NETD Explained: What Thermal Sensitivity Really Means for System Designers

NETD is the single number most commonly used to describe infrared camera sensitivity, and it is also one of the most frequently misunderstood. Datasheets specify it, procurement officers ask for it, and integration engineers use it to compare modules — but it only tells part of the story. Here’s what you need to know.

What NETD Actually Measures

NETD (Noise Equivalent Temperature Difference), sometimes written NEDT, is the minimum scene temperature difference a detector can distinguish from noise. Expressed in millikelvin (mK), it represents the point where signal equals noise — a lower number means a more sensitive detector.

Formally: NETD is the blackbody temperature difference that produces a signal-to-noise ratio of 1.0 at the detector output.

In practice: an uncooled LWIR module with NETD = 50 mK can reliably distinguish a target that is 50 mK warmer than its surroundings, under the specified test conditions.

How NETD Is Measured — and Why Test Conditions Matter

The standard NETD measurement uses a blackbody source at a reference temperature (typically 30°C or 35°C) with a defined aperture producing a target filling a specific solid angle. The module images the blackbody; the test system measures the RMS output noise and the signal increment from a small temperature step.

What’s often buried in the footnotes:

• F/number of the optics matters. NETD scales with the square of the F-number. A module measured with F/1.0 optics will read significantly better than the same detector measured at F/1.6. Always check whether the datasheet NETD is specified with optics included or detector-only.
• Temperature matters. NETD degrades at high ambient temperature because the detector’s thermal noise floor rises.
• Integration time matters. Longer integration improves SNR but reduces frame rate.

When comparing competing modules, insist on consistent test conditions — same F-number optics, same ambient temperature (25°C is standard), same integration time.

NETD Values in Context

IRmodules’ SPECTRA L06 achieves NETD ≤ 50 mK at f/1.0; the SPECTRA L12 targets ≤ 40 mK. The cooled SPECTRA H10 (MCT, MWIR) reaches < 20 mK.

The NETD-to-Range Connection

Better NETD translates directly into longer detection range for low-contrast targets. However, the relationship isn’t linear — because you’re fighting both detector noise and scene clutter.

A rough rule of thumb: halving NETD (e.g., 50 → 25 mK) extends detection range for a low-contrast target by roughly 30–40%, assuming optics remain constant. Beyond a certain point, scene clutter — thermal variations in the background — dominates over detector noise, and further NETD improvements yield diminishing returns.

This is why a 20 mK cooled MWIR module doesn’t automatically outperform a 40 mK uncooled LWIR module in all scenarios. In a cluttered urban background, the uncooled module’s inferior NETD is often not the limiting factor.

Three Things NETD Doesn’t Tell You

1. Spatial resolution: A module with excellent NETD but poor MTF (modulation transfer function) will detect a temperature difference but fail to resolve the shape of the target. Both sensitivity and resolution matter for identification tasks.

2. Fixed-pattern noise (FPN): High FPN — vertical or horizontal stripe artifacts — can mask small temperature differences even when pixel-level NETD is good. Ask for a flat-field image, not just a NETD number.

3. Dynamic range: NETD is measured at a specific target temperature. Scenes with very hot objects (engine exhaust, fires) can saturate the well, degrading effective sensitivity on nearby cooler targets.

For a complete sensitivity picture, request NETD, MTF at Nyquist, and a FPN characterization image from your module supplier. IRmodules provides all three in the engineering data package for each module family.