The Repair LibraryRead · Learn · Master

Identifying EMI Problems After Repair

The two sections before this one built the model and the reason it matters; this one turns diagnostic, because the single most common way a compliant device becomes an interfering one is a repair, and a technician needs a method for finding the interference their own work created. The method begins with the most powerful clue a bench ever gets: the regression. The device worked, was opened, and now emits or is disturbed — so the repair is the prime suspect, and the search starts not with the whole universe of possible faults but with the short list of things the work actually touched. From there the section teaches the three shapes a repair-created problem takes, each a face of the emission-and-susceptibility split from the chapter's opening. There is self-interference, where the device now degrades its own function — a reworked switching regulator that deafens the very radio it shares a board with. There is emission outward, where the device now disturbs its neighbors. And there is new susceptibility, where the device is now disturbed by them. All three trace to the same root, a coupling path the repair opened, and the section's core skill is finding it: the emission sweep, running a near-field probe across the board to locate where the unwanted energy is strongest and so to name the aggressor — the source now radiating that the containment used to hold. The sweep is read against a known-good board, the golden reference from the measurement chapter, because the question is never 'does this board emit' — every board emits — but 'does this board emit more than its healthy twin, and where.' The strongest point on the sweep names the coupling path, and the coupling path names the disturbed containment: the shield left off over that exact spot, the ground stitch not restored beneath it, the lead re-dressed past it, the ferrite missing from that line. The section closes on the diagnostic creed the whole chapter has been building toward: after a repair, interference is a regression to be traced, not a mystery to be endured — sweep for the aggressor, compare to the golden board, and follow the strongest emission back to the containment the work disturbed, because that is where the fix will be.

AdvancedLow Risk23 min read

What You Will Learn

  • You will learn the regression logic — worked before, broke after — that makes a repair the prime suspect for new EMI.
  • You will learn the three shapes of a repair-created problem — self-interference, emission outward, and new susceptibility.
  • You will learn the emission sweep — running a near-field probe across a board to locate where the unwanted energy is strongest.
  • You will learn to read a sweep against a golden board, because every board emits and the question is where it emits more.
  • You will learn to trace the strongest emission back through the coupling path to the containment the repair disturbed.

What You Will Be Able To Do

  • You will be able to treat a post-repair interference as a regression and start from what the work actually touched.
  • You will be able to classify a repair-created problem as self-interference, emission, or susceptibility.
  • You will be able to run an emission sweep with a near-field probe to locate a board's strongest emission point.
  • You will be able to compare a repaired board's sweep to a known-good board and read the meaningful difference.
  • You will be able to trace a located aggressor back to the shield, ground, lead, or ferrite the repair disturbed.

Required Tools

  • A near-field probe with a receiver or spectrum analyzer — with a scope serving at close range — the instrument that runs the emission sweep and finds the aggressor
  • A known-good reference unit of the same model — the golden board every sweep is read against
  • A donor device repaired with a shield deliberately left off — a made regression to practice tracing
  • The notebook — this section records the before-and-after baseline and the swept map of each board

When NOT to Attempt This

Do not attempt this section if any of the following apply to you:

  • You are not comfortable working with small surface-mount components.
  • You have not completed the prerequisite sections for this skill.
  • You do not have the required tools in working condition.

Section Overview

Sections 2.1 and 2.2 built the model and its reason; this one turns diagnostic (electromagnetic-interference-basics). A repair is the most common way a compliant device starts interfering, so the strongest clue is the regression. The device worked, was opened, and now emits or is disturbed — so the repair is the prime suspect, and the search starts from what the work touched (common-rf-failure-modes-in-consumer-devices). A repair-created problem takes three shapes. Self-interference where the device degrades its own function, emission where it disturbs neighbors, and new susceptibility where they disturb it — the emission-and-susceptibility split, seen after a repair (emc-standards-and-why-they-matter-for-repair). All three trace to a coupling path the repair opened, and the core skill is finding it. The emission sweep runs a near-field probe across the board to locate where the unwanted energy is strongest and so to name the aggressor — the source the containment used to hold (rf-signal-measurement-basics). The sweep is read against a golden board. Every board emits; the question is whether this one emits more than its healthy twin, and where. The strongest point names the coupling path, and the path names the disturbed containment. Regression, three shapes, sweep, golden-board comparison, and the trace to the fix — the chapter's diagnostic creed.

Why This Matters

This section is where the chapter's model becomes a bench procedure a technician can actually run (electromagnetic-interference-basics). This matters because the regression is the most powerful diagnostic a repair ever offers: a device that worked before the repair and interferes after it has told you where to look — the work — so the technician who starts from the short list of what was touched solves in minutes what a blind hunt through the whole board never finds (common-rf-failure-modes-in-consumer-devices). This matters because the three shapes tell you which way the chain points: self-interference, emission, and susceptibility are found and confirmed differently, so naming the shape early aims the whole search (emc-standards-and-why-they-matter-for-repair). It matters because the sweep turns the invisible into a map: interference has nothing to see, but a near-field probe swept across a board draws a picture of where the energy is strongest, converting a formless complaint into a location on the board (rf-signal-measurement-basics). And it matters because the comparison is what makes the sweep mean anything: every board emits, so an absolute reading proves nothing — only the difference between the repaired board and a known-good one separates the repair's damage from the board's normal hum, and without the golden reference a technician is guessing at which peaks matter. Read the regression, name the shape, sweep for the aggressor, and compare to the golden board — and a repair-created EMI problem stops being bad luck and becomes a fault with an address.

Required Prerequisites

Before starting this section, you should have completed:

  • Isopropyl alcohol and swabs — shield seams and ground contacts are cleaned before a sweep, because a contaminated joint is itself a coupling path and a false aggressor.
  • Low-tack labels and a marker — the strongest points on a sweep are marked on the board as they are found, so the emission map is physical and the trace to the containment is unambiguous.
  • A camera — the before-and-after baseline is photographed, because the sweep means most when the repaired board can be set literally beside its known-good twin.
  • A near-field probe with a receiver or spectrum analyzer — the core instrument of the section, run across a board to find and rank its emission points; a scope serves at close range where the signal is strong.
  • A known-good unit of the same model — the golden board that turns an absolute sweep into a meaningful comparison; without it the sweep is only numbers.
  • A donor device with a shield deliberately left off — a repair-created regression made on purpose, so the trace from strongest emission back to the missing containment can be practiced end to end.

Real-World Applications

This is the section a technician runs whenever a device misbehaves after their own work. A repairer whose reflowed board now deafens its own Wi-Fi recognizes self-interference, sweeps the board, and finds the aggressor at the switching regulator they reworked, whose shield they never refitted (emc-standards-and-why-they-matter-for-repair). A technician whose repaired device now disturbs a nearby monitor reads an emission-outward problem, sweeps for the strongest point, and traces it to a ground stitch left open by the repair (electromagnetic-interference-basics). A bench comparing a repaired unit's sweep to a shelf-spare of the same model sees at once which emission peaks are normal and which are the repair's, because the golden board shows what healthy looks like (rf-signal-measurement-basics). And a tech chasing an intermittent that only appears when the case flexes sweeps while flexing and watches the aggressor come and go with a shield can that no longer seats (common-rf-failure-modes-in-consumer-devices). The confusions this prevents: a repair-created regression hunted as a random fault, a normal emission peak mistaken for damage, an interference chased with no probe and no map, and a symptom treated without tracing it to the containment the work disturbed.

Common Challenges

  • Every board emits, so absolute readings mislead. A sweep of a healthy board is full of peaks — only the difference from a golden board separates the repair's damage from the board's normal hum (rf-signal-measurement-basics).
  • The symptom names the victim, not the source. "The radio got worse" points at what is degraded — the sweep is what finds the source, and the two are often far apart on the board (electromagnetic-interference-basics).
  • The regression window can be forgotten. A technician deep in a board forgets the device worked an hour ago — the single most useful fact, that the repair is the suspect, is the first thing a frustrated search discards (common-rf-failure-modes-in-consumer-devices).
  • The disturbed containment hides its own cause. A lifted shield or an open ground looks like nothing — the sweep points to the spot, but only a technician who knows what containment belonged there reads the emission as a missing part (emc-standards-and-why-they-matter-for-repair).

Safety Notes

Risk Level: Low. This section sweeps and compares powered boards with a probe — it heats nothing — and the standing bench law frames it.

  • The probe observes, it does not poke — a near-field probe reads the field near a board; it is never jabbed into a live switching node, mains section, or charged capacitor.
  • Respect the powered board — sweeping is done on a live board, so ESD discipline and live-board caution apply throughout.
  • The transmit rule still holds — where a swept device can transmit, its antenna and feed are never opened while it is keyed.

Professional Tips Before Starting

  • Start from the regression. Worked before, broke after — the repair is the suspect — begin at the short list of what the work touched, not the whole board (common-rf-failure-modes-in-consumer-devices).
  • Name the shape first. Self-interference, emission, or susceptibility — which shape it is aims the search before the probe comes out (emc-standards-and-why-they-matter-for-repair).
  • Never sweep without a golden board. Every board emits, so a lone sweep proves nothing — the known-good twin is what makes a peak meaningful (rf-signal-measurement-basics).
  • Follow the strongest point to the part. The peak names the coupling path, and the path names the disturbed containment — trace the emission to the shield, ground, or ferrite the work moved (electromagnetic-interference-basics).
  • Mark the map as you go. Label the strong points on the board as the sweep finds them — a remembered map is a guessed map, and the trace deserves a real one.

Finding the Interference a Repair Created

The Regression — Worked Before, Broke After

The most powerful fact in a post-repair EMI hunt is the one a frustrated technician forgets first: the device used to work (common-rf-failure-modes-in-consumer-devices). A repair-created interference problem is a regression, and a regression names its own suspect. The device functioned and complied before it was opened; it was opened; and now it emits, is disturbed, or degrades itself — so the change that caused the problem is overwhelmingly likely to be the change that was just made. This collapses the search. Instead of the whole board and every possible fault, the suspect list is the short set of things the repair actually touched: the component that was replaced, the shield that was removed and refitted, the ground that was reworked, the leads that were dressed, the connectors that were reseated. The discipline is to hold that window. A technician an hour deep in a stubborn board starts treating it as a fresh mystery and forgets the single fact that would solve it — that everything was fine before the work — so the regression is written down at the start and returned to whenever the search sprawls. The regression also dates the fault. If the device was quiet before and interferes now, the interference is the repair's, not a latent design flaw or a customer's imagination, and that certainty is worth defending against the doubt a long search breeds. Worked before, broke after, suspect the work — the first move in every post-repair EMI hunt, and the one most often skipped.

The Three Shapes — Self-Interference, Emission, and Susceptibility

A repair-created problem shows up in one of three shapes, and naming the shape early aims everything that follows (electromagnetic-interference-basics). The first shape is self-interference — the device now degrades its own function, its own emitted noise coupling into its own sensitive circuit, as when a reworked switching regulator whose shield was not refitted raises the noise floor of the radio it shares a board with and the device quietly deafens itself. The second shape is emission outward — the device now disturbs its neighbors, the repaired unit become an emitter that was quiet before, found because something else nearby started misbehaving when the device is on. The third shape is new susceptibility — the device is now disturbed by its neighbors, glitching when a motor or a charger runs nearby that it tolerated before the work. These are the emission-and-susceptibility split of Section 2.1, seen from the repair bench, with self-interference as the special case where the device is both the source and the victim at once. The shape decides the test. Self-interference is reproduced on the bench with the device alone; emission is confirmed by watching a nearby victim; susceptibility is confirmed by bringing an interferer near — so the shape tells the technician what to set up before a probe is lifted. And all three share one root: a coupling path the repair opened, whichever direction the energy now flows. Self, outward, inward — three shapes of one broken containment, and naming which one is on the bench is the second move after the regression.

The Sweep and the Trace — Finding the Aggressor and Following It Home

The core skill of the section is turning an invisible interference into a marked spot on a board, and then into a named missing part (rf-signal-measurement-basics). The tool is the emission sweep. A near-field probe is run slowly across the board while a receiver or spectrum analyzer watches — the sensitive, frequency-resolving instrument for weak emissions, though a scope serves at close range where the signal is strong — and where the reading peaks is where the unwanted energy is strongest — the location of the aggressor, the source now radiating that the containment used to hold. But a sweep alone lies, because every board emits. A healthy board is full of peaks — its clocks, its switcher, its buses all radiate at close range — so an absolute reading proves nothing, and the diagnostic question is never "does this board emit" but "does this board emit more than its healthy twin, and where." This is why the sweep is read against a golden board: the same sweep is run on a known-good unit of the same model, and the meaningful signal is the difference — the peak that is tall on the repaired board and normal on the golden one is the repair's aggressor, while the peaks that match are the board simply being itself (common-rf-failure-modes-in-consumer-devices). Then comes the trace, which is the whole point. The strongest differing point names the coupling path, and the coupling path names the disturbed containment: the shield left off directly over that spot, the ground stitch not restored just beneath it, the lead re-dressed so it now passes it, the ferrite missing from the line that runs to it. The sweep leads from symptom to source to the exact thing to fix, and the fix — restoring that containment — is the work of the section that follows. Sweep for the peak, subtract the golden board, and follow the difference to the missing part — the method that makes repair-created EMI a fault with an address.

Common Mistakes

  • Abandoning the regression. The device is treated as a fresh mystery — it worked an hour ago, and the repair is the suspect the frustrated search forgot (common-rf-failure-modes-in-consumer-devices).
  • Sweeping without a golden board. An absolute reading is trusted — every board emits, so only the difference from a known-good twin separates damage from normal hum (rf-signal-measurement-basics).
  • Chasing the victim instead of the source. The degraded circuit is probed — the sweep finds the source, which is often far from the victim on the board (electromagnetic-interference-basics).
  • Stopping at the spot instead of the part. The strongest emission point is found and the search ends there — the peak names a coupling path that names a missing shield, ground, or ferrite, and the fault is the missing part, not the spot (emc-standards-and-why-they-matter-for-repair).
  • Not naming the shape. The problem is chased without deciding whether it is self-interference, emission, or susceptibility — the shape decides the test, and skipping it aims the search at nothing.

Troubleshooting Guidance

  • A device misbehaves after your own repair — start from the regression: it worked before, so the repair is the suspect, and the search begins at the short list of what the work touched rather than the whole board (common-rf-failure-modes-in-consumer-devices).
  • The device degrades its own function — self-interference, reproduced alone: the device's own noise is coupling into its own circuit, so reproduce it on the bench with the unit isolated and sweep for the internal aggressor, most often a reworked switcher missing its shield (electromagnetic-interference-basics).
  • A sweep shows many peaks and you cannot tell which matters — read against a golden board: run the same sweep on a known-good unit and compare, because the meaningful signal is the peak that is tall on the repair and normal on the twin (rf-signal-measurement-basics).
  • The strongest point is found but the fault is not obvious — trace the coupling path to the containment: the peak names a path, and the path names the shield, ground, lead, or ferrite the repair disturbed at that spot, which is the part to restore (emc-standards-and-why-they-matter-for-repair).

Verification & Testing Methods

Confirm your diagnostic skill before moving on:

  • [ ] I can treat a post-repair interference as a regression and start from the short list of what the work touched.
  • [ ] I can classify a repair-created problem as self-interference, emission outward, or new susceptibility, and name the test each needs.
  • [ ] I can run an emission sweep with a near-field probe to locate a board's strongest emission point.
  • [ ] I can read a sweep against a golden board and separate the repair's aggressor from the board's normal emission.
  • [ ] I can trace a located aggressor back through its coupling path to the shield, ground, lead, or ferrite the repair disturbed.

Then try the practice exercises below — regression reasoning and sweep-and-trace only; scenarios differ from the quiz.

Practice Exercises

  1. Name the regression (5 minutes, desk work). For a described post-repair complaint, write the regression explicitly — what worked before, what changed, and the short suspect list of what the repair touched — so the most powerful clue is stated before any probing begins (common-rf-failure-modes-in-consumer-devices).
  2. Classify the shape (5 minutes, desk work). For three described post-repair problems, label each self-interference, emission, or susceptibility, and state the bench setup each would need to confirm — the unit alone, a nearby victim, or a nearby interferer — so the shape aims the test (electromagnetic-interference-basics).
  3. Run the sweep against a golden board (5 minutes, donor plus known-good). Sweep a donor board and a known-good unit of the same model with a near-field probe, and record which peaks match and which differ, so the difference — not the absolute reading — becomes the signal you trust (rf-signal-measurement-basics).
  4. Trace the source home (5 minutes, made-regression donor). On a donor with a shield deliberately left off, sweep to the strongest differing point and follow it to the missing containment — the shield, ground, or ferrite that belonged there — completing the path from symptom to the exact part to restore (emc-standards-and-why-they-matter-for-repair).

These core steps — the named regression, the classified shape, the golden-board sweep, and the traced aggressor — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • A post-repair interference is a regression that names its own suspect — the device worked, was opened, and now interferes, so the search starts from the short list of what the repair touched, not the whole board (common-rf-failure-modes-in-consumer-devices).
  • A repair-created problem takes three shapes — self-interference where the device degrades its own function, emission where it disturbs neighbors, and susceptibility where they disturb it — and naming the shape decides the test (electromagnetic-interference-basics).
  • The emission sweep runs a near-field probe across the board to find where the unwanted energy peaks and so to locate the aggressor — the source the containment used to hold (rf-signal-measurement-basics).
  • Every board emits, so a sweep is read against a golden board — the meaningful signal is the peak that is tall on the repair and normal on the known-good twin, not the absolute reading (emc-standards-and-why-they-matter-for-repair).
  • The strongest differing point names the coupling path, and the path names the disturbed containment — a shield, ground, lead, or ferrite the repair moved — so the sweep leads from symptom straight to the part to restore.

Skills Learned

After completing this section, you can:

  • Treat a post-repair interference as a regression and start from what the work touched.
  • Classify a repair-created problem as self-interference, emission, or susceptibility, and set up the right test.
  • Run an emission sweep with a near-field probe to locate a board's strongest emission point.
  • Compare a repaired board's sweep to a golden board and read the meaningful difference.
  • Trace a located aggressor back to the shield, ground, lead, or ferrite the repair disturbed.

Glossary Additions

New terms introduced in this section:

  • self-interference — a repair-created interference problem in which a device degrades its own function, its own emitted noise coupling into its own sensitive circuit so that the device is at once the source and the victim of the interference. The classic case is a switching regulator whose shield was not refitted after a repair raising the noise floor of a radio on the same board until the device deafens itself. It is the special case of the emission-and-susceptibility split where both roles belong to one device, and it is diagnosed by reproducing the fault with the unit isolated on the bench — no neighbor required — and sweeping for the internal aggressor, which almost always traces to a coupling path the repair opened between the device's own noisy and sensitive sections.
  • aggressor — the source in an interference coupling, the circuit or node actively emitting the unwanted energy that degrades a victim; the counterpart to the victim in the source-path-victim model. Identifying the aggressor is the goal of an emission sweep, because a symptom names the victim — the circuit that is degraded — while the fix lives at or between the aggressor and the coupling path, often far from the victim on the board. After a repair the aggressor is typically a source the containment used to hold quiet — a switcher, a clock, a fast bus — that a lifted shield, an open ground, or a moved lead has newly exposed, so naming the aggressor is the step that turns a complaint about a degraded victim into a location on the board.
  • emission sweep — the diagnostic technique of running a near-field probe slowly across a board while a receiver or spectrum analyzer watches — a scope serving at close range where the signal is strong — mapping where the board's emitted energy is strongest in order to locate an aggressor. Because every board emits — its clocks, switcher, and buses all radiate at close range — a sweep is meaningful only in comparison: the same sweep run on a known-good golden board of the same model separates the peaks that are normal from the peak that is the repair's damage, and the diagnostic signal is the difference between the two, not the absolute reading. The strongest differing point names the coupling path and, through it, the specific shield, ground, or filter the repair disturbed, making the sweep the bridge from an invisible symptom to the exact part to restore.

Suggested Next Sections

Must read next:

  • Shielding and Filtering in Board Repair — Section 2.4 closes the chapter at Professional depth on the fix this section's trace leads to: how to restore the shield cans, gaskets, ground stitching, ferrites, and filter components that hold a coupling path closed, exactly as the design intended.

Recommended: