Thermal 101

Four kinds of device share the same cores. A riflescope has its own reticle and zero and replaces your day scope. A handheld monocular is the scanner: light, wide, not attached to anything. A clip-on mounts ahead of your day scope’s objective and projects a thermal image into it, so you keep your zero. A binocular feeds one sensor to two eyepieces for long sits. The catalog covers 4 types; the type pages compare within each, because a 2 lb scope and a 12 oz monocular are not competing for the same job.

A thermal core is a grid of microbolometers — tiny resistors whose resistance changes with the infrared energy landing on them. The market names the grid by its width: 1280, 1024, 640, 400, 384, 320 in this catalog. More pixels across means a wider view behind the same lens, or the same view at longer range behind a longer one — and nothing else. Resolution does not set sensitivity, and it does not make each pixel sharper.

A 640 has 2.8 times the pixels of a 384 and 1.67 times the width, so behind identical lenses it reaches 1.67 times further at every Johnson level. That ratio is the whole comparison. Every unit’s page shows the ranges, so the comparison is two numbers, not a belief.

Pitch is the physical size of one pixel. 58 units here run 12 µm cores and 8 run 17 µm. It matters because pitch divided by focal length is the angle one pixel sees, and that angle is what sets range: 12 µm behind 35 mm and 17 µm behind 50 mm see the same 0.34 mrad per pixel and reach the same distances. The 12 µm unit just does it with a shorter, lighter lens.

The cost of the smaller pixel is energy: half the area collects half the infrared, so a 12 µm core of the same generation is noisier. Current 12 µm cores have mostly closed the gap, but a well-made 17 µm core can still give the cleaner image on a humid, low-contrast night. Buy the pitch-and-lens pair, not the pitch.

Noise-equivalent temperature difference: the smallest temperature difference the core can distinguish from its own noise, in millikelvin. Lower is better. Around 40 mK is entry level, 25 mK is the current mainstream, and the best published figures sit below 20 mK. It is the spec that decides how much a deer stands out from warm ground in September, and it is the spec most makers are least willing to publish: 16 of the 67 units here do not.

Compare NETD across makers with care. The figure depends on the test conditions — f-number, scene temperature, frame integration — and not everyone states them. The lowest published figure in the catalog is the AGM Adder V2 LRF 35-640 at 15 mK.

How many times a second the core redraws the scene. Most cores run 50 Hz, a leftover of the European video standard; 31 of the units here publish 60 Hz. The difference shows when something moves — a running hog at 30 Hz smears, at 50 Hz it tracks, at 60 Hz it tracks a little more smoothly. Rates in this catalog: 50, 60 Hz.

You will also see 9 Hz versions of many of these units. Those exist because US export rules restrict thermal cores faster than 9 Hz; the 9 Hz variants are for export markets and are noticeably laggy to pan. None are listed here.

Thermal lenses are germanium, which is why they are expensive and why a longer one adds weight fast. Focal length — 9.1 to 60 mm in this catalog — narrows the view and stretches every range figure in proportion. It is also why a “640” can be either a wide-angle scanner or a long-range scope: the sensor is the same, the lens is not.

One number ties pitch and lens together: pitch divided by focal length, the IFOV, in milliradians. The finest in the catalog is the ATN ThOR 5 XD LRF 4-40x at 0.120 mrad; the coarsest is the Teledyne FLIR Breach PTQ136 at 1.319 mrad. Lower is finer, and it is the single number that sets range. The lens pages group units by focal length so you can see how much of a unit’s reach is glass.

Base (optical) magnification is set by the lens, the sensor size and the eyepiece, and is usually 1× to 4× on a thermal — lower than a day scope, because you cannot magnify past what the pixels resolve. Digital zoom is cropping: 2× digital shows a quarter of the pixels at twice the size, and by 4× to 8× you are looking at blocks. It is useful for framing a shot on an animal you have already recognized, and useless for recognizing one. The “up to 32×” on a box is the base times the maximum digital step, and the upper half of that range is decorative.

Horizontal field of view follows from sensor width, pitch and focal length — 2 × atan(sensor width ÷ 2 × focal length) — and every unit’s page shows the derived figure beside the published one. Where the two disagree by more than 15 % the page says so; the usual reasons are a listing that quotes a diagonal, or a spec sheet copied from a sibling model with a different lens.

Wide is for scanning, narrow is for range, and you cannot have both from one lens. Pixels per degree — sensor width divided by horizontal FOV — is the number that says how much detail lands on the display regardless of magnification.

The microdisplay you actually look at: OLED and AMOLED are the norm, LCOS appears on some older units. Resolution matters less than it looks — a 1024×768 display showing a 640×512 image is already upscaling — but it decides how sharp the reticle and the menus are, and OLED’s black level is what makes a white-hot image look like an image rather than a grey screen. Display resolution is listed on every unit; it is not used in any range estimate.

Almost every maker publishes a detection range, usually for a 1.8 m × 0.5 m standing human, sometimes for a deer, occasionally for nothing stated. 60 of the 67 units here publish one; 7 do not. The figure is nearly always computed against the 1.8 m height: the distance at which a person spans about two pixels top to bottom. That is a real, defensible number, and it describes a warm smudge.

The claim is shown on every page, labelled as the maker’s. It is not used to sort anything.

In 1958 John Johnson at the US Army’s Night Vision Laboratory measured how many resolvable line pairs across a target observers needed to detect it, recognize what it was, and identify which one it was. At 50 % probability the answers were roughly 1, 4 and 6.4 line pairs — 2, 8 and 13 pixels — across the target’s critical dimension, the narrower one, because the width of an upright figure is what separates it from the ground. This is the basis of STANAG 4347 and of every serious thermal range specification.

Every range estimate on this site is that calculation: 0.5 m ÷ (pixels × IFOV) × 1000, for a standing human, in clear air. Across the catalog the estimates run from the Armasight Sidekick 320 (190 m detect) to the ATN ThOR 5 XD LRF 4-40x (2,083 m); the median is 729 m. The box claim averages 2.3× our detection estimate; the widest gap is the ATN ThOR 5 320 3-12x at 5.1× and the narrowest the Burris BTS35 v3 640 at 1.4×.

Treat recognition as the working range and identification as the distance at which you should be sure what you are shooting. Both are upper bounds: humidity, rain and dust shorten every one of them, and 50 % probability means you miss half the time at the stated distance.

Cold kills lithium cells. A published run time is a room-temperature figure with Wi-Fi off; expect meaningfully less in January. The spec that matters as much as the number is whether the pack comes out: 23 of the units here take a removable pack, which turns a spare in your pocket into double the run time. Internal-only units need a power bank on a cable. The longest published run time in the catalog is the ATN BlazeHunter 635 LRF at 14 h.

The range here runs from the Teledyne FLIR Breach PTQ136 at 7.4 oz to the ATN ThOR 5 XD LRF 4-40x at 52.6 oz. Weight is the maker’s figure for the unit alone — mounts and batteries are sometimes included and sometimes not, and where a maker says which, the unit’s page does. On a clip-on every ounce hangs off the far end of the rail; on a handheld it is what you hold up for an hour.

Judging distance through a thermal is hard — no ground texture, nothing to scale against — so a built-in laser rangefinder is the feature most people wish they had bought first. 22 units here have one. A ballistic calculator uses that range plus your load data to move the reticle; it is only as good as the rangefinder feeding it and the data you typed in. Neither affects any range estimate on this site.

IPX7 means survives 30 minutes under a metre of water; IP67 adds a dust rating; IP66 means powerful jets but not immersion. Some makers say only “weather resistant”, which is not a rating, and the field is blank for those. Rain is the ordinary case for a night hunt, so anything below IPX6 is worth thinking about.