MIG Welding for Beginners: The Complete Step-by-Step Guide (2026)
Learn MIG welding from scratch: machine setup, settings charts, safety, common mistakes, and practice projects. A complete beginner's…
READ ARTICLE →A homeowner pays a handyman to weld a trailer hitch, takes it off the driveway, and watches the tongue crack at the first pothole — the bead looked shiny, but the metal underneath never fused. That is exactly why people search for a terrible welding job: they are either staring at work they already paid for, trying to decide whether to accept or reject it, or trying to never hire the wrong person twice.
A homeowner paid a handyman $180 to weld a trailer hitch mount. The bead looked shiny, even and proud. Two weeks later, on the first real pothole, the tongue cracked clean off. The metal underneath had never actually fused — the arc had melted the surface and laid a pretty cap over two pieces of steel that were, structurally speaking, still strangers to each other.
That gap between how a weld looks and what a weld does is the whole reason this article exists. Nobody types “terrible welding job” for fun. There is almost always a decision sitting behind the search.
1. You paid for it, and something looks wrong. You’re staring at a handrail, a gate, a hitch or a bracket someone else welded, and you’re trying to work out whether you’re being picky or whether you should pick up the phone and call them back. Diagnosing work you already accepted — or rejected — is the most common reason people land here.
2. You have to accept or reject a weld. Maybe you’re a foreman, a quality control tech, a site supervisor, or a welding student running your first pass through inspection. You need language and criteria, not opinions, because on a job site the words reject, repair, rework and reinforce each mean something specific in code-driven work. Using them correctly is the first sign you know what you’re talking about.
3. You want to never do it again. Either you’re hiring and got burned once, or you’re a DIY welder who made the mistake yourself and want the honest fix instead of another forum argument.
A terrible welding job is weld work that does one or more of four things: it fails appearance criteria, it fails the applicable code or acceptance criteria, it fails while in service, or it creates a safety hazard for whoever stands near it. Any one of those is enough. You do not need all four.
And it’s worth saying plainly: bad work is judged against published code, not opinion. The American Welding Society notes that demand for skilled, qualified welders is at an all-time high, and AWS alone has authored more than 350 standards for welding practices and procedures. Behind most professional acceptance decisions sits a written standard — not a gut feeling, and not your buddy’s opinion from the shop floor.
Quick Note: Ugly but sound vs. pretty but dangerous. Excessive spatter, heat tint, surface oxidation and a slightly uneven bead are appearance problems. A weld with all of those can still be perfectly strong. The opposite is the dangerous case: lack of fusion, cold lap and toe cracking are often hidden under a clean, good-looking cap. A bead can pass a glance and still hide a crack inside the joint. Never accept or reject on color and shine alone — check the toes, the tie-in and the root.
Appearance and strength are simply not the same variable. A weld can look acceptable and still hide lack of fusion or a crack at the toe where the stress concentrates. That is also why inspectors never stop at the eyeball test.
We’ll move through the problem in the order a professional actually works it:
Headings are bolded so you can scan, and Quick Note callouts appear throughout for the one-line takeaways you’ll actually remember in the shop. Outbound links point to authoritative sources such as OSHA’s cutting, welding and brazing standard, NIOSH’s welding research and the AWS, so you can check any claim yourself.
You’ll also find internal guides woven in where they genuinely help: our complete guide to welding PPE for protection basics, how to set up a MIG welder when your machine settings are the actual culprit, our deep dive on undercut for acceptance limits, pipe welding procedures for how a real WPS works, and our guides to what welders wear and MIG welding safety when health and hazard questions come up.
Note: This article is written for judging work, not for certifying it. Only a qualified inspector working to a written procedure can formally accept or reject production welds. Use what follows to ask better questions, catch problems early, and know when to call someone in.
Whether you’re holding a cracked part or a set of blueprints, the goal is the same: replace “it looks bad” with “here’s what’s wrong, here’s why it happened, and here’s what the fix actually requires.” Let’s start by learning to see the defects.
Before you can reject a weld, fix one, or refuse to pay for it, you need a vocabulary. Acceptance tables never say “ugly” or “botched” — they name measurable conditions. What follows is a field checklist: read the description, look at the weld in front of you, match what you see.
| Defect | What you see | Structural effect | Verdict |
|---|---|---|---|
| Porosity (blowholes) | Craters, bubbles, or a spongy surface; surface, internal, or clustered | Reduces load-bearing cross-section; pore chains in a root are severe | Acceptable within limits; clusters rejected |
| Pinholes | Tiny pinprick holes, often visible only after deslagging or brushing | Minor individually; a dense field signals a gas or contamination problem | Often accepted in non-critical areas |
| Undercut | A groove melted into the base metal along the weld toe | Severe stress raiser; the classic initiation point for fatigue cracks | Reject above limit |
| Overlap / cold lap | Weld metal rolled over the base plate without wetting into it | No fusion at the toe; load never transfers into the plate | Reject |
| Lack of fusion | Not visible externally; found by NDE or sectioning | Zero bond between weld and base metal along a plane | Always reject |
| Incomplete penetration | Root not fully joined through the joint thickness | Deep notch at the root of a load path | Always reject in full-penetration joints |
| Slag inclusions | Trapped slag visible on X-ray or at a broken root | Weak plane inside the weld | Reject beyond limits |
| Cracks (toe, crater, longitudinal, hydrogen) | Fine lines at crater ends or weld toes; may not show for hours | Unstable and propagating under load | Always reject |
| Excess concavity / underfill | Weld face dips below the plate surface | Thinned cross-section at the joint | Reject over limit |
| Excessive convexity | Bulging, ropey bead with a sharp toe angle | Stress concentration at the toe; poor fatigue life | Reject over limit |
| Burn-through / holes | Blown-out craters, visible gaps in thin material | Missing metal and severe distortion | Reject |
| Misalignment / mismatch | Two members not flush at the joint | Bending load introduced where none was designed | Reject above tolerance |
| Arc strikes | Stray arc marks and spatter outside the weld zone | Hard spots that nucleate cracks in hardenable steel | Reject; grind smooth |
| Spatter | Balls of metal stuck to the surrounding plate | Cosmetic, unless embedded or ground off carelessly | Usually cosmetic |
| Distortion / warping | Plate pulled out of flat; angular change across the joint | Geometry out of tolerance; fit-up problems downstream | Reject over tolerance |
| Unblended grind marks | Deep parallel scratches across a finished weld | Surface notches if left on a finished part | Reject on finished surfaces |
Color is the least reliable clue in any photo of terrible welding work. Heat tint and blue straw colors are oxide films, cosmetic by themselves, and plenty of structurally perfect welds are ugly. What matters is geometry: ripples should be evenly spaced, and the toes should show a smooth wetting transition into the base metal rather than a sharp, cold edge where filler simply stacked up. Where the weld metal meets the plate, does it fade out, or does it sit on top?
Three defects account for most in-service failures, and none are fixed by a grinder and a bit of paint: undercut, cracks, and lack of fusion. Undercut removes metal exactly where stress concentrates. Cracks grow under cyclic load. Lack of fusion is invisible until the joint comes apart.
Quick Note: Judge welds by ripple spacing and toe wetting, not color. A dull gray, slightly rippled bead that ties in smoothly at both toes is structurally better than a mirror-smooth, rainbow-tinted bead with a rolled-over edge. Our visual guide to undercut, including its acceptance limits and repair method, is the best place to start training your eye.
Here the acceptance limits become concrete. The American Welding Society’s structural code, AWS D1.1, treats cracks of any kind as rejectable conditions requiring removal and rewelding, and it treats lack of fusion and incomplete penetration as rejectable defects rather than cosmetic blemishes. Undercut gets a hard number: roughly 0.010 in (0.25 mm) for members subject to fatigue loading and about 0.030 in (0.8 mm) for other members under D1.1 visual examination criteria. That is thinner than a sheet of paper for the fatigue case.
Porosity is judged by where it sits. A single isolated pore in a non-critical fillet is frequently acceptable. A cluster or chain of elongated pores in the root of a load-bearing joint removes cross-section precisely where stress runs highest, and that gets rejected. Context is the whole game.
Two more notes on how these defects really appear. Slag inclusions and internal lack of fusion never show up in a phone photo, which is why visual inspection alone cannot certify a structural weld. And cracks are often delayed: hydrogen-induced cracking can appear hours after the weld cools, so a joint that passed a Friday-evening visual check may need re-inspection on Monday. Welding outdoors, on thin material, or on critical joints? Our MIG welding safety guide covers the setup habits that prevent most of these conditions at the source.
Keep this catalogue open the next time you photograph your own work or inspect a contractor’s. If you can name the defect, you can argue about the fix.
A terrible welding job is almost never one dramatic mistake. It is usually two or three small ones stacked together. That is why the bead looked fine from four feet away but failed the moment it carried a load.
Experienced welders and inspectors troubleshoot in a fixed order: shielding first, then base-metal cleanliness, then heat settings, then the welder’s hand. Learn that sequence and you will stop blaming yourself for defects actually caused by a drafty doorway or a rusty plate.
Oil, grease, paint, rust, moisture, mill scale, mill lubricants and primer are the leading cause of porosity and blowholes. The arc breaks these contaminants down chemically, and the gas has nowhere to go except through the molten pool. Frozen in place, you get a pinhole; clustered, you get a blowhole that destroys the weld’s cross-section.
Zinc-coated and galvanized steel are the classic trap. Zinc vaporizes at around 907 °C, well below steel’s melting point, so it boils out while the pool is still liquid. The result is heavy porosity plus a white zinc-oxide smoke that causes metal fume fever. Grind or chemically remove the coating at least 2 in (50 mm) either side of the joint, keep the area dry, and set up real ventilation — OSHA’s general ventilation rule for welding, cutting and brazing is 29 CFR 1910.252, and fume extraction is covered in our MIG welding safety guide.
Shielding problems produce a recognizable bead: porous, discolored, sugary-looking, brittle. The usual suspects are a wrong gas mix (100% CO2 where a 75/25 argon-CO2 blend was specified), flow set high enough that the nozzle vortex pulls in room air, flow too low, a draft from an open door, an empty or leaking cylinder, excessive contact-tip-to-work distance, a nozzle and liner packed with spatter, or a worn contact tip. Getting the machine dialed in is the whole job — our MIG welder setup guide walks through it step by step, and NIOSH’s welding topic page explains why shielding and ventilation go hand in hand.
Wrong settings fail in predictable directions. Too little heat leaves cold lap, lack of fusion at the toes, and a bead that sits on the plate instead of tying into it. Too much heat causes burn-through, excessive penetration, heat-affected-zone grain damage and warping. Wrong polarity (DCEN where DCEP was needed) compounds both by shifting penetration depth. Sizing amperage to thickness is not guesswork: for a 3/32 E7018 rod see 3/32 E7018 settings and polarity, for rod selection logic what electrodes actually do, and thin material punishes both cold and hot settings fastest, as covered in stick welding thin metal.
Excessive travel speed starves the toes and leaves undercut; crawling too slowly piles up metal and traps slag. Work and travel angles drift, arc length stretches, and weaves get wider than the joint allows. Long arcs also pull in air and create porosity. Craters at stop points are a frequent source of crater cracks, and a careless start can leave an unfused cold start. Even the push-or-pull choice changes gas coverage and penetration.
This is the cause people never suspect. A wide root gap, a missing bevel or land, or joint geometry that doesn’t match the filler size all change how the arc penetrates — even when the hand is perfect. Heavy, oversized tack welds act like heat sinks and create cold-start craters; keep tacks modest and properly spaced, as covered in how to tack weld with a MIG welder. Clamps that trap heat in one area drive distortion in another.
Wrong filler is quietly common: mild-steel filler on unknown scrap alloy, or stainless wire on galvanized mild steel. Recycled stock with unknown chemistry is a gamble. Damp low-hydrogen electrodes are worse — the 7018-type flux coating absorbs moisture from shop air, and that moisture becomes hydrogen in the arc. Store rods dry, re-bake per the manufacturer’s spec, and buy from suppliers who turn stock fast.
Production pressure, an untrained helper on critical passes, fatigue and the absence of a written procedure cause more rework than any single machine fault. A Welding Procedure Specification tells the welder what filler, heat, technique and acceptance criteria apply; our pipe welding procedures guide shows how it is written and applied. Welding by feel is how “my buddy said it’s fine” becomes a structural failure.
Quick Note: Triage with one question: is this a gas problem, a dirt problem, a heat problem, or a hand problem? Pinholes scattered evenly = gas or shielding. Large clustered blowholes near the start = moisture or oil. Cold lap and sticky arcs = heat too low. Burn-through and warping = heat too high. Undercut only on fast sections and ragged toes = travel speed and angle, the hand.
This is the defect that shows up hours after the welder has gone home. It takes three ingredients at once: hydrogen (from damp electrodes, oil, moisture or mill scale), a hard, susceptible microstructure (fast-cooling high-carbon steel), and restraint (a joint that cannot move, so tensile stress builds). Remove any one and the crack usually doesn’t form. That’s why low-hydrogen consumables, preheat on thick sections, cleaning, and slow cooling with a blanket matter more than any amount of grinding afterward.
Once you can name the cause, you can decide what to do about it. Next, we look at how professionals actually judge a weld — codes, inspection methods and the acceptance tables that replace opinions with criteria.
Here is the uncomfortable truth about welding: there is no universal standard for “looks good enough.” Two welders can stare at the same bead and argue for an hour. What separates professional judgment from opinion is a system: written procedures, defined acceptance criteria, and a credentialed person who signs the verdict.
Every structural weld made in a professional shop or on a job site traces back to paperwork. The chain works like this:
The governing codes set the rules. Structural steel in the United States falls under AWS D1.1/D1.1M:2025-AMD1 from the American Welding Society, the currently listed edition, covering preheat through visual acceptance tables. Pipework commonly falls under API 1104, pressure vessels and pipe under ASME Section IX, and bridges under AWS D1.5. Our companion guide on pipe welding procedures and inspection shows how this paperwork reads in the field.
Visual examination is the mandatory first check. OSHA requires it for welding, cutting and brazing operations under its hot-work general industry standard (1910.252), and virtually every acceptance code does the same. VT is cheap, immediate, and catches the majority of defects — cracks, undercut, overlap, misalignment, porosity and incomplete joint penetration visible from the accessible side — before anyone pays for fancier methods. The inspector cleans the weld, lights it directly, often with a mirror and magnifier, and checks it against a published acceptance table.
When visual says “maybe,” non-destructive examination answers the question:
Quick Note: Pretty is not proof. A mirror-smooth bead with perfect ripples can hide lack of fusion or a toe crack that only MT, PT, RT or UT will reveal. Appearance proves the welder had good hand control. It does not prove the weld is sound.
What surprises most homeowners and DIYers: some “ugly” is code-legal, and some “clean-looking” is an automatic rejection. A typical structural visual inspection table works roughly like this:
| Condition | General treatment |
|---|---|
| Cracks of any kind (toe, crater, longitudinal, transverse) | Never acceptable. Remove, fix root cause, re-weld. |
| Lack of fusion / incomplete penetration (where full penetration is required) | Reject. |
| Undercut beyond a stated limit (commonly 0.010 in / 0.25 mm on cyclically loaded members, more on static) | Reject, or dress and re-inspect within limits. |
| Porosity beyond the allowed number/spacing or any elongated pore | Reject. |
| Misalignment beyond a specified offset for the joint thickness | Reject. |
| Spatter, heat tint, minor bead asymmetry, slight convexity | Usually acceptable cosmetically. |
Cracks deserve emphasis: structural codes never accept them, and welding over one is futile — it re-opens. See our deep dive on undercut causes and acceptance limits.
On anything structural, a third party stands between the welder and acceptance: an AWS Certified Welding Inspector, who applies the acceptance criteria in the code, the WPS and the drawing. Endorsements show what an inspector actually knows — Structural Steel (D1.1), Bridge Welding (D1.5), Pipeline (API 1104) and Aerospace (D17.1). For a bridge, an inspector endorsed only in structural steel is not the right signature.
Homeowners rarely need a CWI — but for anything load-bearing, a handrail, a trailer frame or a pressure connection, a third-party visual inspection before final payment is the cheapest insurance you will buy. Quality habits start with the welder, so it pays to know what welders wear.
Bottom line: when someone calls a weld fine, ask what code applies and who inspected it against that code. No answers means opinions, and opinions hold no weight.
A bad-looking weld is annoying. A weak one is expensive. The problem with a terrible welding job is that the bill rarely shows up on the invoice for the weld itself. It shows up later: a cracked bracket, a failed inspection, a leaking vessel, an injury report, or a customer who never calls again. Here is where the real cost actually lands.
Welds carry load, and defects change how that load travels. Lack of fusion, porosity, undercut, overlap and incomplete penetration all act as stress raisers. Undercut is a good example: it notches the toe of the weld and thins the effective cross section at exactly the point where stress concentrates. The nasty part is that such flaws often behave like a notch in a brittle material, so the crack starts quietly and the final failure arrives suddenly rather than as a slow, visible sag.
Where that happens matters. A cracked decorative gate is a nuisance. A cracked trailer hitch, guardrail post, lifting lug, pressure line or tank seam is an incident with witnesses. This is why structural work is governed by written procedures and inspection rather than opinion, from AWS structural codes such as D1.1 to pressure vessel codes that require documented procedures and qualified welders.
There is a second, less obvious safety cost. Sloppy work often means sloppy surroundings: a hot arc placed near cardboard or oily rags, a damaged ground clamp or frayed lead, a welder rushing overhead work without proper shielding. Burns, electric shock and arc flash are common on jobs where the welding itself is already a mess.
The arc produces ultraviolet and infrared radiation that causes photokeratitis, commonly called arc flash or “welder’s flash,” plus skin burns that behave like a bad sunburn. Anyone watching an unshielded arc for a few seconds learns this lesson quickly.
The fume side is slower and less obvious. Welding fume is a mixture of metal oxides from the base metal, filler and coating, plus flux and shielding gas byproducts. Welding stainless steel can generate hexavalent chromium, which OSHA regulates at a permissible exposure limit of 5 micrograms per cubic meter of air. Manganese, nickel, zinc and cadmium appear in fume depending on the material, and zinc-heavy work produces the classic metal fume fever. OSHA’s welding, cutting and brazing standard (1910.252) requires ventilation and protective equipment to keep exposures under control, and NIOSH publishes detailed guidance on controlling welding fume.
Poor technique makes exposure worse. Long arcs, wandering arcs, wrong polarity, excessive current and burning through paint or plating all drive more fume into the breathing zone. So does skipping local exhaust on a job that “will only take five minutes.” For prevention basics, see our welding safety overview.
The financial damage almost always exceeds the cost of doing the weld correctly the first time, because the original labor is paid twice.
| Cost Category | What It Actually Costs |
|---|---|
| Rework labor | Cut out, grind back to sound metal, re-preheat, re-weld, re-inspect. Often doubles the hours on the joint. |
| Scrap and consumables | Destroyed plate, wasted filler rod, grinding discs, gas and electrodes. |
| Downtime | The machine, jig or production line sits idle while the repair happens. |
| Inspection and testing | Re-inspection, dye penetrant or radiographic re-checks, and code paperwork. |
| Schedule and contracts | Missed handover dates, liquidated damages, overtime to catch up. |
| Warranty and liability | Field failures, returns, injury claims and insurance premiums. |
| Reputation | The quiet cost: the customer who simply stops referring you. |
Quick Note: The filler metal is the cheapest line item on the job. The expensive part is everything that follows a defect into the field, where access is limited, the parts are already assembled and the clock is running against a deadline.
The math is straightforward. Better preparation, correct parameters, clean base metal and a qualified procedure cost minutes. Chasing a bad weld costs hours, sometimes days, plus the credibility you cannot invoice for. Preventing a terrible welding job is not a matter of perfectionism; it is the cheapest option on every job sheet.
Finding a defect is the easy part. What separates a welder who fixes problems from one who makes them worse is the next thirty minutes. Every serious code follows some version of the same repair sequence: clean, identify, remove to sound metal, verify, re-weld to the qualified procedure, then re-inspect and document. Skip a step and you haven’t repaired anything. You’ve buried a defect and bought yourself a second failure.
Start with a wire brush or a grinder until you’re looking at bare metal. Spatter, slag, paint and heat tint all hide cracks, so a bead that looks “clean” can still have a toe crack running under the cap. Once it’s bright, check the full length, not just the spot someone complained about. If the joint matters, run a dye penetrant test: clean it, spray the red indicator, let it soak, wipe it off, dust on the developer, and wait. Any red bleeding back through the white developer is a surface-breaking crack or a pore chain, and it tells you exactly where to cut.
Now sort it honestly into two buckets:
Undercut looks minor and behaves badly, so know the actual depth limits before you decide it’s fine — our guide to undercut causes, acceptance limits and fixes lays out what inspectors measure. And note that even a tiny crater crack at a stop has to be ground out, never filled over.
How you remove the defect depends on how deep it lives:
Carbon-arc gouging reaches internal defects much faster than grinding, but it needs compressed air, real PPE and strict fire control. OSHA requires fire prevention measures, including protecting or removing combustibles, whenever cutting or welding goes on (OSHA 1910.252). Grinding is slower and perfectly adequate for surface-level repairs.
Quick Note: Grind it, don’t hide it. Welding over a crack or porosity without removing it first is both ineffective and prohibited by structural codes — the original defect stays exactly where it was, still growing, now hidden under a fresh cap. A thin cover pass is a disguise, not a repair. Before the grinder starts, check the basics: guard on, wheel rated for the speed, face shield over your helmet, and fire watch for the sparks. The full kit is covered in our welding PPE guide.
Don’t start arcing the moment the grinding stops. Brush the cavity bright, look into every corner, and run penetrant again if the original defect was a crack. Confirm there’s no dark line, no undercut wall, no trapped slag. Then re-weld to the qualified procedure: correct filler, correct polarity, correct heat, and controlled interpass temperature. If you’re running low-hydrogen stick, respect the oven and exposure rules, because a damp E7018 rod reintroduces the very hydrogen cracking you just ground out. Fill the crater on every stop.
Clean the finished repair, run visual examination against the acceptance table, and photograph it close up with a ruler for scale. Structural, pressure and code-regulated parts usually require a written repair procedure, a qualified welder, and re-inspection by visual plus the applicable NDE method. That’s why pressure work runs on documents rather than memory, and it matches what AWS publishes in its structural welding codes. Keep the before, during and after photos together — they end arguments later.
Sometimes the honest answer is scrap. Thin-walled tubing that’s been burnt through, a crack sitting in a critical load path, or a piece of metal whose chemistry you don’t know — welding on those trades a small fix for a large unknown. Cut the part out and replace it. A cheap repair that fails twice costs more than the new piece ever would.
Most terrible welding jobs do not happen in fabrication shops with written procedures and inspectors. They happen in garages and backyards, on a Saturday, with a borrowed welder and a deadline. The good news: nearly every home-shop failure traces back to a short, predictable list of mistakes. Fix these twelve and your welds will improve faster than any amount of practice alone.
Paint, oil, and grease decompose in the arc and release gas straight into the puddle. The result is porosity and a bead that looks solid on top but is Swiss cheese underneath. Grind or solvent-clean two inches back from the joint on both pieces, then wipe it dry. If it smells chemical rather than metallic, it is contaminated.
Zinc boils off long before steel melts. That white cloud is zinc oxide fume, the cause of metal fume fever, a flu-like chill and headache that hits hours after a “fine” session. Strip the coating back, keep your face out of the plume, and ventilate. OSHA requires controls to keep welding fumes below the permissible exposure limit, with respiratory protection where controls are not enough (29 CFR 1910.252). Our MIG welding safety precautions guide covers fume control in detail.
Porch-screen porosity — a broad field of surface pinholes — is almost always a gas problem. Suspects: a cylinder running empty, a cracked nozzle, a loose flowmeter, a leaking hose connection, or straight CO2 where the chart calls for 75/25. Bump the flow rate for a second on scrap plate; if nothing changes, you have no gas. Full walkthrough in our MIG welder setup guide.
DCEN versus DCEP is not a preference; it changes penetration, deposition, and arc stability, and the wrong polarity on stick electrodes gives an erratic arc and weak fusion. Too-cold settings on thick steel cause cold lap and lack of fusion; too-hot settings blow through thin sheet and warp everything. Set from a chart, then verify on scrap of the same thickness.
A long arc wanders, pulls in air, and leaves undercut along the toes. A weave wider than about two to two-and-a-half electrode diameters stacks heat on the edges and cools the center, inviting cracking. Keep the arc short and use a tight weave or straight stringers. See undercut causes and acceptance limits for why the toe is the danger zone.
This failure is invisible. Flux coatings absorb moisture, and that moisture puts hydrogen into the weld; hydrogen plus restraint plus hard steel is the recipe for cold cracking. Low-hydrogen 7018 rods are especially sensitive, so store them in a rod oven or sealed bucket and re-bake per the maker’s instructions. Amperage matters too — start with our 3/32 E7018 amperage and polarity guide.
Cheap consumables cause most home-shop failures. A worn contact tip changes stick-out and makes the arc wander, a kinked liner drags the wire and makes it pulse, a spatter-clogged nozzle defeats the shielding gas, and a cracked nozzle leaks it. Replace tips liberally, clean the nozzle every few passes, and check your ground clamp and cable connectors — a poor return path makes the machine hunt.
No bevel, no land, and a sloppy half-inch root gap means the arc never reaches the joint root, so incomplete penetration sits under a bead that looks finished. Bevel thick edges, leave a land of about 1/16 in (1.5 mm), control the root gap to the procedure, and tack before you run.
Thick sections and clamped, high-restraint joints pull heat away fast and cool weld metal into a hard, crack-prone microstructure. Even a modest preheat — checked with a temperature stick or IR thermometer — slows cooling dramatically, and coated sections often need it anyway to drive off moisture.
Vertical and overhead are legitimate skills, but if the part can be flipped, weld it flat or horizontal. Out-of-position welding demands lower heat, faster hands, and more control than most beginners have, and gravity works against you, producing sagging fillets, saggy roots, and trapped slag.
Arc eye, burns, and fume exposure are not badges of toughness. You need an auto-darkening helmet rated for your process, a flame-resistant jacket, leather gloves, safety glasses under the helmet, and ventilation at the source. Our welding PPE guide lists exactly what each process requires.
Slag left between passes gets buried, creating inclusions that come out later as a grinding job. Deslag and wire-brush every pass, watch your interpass temperature, and stop treating a multi-pass weld as one continuous bead. Rushing is also where crater cracks come from: hold the arc at the end and fill the crater.
Quick Note: Pre-weld checklist. Before striking an arc: material identified, joint cleaned two inches back, fit-up correct, filler and gas confirmed, parameters set from a chart, PPE on, ventilation running. TIG rule of thumb: start near 1 amp per 0.001 in (0.025 mm) of steel thickness — about 40–45 amps for 0.040 in sheet — then dial down when a heavy fixture is pulling heat out.
None of these fixes require a better welder. They require slowing down for twenty minutes before the arc starts: clean the metal, check the consumables, set from a chart, put on your gear. That is the whole difference between a Saturday repair you trust and one you grind off next spring.
Whether you are a homeowner hiring for a handrail or a general contractor subbing out structural steel, the buying decision is the same: you are paying for a weld you cannot personally evaluate. That makes the hiring conversation your real inspection tool. Here is how to use it.
Watch for these signals before money changes hands:
Quick Note: For code work, ask three documents up front: the Welding Procedure Specification (WPS), the welder’s performance qualification records, and the name of the inspector. Per the American Welding Society, a contractor who cannot produce these is a documented risk, not a paperwork inconvenience.
Emotion loses arguments; criteria win them. Use a script like this: “I’d like you to remove the undercut at the toe here per the visual criteria and re-run this pass.” Then point to the written scope. An argument about whose fault it is should follow the contract, not the phone call. Keep it factual, reference the standard, and put the request in writing the same day.
Any welder can claim AWS certification. The AWS Certification QuickCheck verifies that a Certified Welding Inspector credential is current — use it before relying on a CWI’s sign-off. Hire a third-party inspector when the work is structural, the load path is critical, or the welder’s own inspector is also their only employee.
If you need to document defects, clean the weld first. Then shoot a tight close-up with a ruler or coin for scale, using raking light that skims across the bead. Glare and spatter hide undercut and porosity in flat, direct lighting. One clean photo beats ten blurry phone snaps in any dispute.
Note — Contract Language Worth Writing Down: specify the acceptance standard, state who pays for rework, include a payment holdback until inspection passes, and add a workmanship warranty period. A good welder corrects workmanship defects at their own cost — that clause separates pros from pretenders.
Finally, keep your own standards realistic. Knowing what a properly executed weld looks like is half the battle, and our guides on welding PPE and MIG welding safety will help you spot a contractor who cuts corners on protection too — because the same person who skips fume extraction usually skips root fusion as well.
The process you choose does more than set your pace. It decides what kind of terrible weld you are likely to make. Every process has a signature failure mode: MIG gives you porosity, TIG punishes a dirty joint, stick buries slag and undercut under a bead that looks fine until it breaks. Name the process and you can usually predict its top three defects before anyone strikes an arc, which is useful when someone hands you a finished weld and asks whether to accept it.
Gas metal arc welding is what most people learn first, and for good reason: it is quick, the spatter is low, and the machine does much of the thinking. Its most common field failure by a wide margin is porosity from shielding-gas loss. A leaking hose, an empty cylinder, a drafty shop door, or holding the nozzle too far from the arc lets atmosphere into the puddle, and you get the classic cluster of blowholes that appears in nearly every photo of terrible MIG work.
The second MIG trap is settings run too cold on thicker steel, which produces lack of fusion and cold lap at the toes; the bead sits on top of the plate instead of tying in. Getting the setup right is half the battle, and our complete MIG welder setup guide walks through gas flow, stick-out, and wire speed step by step. MIG also means compressed gas cylinders and live electrical circuits, so it carries its own safety checklist in our MIG welding safety guide.
TIG produces the welds people post online and call perfect: clean bead, even ripples, no spatter. The trade-off is speed and sensitivity. GTAW gives you the most control of any process and demands the most discipline in return. A joint not scrubbed free of oil and oxide, or a filler rod that does not match the base metal, will crack no matter how beautiful the cap looks. Heat tint and a flawless ripple pattern prove nothing about what is happening underneath. TIG also has the steepest learning curve on thin material, where settings and pedal control matter enormously; see our TIG sheet metal guide.
Stick welding remains the most portable process there is and tolerates wind and light surface contamination far better than MIG or TIG, which is why it still dominates field work, repair, and pipeline construction. Its signature defects are slag inclusions, undercut from travelling too fast, and porosity caused by a damp coating on the electrode. Electrode choice matters more here than anywhere else: deep-penetration 6010 and 6011 rods are the go-to for root passes and out-of-position work, while low-hydrogen 7018 types are preferred for structural strength but are moisture-sensitive and will punish you with porosity if they have absorbed water. New to it? Our honest beginner’s guide to stick welding sets realistic expectations.
Flux-cored arc welding deposits metal fast, which makes it the natural choice for thick sections and structural fabrication. Self-shielded wire handles wind far better than gas-shielded MIG, which is why you see it on outdoor jobsites. The cost is more spatter and far more slag, and gun hygiene becomes critical: a dirty nozzle or a contaminated liner is usually the difference between an acceptable bead and a terrible one. Our flux core welding tips cover the cleaning routine in detail.
| Process | Forgives | Punishes with |
|---|---|---|
| GMAW (MIG) | Speed, thin-to-medium stock | Porosity, cold lap |
| GTAW (TIG) | Appearance, precise control | Cracks from contamination |
| SMAW (Stick) | Wind, dirty base metal | Slag inclusions, undercut |
| FCAW (Flux Core) | Thick sections, deposition rate | Spatter, slag, dirty guns |
Quick Note: Ugly but sound is a real thing. Heat tint, surface oxidation, a bit of spatter, or a slightly uneven ripple pattern do not reduce strength. Pretty but weak is the dangerous combination: a clean, ground-smooth cap hiding lack of fusion, a toe crack, or a line of undercut at the weld edge. When judging someone’s work, look at the toes and edges, not the middle of the bead.
Undercut deserves attention in every one of these processes and is among the defects most often responsible for in-service failure at the weld toe. Our undercut causes and acceptance limits article explains the numbers behind that. Whatever process you run, protective gear and ventilation are non-negotiable; the complete welding PPE guide covers what you actually need on your body and in your air.
The short version: choose MIG for speed, TIG for precision, stick for portability, and flux core for thickness. Then judge the result by its toes, not its shine.
After seeing the causes, costs, and fixes for a bad weld, the most impactful action is prevention. Instead of relying on luck, use a repeatable system that catches issues at the right time. Focus on these five areas to stop defects before they reach an inspector or a load.
The vast majority of weld failures are decided before you strike an arc. Get these tasks out of the way first:
Once the arc is running, your goal should be control. Keep a steady arc length, maintain consistent travel and work angles, and pay extra attention to the toes of the weld, as these are common spots for undercut and lack of fusion. Fill your crater at every stop to avoid crater cracks, clean between passes, and monitor interpass temperature. If something looks off, stop. A half-inch correction will always be cheaper than reworking a finished joint.
Visual inspection is the cheapest form of quality control in this trade. Deslag, brush, and compare your work to the applicable acceptance table. Photograph it with a ruler for scale, grind any high spots, and schedule NDE if the code requires it. For code-driven projects, our pipe welding procedures guide shows how to document these steps properly.
Deliberate practice beats random welding. Run a bead on plate, section it, and inspect your penetration firsthand. Change only one variable at a time, such as amperage, travel speed, or arc length, and repeat the test. Over the course of a month, you will know your settings for certain. If you are starting with SMAW, our honest beginner’s guide to stick welding is a great place to build that feel.
If a critical part is cracked or badly burnt through, cut it out and start again. Do not cap over porosity, and do not bridge a crack with a hot pass. No amount of grinding can hide a defect that will be found under load. When it comes to safety in the shop, our MIG welding safety guide will help reinforce these boundaries.
Quick Note: Print this checklist and tape it in your hood case: (1) Identify material, (2) Clean 2 inches back, (3) Check fit-up, (4) Confirm filler and gas, (5) Set parameters from a chart or WPS, (6) PPE and ventilation on, (7) Consistent arc and crater fill, (8) Clean between passes, (9) Deslag and visually inspect, (10) Document and grind high spots.
At its core, prevention is simple: a written procedure, a pre-weld checklist, and a thirty-second post-weld visual check. Those three habits will catch nearly every common defect, saving you time, money, and liability. That is not overkill. That is just sound welding.
Clean it and look for cracks, undercut at the toes, lack of fusion, porosity clusters and burn-through rather than spatter or discoloration. Color and spatter are cosmetic; cracks, undercut over the code limit and unfused edges are structural. When in doubt, have a qualified inspector do a visual exam and, for critical joints, an NDE test such as penetrant or radiographic.
Most surface-level defects can be ground out to sound metal and re-welded to the original procedure. Cracks, root porosity and lack of fusion require gouging or cutting back to clean metal, and severely burnt-through thin material is often better replaced. Pressure and structural parts may need a written repair procedure and re-inspection.
Appearance and soundness are related but not identical — a lumpy, spattered bead can still be fully fused, while a glassy-smooth bead can hide lack of fusion. That said, ugly welds usually show inconsistent parameters or technique that also raise defect risk. Test the joint under load only in safe, non-critical situations, and inspect it visually or with penetrant dye before relying on it.
Porosity comes from gas trapped in the puddle, most often caused by oil, paint, rust, moisture or zinc on the base metal, or by shielding-gas loss from leaks, drafts, wrong flow or excessive stick-out. Clean the joint thoroughly, check your gas connections and nozzle, and slow your travel speed enough for the puddle to outgas. Damp electrodes are a separate hydrogen-related cause of porosity and cracking.
Reject work showing cracks, undercut beyond the acceptance limit, lack of fusion, holes, deep or clustered porosity, misalignment, or any weld outside the agreed scope or code. Ask for the written acceptance standard, photograph the defects with a scale for reference, and require removal and re-welding at the contractor’s cost. For structural or pressure work, request independent visual inspection and the required NDE before final payment.
Only after the defect has been fully removed — you can weld over a cleaned, ground-down repair area, but never cap over a crack, porosity pocket or unfused region. Welding on top hides the defect, traps contamination and concentrates heat. Grind or gouge to sound metal, verify the cavity with visual or penetrant inspection, then re-weld.
For code-regulated structural, pressure or bridge work, yes — the welder should be qualified to the applicable procedure and the work inspected by a Certified Welding Inspector or equivalent third party. DIY projects like a garden gate or shelf do not require certification, but they still deserve proper preparation and inspection. When money or safety is on the line, hire qualified people and verify credentials.
Undercut is a groove melted into the base metal beside the weld toe that was never filled, and it acts as a stress raiser. It matters most in fatigue-loaded joints, where codes typically cap it at about 0.010 in (0.25 mm), with a slightly looser limit around 0.030 in (0.8 mm) elsewhere. Because it initiates cracks under repeated loading, undercut is one of the most common reasons a good-looking weld gets rejected.
It depends on access, material and whether NDE or a remake is required — repair invoices usually combine a trip or minimum charge with hourly labor, and re-inspection adds more. A rejected weld that forces a part to be scrapped and rebuilt costs several times what proper prep and settings would have cost up front. Getting a written scope and acceptance criteria before work starts is the cheapest insurance.
Clean the joint, fix your settings, and control arc length — those three eliminate most beginner defects before technique even matters. Practice beads on plate with a settings chart, then cut or grind into your welds to see penetration, changing only one variable at a time. Inspect every bead against a simple acceptance checklist so you catch undercut, porosity and crater cracks while they are still easy to fix.