The Ultimate Guide to Thermal Scopes in Australia →
Thermal Academy · Core Curriculum
Imagine holding a sheet of paper. Now imagine covering that sheet with a third of a million tiny thermometers — on a real 640×512 thermal core, exactly 327,680 of them, packed in a perfect grid. Each thermometer is smaller than a grain of sand. Each one measures the infrared energy landing on it. Each one reports its reading fifty times every second, to a processor that assembles all 327,680 numbers into a single picture — then does it again, and again, all night. That sheet is, near enough, a thermal sensor.
Every other guide in Thermal Academy has been standing outside the scope, looking at the world: why heat exists, why contrast exists, why emissivity bends it, why NETD matters. This one finally goes inside and answers the question all of that was building toward: what is actually in there, doing the seeing? The answer is that grid of tiny thermometers — except they aren’t really thermometers at all. They’re microbolometers, and understanding how one works quietly explains almost everything else about how your thermal behaves.
A microbolometer doesn’t measure heat.
It measures how much a speck of material warms when infrared lands on it — hundreds of thousands of times at once.
Thermal Academy
That’s the whole idea in two lines, and the rest of this article is just unpacking it. Once you see that a thermal sensor is a grid of independent detectors, each doing one tiny job, the mysteries dissolve: why thermal needs no cooling and no light, why resolution isn’t zoom, why the image needs constant calibration, why some pixels die, and — the big one — where that famous NETD number actually comes from. None of it is magic. All of it falls out of how one microbolometer pixel works.
Key takeaways
- A microbolometer is an uncooled infrared detector. A thermal sensor is a grid of them — one per pixel, hundreds of thousands in total.
- Each pixel is a tiny, thermally-isolated bridge that absorbs infrared, warms a fraction of a degree, and changes its electrical resistance. The camera reads that change and turns it into a pixel value.
- Because it senses emitted infrared, not reflected light, it needs no illumination — and because it works at room temperature, it needs no cooling. That’s what makes affordable, pocket-sized thermal possible.
- Resolution is detector count, not magnification. 640×512 means 327,680 detectors — more of the scene sampled, not a closer view.
- Every pixel is slightly different and drifts with temperature, which is exactly why calibration (NUC) exists — and why a few pixels can fail as dead pixels.
- NETD is simply how small a temperature change one of these detectors can pull out of its own electronic noise. It isn’t a range spec — it’s a property of the microbolometer.
What Is a Microbolometer?
A microbolometer is a device that detects infrared radiation by letting it warm a tiny piece of material and then measuring how much that material’s electrical resistance changes. That’s it. It doesn’t “take a temperature” the way a thermometer does; it catches infrared, warms up a hair, and reports the warming as an electrical signal. Do that with one detector and you have a single light-meter for heat. Do it with a grid of hundreds of thousands, read them all at once, and map each reading to a pixel — and you have a thermal image.
The word that matters most in that definition is uncooled. For most of infrared history, “seeing heat” meant exotic detectors chilled to nearly −200 °C. The microbolometer’s entire reason for existing is that it works at ordinary room temperature — which is precisely why you can hold a thermal scope in your hand, run it off a small battery, and buy one without a research grant. We’ll come back to that, because it’s a bigger deal than it sounds.
Why “Micro,” and Why “Bolometer”?
The name is two ideas stuck together, and each one tells you something real about the device.
“Bolometer” is the old part. A bolometer is any instrument that measures radiant energy by the heating effect it produces — the word comes from the Greek bolē, a ray or something thrown. It was invented in 1878 by the American astronomer Samuel Pierpont Langley, who wanted to measure the faint infrared warmth of the sun with far more sensitivity than a thermometer could manage. His original bolometer was a blackened strip of platinum whose resistance changed as sunlight warmed it — the exact same principle, nearly 150 years old, that sits behind the sensor in a modern thermal scope.
“Micro” is the new part, and it’s what turned Langley’s single strip into an imaging device. Modern microfabrication (the same broad toolkit that makes computer chips) lets you build one of those little heat-sensing elements at a microscopic scale — around 12 or 17 microns across, smaller than a grain of sand — and then print them by the hundreds of thousands in a grid. “Micro” plus “bolometer”: a tiny radiant-heat detector, mass-produced small enough that a whole picture’s worth of them fits on a chip the size of your fingernail. The leap from Langley’s strip to an imaging array happened in the late 1980s and 90s, and it’s why thermal imaging went from laboratory to hand-held within a generation.
Drawn to scale, a single microbolometer detector is dwarfed by a human hair and a grain of sand. That’s the “micro” in microbolometer — and why a whole picture’s worth of them fits on a fingernail-sized chip.
The Anatomy of a Single Pixel
To understand the whole sensor, you only have to understand one pixel, because the array is just that pixel copied a few hundred thousand times. And the design of that pixel is a small marvel of getting one thing right: keeping the detector as thermally alone as possible, so that the only thing warming it is the infrared you want to measure.
A single microbolometer pixel is a tiny bridge floating in a gap. On top is an absorber — a thin membrane tuned to soak up long-wave infrared. Built into it is the sensing element: a material whose resistance changes sharply with temperature (a thermistor, usually vanadium oxide or amorphous silicon). The whole bridge is held up over the chip below by two thin support legs, which do double duty: they carry the electrical signal out, and — being thin — they deliberately make it hard for heat to drain away into the chip. Underneath sits the readout circuit (the ROIC), the silicon that measures each pixel’s resistance and passes the number on.
The trick of the design is isolation. The absorber is suspended in a gap on thin legs so almost nothing can warm it except the infrared from the scene. A grain-of-sand-sized bridge, floating, waiting to be warmed by a fraction of a degree — repeated hundreds of thousands of times across the chip.
| Part | What it is | Its job |
|---|---|---|
| Absorber | Thin membrane tuned to long-wave IR | Soaks up incoming infrared and warms slightly |
| Sensing element | Temperature-sensitive resistor (VOₓ or a-Si thermistor) | Turns that warming into a change in electrical resistance |
| Support legs | Two thin arms holding the bridge up | Carry the signal out and block heat from draining into the chip |
| Vacuum gap | The sealed empty space beneath the bridge | Removes the air that would otherwise steal the heat |
| Readout circuit (ROIC) | Silicon chip below the array | Measures each pixel’s resistance and passes on a number |
Five parts, one purpose: let only the scene’s infrared warm the sensing element, then read the result. Every microbolometer pixel — and there are hundreds of thousands — is this same little machine.
How One Pixel Turns Infrared Into a Number
Now watch a single pixel do its job, start to finish. The infrared arriving is long-wave — roughly 8–14 microns, the band that objects at everyday temperatures radiate most strongly, and exactly the band the absorber is tuned to soak up. Every step after that is ordinary; the cleverness is only in the isolation and the speed.
Infrared lands on the absorber; the absorber warms by a tiny fraction of a degree; that warming changes the resistor’s resistance; the readout circuit turns the change into a voltage and then a number; that number becomes one dot in the picture. Every pixel runs this loop in parallel, dozens of times a second.
Two things about that loop are worth sitting with. First, the temperature changes involved are minuscule — we’re talking thousandths of a degree of warming on the membrane — which is why the thermal isolation matters so much and why the whole thing has to be shielded from stray heat. Second, and this is the line to remember: the detector has no idea what it’s looking at. It doesn’t know it’s aimed at a fox or a fence post, a warm engine or a cold sky. It only registers how much infrared reached it and warmed it. As the emissivity guide put it from the other side: the detector doesn’t know whether a surface is being truthful or reflective — it simply measures the infrared energy that arrives. All the meaning gets added later.
Why Thermal Sensors Don’t Need Cooling
Here’s a fact that trips up almost everyone who reads a little about infrared: for decades, “thermal detector” meant a device chilled to cryogenic temperatures — around 77 kelvin, roughly −196 °C, cold enough to liquefy nitrogen. Those detectors (materials like indium antimonide or mercury-cadmium-telluride) count individual infrared photons, and to do that without being swamped by their own heat, they must be brutally cold. They’re superb — and they’re bulky, power-hungry, expensive, and slow to start because they need a cryocooler running first.
The microbolometer sidesteps all of that. It doesn’t count photons; it measures warming. And measuring warming works perfectly well at room temperature — you just have to isolate the detector so the only warmth it feels is the scene’s. That single design choice is why the thermal scope on your rifle is the size it is, costs what it costs, and switches on in a couple of seconds instead of spooling up a cooler. When people ask why consumer thermal only arrived recently, this is the answer: the uncooled microbolometer array is what made “thermal you can actually carry” possible.
| Cooled photon detector | Uncooled microbolometer | |
|---|---|---|
| How it senses | Counts infrared photons | Measures tiny warming (resistance change) |
| Operating temp | ~77 K (−196 °C) | Room temperature |
| Size & power | Large, high power (needs a cryocooler) | Compact, low power |
| Start-up | Must cool down first | Near-instant |
| Cost | Very high | Affordable |
| Typical use | Military, science, long-range research | Hunting, security, industry — nearly all civilian thermal |
Cooled detectors are more sensitive at the extreme, but the uncooled microbolometer is the reason thermal imaging is something you can hold in one hand. Every Pixfra optic — every mainstream hunting thermal — uses an uncooled microbolometer.
Every Pixel Is Its Own Detector
This is one of the most useful ideas in the whole guide, and it’s easy to miss. A thermal sensor is not a single “eye” that takes one picture. It is a grid of completely independent detectors, each one a self-contained microbolometer measuring only the little cone of the scene focused onto it. Pixel number 200,000 doesn’t know or care what pixel number 1 is doing. Each simply catches its own sliver of infrared, warms, and reports a number. The image is what you get when you lay all those independent numbers out in a grid.
There’s no single “camera eye.” Every dot in your thermal image is a physically separate detector reporting its own reading. Hold that picture in your head and the next three sections — resolution, calibration, dead pixels — all explain themselves.
Why Resolution Is More Detectors — Not More Zoom
Because each pixel is its own detector, sensor resolution means something very physical: it is literally the number of detectors on the chip. A 384×288 sensor has 110,592 of them. A 640×512 sensor has 327,680. A 1280×1024 has over 1.3 million. Going from 384 to 640 isn’t a closer view of the scene — it’s the same view chopped into three times as many samples, so finer detail survives and a warm shape holds together at greater distance before it collapses into a blur.
Same silhouette, three sampling densities. More pixels means the scene is measured more finely, so detail holds together further out — but the picture is not magnified. Zoom is the lens’s job; resolution is the detector count’s. The trade-offs (and why 640 isn’t always “better”) get the full treatment in 384 vs 640.
Why Pixels Drift — and Why Calibration Exists
Now for the consequence of building hundreds of thousands of tiny detectors: no two of them are exactly alike. Manufacturing can get astonishingly close, but each microbolometer has a slightly different sensitivity and a slightly different baseline reading. Worse, that baseline moves. Because a microbolometer works by sensing warming, it’s sensitive to its own temperature too — as the sensor and the camera body heat up in your hands or drift with the night air, every pixel’s zero point creeps. Left uncorrected, you’d see the scene overlaid with a faint, shifting fixed pattern and a slow fog of drift.
Left uncorrected, the small differences between pixels — and their drift with temperature — would print a faint fixed pattern over everything. Calibration measures that pattern against a uniform reference and subtracts it, so the scene comes back clean.
That correction is non-uniformity correction (NUC), and it’s the reason for a behaviour every thermal owner notices: the brief freeze and soft “click” the camera makes every so often. That’s an internal shutter swinging across the sensor to give every pixel a uniform, known target, so the camera can re-measure each one’s offset and flatten the image again (a flat-field correction, or FFC). Far from a fault, it’s the sensor doing exactly what a grid of slightly-mismatched detectors must do to keep showing you a clean picture. This deserves — and will get — its own cornerstone; it’s the natural next step from here.
Where NETD Actually Comes From
Now the payoff — the moment the abstract spec on the box turns concrete. You already know from the NETD guide that NETD is a sensitivity figure quoted in millikelvin, and that smaller is better. What the microbolometer adds is where that number lives. Detection, remember, is a tiny warming producing a tiny change in resistance. But every electronic circuit has a little random noise of its own — a faint electrical hiss. So there’s a floor: a temperature difference in the scene so small that the resistance change it produces is lost in the detector’s own noise. That floor is NETD.
That’s the whole “aha.” NETD isn’t a mysterious quality score and it certainly isn’t a range — it’s simply how small a temperature change one of these detectors can reliably notice. A more sensitive microbolometer has a better signal-to-noise floor, so it resolves fainter differences: the subtle warmth of an animal against a background that’s nearly the same temperature, the last thin edge of contrast at dusk. Everything the NETD article told you about why low-contrast scenes are the real test now has a physical home — it’s happening in the resistance-versus-noise of each tiny bridge. The spec on the box is a statement about the hardware you’ve just met.
Why Dead Pixels Exist
When you’re printing hundreds of thousands of microscopic suspended bridges, a few won’t come out perfect — or a few will fail over the sensor’s life. A pixel whose leg, absorber or connection is faulty stops responding: it gets stuck bright or stuck dark and no longer tracks the scene. That’s a dead pixel (or a “stuck” or “hot/cold” pixel), and it’s a completely normal, expected part of how these sensors work. Manufacturers specify an allowable number, and it doesn’t mean your sensor is broken.
A faulty detector can’t be un-faulted, so the camera maps the bad ones and quietly fills each with the average of its healthy neighbours. You never see it, and a handful within spec has no practical effect on the image.
Why Thermal Sensors Cost So Much
If a microbolometer works at room temperature and prints like a chip, why isn’t thermal as cheap as a phone camera? Because keeping those tiny bridges thermally alone is genuinely hard, and the parts that make it possible are exotic. Three things drive the cost, and you’ve already met two of them.
First, the array has to live in a vacuum package. Remember that the whole design depends on the absorber being warmed only by the scene — but ordinary air would carry heat to and from those membranes and ruin the isolation. So the sensor is sealed in a tiny, precise vacuum, and holding that vacuum reliably for years is a demanding piece of engineering. Second, the detectors themselves are MEMS devices — micro-machined suspended structures built to sub-micron tolerances. When you’re fabricating hundreds of thousands of microscopic suspended bridges on a single wafer, even a small percentage failing during manufacture means whole chips discarded or downgraded — and that yield loss is quietly built into the price of every good sensor that ships. Third, thermal light won’t pass through ordinary glass, so the optics in front can’t be glass at all; they’re germanium, an expensive infrared-transparent material we met in the lens guide and again in emissivity. A precision vacuum, micro-machined detectors, and a germanium lens: none of it is cheap, and together they set the floor on what a real thermal costs.
A thermal core is a sealed vacuum package with micro-machined detectors behind a germanium window — closer to a piece of lab instrumentation than a webcam. That’s the honest reason good thermal isn’t pocket-change, and why sensor grade is where most of a scope’s cost lives.
Why Thermal Needs No Light at All
One last consequence, and it circles back to the very first idea. A microbolometer responds to emitted infrared — the heat radiating off objects themselves — not to reflected visible light. A night-vision tube amplifies the faint light that’s already there and is blind in true darkness; a thermal sensor doesn’t care whether there’s any light in the scene at all, because it isn’t using light. Pitch black, thick shade, a moonless paddock — the fox is still radiating its warmth, and 327,680 little detectors are still catching it. That’s why thermal “sees in the dark” in a way night vision fundamentally can’t: it was never looking at light in the first place.
What those detectors hand off is just a grid of raw numbers — warmer here, cooler there. Turning that into the crisp, palette-mapped picture on your display, with its edges sharpened and its scene stretched for contrast, is a separate job done downstream by the camera’s software. That’s image processing, and it’s where the microbolometer’s honest, noisy little numbers become something your eye can read at a glance.
Where the Microbolometer Sits — Now We Go Inside
This is the moment the whole curriculum pivots. Every cornerstone before this one lived out in the world: the atmosphere the signal crosses, the surface that radiates it, the scene’s contrast, the clock’s crossover, the sensitivity needed to catch it. The microbolometer is the threshold where the journey crosses from the world into the device. Everything above it is what happens to infrared before it arrives; everything below it is what the scope does with it once it’s caught.
The microbolometer is the doorway. Above it, the reasons a signal is strong or weak by the time it arrives (weather, emissivity, contrast, crossover, NETD). Below it, what the hardware and software do to render it (calibration, processing, image). Understand the detector and both halves finally connect.
The sensor is where the money — and the seeing — lives
Everything in this article is why sensor grade matters more than any other spec. Pixfra’s Pegasus 2 and Cetus run 640×512, 12µm uncooled microbolometers at ≤15 mK NETD, with genuine Australian stock and C.R. Kennedy service behind them.
Common Microbolometer Myths
The Whole Scope, in One Tiny Bridge
Go back to that sheet of paper covered in tiny thermometers. You now know they aren’t thermometers but microbolometers — suspended micro-bridges that catch infrared, warm a few thousandths of a degree, and turn that warming into a number, hundreds of thousands of times over, fifty times a second, at room temperature and in total darkness. And you know that almost every other behaviour of a thermal scope is just this one fact playing out: no cooling because it senses warming, not photons; no light needed because it senses emitted heat, not reflection; resolution as detector count; calibration because the detectors drift; dead pixels because a few of so many will fail; and NETD as the smallest difference one of them can hear over its own hiss.
A microbolometer doesn’t measure heat.
It measures how much a speck of material warms when infrared lands on it — hundreds of thousands of times at once.
Thermal Academy
That’s the thing that separates understanding thermal from merely owning it. Everything the world does to the signal, and everything the scope does with it, meets at this one small device. Learn the microbolometer and the spec sheet stops being a list of mysterious numbers and becomes a description of a machine you can actually picture.
Every one of those hundreds of thousands of detectors reads a little differently.
Keeping them all in agreement is the job of calibration.
Continue learning — the Core Curriculum cornerstones
You’re inside the scope now. From the detector, the trail runs on:
Test your knowledge · 6 questions
Check what you've learned
Open book — scroll back up any time. Each question appears as you go; the test marks itself once all six are answered. Score 100% to claim your Thermal Academy certificate.
Enter your details and we'll email your personalised Thermal Academy certificate for Inside the Microbolometer as a PDF.
Frequently Asked Questions
What is a microbolometer?
A microbolometer is an uncooled infrared detector. It absorbs infrared radiation, warms a tiny fraction of a degree, and changes its electrical resistance in proportion — and the camera reads that change as a pixel value. A thermal sensor is a grid of these detectors, one per pixel, read many times a second to build an image.
How does a thermal sensor actually see heat?
Each pixel is a tiny suspended bridge that absorbs infrared from its patch of the scene and warms slightly. That warming changes the bridge’s resistance; the readout circuit converts the change to a voltage and then a number. Do this for every pixel, 50–60 times a second, and the grid of numbers becomes a moving thermal image.
Why don’t thermal sensors need cooling?
Because a microbolometer measures warming rather than counting individual photons, it works at room temperature — it just has to be thermally isolated from stray heat. Only high-end photon detectors (used in military and scientific systems) need cooling to around −196 °C. The uncooled microbolometer is what makes affordable, hand-held thermal possible.
Does higher resolution mean more zoom?
No. Resolution is the number of detectors on the chip — 384×288 is 110,592, 640×512 is 327,680. More pixels sample the scene more finely so detail holds together at greater distance, but they don’t magnify the view. Magnification comes from the lens, not the sensor.
Why does my thermal freeze and click every so often?
That’s calibration — non-uniformity correction (NUC), often via an internal shutter (flat-field correction). Because each microbolometer pixel reads slightly differently and drifts with temperature, the camera periodically re-references them all against a uniform target to keep the image clean. It’s normal and correct behaviour, not a fault.
What causes dead pixels on a thermal sensor?
When you build hundreds of thousands of microscopic detectors, a few won’t be perfect or will fail over time, getting stuck bright or dark. That’s a dead pixel. It’s normal and specified within limits — the camera maps the bad ones and fills each with the average of its neighbours, so you don’t see them.
Where does NETD come from?
NETD is set by the microbolometer. Detection is a tiny resistance change, and every circuit has some electronic noise, so there’s a floor: the smallest temperature difference whose signal can still be pulled out of that noise. That floor is NETD — a sensitivity spec (in millikelvin), not a range figure.
Why are thermal sensors so expensive?
Three reasons: the detector array must be sealed in a precise vacuum (air would ruin the thermal isolation), the detectors are micro-machined MEMS structures with real yield loss, and the optics must be germanium because thermal infrared won’t pass through ordinary glass. A precision vacuum package plus a germanium lens sets the cost floor.
Why doesn’t thermal need any light?
A microbolometer senses infrared that objects emit as heat, not visible light they reflect. So it works in total darkness — there’s nothing for it to be “too dark” for. Night vision, by contrast, amplifies existing light and fails in true darkness. That’s the fundamental difference between the two.
Is a microbolometer the same as a thermal camera?
The microbolometer is the detector — the sensor chip at the heart of the camera. A complete thermal camera adds the germanium optics in front, the readout and calibration electronics, the image processing that turns raw pixel values into a viewable picture, and the display. The microbolometer is where the seeing happens; the rest makes it usable.
Educational guide only. Simplified for the thermal owner: a microbolometer is an uncooled infrared detector whose absorber warms and changes resistance; a sensor is a 2-D array read out 50–60×/second. Background: bolometer principle (Langley, 1878); uncooled microbolometer arrays; vanadium-oxide / amorphous-silicon thermistor materials; vacuum packaging and germanium optics; NUC / flat-field correction; NETD as noise-limited sensitivity (thermal-imaging references).
The Ultimate Guide to Thermal Scopes in Australia
