When the “good-part-only” method becomes too good to be true, you often end up with a bigger mess later.
On the surface, the concept sounds appealing. Simply scan good parts and let the inspection system learn what “normal” looks like. In theory, that approach should work.
In practice, it is a bit like baking a cake without measuring the ingredients and hoping it turns out okay because you added all the right things. No one wants a salty cake, and no manufacturer wants faulty components to reach customers.
Good results usually come from good preparation. Taking the time to properly configure your non-destructive testing (NDT) equipment protects far more than inspection results. It protects product reliability, regulatory compliance, warranty exposure, brand reputation, and overall inspection efficiency.
Eddy current is a powerful technology, but its effectiveness depends on how the system is configured. The difference between a reliable inspection process and a risky one often comes down to whether the system is optimized for actual defects or simply trained to recognize statistical normality.
So, it is worth asking a simple question, “Are you optimizing your inspection to detect defects, or are you modeling what normal production looks like and hoping defects stand out?”
That difference matters far more than most people realize.
Real production environments introduce many variables that can influence inspection signals:
A good inspection system must be able to distinguish between harmless variations and actual defects. When a system is configured using only good parts, several problems appear quickly.
If cracked, pitted, or metallurgically flawed parts are never introduced during setup, the inspection system has no reference for defect signals. It cannot distinguish between a real defect and a harmless signal variation caused by positioning or geometry. In many cases, the system ends up measuring variation among acceptable parts rather than detecting true defects.
Reliable inspection setups normally compare conforming parts, known defective parts, and calibrated defect standards. This allows engineers to place thresholds where good parts consistently pass, and defective parts consistently fail. When only good parts are used, thresholds become statistical assumptions rather than physics-based decisions.
Eddy current systems are susceptible to drift. Electronics warm up, coils age, probes wear, and mechanical alignment shifts. When defect references are included in verification routines, drift becomes visible because known reject components stop triggering the system correctly. Without those references, the baseline quietly shifts until defect sensitivity may be reduced.
ASTM E243, ASTM E309, ASME Boiler and Pressure Vessel Code Section V, and ISO 15548 all emphasize the use of known defect references during calibration and verification. Without these references, it becomes harder to demonstrate that your inspection system can detect the required defect types.
A properly configured eddy current inspection system can avoid these problems. When both conforming and nonconforming samples are used during setup, you gain confidence that the inspection system can reliably separate good parts from defective ones. This means you can detect real defects such as micro cracks, splits, dents , or heat treatment anomalies while minimizing false rejections that interrupt production.
This is where a physics-based inspection approach makes a measurable difference.
And it is exactly the approach FOERSTER has built into its eddy current inspection systems.
We approach eddy current inspections from a different starting point. Instead of modeling what normal production looks like, we focus on what matters most: the defect itself.
Every inspection challenge has a physical solution. Conductivity, permeability, penetration depth, and defect geometry all influence the behavior of an eddy current signal. When tuning inspection parameters around these physical properties, the inspection system becomes more reliable and stable over time.
In other words, the inspection is tuned to detect the defect directly rather than assuming the defect will fall outside a statistical model of “good” parts.
The FOERSTER STATOGRAPH and MAGNATEST product families are designed with this physics-first philosophy built directly into the platform.
For any eddy current inspection application, there is an optimal frequency that produces the strongest defect signal relative to background noise. That relationship depends on material conductivity, magnetic permeability, penetration depth, and expected defect geometry. FOERSTER component testing systems utilize an integrated Setup Wizard to guide frequency selection and signal optimization using known reference defects. This process ensures the system is tuned to maximize the signal-to-noise ratio of the defect itself rather than simply measuring variation among acceptable parts.
Inspection methods that rely only on conforming parts define what “normal” looks like and assume defects will produce signals large enough to fall outside that model. FOERSTER systems instead anchor the inspection to the physics of the defect, ensuring sensitivity is built into the setup from the start.
Probe lift-off variation is one of the most common sources of false rejects or false positives in eddy current inspection. Small changes in probe distance caused by part geometry, vibration, or positioning can create signal changes that can impact your results. Our inspection systems account for these changes by including a built-in clearance compensation that automatically adjusts for variations. By compensating probe distance at the signal level, the inspection remains focused on the defect rather than reacting to mechanical variation.
Rather than widening statistical tolerance windows to absorb this variation, FOERSTER systems work to increase or decrease the signal amplitude all relative to probe liftoff. This keeps the inspection focused on the real defect signals instead of mechanical variation.
Real manufacturing environments are rarely perfectly stable. Material batches change, surface conditions vary, and production environments introduce new variables over time. Because FOERSTER systems are optimized around the defect signal itself, inspection sensitivity remains anchored to the defect rather than drifting with statistical production variation. This helps maintain long-term inspection stability without frequent retraining of the system.
A reliable inspection system should not rely on assumptions. It should demonstrate its ability to detect the required defects before it is deployed in production. That is why FOERSTER adheres to standards and regulations. We also offer application studies using both conforming and nonconforming parts. These studies verify that the inspection system can clearly separate good parts from defective ones while optimizing sensor selection, inspection parameters, and system configuration.
Without defective references, inspection systems can only assume they will detect a flaw. By validating inspection performance with both conforming and nonconforming samples, FOERSTER verifies that the system can detect the defects it is responsible for finding.
When you use a FOERSTER system, you get the confidence that the inspection system can detect the defects it was designed to find. Confidence that production variation will not hide critical flaws. And confidence that the inspection process can withstand internal quality reviews and external audits.
When inspection systems are optimized for the defect instead of statistical normality, the result is an inspection system and process that remains reliable for the long term.
Sure, the idea of configuring an inspection system using only good parts can sound efficient.
But inspection systems are meant to identify defects before they reach your customers, not just ensure components match the “good part.” If the system has never been shown what a defect actually looks like, it cannot truly confirm that it will detect one.
At that point, the inspection process is no longer based on proven detection capability. It assumes that defects will behave differently enough from good parts to be noticed. In safety-critical manufacturing environments, assumptions are not a substitute for verification.
A defect-driven approach removes that uncertainty. By optimizing inspection parameters around known defect signals, manufacturers can ensure the system is capable of detecting the flaws that matter most. This approach also improves long-term inspection stability, supports audit readiness, and helps reduce both missed defects and unnecessary false rejects.
If your current eddy current inspection setup was created using only conforming parts, it may be worth asking a few important questions:
Can your inspection system reliably detect the defect sizes you require?
Was the inspection frequency optimized based on material physics and defect characteristics?
Are defect reference standards used to maximize the signal-to-noise ratio during setup?
Does your system compensate for probe lift-off variations caused by part geometry?
Will inspection sensitivity remain stable if material batches or production conditions change?
Can your calibration method be clearly defended during a quality audit?
If these questions are difficult to answer with confidence, your inspection process may be modeling what normal production looks like rather than proving it can detect what is wrong. If you want to strengthen your inspection reliability, here are a few practical steps you can take today:
Interested in testing our inspection systems? Let us test your parts in an application study to evaluate both conforming and nonconforming sample parts. Our team can help you optimize frequencies, probes, and inspection parameters for your component to ensure the system reliably separates good parts from defective ones before it is deployed in production.
When safety, compliance, and product quality are on the line, the most reliable inspection strategy is clear. Optimize your inspection system for the defect. Just like baking a cake, the best results come from measuring the ingredients, not guessing and hoping for the best.