DIY Narrow-Band Analog Thermometer

[Lester]

Below is the circuit I am using. I make no claims on accuracy. This is what I use and I find in accurate enough for distilling. The circuit is simple enough that it can be built on a hobby scale. With precision parts and careful calibration the circuit is good enough to be used in an industrial environment.

(to be continued)

[Lester]

Temperature sensor:
I used 5 pieces of glass diodes (1N4148) wired in series. A diode’s forward voltage has a temperature coefficient of -2.1mv/˚C. What that means is that the diode’s forward voltage (typically 0.6v at room temp) DECREASES as the temperature is increased. Having 5 of these diodes in series gives me a temperature coefficient of -10.5mV/˚C. That is a big change in voltage and it makes the circuit very easy. And these diodes are very cheap. Plus, the glass package makes it ok for use in alcohol vapors.
After soldering, the diodes are wrapped with PTFE tape for insulation then placed inside a length of 5/32” copper tubing with one end sealed.

Cable:
You may have some unused LAN cable lying around somewhere. This has 5 pairs of twisted wires inside and it makes a very good cable for your probes.

Voltage Reference:
I used LM336-5.0 voltage reference and as the name suggests it is 5.0v. This reference is VERY stable which means the thermometer will also be very stable. Once calibrated you can expect it to hold it’s calibration for more than a year. DO NOT use an ordinary zener diode in place of this reference because your readings will drift over time and ambient temperature change. Your thermometer is only as good as your reference.

Circuit Description:
The Temp sensor and voltage reference plus the trimming resistors form what is known as a “bridge” circuit. The op-amps (TL-082) serves to present a very high impedance to the bridge so as not to upset its balance. The diode sensor when forward biased will have a voltage drop of approximately 3v. The trimmer resistor is adjusted so that at 100˚C we get a full scale reading on the meter. The op-amp also provides gain, in this case 20x. The gain will multiply the tempco of the sensor to:
10.5mV/˚C x 20 = 210mV/˚C
And it is this voltage that we feed to the meter to get a reading.

Meter:
I am using a 0~100uA analog meter. It has an internal resistance of 1.9k ohms. (Do a search to see how you can measure the internal resistance of your meter). I wanted it to read from 75˚C to 100˚C because this is the range of temperature that is of interest to distilling. Above and below that temperature range is of no use to me so there’s no need to display it.

I will write on the meter’s dial “75....80....85...90...95....100” so I can read the temperature right off the scale on the dial. Since the sensor’s tempco has been gained by 20x we get 210mV/˚C. The difference between 75˚C and 100˚C is 25˚C. So that means the voltage across the meter will be 5.25v at full scale reading. To get 100uA (my meter) out of 5 volts means a resistance of
5v / 100uA = 50kΩ
Since my meter has an internal resistance of 1.9kΩ I need to put a
50kΩ - 1.9kkΩ = 48.1kΩ
Resistor in series with the meter. It can be seen on the schematic diagram. If you meter has a different internal resistance then you must compute for what series resistor you need to put in there so that the total resistance of the meter and series resistor is equal to 50kΩ.

12v Power Supply:
I use a “wall wart” power supply which you can buy from computer shops. You can also use a small 12v motorcycle battery or even a 12v car battery. Power consumption of the circuit is very low, less than 1 watt.

Calibration:
Immerse the probe in boiling water and turn the trimmer so that you get a full scale reading. Once you take the probe out of the water you can see that the meter reading comes down also. The component’s values were selected so as to get a 75˚C to 100˚C reading on the meter.
If you have another water bath and can hold the temp to a constant 75˚C you can alternately immerse the probe in each bath and adjust for gain and offset. You now have a VERY accurate thermometer (You need to have the gain adjustable in order to do this). I did not bother to do this extra step.

(to be continued)

[Lester]

Quirks:
Because the diodes are encased in a copper tube sheath the response time is slow, about 20 seconds or more. I find it good enough for my boiler & column but if you need a faster response (like when taking readings on the reflux vapour temp) you can expose the diodes directly to the vapors and the response time will be less than 2 seconds. You must wrap the diode’s leads with PTFE tape to prevent the condensate from shorting out your diodes, which gives you a false reading.

Variations:

I have shown above how to make a thermometer that reads from 75˚C to 100˚C. You can change the gain and offset so that the meter will read from 78˚C to 79˚C,for example. I have such a meter on my reflux and I can see very small changes in vapour temperature, down to 0.01˚C. That’s a lot better resolution than you can get from any digital thermometer you can buy. I leave it up to you to figure out how it’s done. It is still the same basic circuit I have drawn above, only some resistor values need to be changed.

Enjoy!

Who am I:
I am a retired circuit designer. I used to do this for a living for the past 20 years.

[Lester]

Rear view shows the rat's nest wiring. I did not bother to make it pretty.

Front view shows my meters, selector switch, gain switch, and coarse & fine adjustments for the 0.4-deg C full scale reflux thermometer. I still need to write on the meter's faces so it reads in deg C.

[Lester]

Links to datasheets:

http://www.ti.com/lit/ds/slos081g/slos081g.pdf" onclick="window.open(this.href);return false;" rel="nofollow
http://www.ti.com/lit/ds/symlink/lm136-5.0.pdf" onclick="window.open(this.href);return false;" rel="nofollow
http://www.nxp.com/documents/data_sheet ... 1N4448.pdf" onclick="window.open(this.href);return false;" rel="nofollow

If you do build one of these thermometers please let me know.

[Lester]

I've been looking at your thermometer, and I'm thinking of using it to measure vapour temperature. You say you wrapped it in PTFE tape. I use that tape on the joins in my column--I don't have tri-clamps, I just insert the sections of the column into couplings. The tape gets rather stringy and disintegrated, so I put a few turns of electrical tape over it. I'm wondering if you get the same effect, and if so, does vapour get to the diodes? I don't think vapour on the diodes would matter very much. They are forward-biased, so have low resistance and pure ethanol vapour has very low conductance (and probably water vapour, too). And even if a very high resistance is in parallel with the diode, its voltage wouldn't change much. Thoughts?

M

For sensing temperature near the take off valve I use bare diodes, no PTFE tape wrap, no copper housing. The tape is an insulator and it slows down the response time so I took it off. I staggered the location of the diodes relative to each other so there's no chance for the leads to touch each other inadvertently. The diodes are bathed in vapor all the time.

For temp sensors embedded in the column packing I needed to house the diodes in copper tube, PTFE tape for insulation, just enough tape to prevent shorts while inserting the diodes into the tube housing. Even so the response time is a lot slower. You might want to try filling the copper housing with cooking oil to aid in temperature conductivity.

Please add a 100nF capacitor across R3 in the schematic. This is needed otherwise the thermometer needle will be jittery. Haven't edited the schematic yet.

Alcohol & water vapor are non-conductive so you are correct in your thinking, these do not affect the forward voltage of the diodes. At one time I drove the still well into the tails and shorted out the diodes. Had to pull the diodes out and dunk it in foreshots to get rid of the short. It seems I can use conductivity as an indicator for tails but I haven't gotten around to making an instrument for it, or an alarm. Later.

[Lester]

For calibrating bare diodes use your still but load it with plain water and allow for some time to drive the remaining alcohol out. Place the sensor where they will be used and calibrate it right there, on the spot, using water vapor.

As with soldering anything in the vapor path: just a friendly reminder to use lead-free solder for the diodes.