What Does Good Welding Look Like? A Visual Inspection Guide

A fabricator once held up two brackets for a foreman: one bead looked like a row of evenly spaced fish scales flowing smoothly into the plate, the other looked like a lumpy stack of burnt toast. Only the second one failed inspection — and the first one cracked in service weeks later, because a weld can look beautiful and still be wrong. That gap between ‘looks good’ and ‘is good’ is exactly what this guide teaches you to close.

What Good Welding Actually Looks Like (and Why Looks Are Only Step One)

Two brackets, one foreman, one honest question. One bead looked like a neat row of evenly spaced fish scales flowing into the plate; the other looked like a lumpy stack of burnt toast. The lumpy one got rejected on sight. The pretty one cracked in service a few weeks later. Both were welded. Neither proves much by appearance alone. So what does good welding actually look like? The short answer, and the one this whole guide builds on: a good weld is judged on three layers at once — the visible surface, the internal soundness you cannot see, and conformity to the code referenced on the drawing. Nail only the first and you have decoration, not a joint.

The Three Layers of a Good Weld

Layer one: visual condition. This is what you see with your eyes and a flashlight — uniform bead width, even ripple spacing, smooth tie-in at the toes, no cracks, no trapped slag, no heavy spatter, correct leg size on a fillet. It is the fastest, cheapest check, which is why visual inspection is the baseline every other method builds on.

Layer two: internal soundness. A handsome cap can sit on top of an unfused root, a cluster of porosity, or a slag pocket. Penetration and fusion — whether the weld metal actually bonded to the base metal through the full joint — decide whether the part holds. This layer needs technique history or actual testing to verify.

Layer three: code compliance. “Looks fine to me” is not an acceptance criterion. Good means it meets the standard the contract names, most often AWS D1.1, the Structural Welding Code — Steel. The current AWS D1.1/D1.1M:2025 is the 25th edition of that code, released in April 2025, and it applies to structural steel joints 1/8 in (3 mm) thick and thicker with minimum specified yield strength up to 100 ksi (690 MPa). Outside that envelope, other documents take over.

Workmanship Is Not the Same as Quality

Workmanship is about how neatly the weld was laid down: consistent travel speed, clean prep, a steady hand. Quality is about whether the finished joint does what the design requires — strength, leak-tightness, fatigue life, corrosion resistance. A tidy bead on contaminated aluminum has poor quality despite decent workmanship, and a slightly rough-looking weld that meets every acceptance limit is a quality product. Confusing the two is why beginners over-grind attractive welds and why some shops ship pretty, weak work. The distinction matters most in Aluminium Welding Course: What You Learn, Costs & Careers settings, where learners are graded on both the look and the test coupon behind it.

Who This Guide Is For

This is written for beginner welders trying to read their own beads, hobbyists who have no inspector looking over their shoulder, shop leads who sign off on work, and inspectors-in-training learning what to hunt for. The purpose is practical and people-first: good welds keep structures standing and welders employed, and the gap between “looks good” and “is good” is where accidents and rework live. Whether you run MIG, stick, TIG, or flux-cored — and each process has its own visual signature, as Advantages of Stick Welding: Why SMAW Still Wins makes clear — the judging framework is the same.

Here is the road ahead. We will break down the anatomy of a quality bead, then go beneath the surface into fusion and penetration, walk through a defect-by-defect field guide, meet the codes that define “good,” look at how inspectors actually verify welds, troubleshoot the mistakes that ruin good work, compare visual expectations across processes, cover the shop conditions behind reliable welding, and finish with a checklist you can use on your own welds tomorrow morning.

Anatomy of a Quality Weld: What the Bead Should Tell You

Pick up a welding magazine or scroll through any welder’s forum and you’ll hear the same advice: a good weld looks like a stack of dimes. That phrase is a useful starting point, but it’s only a starting point. A quality bead is a story written in metal, and every feature of it tells you something about what your hands, your settings, and your technique were doing six inches ago. Learn to read that story and you can catch problems before a code inspector ever walks into your shop.

The Visual Signature of a Sound Weld

Run your eye along a properly made fillet or groove weld and you should notice a short list of traits, consistently present from the arc strike to the arc break:

• Uniform width. The bead maintains roughly the same width start to finish. Width wander is travel speed wander, and it usually means the welder was adjusting body position or fighting fatigue mid-pass.
• Even ripple spacing. The solidified ripples are evenly spaced and consistent in shape, the fingerprint of a steady travel speed and a stable arc length.
• Smooth tie-in at the toes. The edges of the weld blend into the base metal without a sharp notch, a cold step, or an undercut groove. This is the single most important visual feature, and we’ll get to why in a moment.
• Slight or specified convexity. A small amount of crown on a fillet is normal. Flat-to-slightly-convex profiles are the target; a humped, ropey bead signals travel speed too slow or current too low.
• Clean, slag-free surface. No trapped slag, no oxide film, no heavy spatter clinging to the surrounding plate.
• Proper wetting on both members. The weld metal wets into both plates equally rather than sitting on top of one side.

Those ripples deserve a closer look. Each ripple marks one solidification event as the puddle advances. If the spacing widens, the welder sped up; if it tightens, they slowed down. An inconsistent arc length changes the heat and the bead width in the same rhythm. So when you see a bead with irregular ripples, you’re not just seeing a cosmetic flaw — you’re seeing a record of inconsistent heat input, which also means inconsistent penetration.

Why the Toes Matter Most

Most fatigue cracks in welded structures don’t start in the middle of a weld. They start at the toes, where the weld face meets the base metal. A sharp, poorly blended toe acts like a built-in notch, concentrating stress exactly where cyclic loading likes to pry. A smooth, well-feathered tie-in spreads that stress out and dramatically improves fatigue life.

The classic toe defect is undercut: a groove melted into the base metal right beside the weld toe and never filled by weld metal. It’s a textbook fatigue initiator, and it isn’t a matter of opinion — AWS D1.1 Table 8.1 sets explicit limits on undercut depth for structural steel, with stricter limits under cyclic loading. The same table governs weld profile, so “it looks a little rough” sometimes crosses the line into “it’s a reject.”

Size Is an Acceptance Criterion, Too

Appearance isn’t only about texture — it’s about geometry. A fillet weld has a required leg size (and often a throat thickness) called out on the drawing, and an undersized weld fails no matter how pretty the ripples are. Groove welds have acceptable reinforcement limits as well: too little reinforcement can mean a starved joint, too much creates a stress riser at the toe. Undersized welds and profile problems are among the most commonly cited issues in AWS Inspection Trends reports, which is why inspectors carry a weld gauge, not just their eyes.

What to Measure, Not Just Admire

When you judge your own work, look at it the way a certified inspector would: check leg size with a gauge, measure undercut with a rule or gauge against the table limits, look at the toes under good light, and check the start and stop for craters. If you’re welding aluminum, remember that appearance standards shift with the process — a Aluminum MIG Welding guide is useful for reading bead profiles on non-ferrous joints, and cleaning discipline matters even more than on steel, as covered in Aluminum Cleaner for Welding.

The takeaway: uniformity, tie-in, and correct geometry are the three pillars. Master them visually, then verify what your eyes can’t confirm — because a bead that tells a consistent story on the surface is the proof of a consistent process underneath.

Beyond the Surface: Fusion, Penetration, and What Your Eyes Can’t See

A weld bead can sit on top of a joint like a plump caterpillar and never become part of it. That is the uncomfortable truth about visual welding: the face of the bead shows the welder’s hand, not what happened at the bottom of the joint. Two ideas separate “looks tidy” from “actually joined,” and confusing them is why beginners misjudge their own work.

Face Fusion vs. Root Fusion

Face fusion is the bond between the visible weld face and the base metal along the weld toes. You can usually see it: a smooth tie-in where the bead wets into the plate, with no visible line or notch at the toe. Root fusion is the bond at the very bottom of the joint, where the first pass meets the innermost faces of the groove. You generally cannot see it at all once the joint is capped.

Both matter, but they fail differently. Poor face fusion shows up as cold lap or a visible unfused toe line. Poor root fusion stays invisible until the part is loaded. This is precisely why AWS structural guidance treats fusion to the base metal and fusion between passes as distinct visual acceptance checks, alongside crater condition, weld profile, undercut, and piping porosity. A visual inspection walkthrough of AWS D1.1 acceptance criteria makes the same point: visual inspection is thorough where it can reach and silent where it cannot.

Full Penetration vs. Partial Penetration

• Full penetration (CJP) means the weld metal runs completely through the joint thickness, root face included. The weld is as strong as the plate across its whole depth.
• Partial penetration (PJP) is a deliberate design choice. The joint is only fused to a specified depth, and the drawing’s effective throat defines what is acceptable.

The trap is that an intentionally partial joint and an accidentally unfused joint can look identical from the outside. If the drawing calls for complete joint penetration and the root is bridged rather than fused, you have incomplete joint penetration — a discontinuity that limits the load path and turns the unfused root tip into a crack starter. This is why inspection guidance under D1.1 requires verifying root opening, groove angle, root face, and member alignment before welding, plus a root pass inspection while the joint is still open. A joint fitted too tight or misaligned makes full penetration physically impossible, no matter how skilled the welder is.

What Actually Drives Penetration

Penetration is not a personality trait; it is a physics outcome controlled by four variables:

• Heat input: voltage, amperage, travel speed, and process. Too little heat and the arc never reaches the root; too much and you burn through. On heat-sensitive materials such as aluminum, the window is narrow enough that settings choices deserve their own study; see Aluminum MIG Welding: The Complete Practical Guide.
• Joint prep: root opening, groove angle, and root face determine how easily the arc reaches the bottom. Cleanliness matters equally, since oxide and contamination block wetting; see Aluminum Cleaner for Welding: How to Prep Aluminum Right.
• Travel angle and arc length: a slight push or drag angle directs the arc into the joint, while a long, wandering arc scatters heat sideways instead of down.
• Pass sequence: the root pass does the structural work; subsequent passes only build size and profile.

Craters, Cold Lap, and Piping Porosity

Two defects here sit right on the line between surface and interior. A crater left unfilled at the termination shrinks as it solidifies and can crack — a crater crack is still a crack, and cracks are never acceptable workmanship. Cold lap (overlap) is molten metal that rolled over solid base metal without fusing, often caused by travel that is too fast or an arc held too far back.

Piping porosity is the deceptive one: a gas bubble rises through the still-liquid trailing edge of the solidifying pool and draws a vertical channel from the weld face toward the root. It may show only as a tiny surface pore, yet it reduces the joint’s cross-section along its length.

Why NDT and Destructive Tests Exist

Since incomplete fusion and incomplete penetration live inside the joint, visual acceptance criteria are always paired with other methods when the application demands it. Dye penetrant and magnetic particle testing chase surface-breaking flaws; ultrasonic and radiographic testing look through the volume. Destructive testing — sectioning a coupon and bending or breaking it — gives an undeniable picture of root fusion and anchors procedure qualification. In practice, the required verification level is written in the contract documents, not decided at the bench by the welder or the inspector.

The Defect Field Guide: Reading Cracks, Porosity, Undercut, Overlap, and Inclusions

Every welder learns to describe a bead. Learning to diagnose one takes longer, because defects hide in a toe, in a root, or inside a pass that looks perfect from the outside. This is the pocket reference: what each common defect looks like, why it forms, how the codes treat it, and what to do about it.

Two words first, because they carry legal weight. A discontinuity is any interruption in the normal structure of a weld; it becomes a defect only when it exceeds the acceptance criteria of the governing code. That is exactly what the AWS D1.1 visual acceptance criteria exist to do: turn “looks bad to me” into a measurable standard.

Cracks: the defect that never gets a second chance

Start here, because cracks are the only category with no gray area. Under AWS D1.1 visual acceptance criteria, any crack of any size, in any location, in any direction, is cause for rejection — longitudinal, transverse, toe, root, and the star-shaped crater cracks left behind at the stop. A crater crack looks harmless; metallurgically it is a notch waiting to grow. Causes include restraint and rapid cooling, hydrogen from damp coatings or contaminated metal, and simply letting the arc stop without filling the crater. Repair means removing the crack entirely — carbon arc gouging with a scratch rod is the standard method, as covered in our guide to air arc welding and carbon arc gouging — then re-welding with corrected settings. Never weld over a crack: the overlay heat can extend it deeper into the base metal.

Porosity: what the surface shows and what it hides

Porosity is trapped gas that never escaped the puddle. It appears as surface pinholes, as worm-track piping porosity running into the weld, or as clusters that follow a stop-start or a bad gas event. Acceptance treats them differently: scattered and elongated porosity is generally a workmanship and cosmetic issue with defined size and spacing limits, while piping porosity reduces the effective weld cross-section and strength, so AWS D1.1 gives it its own, tighter limits. Four triggers cover most cases: welding on rusty or painted surfaces, moisture trapped in lap joints, poor gas shielding from wind or wrong flow rate, and contaminated filler or base metal. One subtle tell in GMAW: elongated porosity can indicate electrode stickout is too short, destabilizing the arc and gas envelope. Aluminum punishes every shortcut hardest, which is why proper aluminum cleaning and prep and correct technique, detailed in our aluminum MIG welding guide, matter so much.

Undercut and overlap: two edge failures in opposite directions

Undercut is a groove melted into the base metal at the weld toe and not filled back in — a notch, and a classic fatigue stress raiser. Overlap, or cold lap, is the opposite: weld metal rolls over the base metal without fusing to it. Both usually trace back to travel speed, current, and work angle. Shallow overlap can be ground back to sound metal; undercut needs a genuine repair pass so new filler can bond into the notch.

Inclusions and incomplete fusion: the failures you can’t always see

Slag inclusions are trapped flux left between passes, common in SMAW and FCAW when interpass cleaning is skipped or the bead profile creates a pocket the arc can’t reach. Incomplete fusion (lack of fusion) is weld metal that never bonded to the base metal or the previous pass; on the surface it shows as a cold, unfused toe or a visible gap between bead and plate. Both are the reason volumetric NDT exists, and why high-reliability fields such as aerospace TIG welding set far stricter limits than general structural work.

Profile problems: reinforcement, burn-through, misalignment, and spatter

• Excessive reinforcement: a tall cap is not extra strength. It creates a sharp toe transition that concentrates stress, and AWS D1.1 caps groove reinforcement at 1/16 in (1.6 mm) above the base metal surface for typical non-tubular work.
• Burn-through: current too high, travel too slow, or a root gap too big. Fill it or gouge and re-weld; never scab over the hole.
• Misalignment (high-low): poor fit-up forces the load path to jog, adding bending where the design assumed tension. Thin sections are unforgiving — a reason alloy wheel welding demands such careful joint prep on cast pieces.
• Spatter: rarely rejectable on its own, but a splattered bead usually signals voltage, travel speed, or polarity problems that may be hiding elsewhere.

The ten defects at a glance

Defect	What you see	Main cause	Typical verdict
Crack (any)	Line or crater star	Restraint, hydrogen	Always rejectable
Surface porosity	Pinholes on face	Rust, paint, shielding	Size/spacing limits
Piping porosity	Worm tracks inward	Moisture, short stickout	Tighter D1.1 limits
Undercut	Notch at the toe	Fast travel, high heat	Reject beyond limits
Overlap	Rolled unfused edge	Low heat, slow travel	Reject if unbonded
Slag inclusion	Trapped between passes	No interpass cleaning	Reject in structural joints
Incomplete fusion	Cold gap at toe/root	Low heat, tight fit-up	Reject; often internal
Excessive reinforcement	Tall, overbuilt cap	Slow travel, extra filler	Reject over 1/16 in
Burn-through	Hole through the joint	High heat, big gap	Repair required
Misalignment	High-low between plates	Poor fit-up	Reject beyond tolerance

The goal is not memorizing code clauses on day one. It is catching the notch at the toe before the inspector does. When you want the detail, the welding defects acceptance criteria reference and the AWS D1.1 visual inspection guide are both worth one proper read — not a hurried skim during an argument at the fit-up table.

Who Decides What ‘Good’ Means? Codes, Standards, and Acceptance Criteria

Here is the truth that trips up nearly every beginner welder: there is no universal definition of a “good weld.” A bead that passes on a backyard handrail might be rejected outright on a bridge girder or a pressure vessel. The referee is not the welder, the foreman, or even the inspector’s taste. It is the code written into the contract documents for the job in front of you.

Knowing which code governs your work, and where its acceptance limits live, separates someone who lays down nice-looking beads from someone trusted on structural work. These are the documents that decide what “good” means.

AWS D1.1: The Structural Steel Rulebook

In the United States, structural steel welding runs on AWS D1.1, the Structural Welding Code – Steel. Weld buildings, bridges, or ordinary steel frames and D1.1 is almost certainly the contractually referenced standard, and its acceptance criteria table is where disputes get settled.

The key document inside D1.1 is Table 8.1, Visual Inspection Acceptance Criteria, which sets measurable limits for cracks, undercut, weld profile, porosity, and reinforcement. D1.1 provides two separate columns of limits: one for statically loaded connections and one for cyclically loaded connections. A joint that experiences repeated loading and unloading (cranes, bridges, machinery frames, anything that vibrates) falls under the stricter cyclic criteria, because fatigue cracks initiate at small surface discontinuities that static loading would tolerate. Undercut acceptable under static criteria can be an automatic reject under cyclic criteria.

D1.1 also imposes documentation duties many shops overlook. The code requires inspectors to mark inspected and accepted parts or joints, and to generate an inspection report for every weld inspected. An unmarked, undocumented weld on a D1.1 job is not a finished weld, however perfect the ripples look. For a closer look, this D1.1 acceptance and inspection overview and this visual inspection guide (PDF) walk through Table 8.1 criterion by criterion.

The Discontinuity vs. Defect Line

Codes force a precise vocabulary. A discontinuity is any interruption in the normal weld structure: a small porosity cluster, a shallow undercut, an off-profile bead. A defect is a discontinuity that exceeds the acceptance limits of the governing code, or one that makes the weld unfit for service.

A weld can carry minor discontinuities and still be code-compliant and acceptable; acceptance criteria are tolerances, not aspirations. Conversely, a flawless-looking weld is still a defect if it lacks required penetration or fusion, because Table 8.1 cannot judge what your eyes cannot see. Acceptance criteria references make the same point across industries: limits, not perfection, are the standard.

Pressure Work: ASME Boiler and Pressure Vessel Code

Anything holding pressure at elevated temperature or hazardous contents plays by a different book: the ASME Boiler and Pressure Vessel Code. Section IX governs the qualification of welding procedures and welders (the WPS/PQR/WPQ system), while Sections I, VIII, and the applicable B31 piping codes carry the examination and acceptance requirements. This world is stricter, fully documented, and non-negotiable. In practice, Alamo Welding & Boiler Works: Services & Boiler Expertise shows the service side of boiler fabrication.

ISO 5817: Quality Levels B, C, and D

Outside the US, or on international projects, the common reference is ISO 5817, which defines three quality levels for fusion-welded joints in steel:

• Level B (Highest): fewest and smallest imperfections allowed. Used for highly stressed cyclic and dynamic joints.
• Level C (Medium): the general-purpose industrial level for most static structures.
• Level D (Lowest): tolerates the most imperfection. Suitable only for lightly loaded or non-critical work.

The drawings state which level applies; each maps to specific limits on cracks, porosity, undercut, and misalignment. Aerospace work sits at the demanding end of this spectrum; Aerospace Welding: Processes, Standards, and Careers Guide explains how tight those requirements become when flight hardware is involved.

Who Actually Applies the Criteria

Inspection level is a design and contract input, not an inspector’s judgment call. The contract documents specify what examination method is required, at what extent, and to which acceptance criteria. Inspectors get frustrated when welders argue acceptance based on “how it looks” rather than the referenced table: visual appeal is evidence, not the criterion. Training programs such as the Aluminium Welding Course: What You Learn, Costs & Careers increasingly emphasize code literacy alongside torch skills.

Certification has a defined scope, too: an AWS Certified Welding Inspector is qualified to a specific code, material group, and position range, and accepts work only inside that scope. Outside it, they escalate rather than decide.

Safety Standards: ANSI/AWS Z49.1

One more document belongs here: ANSI/AWS Z49.1, Safety in Welding, Cutting and Brazing, the consensus safety standard that OSHA’s welding standard 1910.252 references. It covers ventilation, fire prevention, cylinder handling, and PPE. It never judges bead appearance, but it governs the conditions that make acceptable work possible, which is why our later standards-focused welding discussions treat it as foundational.

Bottom line: good welding is code-compliant welding, verified by the required examination, and documented. Appearance is one input; the contract’s code is the decision.

How Welds Are Verified: Visual, Dye Penetrant, Magnetic Particle, Ultrasonic, and Radiographic Testing

Seeing a bead is one thing; proving it sound is another. In a working shop, “it looks fine to me” is never the last word, because codes require evidence that the weld meets the contract requirements. That verification happens through a hierarchy of examination methods, each with a specific job, cost, and place in the sequence.

Visual Testing (VT): The Baseline for Every Weld

Visual inspection always comes first, and nothing replaces it. Before any instrument is used, an examiner confirms dimensions, profile, workmanship, cleanliness, and joint preparation against the drawing and applicable code. If a weld fails visually, there is little point paying for ultrasonic testing to find internal flaws in a joint already being rejected.

Under AWS D1.1, Structural Welding Code – Steel, Section 8 places real obligations on this process: the contractor must give advance notification of the inspector before operations subject to inspection, and the inspector examines the work against the code’s requirements and records acceptance with distinguishing marks or another agreed recording method. For the detailed mechanics of the step itself, this AWS D1.1 visual weld inspection guide is worth reading.

The Surface Methods: Dye Penetrant and Magnetic Particle

When a surface-breaking flaw is suspected but the eye can’t confirm it, two liquid or particle methods take over.

• Penetrant Testing (PT) uses a dyed liquid that seeps into open cracks, is wiped off the surface, and is drawn back out by a developer, leaving a visible indication. PT works on any nonporous material, including stainless steel and aluminum, which makes it common on Aluminum MIG Welding joints where you can’t run magnetic particle. Its limit is that it only finds open, surface-breaking flaws.
• Magnetic Particle Testing (MT) magnetizes ferromagnetic parts and sprinkles fine iron particles over the area; particles cluster where a crack interrupts the magnetic field. MT catches surface and near-surface cracks in steel and other magnetic metals, and it’s faster than PT in a production shop, though it won’t work on nonmagnetic materials.

The Volumetric Methods: Ultrasonic and Radiographic Testing

Internal flaws require energy that can travel through the metal. Ultrasonic Testing (UT) bounces high-frequency sound waves through the weld and reads the echoes; reflectors such as lack of fusion, slag, or cracks show up as signals on the display. It is deep-reaching, portable, and generates no radiation, but it is heavily operator-dependent and produces a signal interpretation rather than a permanent visual record.

Radiographic Testing (RT) passes X-rays or gamma rays through the joint onto film or a digital detector. Denser material absorbs more radiation, so cracks, voids, and inclusions appear as darker or lighter areas on the image. Radiography gives a permanent, reviewable record and is excellent at showing porosity and slag, but it is comparatively slow and expensive, requires controlled areas for radiation safety, and can miss tight planar cracks that lie parallel to the beam.

Radiographic and ultrasonic methods exist precisely because visual inspection cannot see volumetric discontinuities hidden inside the weld, as acceptance-criteria references make clear.

Who Does the Checking: Qualifications and Timing

Inspection isn’t just a method, it’s a qualified person. The common professional benchmark is the AWS Certified Welding Inspector credential, which requires documented welding experience, formal training, and passing both a code-fundamentals and a practical examination. Engineers may supplement this with Level II or III NDT certifications for UT, RT, PT, or MT specifically.

Inspection happens in three windows:

1. Pre-weld — base metal condition, fit-up, root opening, alignment, cleanliness, preheat, and WPS availability.
2. In-process — travel speed, interpass temperature, pass cleaning, and technique verified while the weld is still correctable.
3. Post-weld — visual acceptance first, then NDT where the code or drawing specifies it.

The sequence follows economics: VT is inexpensive and catches enough problems to justify running it on 100% of welds. PT or MT adds moderate cost for surface-critical joints. UT and RT are reserved for joints where internal soundness matters — pressure boundaries, heavy structural members, cyclically loaded details — because they cost more, take longer, and may require pulling the component out of production. Boiler and pressure vessel work such as Alamo Welding & Boiler Works builds this escalation straight into the workflow.

For pressure equipment, the governing reference shifts from AWS to the ASME Boiler and Pressure Vessel Code, and the examination extent is still defined by the governing code and the contract documents — never chosen at the inspector’s discretion.

Next, we cover the mistakes that quietly undermine good welds even when the inspection plan itself is sound.

Common Mistakes That Ruin an Otherwise Good Weld — and How to Fix Them

Here is the uncomfortable truth about most rejected welds: the welder could actually run a decent bead. The trouble started earlier, with a joint that was dirty, misaligned, or shielded badly, or ended later, with an arc that was snapped off instead of filled in. Good welding is a chain, and it only takes one weak link to break it. The table below reads like a shop-floor triage sheet: what you see, why it happened, what to do about it now, and how to stop it on the next joint.

Symptom, Cause, Fix, Prevention

What you see	Root cause	Fix on this weld	Prevent it next time
Blistered porosity, bubbles under the skin, sooty surface	Welding over rust, paint, oil, mill scale, or moisture	Stop, clean back to bright metal, gouge out the affected area, re-weld	Grind and degrease the joint before arc start; follow the OSHA 1910.252 cleaning requirements
Rough, erratic bead or excessive spatter despite correct amps	Wrong polarity or machine settings for the electrode and process	Verify polarity against the electrode box, re-test on scrap, re-weld	Check the WPS or electrode data sheet before striking an arc; the Advantages of Stick Welding guide covers polarity basics for SMAW
Narrow, tall, frozen-looking bead with poor wetting	Travel speed too fast or weave too wide for the heat input	Slow down, add a pass where throat thickness falls short	Match travel speed to amperage; keep weaves narrow, especially on thin plate
Long arc, hissing, spatter, wandering bead	Excessive stickout on MIG, or arc length too long on stick	Reset stickout, restart the pass	Hold roughly 3/8–1/2 in. on solid wire and about 1 in. on flux-cored
Porous, discolored, or icy-looking bead on aluminum	Lost shielding gas from wind, wrong gun angle, or an empty cylinder	Shield the area, purge the line, remove the bad weld	Baffle drafts and verify flow; Airgas Welding Supplies resources explain gas handling and cylinder setup
Slag trails, lack of fusion between beads, cold spots	No wire brushing or slag removal between passes	Chip, brush, and grind the contaminated zone before the next pass	Make inter-pass cleaning part of the pass sequence, not an afterthought
Crater crack at the stop, star-cracking in the shrinkage pipe	Arc pulled away instantly with no dwell or back-step	Gouge out the crater and re-start past it	Hold position for a beat at the stop and step back to fill the crater
Uneven profile, off-size legs, cold root	Excessive root opening, poor alignment, or wrong groove angle	Correct the fit-up before welding continues; re-work completed defects	Tack and shim to the drawing; fit-up practice is exactly what the AWS D1.1 visual inspection guide examines
Undercut, brittle starts, hydrogen cracking after the shift	Cold or damp work, wet electrodes, low preheat	Preheat per procedure, dry consumables, re-inspect after cooling	Check base-metal temperature and store low-hydrogen electrodes in a holding oven

Fit-Up Is Not a Filler Problem

Joint fit-up issues rarely announce themselves while you are welding. A root opening that is too wide, a mismatched joint alignment, or a groove angle that does not match the drawing shows up later as profile defects, undersized welds, and incomplete fusion at the root. AWS inspection practice treats root opening, groove angle, root face, joint alignment, and whether you ran a stringer or a weave as documented, inspectable items, so the fix belongs at the fit-up table, not at the grinding wheel. If a gap will not close, weld it as designed and sized, then flag it for the inspector rather than packing it with extra passes that hide the geometry problem.

The Crater Habit

Crater cracks are the most preventable defect in this entire list, and they form in the last half-second of a weld. When the arc is snapped off, the shrinking pool solidifies from the outside in and tears open in the center. A momentary dwell, a slight back-step, or a run-on/run-off tab fills the crater with metal that can shrink harmlessly. Because cracks are never acceptable under common weld acceptance criteria, one crater crack can send an otherwise perfect bead back through gouge-and-repair.

None of these fixes require better talent. They require a habit: clean the joint, check the fit-up, confirm the settings, hold the arc steady, fill the crater, and clean between passes. Do that consistently and the repair grinder stops being part of your routine.

What Good Welding Looks Like in Each Process: MIG, TIG, Stick, and Flux-Cored

Strip away the process names and a sound weld announces itself the same way every time: smooth, uniform, clean, and consistent from the start of the bead to the stop. What changes is what “uniform” looks like under each arc. MIG lays down a continuous, gently rippled bead. TIG produces tightly spaced ripples that resemble a stack of coins. Stick welds carry broader ripples under a layer of slag you chip off before you can judge anything at all. Knowing the expected texture for your process is what lets you spot trouble in seconds, because porosity, cracks, and undercut are never signs of quality in any process, on any metal, in any position.

GMAW (MIG): Smooth, Even, and Almost Spatter-Free

A good MIG bead is the one most beginners recognize: a continuous ribbon of metal with fine, even ripples, consistent width through the whole length, and a smooth tie-in where the weld toes melt into the base metal. The surface should be nearly free of spatter, with only light, easily brushed-off beads if any. Little balls of spatter stuck hard to the plate usually point to a root cause: voltage and wire speed out of balance, contact tip too far from the work, contaminated base metal, or gas coverage disrupted by a draft or wrong gun angle.

Stickout matters more than most people expect. For solid wire it’s typically around half an inch, and drifting longer changes the amperage at the arc, which changes penetration and profile. You’ll also hear a good MIG arc more than you’ll see it — a steady frying-bacon crackle, not a harsh stutter or a soft hum. When the bead necks down or wanders in width, the first suspects are travel speed and hand position, not the wire itself. For aluminum specifically, the prep and machine setup demands are higher, and Aluminum MIG Welding: The Complete Practical Guide walks through the profile you should be aiming for.

GTAW (TIG): The Stack of Dimes — and What It Really Means

On steel and stainless, a good TIG weld shows tightly spaced, evenly sized ripples, flat to slightly convex, with toes that wet in smoothly rather than standing proud. Those “dimes” are the visual record of dabbed filler or a pulsed current: each dab freezes into a ripple. Vary your timing and the ripples go uneven; vary your heat and the width wanders. That’s why the bead is such an honest readout of the operator’s rhythm.

Aluminum TIG looks deliberately different. The toes run wider and flatter, the profile is smoother, and pronounced coin ripples are less the goal than a fluid, wetted-in appearance. Cleaning dominates everything here, since the oxide layer on aluminum melts far higher than the metal underneath — poor cleaning shows up as dirty, sooty toes and erratic bead edges. On the rejection side, tungsten inclusions appear as dark specks or a ragged line where the electrode touched the puddle, and cathodic etching shows as a frosted, milky band beside the weld on aluminum. TIG work in demanding fields is a discipline of its own; see Aerospace TIG Welding: Standards, Techniques, and Training for how tightly that appearance is specified, and Aluminum Cleaner for Welding: How to Prep Aluminum Right covers the cleaning step that makes good-looking aluminum welds possible.

SMAW (Stick): Uniform Ripples and Slag That Lets Go

A good stick weld has even ripples of consistent spacing, a bead width that stays put for the whole run, and toes blended into the plate with no biting notch left behind. The arc length is written into the bead: too long and the bead flattens, widens, gets spattery and may show porosity; too short and the bead climbs high, narrow, and rough with trapped slag. Slag removal is the second test. Quality slag lifts off in clean curls with light tapping; if you’re grinding and picking at it, the weld underneath often has slag islands or the flux and current were mismatched. There’s a reason this process still wins certain jobs outright — Advantages of Stick Welding: Why SMAW Still Wins explains where it stays unbeatable.

FCAW and Gas-Shielded Flux-Cored: Reading the Slag

Flux-cored arcs hide the bead under slag, so your read starts with how that slag behaves. It should cover the bead evenly and peel or lift cleanly. Stubborn, patchy slag means trouble with travel speed, heat, or coverage.

• Self-shielded FCAW generates its own shielding from the flux, so it tolerates wind and drafts better, but it runs dirtier: heavier slag, more smoke, and more cleaning between passes and before inspection.
• Gas-shielded (dual-shield) FCAW needs an external gas, hates drafts, and gives back a cleaner bead, lighter slag, less smoke — plus the smooth, tight profile you’d recognize from MIG. It’s also less tolerant of skipped interpass cleaning.

Both variants can produce excellent welds, but “clean” means something different in each. If you’re working with aluminum flux-cored wire, the slag chemistry and cleaning demands shift again; Aluminum Flux Core Welding: What Actually Works in 2026 and Aluminum Flux Welding: How It Works and When to Use It cover where the process genuinely performs and where it doesn’t.

Process	Good bead looks like	Common visual red flag
GMAW (MIG)	Smooth, fine ripples, consistent width, minimal spatter	Beaded spatter, wandering width, uneven toes
GTAW (TIG)	Even stacked ripples on steel; flat, wetted toes on aluminum	Uneven dimes, dark tungsten specks, frosted band
SMAW (Stick)	Uniform ripples, clean slag release, blended toes	Slag that won’t lift, flat wide bead, undercut
FCAW	Even slag coverage that peels cleanly, consistent profile	Patchy slag, smoke-stained rough surface, porosity

Whichever arc you’re running, texture is only the first read. The final verdict comes from the governing acceptance criteria, and AWS publishes the process and quality standards the industry writes contracts against. A weld that looks right for its process and meets those criteria is the only combination worth calling good.

Safety, Ventilation, and the Shop Conditions Behind Reliable Welds

Ask most welders what makes a bead look good and you will hear about amps, travel speed, and rod angle. Almost nobody first mentions the room they are welding in. Yet the shop is a silent co-author of every weld you make. A welder breathing fumes at the end of a long shift does not see the same bead a fresh, clear-headed welder sees. Bad lighting hides the puddle. A missing cylinder strap turns a routine afternoon into an emergency. Safety and weld quality are not two separate conversations; they are the same conversation.

Why Unstable Conditions Produce Unstable Welds

Welding precision depends on the welder’s nervous system. Steady hands need oxygen, clear vision, and comfort. Welding fumes — a mixture of metal oxides, flux decomposition products, and gases — displace fresh air and cause symptoms long before anyone reaches a medical threshold: eye irritation, headaches, dizziness, and a foggy sense of timing that shows up as inconsistent arc length and erratic travel speed.

Visibility matters just as much. A helmet lens that is too dark for the amperage, a workpiece shadowed by poor shop lighting, or smoke lingering in a still-air shop will all push a welder to rely on habit instead of observation. That is exactly how the beautiful-looking, badly fused weld from our earlier example gets made. Fatigue from noise, heat, and dehydration compounds it. By hour nine, nobody is running the tight, even ripples they ran at hour two.

The practical takeaway: if your welds are deteriorating across a shift, look at your conditions before you blame your technique. Serious fabrication shops treat proper aluminum cleaning and prep and fume control as one discipline, because both remove variables that hide inside the finished joint.

Ventilation: The Numbers OSHA Actually Uses

The governing U.S. rule for welding, cutting, and brazing is OSHA 29 CFR 1910.252, and its ventilation requirements are specific enough to plan a shop around.

Those thresholds are not suggestions. A two-car garage with three welders running exceeds almost every trigger on the list. Local exhaust ventilation at the arc — a fume arm positioned close to the puddle — captures fumes far more efficiently than general room dilution, which is why it is preferred wherever the process allows it.

PPE: The Baseline Under 29 CFR 1910 Subpart I

PPE is the last line of defense, not the first, but it is legally defined. Under OSHA 29 CFR 1910 Subpart I, welding PPE includes:

• Helmets and eye protection — ANSI-approved helmets with the correct filter lens shade for the process and amperage, plus safety side shields. An incorrectly shaded lens is a vision problem and a quality problem at once.
• Flame-resistant clothing — leather or treated cotton jackets, sleeves, and aprons. Synthetic fabrics melt into skin, and they melt into ruined joints when sparks land in the wrong place.
• Gloves — process-specific: heavy stick gloves for SMAW, thinner and more dexterous gloves for GTAW.
• Safety footwear and hearing protection — closed-toe, metatarsal-capable boots, plus ear protection in any shop where chipping and grinding run all day.

Cylinder Handling, Fire Watch, and Gas Supply Discipline

Compressed gas storage is where shops get hurt. OSHA 1910.252(e) requires oxygen cylinders and fuel-gas cylinders to be separated by at least 20 feet, or by a fire-rated barrier at least 5 feet high with a half-hour fire rating. Cylinders stay capped when not in use, secured upright against rolling, and moved with caps on and carts — never dragged or lifted by the cap. Suppliers such as Airgas welding supply operations publish handling guidance that mirrors these rules, and it is worth reading even if you buy local.

Fire watch is the other half. Hot work rules require a cleared radius around the work, combustibles removed or protected, and a fire watch with an extinguisher maintained for a period after work stops — smoldering sparks hide in insulation and wall cavities long after the arc is off.

How Safety Failures Correlate With Workmanship Failures

The correlation is rarely dramatic. It is cumulative. Skip ventilation and fume exposure accumulates across a shift. Skip PPE and a welder hurries, positions awkwardly, and loses puddle visibility. Skip cylinder discipline and a shop spends its attention on hazards instead of fit-up. Each shortcut steals attention from the one thing that produces good welds: sustained, focused observation of the puddle. If you are asking what does good welding look like from the operator’s side, this is part of the answer — a well-lit, well-ventilated station where the arc never has to compete for attention. The strongest shops treat the ANSI and OSHA safety framework and the AWS workmanship standards as one system, because the conditions that keep welders safe are the same conditions that let them produce welds that pass inspection the first time.

A Practical Checklist: How to Judge Your Own Welds Like an Inspector

You know what a quality weld looks like, what hides beneath the surface, and how codes define acceptance. This last section turns all of that into a routine you can run at your own bench, in under five minutes, every time you strike an arc.

The Five-Minute Self-Inspection Routine

Inspectors do not judge welds by gut feeling. They follow a sequence, and you should too. Work through these steps in order:

1. Clean first. Wire-brush or chip away slag, spatter, and smoke residue. You cannot inspect what is covered: a slag-heavy flux-cored bead hides porosity and undercut as effectively as paint hides rust.
2. Look at the whole weld, then the details. Step back for the big picture: is the bead roughly uniform in width, with consistent ripple spacing? Then move in close for the trouble spots.
3. Check the toes. Inspect both edges where the weld meets the plate. Look for undercut, a groove melted into the base metal, and overlap, often called cold lap, where filler spilled over without fusing. Fatigue cracks start at the toes, so this matters more than bead shine.
4. Check the start, stop, and crater. Craters are prime real estate for star cracks. Tie-ins between passes should be smooth, not stacked like separate lumps.
5. Measure, then compare. Use a fillet weld gauge or caliper to check leg size or reinforcement height, then compare what you measure against the drawing and the applicable code limit. A bead can look perfect and still be undersized.

Questions to Ask Before You Call a Weld “Done”

• Does the bead tie in smoothly at both toes, with no visible groove or unfused edge?
• Is the size what the drawing calls for, measured rather than eyeballed?
• Are the start, stop, and crater filled and crack-free?
• Is the surface free of porosity, spatter, and trapped slag?
• Am I welding the joint I was given, or compensating for bad fit-up with a fat bead?

If you cannot answer “yes” to all five, the weld is not finished, no matter how good it photographs.

When to Escalate to NDT, and When to Repair

Self-inspection has a boundary: your eyes cannot see the root. If a joint is pressure-retaining, cyclically loaded, or specified for volumetric examination, visual checks are only the entry ticket. Ultrasonic and radiographic testing exist because a beautiful cap can hide an unfused root. When your procedure or contract calls for NDT, that is a design input from the contract documents, not something anyone waives on the shop floor.

When a weld fails visual inspection, resist the urge to run a cosmetic pass over the top. Overwelding does not fix lack of fusion; it adds heat, distortion, and a fat bead hiding a bad joint. Repair of rejectable discontinuities generally means removal by grinding or air-carbon-arc gouging followed by a proper re-weld, executed to the same procedure as the original.

Keep a Practice Log

Welders who improve fastest treat every joint as data. Log the process, filler, settings, position, material, and what you saw on inspection, then photograph each bead beside your gauge readings. Patterns emerge fast: porosity in wind, undercut when you rush vertical-ups, craters at a careless stop. If you are formalizing your skills, a structured aluminium welding course teaches exactly this disciplined habit.

Why the Pretty Weld Cracked

Remember the two brackets from the start of this article. The “burnt toast” weld failed inspection honestly and visibly; the beautiful one cracked in service weeks later. Having read this far, you can now name plausible causes: a partial-penetration joint that looked fully fused from the face, a micro-crater at the stop, or subtle lack of fusion at the toe that no ripple pattern would reveal. Uniform ripples prove steady travel speed, not soundness. Appearance is evidence about your technique; it is never proof of quality. That proof comes from proper fit-up, a qualified procedure, and verification at the right level.

Key Takeaways

• Good welding is judged at three levels: visual appearance, internal soundness, and code acceptance.
• Inspect in a fixed order: clean, scan, check toes, check start and stop, measure, compare to the drawing.
• Acceptance criteria come from the governing code and contract documents, with cyclic connections held to stricter limits.
• Document every inspection: AWS D1.1 requires a report for all welds inspected and marking of accepted joints for traceability.
• Rejectable welds get removed and re-welded, not covered up.

The gap between “looks good” and “is good” never fully closes through eyeballing alone, but it narrows with a repeatable routine, an honest reading of the codes, and the discipline to escalate to NDT when the joint demands it. Build those habits now and the weld that passes inspection is also the weld that lasts. For continuing practice and reference, our all-position welding rod guide and the rest of the weldgallery.com inspection library are good places to keep sharpening the craft.

Frequently Asked Questions

What are the visual signs of a good weld?

A good weld has uniform width and ripple spacing, smooth tie-in at the toes with no undercut, clean slag-free surface, correct size per the drawing, and no cracks, porosity, or overlap. It should look consistent from start to finish, including the crater.

Can a weld look good and still be bad?

Yes. A smooth, attractive cap can sit over incomplete penetration or lack of fusion at the root. That is why codes pair visual inspection with volumetric methods like radiographic or ultrasonic testing when the application requires it.

What is the difference between a weld discontinuity and a weld defect?

A discontinuity is any interruption in the weld’s normal structure, such as porosity or a slight profile irregularity. It becomes a defect when it exceeds the acceptance criteria of the governing code and must be repaired or rejected.

Is undercut always a reason to reject a weld?

Not always. AWS D1.1 Table 8.1 sets specific undercut limits that vary by connection type, and cyclically loaded joints have stricter limits than statically loaded ones. Cracks, however, are rejected at any size.

What is the ‘stack of dimes’ look, and does it prove quality?

It is the evenly overlapping ripple pattern typical of TIG (GTAW) welding with dabbed filler. It signals consistent technique, but appearance alone does not prove fusion or penetration — internal soundness still needs verification.

Which welding code defines an acceptable weld?

It depends on the work. Structural steel commonly follows AWS D1.1, pressure equipment follows ASME BPVC Section IX, and ISO 5817 defines quality levels B, C, and D. The applicable code is specified in the contract documents or drawings.

What is the most common welding defect?

Porosity is among the most frequently encountered, typically caused by contamination, moisture, or inadequate shielding gas coverage. Undercut and undersized welds are also common workmanship findings in structural inspection.

How do inspectors check welds beyond visual inspection?

They may use dye penetrant testing for surface-breaking flaws, magnetic particle testing for cracks in ferrous metals, and radiographic or ultrasonic testing for internal, volumetric discontinuities. The required method is specified by the applicable code.