The Repair LibraryRead · Learn · Master

Advanced Inspection After Rework

The chapter ends by training the judge, because every bench it built produces work that must be sentenced — passed, reworked, or escalated — and the sentencing skill is its own craft. Fillet reading comes first as a formal discipline rather than a glance: the heel fillet that gull-wing joints must carry and the toe they need not, the end fillets of chip parts, the side fillets castellations and wettable flanks offer, each read against the wetting angle that tells the only structural truth — solder that flowed low and feathered wetted its surfaces, solder sitting tall and beaded did not, whatever its shine says. And shine itself gets demoted: the leaded-era gloss standard misreads healthy lead-free joints as cold every day, because lead-free freezes matte-normal — texture is read per alloy, and the real cold joint confesses by its shape and its wetting, not its finish. The magnification discipline turns looking into inspection: systematic scan patterns that walk every pin and never sample, lighting chosen per question — overhead for texture, grazing for height, side-angled for fillets — and findings photographed into the record as they surface. Then the chain that this chapter has been assembling all along gets formalized: optics claim what each package class actually shows them — everything on a QFP, the edges on a castellated QFN, almost nothing on a flankless DFN — the meters claim continuity, signatures, and the cold shorts check, power-on verification claims function, and the record states which tool verified which claim, because a verification claim is only as honest as its instrument's reach. The section closes with judgment: the acceptance question asked per the work's real stakes — the consumer device and the medical device do not share a pass bar — and the chapter's five benches folded into one discipline, eyes trained for the ball-grid chapters ahead, where optics will claim nothing at all and the chain will carry everything.

AdvancedMedium Risk23 min read

What You Will Learn

  • You will learn fillet reading as a formal skill — heel, toe, side, and end fillets per termination, judged by wetting angle rather than shine.
  • You will learn texture per alloy — lead-free's matte-normal finish, the demoted gloss standard, and the real cold joint's confession by shape.
  • You will learn the magnification discipline — systematic scans that never sample, lighting per question, findings photographed as they surface.
  • You will learn the inspection chain — optics claiming what each package class shows, meters claiming the rest, power claiming function, the record naming which.
  • You will learn acceptance judgment — the pass bar set by the work's real stakes, and the chapter's five benches closed into one discipline.

What You Will Be Able To Do

  • You will be able to read fillets per termination type and judge joints by wetting angle.
  • You will be able to read texture per alloy and name the real cold joint by its shape.
  • You will be able to run systematic magnified scans with lighting per question and findings photographed.
  • You will be able to run the full inspection chain and write the honest verification claim.
  • You will be able to set the acceptance bar per the work's stakes and sentence reworks accordingly.

Required Tools

  • A microscope with a movable light — the inspection's questions each need their own angle
  • The bench camera or scope camera — findings enter the record as they surface
  • The multimeter with fine probes and clips — the chain's second instrument
  • The board diagram — claims are made against nets, not against hope
  • A known-good reference board where one exists — signatures compare or they guess

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

The chapter ends by training the judge — every bench it built produces work that must be sentenced, and sentencing is its own craft (pad-planarity-and-preparation). Fillet reading formalizes first. Fillet reading per termination — heel on gull-wings, end on chip parts, side on castellations — judged by the wetting angle that tells the only structural truth: flowed-low-and-feathered wetted, tall-and-beaded did not (smd-joint-inspection). Shine gets demoted. Lead-free freezes matte and healthy — the gloss standard misreads modern joints daily, and the real cold joint confesses by shape and wetting inside its alloy's normal (digital-microscopes-and-usb-cameras). The magnification discipline turns looking into inspection. Systematic scans that never sample, lighting per question, findings photographed as they surface. The inspection chain gets formalized. Optics claiming what each package class shows, meters claiming the rest, power claiming function — with the verification claim in the record naming which tool verified what (electrical-verification-after-rework). And judgment closes the chapter. The pass bar set by the work's stakes, five benches folded into one discipline, eyes trained for the chapters where optics claim nothing. The fillets, the textures, the scans, the chain, the bar — the judge's bench entire.

Why This Matters

Every rework this chapter taught ends at an inspection, and the inspection's quality decides whether the bench learns the truth now or the customer discovers it later (smd-joint-inspection). This matters because the wetting angle is the only structural witness: a joint's strength lives in whether solder actually bonded its surfaces, the angle at the joint's edges is where that bond confesses, and every other visual cue — including the shine a generation was trained on — is circumstantial (pad-planarity-and-preparation). This matters because the gloss standard actively misleads: lead-free alloys freeze matte by nature, and the bench that rejects matte joints as cold reworks healthy work into actual damage — texture is evidence only inside its alloy's normal (digital-microscopes-and-usb-cameras). It matters because sampling is how the one bad pin ships: fine pitch hides a single failure among a hundred successes, and the scan that walks every pin exists because the glance that samples finds only what it happens to land on. It matters because false claims compound: a record that says 'visually inspected' over joints optics never reached poisons every later diagnosis that trusts it — the honest chain is what makes the record worth reading (electrical-verification-after-rework). And it matters because the ball-grid chapters are coming: BGA will hand optics nothing at all, and the chain this section formalizes — instruments asked only what they can answer, claims recorded per tool — is the entire inspection story there. Read the angle, respect the alloy, walk every pin, claim honestly — and the chapter's benches close into a discipline the rest of the volume stands on.

Required Prerequisites

  • Isopropyl and lint-free swabs — to strip residue before any read, because no fillet confesses through flux gloss (pad-planarity-and-preparation)
  • A printed scan-pattern card per package class — to make the every-pin walk a procedure instead of an intention (smd-joint-inspection)
  • Labeled photo storage per job — to file findings the moment the scope shows them (digital-microscopes-and-usb-cameras)
  • Probe tips and grabber clips kept sharp and clean — to let the chain's second instrument touch fine nets without slips (electrical-verification-after-rework)
  • The chapter's own reworked donors — QFP, QFN, DFN sites from the last four sections — to sentence real work with known histories (pad-planarity-and-preparation)
  • A board with factory joints in both alloys — to calibrate the eye on leaded gloss and lead-free matte side by side (smd-joint-inspection)
  • A microscope with a movable light and a camera — to drill lighting-per-question and photograph findings (digital-microscopes-and-usb-cameras)
  • A known-good reference board — to compare diode signatures instead of guessing them (electrical-verification-after-rework)
  • A deliberately flawed donor — one bridge, one starved joint, one cold joint seeded — to practice finding what is actually there

Real-World Applications

The judge's bench is where reputations are actually made, one sentencing at a time. A technician closing a media streamer's QFP replacement walks all four sides pin by pin under side light — heels climbing every step, one pin flowed low but unfeathered — and reflows that single joint before the meters ever run: the scan that never samples catching what the glance would have shipped (smd-joint-inspection). A bench about to reject a soundbar's amplifier joints as cold stops at the alloy question first — the board is lead-free, the matte is its healthy normal, and the shape reads flowed and feathered — the gloss standard overruled before it reworks good work into damage (digital-microscopes-and-usb-cameras). A shop closing a thermostat's flankless DFN swap writes the claim the package allows: seating verified optically, joints verified by continuity and diode signature against the reference board, function verified at power — three instruments, three claims, one honest record (electrical-verification-after-rework). And a repair café volunteer sentencing a coffee grinder's control board sets the bar where the stakes are — consumer device, dry environment, a slightly generous end fillet passed with a note — while the same volunteer's day job on avionics boards would have failed it: the acceptance bar read off the work, not off habit (pad-planarity-and-preparation). The failures this prevents: the one unfeathered pin shipped inside a hundred good ones, healthy matte joints reworked into cooked boards by a leaded-era eye, a flankless swap logged as visually verified, and a hobby-grade pass bar applied to work that flies.

Common Challenges

  • The eye samples by nature. Attention lands where contrast draws it, and the uniform row hides its one exception — the difficulty is the scan pattern as procedure: every pin, every row, in order, every time (smd-joint-inspection).
  • The gloss reflex dies hard. A generation learned shiny-equals-good, and lead-free broke the rule without telling the eye — the difficulty is re-calibrating texture inside each alloy's normal before judging anything (digital-microscopes-and-usb-cameras).
  • The honest claim feels like weakness. Writing 'joints verified electrically, not optically' reads like an admission — the difficulty is holding that it is the opposite: the claim that names its instrument is the only one worth trusting (electrical-verification-after-rework).

Safety Notes

Risk Level: Medium. Inspection is the bench's gentlest work — and its verdicts gate power, which is where the risk actually lives.

Professional Tips Before Starting

  • Calibrate on known-good before judging suspects. Five minutes on factory joints in the board's own alloy — the eye judges relative to a normal it has recently seen (smd-joint-inspection).
  • Name the question before moving the light. Texture wants overhead, height wants grazing, fillets want the side angle — aimless light answers nothing in particular (digital-microscopes-and-usb-cameras).
  • Write the claim as you verify, not after. Each chain stage logged when it runs — records reconstructed at the end inherit the memory's generosity (electrical-verification-after-rework).

The Judge's Bench — Fillets, Textures, Scans, The Chain, The Bar

Recap and Frame

Four benches built work; this one sentences it — and the sentencing skill is what the rest of the volume will lean on hardest (pad-planarity-and-preparation). The joint-reading foundation arrives from Volume 3. Fillet anatomy and the basic verdicts — promoted here to formal discipline with terminations, alloys, and instruments each given their own rules (smd-joint-inspection). The optics bench arrives configured. The lab volume chose the scopes and cameras; this section aims them with intent — lighting per question, scan per procedure, photograph per finding (digital-microscopes-and-usb-cameras). The meters arrive crowned. Electrical verification has closed every hidden-joint rework since the QFN bench — here it takes its formal place as the chain's second instrument, claiming what optics cannot (electrical-verification-after-rework). The site gate's photographs pay off. Every naked-site shot the chapter demanded becomes the before against which the after is judged (pad-planarity-and-preparation). What is new is the formalization. Reading as skill, scanning as procedure, claiming as honesty, acceptance as judgment — the inspection chain the ball-grid chapters will inherit whole. Hold the frame — foundation promoted, instruments aimed, chain formalized — and the chapter closes into one discipline.

The Fillets — Reading the Joint's Confession

A solder joint confesses its structure at its edges, and fillet reading is the formal skill of taking that confession (smd-joint-inspection). The wetting angle is the truth. Solder that bonded its surfaces flowed low and feathered onto them — a shallow, climbing meniscus — while solder that never wetted sits tall and beaded, a drop resting on a surface it never joined, whatever its shine suggests. Each termination has its required fillet. The gull-wing carries its load at the heel — the fillet climbing the lead's step is required, the toe's is not — chip parts confess at their end fillets, castellations and wettable flanks at the side fillet climbing the groove, and the reader knows which fillet each package must show before judging any of them (smd-joint-inspection). The fault gallery reads by shape. The starved joint's gap where fillet should stand, the bridge's web between neighbors, the disturbed joint's frozen ripple from movement mid-freeze, the cold joint's collapsed grainy bead — each a shape signature first, with texture only corroborating (pad-planarity-and-preparation). Volume reads against the pad. Too little solder starves the fillet; too much hides the lead and pools toward neighbors — the healthy joint shows its termination's outline through a modest, feathered coat (electrical-verification-after-rework). Angle first, required fillet per termination, faults by shape, volume against the pad — the reading entire. The joint cannot lie at its edges — the skill is knowing which edge to depose.

The Textures — Alloys, Gloss, and the Demoted Standard

Texture is real evidence that a generation learned to read wrong, and this bench re-calibrates it (digital-microscopes-and-usb-cameras). Leaded freezes glossy. The tin-lead eutectic solidifies fast and smooth — the mirror finish the old standard canonized, and still the healthy normal wherever leaded work continues (smd-joint-inspection). Lead-free freezes matte — and healthy. The modern alloys solidify slower and rougher by nature: a satin, lightly grained surface is their normal, and the bench that rejects matte as cold reworks good joints into genuinely damaged ones every day the reflex survives. The real cold joint confesses by shape. Beaded instead of feathered, collapsed and grainy beyond its alloy's normal, often paired with the wetting failure the angle already showed — texture corroborates inside the alloy's baseline; it convicts nothing alone (electrical-verification-after-rework). The disturbed joint reads as motion. Frozen ripples or a stepped surface — the part moved as solder solidified — and its fix is a clean reflow, not a texture debate (pad-planarity-and-preparation). And mixed-alloy work reads carefully. Rework that blended leaded into lead-free freezes somewhere between the normals — the calibration is the board's own factory joints, read before any suspect is judged. Leaded glossy, lead-free matte-and-healthy, cold by shape, disturbed by motion, mixed by the board's own baseline — the textures entire. The eye judges against a normal — the discipline is making sure it is this board's normal, not last decade's.

The Scans — Magnification as Procedure

Looking becomes inspection when it follows a procedure, and the procedure has three parts (digital-microscopes-and-usb-cameras). The scan walks; it never samples. Every pin on every side in a fixed order — row by row, corner to corner — because attention drawn by contrast finds only the failures that advertise, and fine pitch's one bad pin among a hundred does not advertise (smd-joint-inspection). The light answers one question at a time. Overhead light reads texture and residue; grazing light reads height and planarity — the site bench's instrument reused on joints; side-angled light throws fillets into profile — and the inspector names the question before moving the lamp, because aimless light answers nothing in particular (pad-planarity-and-preparation). Findings photograph as they surface. The suspect pin shot at magnification with its location logged — before any correction — so the record holds the evidence and not a reconstruction (electrical-verification-after-rework). Corrections re-enter the chain at the top. A reflowed pin is new work: its scan, its reads, its claim — the shortcut that re-inspects only the fixed pin misses what the fix disturbed. Walk everything, light per question, shoot on sight, re-enter after corrections — the scan discipline entire. Sampling is hope wearing a microscope; the walk is the inspection.

The Chain and the Bar — Honest Claims, Fitting Verdicts

The chain formalizes what the chapter has practiced: each instrument asked only what it can answer, and the record naming which answered what (electrical-verification-after-rework). Optics claim by package class. The QFP shows everything and optics may claim it all; the castellated QFN shows its edges and seating; the flankless DFN shows seating alone — the claim shrinks with the daylight, and writing it honestly is the discipline (electrical-verification-after-rework). The meters claim the hidden. Continuity to the diagram's nets, diode signatures against the known-good reference, the cold shorts check before any power — the second instrument's claims, logged as its own (smd-joint-inspection). Power claims function. First power watched through the meter, the device's own behavior as the last witness — function proven is the chain's final line, never its substitute for the earlier ones (digital-microscopes-and-usb-cameras). The bar sets by the stakes. Acceptance is judgment, not reflex: the consumer device in a dry living room and the instrument that rides an ambulance do not share a pass bar, and the sentencing reads the work's real life — environment, vibration, consequence — before calling generous-but-wetted good enough or failing it (pad-planarity-and-preparation). And the chapter folds closed. Daylight craft, hidden-joint humility, handling law, site gate, and now the judge — five benches, one discipline, ready for the chapters where optics claim nothing at all. Claims per instrument, verdicts per stakes, the chapter entire. The record that names its instruments is the only record worth reading — write that one.

Common Mistakes

  • Sampling the uniform row. The glance lands where contrast pulls — the walk is the procedure: every pin, fixed order, every time (smd-joint-inspection).
  • Rejecting matte as cold. The leaded-era reflex applied to lead-free work — texture reads inside its alloy's normal, calibrated on the board's own factory joints (digital-microscopes-and-usb-cameras).
  • Claiming visually what optics never reached. 'Inspected' written over a flankless belly — the claim names its instrument, and hidden joints belong to the meters (electrical-verification-after-rework).
  • Re-inspecting only the fixed pin. The correction's neighbors left unexamined — corrections re-enter the chain at the top, because the fix disturbs more than it touches (pad-planarity-and-preparation).
  • One pass bar for all work. The hobby standard on the avionics board, or the avionics standard billing the coffee grinder — acceptance reads the work's real stakes before sentencing.

Troubleshooting Guidance

The judge's bench troubleshoots its own verdicts: angle, alloy, scan, chain, bar. If a joint reads ambiguous: return to the angle — feathered or beaded is the structural question, and shine is not on the panel (smd-joint-inspection). If a whole board reads cold: the alloy question before any rework — matte lead-free is healthy, and the calibration is the board's own factory joints (digital-microscopes-and-usb-cameras). If the meters fail a net the optics passed: believe the meters and re-scan that region at higher power — the chain disagrees only when the eye missed something (electrical-verification-after-rework). If a correction spawns new failures: the fix disturbed its neighbors — re-enter the chain at the top for the whole affected side (pad-planarity-and-preparation). If function fails after a fully passed chain: the fault is past the joints — the diagnosis moves to the board's own benches, with the chain's record proving where it is not. If sentencing keeps stalling on borderline joints: the bar was never set — name the work's stakes first, and the borderline resolves itself. The throughline: every inspection dispute settles by returning to the instrument that owns the question — angle for structure, meters for the hidden, stakes for the verdict.

Verification & Testing Methods

Confirm the judge's bench before the chapter's work faces it:

  • [ ] I run fillet reading per termination — heel, end, side — and judge by wetting angle, with shine demoted to corroboration.
  • [ ] I read texture inside each alloy's normal, calibrated on the board's own factory joints, and name the real cold joint by shape.
  • [ ] I scan as procedure — every pin in fixed order, lighting per question, findings photographed as they surface, corrections re-entering at the top.
  • [ ] I run the inspection chain in order — optics per package class, meters for the hidden, the cold shorts check, watched first power.
  • [ ] I write the verification claim naming which instrument verified what, and I set the acceptance bar by the work's real stakes.

Then try the practice exercises below — sentencing work on the chapter's own donors; scenarios differ from the quiz.

Practice Exercises

  1. Calibrate and read fillets (6 minutes, factory board in each alloy, the chapter's donors, microscope). Spend the first minutes on factory joints — leaded gloss and lead-free matte side by side — then read the donors' reworked joints per termination: heels on the QFP, side fillets on the castellated QFN, end fillets on any chip parts, judging each by wetting angle and logging one line per package on what the angle confessed (smd-joint-inspection).
  2. Walk a full scan (7 minutes, the QFP donor, movable light, camera). Run the procedure end to end — every pin on all four sides in fixed order, overhead light for texture, side angle for fillets, grazing for height — photographing every finding as it surfaces with its location logged, and count what the walk found that a thirty-second glance had missed (digital-microscopes-and-usb-cameras).
  3. Run the chain on the hidden-joint donor (7 minutes, the QFN or DFN donor, multimeter, reference board). Claim what optics allow — seating, any flank fillets — then run the meters' half: continuity to identifiable nets, diode signatures against the reference, the cold shorts check — and bring the donor to watched first power only after the chain passes, logging each stage as it runs (electrical-verification-after-rework).
  4. Sentence and claim (5 minutes, all donors, desk). Write each donor's verification claim naming instrument per finding, sentence each against two different bars — a consumer bench's and a high-stakes bench's — noting where the verdicts diverge and why, and file the closing record the chapter's method would hand to the next technician (pad-planarity-and-preparation).

These core steps — the calibrated fillet read, the full walk, the honest chain, and the two-bar sentencing — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • Fillet reading is the formal skill of taking a joint's confession at its edges — the required fillet per termination, judged by the wetting angle: flowed-low-and-feathered wetted, tall-and-beaded did not, whatever the shine says (smd-joint-inspection).
  • Texture reads inside its alloy's normal — leaded freezes glossy, lead-free freezes matte and healthy — and the real cold joint confesses by shape, calibrated against the board's own factory joints (digital-microscopes-and-usb-cameras).
  • The scan walks and never samples — every pin in fixed order, lighting per question, findings photographed on sight, corrections re-entering the chain at the top (pad-planarity-and-preparation).
  • The inspection chain asks each instrument only what it can answer — optics per package class, meters for the hidden, watched power for function — and the verification claim names which tool verified what (electrical-verification-after-rework).
  • Acceptance is judgment: the pass bar reads the work's real stakes — environment, vibration, consequence — and the chapter's five benches close into the one discipline the ball-grid chapters will inherit whole.

Skills Learned

  • You can now read fillets per termination type and judge joints by wetting angle.
  • You can now read texture per alloy and name the real cold joint by its shape.
  • You can now run systematic magnified scans with lighting per question and findings photographed.
  • You can now run the full inspection chain and write the honest verification claim.
  • You can now set the acceptance bar per the work's stakes and sentence reworks accordingly.

Glossary Additions

  • fillet reading — the formal skill of judging solder joints at their edges, where structure confesses: the wetting angle as the primary witness — solder that bonded flowed low and feathered onto its surfaces, solder that never wetted sits tall and beaded, whatever its shine — read against each termination's required fillet: the heel climbing a gull-wing's step, the end fillets of chip parts, the side fillet in a castellation's groove. The fault gallery reads by shape first — starved gaps, bridged webs, the disturbed joint's frozen ripple, the cold joint's collapsed grainy bead — with texture corroborating only inside its alloy's normal, and volume judged against the pad: the healthy joint shows its termination's outline through a modest, feathered coat.
  • inspection chain — the ordered sequence that sentences a rework by asking each instrument only what it can answer: optics first, claiming what the package class actually shows — everything on a QFP, edges and seating on a castellated QFN, seating alone on a flankless belly — then the meters claiming the hidden: continuity to the diagram's nets, diode signatures against a known-good reference, the cold shorts check; then watched first power claiming function, never substituting for the stages before it. The chain runs cold before any power, corrections re-enter it at the top because a fix disturbs more than it touches, and its output is the verification claim — the record line naming which instrument verified what, which is the only kind worth reading.
  • verification claim — the record's honest statement of what was verified and by which instrument: seating verified optically, joints verified by continuity and signature, function verified at watched power — written stage by stage as the inspection chain runs, never reconstructed afterward. The claim's discipline is its refusal to overreach: 'visually inspected' is a false statement about joints optics never reached, and the claim that names its instrument's actual reach — shrinking as packages hide more — is what makes a repair record trustworthy to the next technician, to the callback diagnosis, and to the bench's own reputation. On the ball-grid work ahead, where optics claim nothing, the verification claim carries the entire inspection story.

Suggested Next Sections

Must read next:

  • What Is BGA? — Chapter 2 opens the territory this chapter trained for: the ball grid array — joints as solder spheres under the package, the geometry that ended optical inspection, and the fundamentals every reballing and replacement bench in this volume stands on.

Recommended: