Let’s face the facts here, in the middle of a nice, warming spring. Did everything you have survive the winter? If not, how much did you have to throw away before buying new stuff to replace it? I know that my paint locker contains more solid and spoiled paint than it should. I just hope the radios I left sitting on the bench in the hangar didn’t suffer the same fate.
Most of our items we use in aviation come with two temperature ranges. There is generally an operating temperature range and a storage temperature range. Exceeding either of these limits negates any sort of guarantee that the manufacturer or designer provides for the product. Heat causes things to break down, and cold causes things to dry out and crack. Neither is particularly good, but in my experience things that got too hot just sitting on the shelf will generally survive, while things that got too cold simply refuse to work at all. I can do something about the cold relatively simply, but I can’t do much about too much warmth. Ovens are a lot easier to build than refrigerators.
The circuit on the floor of the hangar paint locker. Note that the light bulb (which is going to get hot) is sitting on an aluminum radiator (muffin pan) and insulated from the plastic floor with plywood. Note also that it is on the floor, so that when the heat rises, it warms the lower areas first and then eventually gets to the top shelves.
So, here we go with a little goodie that you can build to keep your stuff from freezing in the winter. Why did I wait until spring to show you this? Because maybe, just maybe, you’ll find the time to build this device in time for next winter.
Here is the crux of the design. The first item we need is something to sense temperature electronically. My component of choice is and always has been a simple silicon diode. The diode voltage versus current curve is about the straightest line you can imagine and is almost always near about 1.2 millivolts change with each degree Fahrenheit (F), or 2.2 millivolts change with each degree Celsius (C).
The problem comes with each individual component having its own individual variation with temperature. Using a circuit-analysis program with the published individual variations, we find that this whole circuit, using the matched resistors and diodes inside of the integrated circuit, varies less than 10% of the variation of a single isolated diode. Thus, with the inclusion of one variable “set temperature” control, we can set the device to come on and go off reliably well within 1.8˚ F (1˚ C).
One consideration to take into account is what sort of setting accuracy we need. Realizing that most of what we want to protect is from room temperature (about 68˚ F or 20˚ C) to freezing (32˚ F or 0˚ C), a control designed to be adjustable between those ranges should work well.
Taking all this into account, the circuit shown in the schematic should do this for us, and amazingly enough, it can be set to trigger from the temperature of the design lab (room temperature) to the inside of the fridge, which showed me that indeed the device can be set to either of these extremes.
I’ll assume you’re using a 12-volt power supply. I don’t care if you get it from the aircraft battery, a small external power supply, a wall wart or a hamster wheel with a generator attached. Given that 12-volt power supply (and I assumed that it could go from 11.5 to 14.2 and still be adequate), I will regulate down to 10 volts with D1 and then with U1A connected as a power unity gain (i.e., a gain of 1), will let the output uf U1A remain at 10 volts within 1% over fluctuations of temperature and input voltage. Going one step down with even more precise regulation, U1B with D2 forms a precise 5-volt source.
Remembering that our diode will vary from 550 millivolts at 68˚ F (20˚ C) to 600 millivolts at 32˚ F (0˚ C), we can construct a three-stage voltage divider with the center control being the temperature set potentiometer. If we choose a common value 1K potentiometer for this control, then Mr. Ohm tells us that the top resistor needs to be 82K and the bottom resistor needs to be 10K.
Now we resort to a little trickery. U1C is connected as what we call a “comparator.” It compares the voltages on the (+) non-inverting input and the (-) inverting input. If the voltage at the (+) input is greater than the voltage at the (-) input, the output voltage is “high,” meaning near the +10 supply voltage. Conversely, if the voltage at the (+) input is less than the voltage at the (-) input, the output voltage is “low,” meaning near 0 volts or ground.
How It Works
Let’s presume that the device has been adjusted to trigger at a nice chilly 50˚ F (10˚ C). Since the diode voltage starts off at room temperature at its lowest level, about 550 millivolts, the output of U1C will also be low. As the temperature drops, the diode voltage starts to rise, and when it gets colder than 50˚ F (10˚ C), the output will suddenly swing from low to high.
For some of our heating circuits next month, we need an output voltage that does the opposite of this. We need the output voltage when warm to start off high and then swing low as the temperature decreases. U1D does this. When the output of U1C is low, the output of U1D will be high. When the output of U1C is high, the output of U1D will be low.
If we left it right here, the operation would be unacceptable. Let me pose a question: What happens when the temperature of the sense diode is exactly 50˚ F (10˚ C)? The answer is that the output goes crazy, not knowing from one microsecond to the next whether to go high or low. Remember the old fish aquarium heaters with this problem? They got to the exact temperature and then the contacts on the heater started to chatter and you could hear them on any radio or TV in the house.
Your neighbors with their aviation-band radios would take a dim view of this. In the parlance of amateur radio, you have just made a very efficient spark gap transmitter and, frankly, the last person to use a spark gap transmitter without raising hate and discontent in the neighborhood was a fellow by the name of Marconi in Newfoundland, a good distance away from the nearest town.
Now we enter the wonderful world of hysteresis: R7 senses the output, and when the output changes (high or low), R7 drives the (+) input just a little more high or low, thus snapping the output to either high or low and keeping it there.
Almost. Somehow that U1C has to reverse output state when the temperature changes, and the operative word here is “changes.” The temperature has to change a degree or so and overcome this hysteresis. Thus, although we’ve set the control to come on at 50˚ F (10˚ C), it may have to go down to 48.2˚ F (9˚ C) to turn on and then up to 51.8˚ F (11˚ C) to turn off. That’s the price we pay for no chatter.
Calibration is pretty easy. It takes a good digital voltmeter and an indicator of some sort on the output of U1C. A bargain basement LED works well, and I’ve shown one in the diagram. Put the temperature sensor and a known good thermometer in a water bath that has been chilled to your desired trigger point. Put the diode into the water and wait for a minute or two. Then set R5 so that the indicator light just barely stays on. Another way of doing this is to temporarily disconnect R7 and set R5 for half-brightness of the LED.
The hard part comes next month. Doing it using wall current (110 AC) is relatively simple, but it’s also easy to create a dangerous condition. Using a 12-volt battery is also fairly straightforward, but getting battery juice to last more than a couple of hours of heating isn’t. Hey, if it were super simple, anybody could do it! Stay tuned.