Calibration 101

There was a rather lively thread the other night about the checking and adjustment of dial-type bimetal dome thermometers like this one:



As we say in the electrical engineering biz, the thread wound up with a pretty low signal-to-noise ratio, so to clear the air about the topic of calibration, and at no small risk of recurring volcanic activity, here's a short primer on calibration and what can - and cannot - be done to improve the performance of one's Egg temperature instrumentation.

First, let's define what calibration is in our case. Calibration is the adjustment of an instrument's available performance parameters in order to make it accurately reflect a process characteristic. In this case, our characteristic is dome temperature and the objective is to make our thermometers read the temperature of the BGE dome to within an acceptable tolerance. If we were to check our thermometers using a known temperature standard and plot the results on a simple x-y graph, the ideal would be a straight line, much like the graph below:



I didn't scale this, but you can see the line slopes upward to the right, with one degree of actual temperature change in the Egg registering as one degree of change on the thermometer. The heavy sloped line lays perfectly over the "ideal" line.

Small Errors

Unhappily, our thermometers do not always perform ideally, sometimes requiring adjustment or replacement. One way a thermometer exhibits error is to develop an offset, as shown in the graph below:



As you can see, this thermometer's response has shifted a small amount over the useful range, let's say 10*F. A reading of 250 on the dial means the dome is at 260*, 350* means 360*, etc. Usually, this is detected using boiling water, around 212*F and the deviation found is assumed to be just this sort of offset, i.e. an error that is consistent over the displayed range. The remedy is to turn the nut on the back of the thermometer until it reads 212* when dunked in the 212* water and, when small deviations like this are found, it probably does the trick.

This is known as "zeroing" an instrument, and it's the only arrow we have in our quiver to fix these thermometers. Keep that in mind: We can only adjust the nuts on the rear of the units, moving the heavy line up and down.

Larger Errors

The same sort of error, but of a different scale, is shown below:



Here, the error detected when we plop the thermometer into boiling water may be very large relative to the temperature being measured. The member who started the aforementioned thread found a ~70* error, for example. The graph actually overstates the information extracted from the boiling water test, since the only thing we are sure of is that the reading is ~70* low at 212*. We just assume that it's 70* across the range of the instrument.

We can, of course, adjust the nut so the meter reads 212* and go on assuming everything is okay, but it's more likely that the 70* deviation is symptomatic of a deeper problem and there is something wrong with the unit and it should be tested to make sure it's not acting like one of the heavy lines on this graph:



As you can see, each of these "calibrations" resulted in a reading of 212* in boiling water, but each will result in a gradually greater error as one increases dome temperature. The responses are still linear (they are straight lines, after all), but the slopes of the lines are such that they only produce erroneous results outside of a very small band. Worse, the Egger will never know what went wrong, believing his or her dome temp to be "calibrated." Worse yet, the response might be like this:



How to dig deeper

To see if your thermometer is accurate at some point other than 212*, you'll need two things: A heat source and a reasonably accurate reference temperture sensor to compare your dome thermometer to. Fortunately, we all have the Egg to use as a temperature chamber and I've used this cheap Accu-Rite digital as my reference:



It's about $12 at Walmart, has a probe that is thin and long enough to slip into the same hole at the dome unit and is, as a thermocouple instrument, accurate to within 3* or so. Not instrumentation-grade calibration equipment, but it's a lot cheaper than a Fluke calibrator and is close enough for our purposes.

The process is straightforward: Simply bring your Egg up to a certain temp, stabilize it, and read the dome temp with the dial thermometer. Then slip the dome unit out and put in the digital's probe, allowing it to settle. Record both readings. Repeat that process over several temperatures, minding the maximum range of the digital (I think it goes up to 392*F, or 200*C). Finally, make a simple x-y plot of the readings with the digital's figures on the x-axis and the dial thermo's on the y-axis. Draw a best-fit line through the points you've plotted.

If the line you've drawn is on top of, or parallel to, the "ideal" line, then congratulations, your unit can be zeroed successfully by turning the nut. You can even do it at your Egg using the digital, in case you don't feel like boiling water.

If the line isn't parallel to the ideal, then you may have a problem. Should the slope be not too different from the ideal, you can use Egg to center the point at which the two lines cross over some temperature that you use a lot, in exactly the same way you'd do it with boiling water. If you do a lot of low-n-slow cooks, this might be around 250*. If you do more high-temp cooking, shoot for something in that range. The important thing is to minimize the reading error in the range you use most.

If the slope of the line is a lot different from the ideal, you're likely out of luck. As mentioned above, we can only move these lines up and down on the graph...we cannot change the slopes of the lines (at least I don't know how). Chuck the thermometer and get a new one. You might even be able to use the digital for low-n-slow cooks in the meantime, as I've done many times.

Large food and beverage firms I deal with calibrate their critical points - ovens, coolers, etc. - on a quarterly basis. This is high-quality, high-cost instrumentation, and it needs tending that frequently. Pharmaceutical firms have even shorter calibration intervals. Given that your dome thermometer is the only thing standing between you and food that either takes forever to cook or is scorched to a cinder in a fraction of the time it should be cooked to perfection, it makes sense to pay attention to what it's doing.