Active Thermography NDT: Lock-in vs Flash vs Pulse Compared

Active thermography is not one technique — it is a family. Choose the wrong member and your camera will see nothing useful, no matter how sensitive the detector. Choose the right one and you will surface subsurface weld defects in seconds, without contact, at production speed.

The confusion is understandable. Vendor literature typically markets a single excitation method as if it were the universal answer. In reality each technique — flash, lock-in, pulse, and the increasingly popular induction variant — answers a different physical question, has a different defect-depth window, and demands different inspection economics.

This guide compares the four main active thermography techniques head-to-head for weld inspection. By the end you will know which one fits your defect catalogue, your throughput target, and your compliance regime under EN 16714 and ISO 9712.

The Common Physics Underneath

All active thermography variants share the same underlying chain:

1. A controlled thermal stimulus is applied to the weld surface or volume.
2. The transient temperature field evolves in time, governed by the heat diffusion equation.
3. A subsurface defect — pore, lack of fusion, crack, inclusion — disturbs that diffusion because it has different thermal effusivity than the surrounding metal.
4. An infrared camera captures the resulting surface temperature contrast at high frame rate.
5. Signal processing extracts the defect signature from the raw thermogram.

What differs is how the heat is delivered and how the response is decoded. That is where flash, lock-in, pulse and induction diverge.

Technique 1 — Flash (Pulse) Thermography

Stimulus: a high-energy xenon flash lamp delivers a millisecond-scale broadband light pulse to the weld surface.

Decoding: the camera records the surface cooldown for 1–10 seconds. Defects appear as transient hotter regions because subsurface voids slow heat diffusion into the bulk.

Strengths

• Fastest acquisition: a single flash + cooldown = full inspection cycle in under 10 s.
• Surface-near defects (≤ 2 mm in steel, ≤ 5 mm in aluminium) imaged with excellent SNR.
• Mature signal processing: thermographic signal reconstruction (TSR), principal component thermography (PCT), pulsed phase thermography (PPT).

Weaknesses

• Limited penetration. Beyond ~2 × thermal diffusion length the transient is buried in noise.
• Surface emissivity must be uniform — bare metal, scale and slag confuse the result.
• Flash lamps degrade and need recalibration; energy uniformity over a large weld area is non-trivial.

Best fit: thin-section welds (≤ 6 mm), automotive sheet, aerospace skin panels, post-weld inspection of stitch and tack welds.

Technique 2 — Lock-in Thermography

Stimulus: a sinusoidally modulated heat source (halogen lamp, laser, or low-frequency induction) excites the weld at a fixed frequency f typically between 0.01 Hz and 1 Hz.

Decoding: the camera demodulates the surface temperature signal at the excitation frequency, returning two images per inspection — amplitude and phase. The phase image is largely independent of surface emissivity and lighting, which is the technique’s defining advantage.

Strengths

• Phase image is emissivity-blind — a huge win on welds where bead, base material and HAZ have very different surface conditions.
• Frequency tuning gives precise depth control: lower f reaches deeper, higher f resolves shallower defects.
• Excellent SNR thanks to narrow-band demodulation; rejects ambient thermal noise.

Weaknesses

• Slow. Each measurement requires several full modulation cycles — at 0.05 Hz that is 60+ seconds per frequency.
• A single inspection rarely covers the full depth range; multiple frequency runs are usually needed.
• Hardware (modulated source + lock-in electronics or software DFT) is more complex than a flash setup.

Best fit: thick-section welds (10–25 mm) where depth profiling matters, structural welds in rail and shipbuilding, batch inspection where cycle time is not the bottleneck.

Technique 3 — Pulse Phase Thermography (PPT)

Stimulus: identical to flash thermography — a single broadband pulse.

Decoding: here is where it diverges. Instead of looking at the time-domain cooldown, the entire thermogram time series is Fourier-transformed pixel-by-pixel. The result is a stack of phase images, one per frequency bin, mathematically equivalent to running lock-in at every frequency simultaneously.

Strengths

• Combines flash speed (one shot) with lock-in’s phase-image emissivity immunity.
• A single acquisition produces depth-resolved phase images — one inspection, multiple depths.
• No specialised modulation hardware; any flash thermography rig can be re-purposed by changing the post-processing.

Weaknesses

• Phase SNR per frequency bin is lower than dedicated lock-in at that frequency.
• Demands a fast camera and a lot of frames (typically ≥ 10 s at 100+ Hz) to populate the FFT spectrum.
• Frequency resolution = 1 / acquisition_time, so deep inspection still requires long cooldown captures.

Best fit: R&D, weld qualification campaigns, mixed-thickness assemblies where you do not know up front which depth carries the defect of interest.

Technique 4 — Induction (Eddy Current) Thermography

Stimulus: a high-frequency electromagnetic coil induces eddy currents inside the weld. Joule heating concentrates around discontinuities — exactly where current paths are forced to detour around cracks or lack-of-fusion zones.

Decoding: thermal contrast appears almost immediately at the defect, not at the surface above it. Cracks light up as they generate the heat themselves, rather than disturbing externally injected heat.

Strengths

• Outstanding sensitivity to surface-breaking and near-surface cracks — even sub-millimetre tight cracks invisible to flash.
• Selective heating: only the conductive material warms up, no parasitic absorption on non-conductive coatings.
• Very fast: inspection cycle 1–3 s.

Weaknesses

• Limited to electrically conductive materials (most weld metals qualify, but heavily oxidised surfaces complicate coupling).
• Crack orientation matters — defects perpendicular to the current path light up; those parallel to it are nearly invisible.
• Coil design and standoff are application-specific; not as turnkey as flash.

Best fit: safety-critical fillet and butt welds in steel and stainless (rail axles, pressure vessels, automotive chassis), in-line crack detection, EN 15085 rail welding inspection.

Side-by-Side Comparison

Decision Framework — Which Technique Should You Pick?

Three questions decide most cases.

Question 1 — What defect dominates your weld failures?

• Surface and near-surface cracks ⇒ induction wins. No competitor matches its sensitivity to tight cracks.
• Porosity, lack of fusion, voids in thin section ⇒ flash is the most economical answer.
• The same defects in thick section (>8 mm) ⇒ lock-in becomes mandatory — flash will not penetrate.
• Mixed catalogue of defects across multiple thicknesses ⇒ PPT, possibly combined with induction for cracks.

Question 2 — What is your throughput?

• In-line, every weld (cycle ≤ 5 s) ⇒ induction or flash. Lock-in is incompatible with line tact.
• Batch sampling (cycle 1–5 min) ⇒ lock-in or PPT, with frequency tuning for depth.
• Lab / qualification (cycle minutes) ⇒ PPT delivers the most information per shot.

Question 3 — What is your compliance scope?

EN 16714 (the dedicated thermographic NDT standard) covers all four techniques explicitly in part 1, with general principles in part 2 and equipment qualification in part 3. ISO 9712 sets operator certification at Level 1, 2 or 3. Lock-in and PPT typically push the operator toward Level 2/3 because of the additional signal-processing decisions. Induction thermography pulls in EN ISO 17643 (eddy-current inspection of welds) as a parallel reference for coil design and reference samples.

If your QMS already aligns with ISO 3834 and EN 1090, all four techniques can be slotted into a documented inspection plan without restructuring procedures — only the equipment qualification record and operator certificates change.

Common Pitfalls When Specifying an Active Thermography System

1. Picking on detector NETD alone. A 20 mK NETD camera is wasted if your stimulus delivers only 5 mK of defect contrast. Source-detector pairing matters more than either component in isolation.

2. Ignoring surface preparation. Flash and PPT live or die by emissivity uniformity. A thin matte coating (e.g., graphite spray) is sometimes the cheapest way to triple your SNR.

3. Underestimating fixturing. Lock-in needs the part stationary for 60+ seconds with stable lighting. If the weld jig vibrates or the ambient light fluctuates, the demodulation filter cannot save you.

4. Forgetting the reference standard. EN 16714-3 demands documented reference samples with calibrated artificial defects (flat-bottom holes, electro-discharge notches). No reference, no qualification, no compliance.

5. Treating active thermography as a replacement for in-process welding monitoring. It is not. Active thermography is post-weld or pre-weld NDT. Real-time process monitoring uses passive thermography on the arc. Most production lines need both.

How Therness Integrates Active Thermography

Therness’s HEAT-INSPECT platform supports flash, lock-in and PPT inspection routines on the same camera hardware, swapping excitation hardware between cycles. Inspection results are captured into the same digital weld record that holds in-process passive monitoring data, MES traceability fields and operator certificates — so a single audit query returns the complete quality genealogy of every weld, in-process and post-weld inspection included.

For full-line deployments we typically pair an induction thermography head for crack detection at the welding cell with a flash or lock-in station downstream for volumetric defect verification on safety-critical joints.

TL;DR

• Flash: fastest and cheapest, shallow inspection, emissivity-sensitive.
• Lock-in: deepest, slowest, emissivity-blind, frequency-tunable.
• PPT: flash speed + lock-in physics, multi-depth in one shot.
• Induction: unbeatable on cracks, only works on conductors, orientation matters.

Match the technique to the defect class, the throughput target and the compliance regime — not to a vendor’s preferred excitation source.