A thermal camera core is the self-contained imaging engine that converts infrared radiation into a digital video signal. It includes the detector array, readout integrated circuit (ROIC), signal processing electronics, and output interface — everything needed to produce a thermal image — in a single compact module. For OEM engineers and system integrators, the core is the critical component that determines the performance ceiling of the entire EO/IR system. The optics, gimbal, and software can be optimized around it, but nothing downstream compensates for a poorly matched core.

What a Thermal Camera Core Contains

At its center is the focal plane array (FPA): a grid of individual detector elements, each responding to thermal radiation in a specific infrared waveband. Most commercial cores operate in the long-wave infrared (LWIR, 8–14 μm), where uncooled detectors based on vanadium oxide (VOx) or amorphous silicon (a-Si) microbolometers achieve acceptable sensitivity without active cooling. Higher-performance applications use mid-wave infrared (MWIR, 3–5 μm) detectors, which require cryogenic cooling but offer significantly lower noise and faster response.

Surrounding the FPA is the ROIC, which reads out the charge or resistance change from each pixel, applies bias, and serializes the signal for downstream processing. The on-board image signal processor (ISP) then performs non-uniformity correction (NUC), bad-pixel replacement, dynamic range optimization, and optional image enhancement. The processed output is delivered via a standardized video interface — MIPI, CameraLink, SDI, BT1120, or GigE — along with a communication channel (RS422, CAN, Ethernet) for configuration and control.

Core Formats and Resolution Classes

Thermal camera cores ship in two broad resolution classes. The 640×512 format is the workhorse: it covers most security, surveillance, UAV, and industrial applications with a good balance of sensitivity, power consumption, and cost. The 1280×1024 format provides four times the pixel count and is used where extended detection range, large-area coverage, or high-resolution identification is required — at the cost of higher power draw, larger board size, and increased data throughput.

Pixel pitch — the center-to-center distance between adjacent detector elements — is a second key dimension. Smaller pitches (12 μm, now 10 μm in leading-edge products) pack more pixels into the same FPA die area, reducing detector cost and enabling more compact optical designs. Larger pitches (17 μm) offer better per-pixel sensitivity and are typically found in older or cost-optimized designs. A 640×512 core at 12 μm fits on a roughly 7.7 mm × 6.1 mm die; the same resolution at 17 μm requires approximately 10.9 mm × 8.7 mm, which drives optics diameter and cost upward.

For example, the SPECTRA L06 — 640×512 Uncooled LWIR Module at 12 μm achieves a board footprint starting from 26×26 mm, making it viable for the smallest UAV and handheld payloads. By contrast, the SPECTRA L12 — 1280×1024 LWIR Module uses a 35×35 mm board and is aimed at systems where detection range and spatial resolution outweigh size constraints.

Cooled vs Uncooled Cores

The choice between cooled and uncooled cores shapes every other engineering decision in the system.

Uncooled cores operate at ambient temperature. The microbolometer detector absorbs infrared radiation and changes resistance proportionally; the ROIC measures these resistance changes at frame rate. Because no cooling is required, power consumption is low (typically 1–5 W for a 640 module), startup is immediate, and MTBF is high. NETD (noise-equivalent temperature difference, the smallest temperature difference a core can resolve) for a modern uncooled 640×512 core at 12 μm is typically 25–40 mK — sufficient for most surveillance, perimeter security, and industrial inspection tasks.

Cooled cores cool the FPA to cryogenic temperatures, typically 77 K for MWIR InSb or HgCdTe detectors. This dramatically reduces thermal noise and enables NETD values below 20 mK — often below 10 mK in research-grade systems — along with faster response, higher dynamic range, and sensitivity to smaller, faster targets. The tradeoff is significant: a cooled MWIR core consumes 20–60 W (mostly driven by the Stirling cooler), requires 2–6 minutes to cool down, and costs an order of magnitude more than a comparable uncooled design. The SPECTRA M06 — 640×512 Cooled MWIR Module illustrates the package size consequence: 240×115×110 mm and DC 28V supply versus the few-centimeter board of uncooled alternatives.

The engineering decision is not about which is “better” but which performance profile matches the application’s requirements. A long-range border surveillance system or a missile seeker tracking fast, low-contrast targets benefits from a cooled core. A perimeter fence camera, UAV search-and-rescue payload, or electrical inspection drone typically does not.

Key Performance Parameters

When evaluating a thermal camera core for integration, five parameters define the performance envelope:

NETD (noise-equivalent temperature difference) quantifies sensitivity: the temperature difference a core can statistically detect. Lower is better, but marginal NETD improvements rarely justify the cost increase when detection range is dominated by atmospheric transmission and target emissivity. NETD below 40 mK is acceptable for most uncooled surveillance applications; below 25 mK for demanding targets.

Frame rate determines whether the system can track fast-moving targets and how much latency the processing chain must absorb. 25 Hz satisfies most perimeter and maritime surveillance. 50–60 Hz is required for fast-moving vehicle tracking or pilot assistance. Export-controlled applications in some markets are limited to 9 Hz.

Interface must match the downstream processing platform. MIPI CSI-2 integrates tightly with embedded SoCs. CameraLink and SDI are preferred for payloads requiring rugged, long-cable connections. GigE supports long-distance, networked architectures. Choosing an interface mismatch at the design stage forces costly adapter boards later.

Power envelope is system-level. A 640 uncooled module drawing 3 W versus one drawing 1.5 W doubles the thermal management burden in a tight UAV payload over a long flight. For battery-operated platforms, power efficiency is often a selection-defining criterion.

Calibration and NUC behavior affects operational usability. Shuttered cores interrupt the image stream momentarily to re-calibrate the detector array; shutterless designs rely on algorithmic NUC but may show slow spatial drift in static scenes. The choice matters for applications where any image dropout is unacceptable, such as weapons sighting or automated detection systems.

For a deeper look at how sensitivity specifications compare across technologies, the EMVA Standard 1288 provides a framework for characterizing camera performance that is increasingly referenced by thermal core manufacturers. The IEEE also publishes relevant work on infrared detector physics via IEEE Xplore for engineers who need to evaluate claims rigorously.

How Thermal Camera Cores Fit Into a System

A thermal camera core is rarely the end product. It is integrated into a larger assembly that adds optics, mechanical housing, environmental sealing, and sometimes a second imaging channel (visible or SWIR). The core’s output drives a processing board — an embedded GPU, FPGA, or SoC — that handles detection algorithms, compression, and communication.

System integrators working on EO/IR gimbals, UAV payloads, or vehicle-mounted platforms typically select the core first, then design the optics and mechanical package around its detector format and interface requirements. Getting the core selection right — resolution, spectral band, pixel pitch, interface, and power budget — before committing to the optical design avoids the most expensive re-spins. The SPECTRA M12 — 1280×1024 Cooled MWIR Module, for instance, defines a very different enclosure constraint than the 640×512 uncooled counterparts.

For application-specific selection guidance, the Airborne / UAV Payloads and Vehicle-mounted Systems application pages outline how core specs map to platform requirements.

Frequently Asked Questions

What is the difference between a thermal camera core and a thermal camera module? A core is the bare imaging engine — detector, ROIC, ISP, and interface board — without a lens or housing. A module typically includes a core integrated with a lens and environmental housing into a ready-to-mount unit. OEM integrators usually work with cores; end-system buyers more often specify modules.

Can I use a thermal camera core without a dedicated SDK? Many cores expose simple RS422 or UART control protocols with a small command set for basic parameters (NUC trigger, palette, digital zoom). A full SDK becomes necessary for advanced features: raw data access, algorithm integration, or real-time calibration updates. Always verify protocol documentation availability before finalizing selection.

How do I compare thermal camera cores from different manufacturers? Request a standardized test report: NETD measured at f/1.0 and operating temperature, uniformity before and after NUC, MTF curve, and start-up time. Datasheets use inconsistent conditions; only controlled side-by-side tests on the same target and ambient reveal real performance differences.

Is a higher resolution core always better? Not for every application. A 1280×1024 core at twice the cost, double the power, and larger physical footprint only outperforms a 640×512 core if detection range or spatial resolution is the binding constraint. For many security, inspection, and robotics applications, a well-chosen 640 core with the right optics outperforms a mismatched 1280 system.

What export restrictions apply to thermal camera cores? Thermal camera cores are dual-use goods subject to export controls in most jurisdictions, including the Wassenaar Arrangement and US EAR (Export Administration Regulations). Frame rate above 9 Hz is a common trigger for higher control classification. Buyers and integrators must verify applicable classifications for their destination country before placing orders.