centigrade one.
The LM334z (T4 & T2) is a readily available device that provides a current source that is directly proportional to the absolute temperature ( ºK ) over the range 0 to 70°C for larger ranges there are other devices in the same family that can be substituted.
Great care was taken to ensure that temperature changes to the body of the instrument do not cause changes to the readings. T2 has changes nulled by P1 but the resistors themselves change value with temperature and this is also canceled out as follows. Any change to the resistance of R1,P2 and P1,R2,R4, R3 with the temperature of the instrument body will cause changes to the current of T4 and T2 but the same change will occur to P3 and R5 and keep the voltage across them constant.
For example if all the resistors went up 10% there would be 10% less current through T4 and 10% less through T2 and the difference in the currents would be 10% less through P3 and R5 but as they would have 10% more resistance the output voltage would stay the same. Any change in the output of T3 with temperature has no effect.
T3 regulates the voltage to 5 Volt and C1 and C2 provide filtering.
The circuit was laid out paying attention to keeping T2 and T1 close together and the resistors that could cause temperature drift in close proximity. An Acrobat reader .pdf file of this is provided and can be printed full size and used to hand make a PCB using a etch resist pen or to photo etch a board if required. atv3.pdf
I have some information on making simple printed circuit boards by manual and photo methods here. Making printed circuit boards.
sensor T4 is
connected by 3 wires. Thin twin shielded wire was used with the
shield connected to the + rail (V+). The connection to the multimeter
made with flying leads fitted with pin sockets insulated by
heat-shrink. Left shows a bottom view of T4.
Power consumption is a couple of milli-amps so it can be powered by a 9 Volt battery. A panel meter if used can be run from the 5 Volt supply but it must be a meter that can be run with a common power supply and has floating inputs.
T1 and T2 should be in close thermal contact. Winding copper wire around them and applying heat-sink thermal grease will aid stability.
The red wire with the knot just passes through the board as the switch is on the opposite side of the board to the battery connector. The 4 unused holes (top right) are are for the power leads to an optional internal panel meter.
Step one is to adjust P1 so that a constant current flows regardless of the temperature of the circuit board.
The probe is made waterproof or contained in a water proof sleeve surrounded by a fine (5mm) crushed ice slurry made from distilled water and contained in an insulated container, the unit (not the probe or the meter) is put in the fridge for about 15 minutes, removed, a reading is taken then the unit is allowed to warm up to room temperature and a reading taken. A hair dryer or other method is then used to raise the units temperature to about 40ºC and a further reading taken. Note it is important that the temperature is even throughout the device. This process is repeated and P1 is adjusted for the minimum change of reading. If the displayed temperature increases with increasing temperatures adjust P1 anticlockwise.
Step 2 is to set P2 to read 0ºC. with the probe at freezing and the unit at room temperature.
Step 3 is to adjust P3 to a standard temperature.
I had about 10 clinical thermometers around the house that gave slightly different readings. I selected the middle one and used it to measure the temperature of an insulated plastic tub of warm water. It is easier to do this step if what you are measuring has a stable temperature so it pays to insulate the tub and keep stirring it. If boiling water is used then some adjustment for altitude may be necessary.
I added a low cost panel meter with the following specifications.
5V Powered, 200mV, True differential input, Impedance 10M ohm, Temco + 100ppm/°C, Accuracy +-0.1%rdg+1 dgt.
coefficient of 100ppm/°C . The sensor circuit has the
drift nulled out for use with an accurate external meter. You can
easily compare the reading from an external and a warmed up internal
meter to see how temperature drift changes the panel meter.
panel meter was
marked CX101B and as I can not presume the one you use will be exactly
the same I can only describe the method I used to connect it which was
to use an 8 and a 6 pin socket with the multiple connections formed by
splicing the necessary leads and insulating the joins with heat-shrink.
Some filing was necessary to the abutting ends of the plugs to keep the
correct pin spacing. Panel meters are not easy to interface to a
circuit. This meter had a common mode range of + - 1
around roughly the middle of the power supply of 5V so it was happy to
work with this circuit.
Pins listed from the right.
BACK TO THE INDEX OF MY OTHER PAGES