Reading heat without sending a beam

Does an ear thermometer send a beam into the ear?

Start with one idea: warm tissue in the ear gives off invisible infrared light, and the thermometer receives it. Follow the warm markers from the ear to the probe. Turning the probe can change which part of the ear it samples.

It receives radiation. The eardrum and canal walls already emit thermal infrared radiation. A probe collects part of it and directs it to a detector; it needs neither a visible laser nor a sound echo. Braun’s description identifies the eardrum and surrounding tissue as the emitting surfaces.

The film follows one probe and one set of tissue temperatures. The drawn section preserves the outer ear, a slightly curved canal and an inclined eardrum, with the probe near the entrance. Warm moving markers always travel from tissue toward the probe. They make invisible radiation visible; their speed does not model the speed of light. Their source, entrance stop and absorber share one geometric model.

The narrow entrance limits the detector’s direct view. In this section, accepted paths pass through an infrared window and a hollow optical channel before reaching the thermopile absorber. Real designs may use reflective waveguides or aperture-limited detectors closer to the tip. This drawing follows direct paths through a narrow opening rather than copying a specific product or silently adding uncalibrated reflections. An ear-thermometer probe patent describes apertures, a thermopile, its cold-junction reference and an infrared-transmitting window.

Absorbed radiation warms the membrane. Thermocouples connected in series produce a small voltage related to their hot–cold junction temperature difference. An amplifier and analog-to-digital converter pass that signal to the processing circuit. A nearby temperature sensor measures the package reference temperature, rather than taking another reading of the eardrum. Analog Devices explains these two inputs.

The net infrared signal alone is therefore insufficient. In the film, the reference changes from 22°C to 30°C while the tissue stays unchanged. The thermopile’s net signal falls, the reference channel rises, and the compensated example reading stays steady. This comparison assumes a uniform package at equilibrium. Real thermal gradients, tissue cooling by a probe and window condition introduce further effects. Melexis distinguishes the absorber and package temperatures.

The turning point comes when the probe turns while the tissue temperatures stay fixed. Its view includes more canal-wall radiation, changing the result. The model mixes radiances before inverting one equivalent temperature. It does not identify the eardrum alone or recover every tissue temperature from a single signal. Original research observed that temperature around the eardrum is not uniform and that sampling location matters. Yeoh and colleagues support that observation. Our 37°C eardrum and spatially varying wall temperatures are illustrative inputs, not reported average body temperatures.

Keep the temperature; change the view

• Change “Probe angle.” Follow each accepted path to the eardrum or canal wall, then compare their signal shares and the reading.
• Enable “Give all visible tissue one temperature.” Orientation still changes the hit locations, but the ideal reading remains unchanged.
• Select “Reference compensation” and change only the package temperature. The net infrared and reference channels jointly determine the result.
• Select “Blackbody calibration.” A cavity at a known temperature provides a reference for the same probe. The curve shows this teaching detector’s finite-band response; the displayed points are not an entire commercial calibration procedure.

We integrate Planck spectral radiance through an ideal rectangular 8–14 μm response. This band is a teaching choice, not a specification for every ear thermometer or a particular Melexis part. Windows, filters and detectors change the actual response. NIST’s radiation-temperature reference defines spectral radiance and Planck’s law.

The unit is W·m⁻²·sr⁻¹: band-integrated radiance, not total power in watts. The full-spectrum Stefan–Boltzmann expression σT⁴ is not a finite-band response. Normalized angular weights mix the radiances along the two-dimensional accepted paths, and the same monotone function inverts the result. Celsius temperatures are not arithmetically averaged.

For black surfaces, ideal transmission and a package at equilibrium, the model uses an illustrative linear gain:

G=0.075 mV/(W·m⁻²·sr⁻¹) makes the small signal easy to observe. It includes geometric and electrical factors not modeled separately and does not specify a real instrument’s sensitivity or accuracy. Calibration shows the relationship between known temperature and signal. The model already knows its gain and band; it does not claim that three points recover a commercial algorithm.

Coordinates are in millimetres, but the canal shape, probe aperture, wall temperatures and materials are illustrative. The accepted angle follows from a 0.65 mm entrance radius and 5 mm channel length. Sixty-one directions use normalized two-dimensional angular weights. Real sensor fields of view often describe declining sensitivity rather than this hard boundary. Three-dimensional angular response, multiple specular reflections, emissivity, cover transmission, thermal shock, wax obstruction and individual anatomy are omitted. The membrane and thermocouples are enlarged; their temperature-difference mechanism is shown without solving the membrane’s transient temperature.

This film explains reception, compensation and spatial sampling. Its example reading does not establish human core temperature or provide diagnosis, treatment decisions or operating instructions.