Sensors, Metrology & Uncertainty

Thermocouple, RTD, or thermistor

Three common sensors could do the job:

• Thermocouple (type K): cheap, rugged, wide range, thin and fast; needs cold-junction compensation and careful wiring (last lesson).
• RTD (resistance temperature detector): a platinum wire whose resistance rises with temperature, roughly R = R₀[1 + α(T − T₀)]. With α = 0.00392 for typical platinum, a 100 Ω RTD reads 100[1 + 0.00392 × 35] = 113.72 Ω at 35 °C ¹. More accurate and stable than a thermocouple, but bulkier, pricier, and slower.
• Thermistor: a ceramic bead with a large resistance change over a narrow range; very sensitive, very non-linear, and many types don’t tolerate smoker heat for long.

For a smoker, a thermocouple wins on ruggedness and price, as long as you know how much to trust it.

Build the budget

Every reading has several independent sources of error. Estimate each, then combine them. For independent errors, add the squares and take the square root (root-sum-square):

The sensor chain contributes about ±4.6 °F combined; where the probe sits contributes ±10. Spending money on a better sensor barely changes the total. Moving the probe to grate height, next to the meat, does. This is the usual lesson of a measurement budget: the biggest term is often not the instrument.

Check it against something you know

Two easy reference points:

• Ice bath: fill a glass with crushed ice, add water, stir, wait a minute. It should read 32 °F.
• Boiling water: water boils at 212 °F at sea level, but lower as elevation rises and air pressure drops, roughly 1 °F lower per 500 feet. Madison sits around 860 feet, so expect about 210 °F.

If both readings are off by the same amount, add an offset in the controller. If they’re off by different amounts, the problem is probably the probe or its wiring, not something an offset should hide.

Response time

A sensor doesn’t report a change instantly. A probe’s time constant is how long it takes to show about 63 percent of a sudden change. Bare (exposed) tips are fastest, grounded tips next, ungrounded tips slowest, often by a factor of two or more ¹. In a smoker with a 25-minute time constant, an ungrounded probe’s few seconds of lag don’t matter. In the control loop of a fast process, they would.

Which probe decides

The chamber probe controls the heat, but it doesn’t say when the food is done. A second thermocouple in the thickest part of the meat does. The bacon method, for example, smokes at 225–250 °F until the bacon reaches 160 °F inside ². The controller’s job ends when the meat probe says so, not when a timer runs out. That probe needs its own check in ice water and boiling water.

What to take from this

Thermocouples suit smokers (rugged, cheap, wide range); RTDs are more accurate (R = R₀[1 + αΔT], α = 0.00392); thermistors are sensitive over narrow ranges. Build a root-sum-square budget: here ±4, 2, 1, and 10 °F combine to about ±11, and placement dominates. Check probes in ice water (32 °F) and boiling water (about 210 °F in Madison). Put the chamber probe at grate height beside the meat, and let the meat probe (160 °F for bacon) decide when it’s done.