The fourth foot of the weld was the problem. Ellis Boone had a 5-by-10 utility trailer frame clamped to the floor of his two-bay shop on a Tuesday, welding 2-inch square tube at 3/16-inch wall thickness with a 160-amp multi-process machine and 0.035 self-shielded flux core wire. The first three joints of the fender rail came out the way flux core is supposed to come out: a tight row of even ripples and slag that lifted in one piece at a light tap. Then he chipped the fourth joint and found three pinholes clustered at the toe, sitting in the same eighteen inches of weld he had already re-run twice before lunch.
What followed was four hours of what Ellis later called professional guessing. He swapped the half-used spool for a fresh one of the same brand, then a different brand, then dropped from 0.035 to 0.030. He changed the contact tip, shortened his stickout from about an inch down to a half, moved the ground clamp to a fresh spot on the frame, cleaned the drive roll groove with a pick, and finally ran two test coupons on a clean plate bolted to his bench. The coupons were flawless. By three o’clock he was seriously entertaining the possibility that the square tube had been made from bad steel.
His uncle Walt, who spent thirty-eight years putting pipelines together in the Gulf, wandered in around four and did not touch a single knob. He watched one weld, listened to it, and then asked two questions. How long has that roll of wire been sitting in the back of your truck? Three days, through two rain showers. Do you keep that door up while you are welding? The roll-up was open, and on that particular afternoon the wind was doing fifteen across the bay, cutting straight through the work area where Ellis was running the fender rail. Walt closed the door, pulled a spool that had been living on a shelf indoors, wiped the joint once more, and Ellis ran the bead again. The slag lifted clean, the ripples were even, and there was not a pinhole in it. The entire fix took eleven minutes.
The door is not the lesson here, though it was the answer that day. The lesson is that Ellis changed the wire, the voltage, the stickout, the tip, the ground, and the technique inside a single afternoon and learned absolutely nothing from any of it, because when you change five things at once a good result tells you nothing and a bad result tells you nothing. What separates a ten-minute diagnosis from a bad afternoon is a fixed order of checks run the same way every time, plus the discipline to change one variable at a time. That is what the rest of this article is: the method first, then every symptom you are likely to meet, the physics behind it, the order to fix it in, and how to prove the fix actually worked.
Troubleshooting flux core is hard when you start from the symptom, because the same symptom has six possible causes and the same cause produces five symptoms. Polarity is first because it takes thirty seconds and invalidates every other reading if it is wrong. The chart is second because it takes one minute. Stickout, direction, wire, feed path, and ground follow, in that order, because each one is a physical input to the arc that you can verify with your eyes and a tape measure.
Self-shielded flux core wire is designed to run on electrode positive, also called DCEP or reverse polarity, where the gun lead is positive and the work clamp is negative. When that is reversed the arc characteristics change fundamentally: less heat reaches the plate and more stays in the wire, so you get shallow penetration, a fast and sloppy melt-off rate, big droplets tearing free of the wire, and a slag layer that never quite covers the bead evenly. Those five consequences are the seed of at least six different symptoms covered later in this guide, which is why polarity has to be check one. If you want certainty rather than trust, run the two-bead test described in the polarity guide or put a DC clamp meter around the work cable. Also read the spool label, since a few specialty wires and most solid MIG wires specify the opposite polarity, and solid MIG wire in a flux core machine is a mistake that hides behind the same confusion.
Most people believe their settings are right because they never changed them, and unchanged is not the same as correct. Pull the chart that came with the wire or the machine, or use a published wire speed and voltage chart, and compare three things: the wire diameter loaded in the gun, the voltage the display is actually showing, and the wire speed measured in inches per minute rather than read off a dial. Measuring real inches per minute takes ten seconds: mark twelve inches of wire, pull the trigger with the gun in free air, and time how long it takes to feed that length. If you are working from amperage recommendations instead of voltage, the flux core amperage chart gives you the current that a given wire diameter should be carrying for the thickness in front of you.
Stickout, properly called contact tip to work distance, is the length of wire visible between the contact tip and the plate. For self-shielded flux core on a small machine the working range is typically a half inch to three quarters of an inch, roughly 13 to 19 millimeters, and the specification sheet for your exact wire is the final authority. This matters because the wire is part of the circuit: lengthen the extension and you add resistance, which changes your actual current and shifts how much of the melting happens in the wire versus in the arc. A stickout that has crept out to an inch and a half, which happens naturally as you get comfortable and lazy in a long seam, will give you a wandering arc, cold lap on the toes, incomplete penetration at the root, and enough extra spatter to make you blame the voltage.
With self-shielded flux core you drag, pulling the gun away from the weld puddle with the wire pointing into the leading edge, at a work angle of about fifteen degrees from vertical for a flat fillet and a travel angle of roughly ten to twenty degrees off perpendicular. Dragging matters because the shielding gases generated by the flux come off the arc behind the wire, and the slag and gas envelope follow the puddle; push that envelope forward and you lay slag and molten flux underneath the arc, which gives you lack of fusion, trapped slag, and an erratic arc that pops as it burns through its own cover. The technique and its exceptions are covered in push versus pull for flux core, and the difference is rarely subtle.
The consumable is doing roughly half the work in a flux core weld, and it is the variable people check last. Look at the wire coming off the spool for three things. First, rust or dull gray discoloration: steel wire that has absorbed moisture will carry that moisture into the arc, where it becomes hydrogen and oxygen in the puddle, and the result is porosity along with a risk of hydrogen cracking in the cold months. Surface rust on a spool that sat through a rain shower is one of the most common and most misdiagnosed causes of exactly the defect Ellis chased, and keeping wire from rusting is mostly about storage discipline. Second, the condition of the tube itself: flux core is a tubular wire filled with powder, and a kink, flat spot, or crushed section from running it over with a cart will interrupt the flux column, causing arc popping and worm tracks further down the weld. Third, diameter versus the job:0.030 is the friendlier choice on a 120-volt machine for thin work, while 0.035 puts down more metal per inch and suits thicker plate, and the trade-off is spelled out in 0.030 versus 0.035 flux core wire.
Between the spool and the arc sits a chain of parts that all have to agree with each other, and a fault anywhere in that chain presents itself as an arc problem. Work backwards from the gun. The contact tip must match the wire diameter, because a 0.035 tip running 0.030 wire gives poor electrical contact and arcing inside the tip; a 0.030 tip running 0.035 wire will simply jam. The drive rolls must have the correct groove for the diameter — a V-groove meant for solid wire running flux core will crush the tubular wire and shave off the flux, which shows up as spatter, popping, and a bead that looks starved. Tension should be just enough that the wire does not slip when you hold it against the plate with the trigger pulled, which is the standard slip test, and no more, because over-tightened rolls deform the wire and score the liner. The specifics live in drive rolls for flux core wire, and a worn set of rolls costs less than a day of chasing a defect they caused.
A ground clamp attached to paint, rust, or a scale-covered flange forces the current to find another path, and when that path is intermittent the arc notices. The joint itself needs the same attention: remove oil, paint, primer, cutting fluid, galvanized coating, and loose mill scale from an inch either side of the seam, because those things decompose in the arc and inject gas and metal oxides straight into the puddle. Wiping with acetone and a clean rag does more for porosity than any voltage change you will ever make, and welding over coatings is covered in detail in welding galvanized steel with flux core. If your work is dirty because of how you cut it, fix the cut: exhaust work in particular fails from contamination far more often than from settings.
This is the check that turns a list of facts into a diagnosis. Decide on a single change, run one test weld, chip it, look at it, and write down what happened on a piece of tape stuck inside the machine door. If the result improved, keep the change and move to the next suspicion. It also inoculates you against the most expensive habit in this craft, which is cranking the knobs to chase a symptom that has nothing to do with the knobs; the full rogues’ gallery of those errors is in flux core welding mistakes to avoid. The method also scales: when a weld fails an inspection or cracks in service, an inspector from an American Welding Society certified program will start from the same place, verifying the procedure before the operator.
| Check | What “good” looks like | Fastest way to verify |
|---|---|---|
| 1. Polarity | Electrode positive on self-shielded wire | Follow the gun lead to the positive post, or run the two-bead test |
| 2. Settings | Voltage and ipm inside the chart range for the thickness | Mark 12 in of wire, time the free-air feed, convert to ipm |
| 3. Stickout | 1/2 to 3/4 in of visible wire | Set it with a ruler once, then trim to length at every pause |
| 4. Direction | Dragging, 10 to 20 degrees off perpendicular | Run three inches dragging and three inches pushing on one coupon |
| 5. Wire | Dry, bright, uncrushed, correct diameter | Wipe a length with a white rag and look at the color |
| 6. Feed path | Correct tip bore, correct roll groove, slip-test tension | Push the wire against the liner with the trigger running |
| 7. Ground and prep | Bright clamp point, coating removed an inch off the seam | Look for shine under the clamp and clean metal beside the joint |
| 8. One change per coupon | Every test weld differs from the last by exactly one thing | Tape a scrap of paper inside the door and write each result down |
Before you diagnose a single defect, confirm that your baseline is sane, because half of all troubleshooting sessions begin with numbers that were never right. The ranges below are sensible starting points for self-shielded flux core wire on a typical 120-volt machine for the thinner rows and a 240-volt machine for the thicker rows. They are starting points, not gospel: the chart printed in your wire’s specification sheet wins every argument.
| Material thickness | Wire diameter | Voltage | Wire speed (ipm) | Stickout | Travel speed | What wrong looks like here |
|---|---|---|---|---|---|---|
| 22 to 20 gauge (0.030 to 0.034 in) | 0.030 in | 16 to 17 V | 100 to 140 | 1/2 in | 15 to 25 ipm | Too hot burns holes; too cold leaves a tall hump and lack of fusion |
| 16 gauge (0.060 in) | 0.030 in | 17 to 18 V | 130 to 180 | 1/2 to 5/8 in | 14 to 22 ipm | Burn-through at stops, crater cracks, warped panels |
| 14 gauge (0.075 in) | 0.030 or 0.035 in | 18 to 19 V | 170 to 220 | 5/8 in | 12 to 20 ipm | Spatter and popping if voltage drops below the range |
| 1/8 in (0.125 in) | 0.035 in | 19 to 21 V | 220 to 280 | 5/8 to 3/4 in | 10 to 18 ipm | Narrow tall bead at the low end, undercut at the high end |
| 3/16 in (0.188 in) | 0.035 in | 21 to 23 V | 280 to 340 | 3/4 in | 8 to 15 ipm | Cold lap and worm tracks from too little heat |
| 1/4 in (0.250 in) | 0.035 or 0.045 in | 23 to 25 V | 340 to 420 | 3/4 in | 6 to 12 ipm | Machine thermal trips if your duty cycle cannot hold this output |
Every experienced troubleshooter has a story about the afternoon lost to a defect that was not actually the defect. Wrong polarity is responsible for more of those afternoons than every other cause combined, because it does not present as one clear failure — it presents as a cluster of five or six moderate problems occurring at once. When self-shielded flux core runs on electrode negative, the plate becomes the cathode instead of the anode. The wire melts off faster than the puddle can absorb it, so droplets arrive large and irregular, tearing away with the kind of explosive detachment that throws coarse spatter across the whole joint. Penetration goes shallow because the arc is no longer driving heat into the base metal, so the bead sits on top of the plate like caulk instead of biting in, and the toes never wet out, which gives you cold lap on both edges. The flux cannot keep up with the melt-off rate, so the slag film is thin, patchy, and locally fused to the bead, and where the film breaks the flux channel leaves a raised line on the surface — the classic worm track.
Run one coupon at your current settings, flip the polarity, run a second coupon directly beside it without changing anything else, chip both, and compare. The correct polarity will show deeper fusion at the root, tighter ripples, a continuous slag covering, and dramatically less spatter. If the difference is not obvious, polarity was never your problem, and you have just eliminated your most expensive suspicion for the price of two inches of wire — which is precisely what a method is for.
| Symptom | Most likely cause | Fix first |
|---|---|---|
| Pinholes in tight clusters | Cross draft, damp wire, or an oily joint face | Block the wind, load a dry spool, wipe with acetone |
| Loose surface pores along the bead | Travel too fast, pool froze before gas escaped | Slow the travel, raise heat one small step |
| Blowholes deep in the weld | Paint, cutting fluid, or zinc burning in the arc | Grind an inch each side, degrease, re-weld |
| Undercut along the top toe | Voltage high, travel fast, angle aimed at the edge | Drop one volt, slow down, aim at the root |
| Cold lap, bead sitting on the plate | Too little heat or stickout past specification | Raise heat, reset stickout to five eighths |
| Heavy coarse spatter, proud bead | Wrong polarity, usually electrode negative | Set electrode positive, run the two-bead test |
| Arc pops, wanders, stutters | Feed restriction or wire speed out of balance | Measure real ipm, then inspect liner and rolls |
| Wire melts back into the tip | Stickout too short or tip bore oversized | Trim to five eighths, fit the correct tip |
| Birdnest at the drive rolls | Blocked output with rolls still pushing | Clear the blockage, then reset roll tension |
| Wire will not feed at all | Kinked liner, wrong groove, worn guides | Free-feed test with the gun disconnected |
| Burn-through on thin metal | More heat input than the thickness holds | Lower voltage, speed up, use 0.030 wire |
| Bead too narrow and tall | Voltage too low for the wire speed | Raise voltage in half-volt steps |
| Bead too wide and flat | Voltage too high, travel too slow | Lower voltage, speed up the travel |
| Slag will not lift, or is fused | Heat too high, welding over old slag | Chip between passes, drop heat one step |
| Raised worm tracks on the surface | Cold pool or wrong polarity | Verify polarity, then raise heat |
| Cracked crater at every stop | Shrinkage in an unfilled concave crater | Hold position and fill before breaking arc |
| Arc weakens after a few minutes | Voltage sag in the cord or a tripping breaker | Shorten and upsize the cord, dedicated outlet |
| Machine quits mid-pass | Thermal overload beyond the duty cycle | Cool down, clear vents, shorten the pass |
| Bead width wanders along the seam | Inconsistent travel speed and work angle | Support the gun, set a steady pace |
What you see. Pinholes are small round holes scattered along an otherwise sound-looking bead. Surface pores are shallow craters in the skin of the bead, usually at a stop-start point. Clustered porosity is the serious one: groups of holes concentrated in one section, sometimes spongy under the slag, because a localized cause always produces a localized patch.
Why it happens. The flux decomposes to generate a shielding envelope of carbon dioxide and carbon monoxide plus a liquid slag film, and both must stay intact until the puddle solidifies, since molten steel dissolves gas eagerly and rejects it violently on cooling. Oil, cutting fluid, paint, and rust generate their own gas under the arc, moisture in a damp spool adds hydrogen, and zinc from galvanized coating boils near 907 degrees Celsius and disrupts the pool.
The fix, in order.
Confirm the fix. Run two inches on clean scrap, chip it warm, and inspect at arm’s length for continuous ripples with no pinpricks at the toes, then repeat on the real workpiece in the real position. If it survives both, the wire is the last suspect, and a fresh, dry spool costs less than another afternoon of guessing.
What you see. A groove melted into the plate along the toe, usually on the upper edge of a fillet, where metal has been cut away and never refilled. It reads as a dark shadow line beside an otherwise decent bead, and because it is a stress raiser that cracks under fatigue, inspection codes treat it as a reject rather than a cosmetic flaw.
Why it happens. The arc melts the edge for fusion and the added wire normally refills the crater it leaves, so undercut means the arc cut faster than the filler replenished: voltage too high, travel too fast for the deposition rate, a work angle aimed at the edge instead of the root, or a wide weave that dwells on the sidewall and swings away before the puddle catches up.
The fix, in order.
If undercut only appears in one position, compare your horizontal position settings before changing anything else.
Confirm the fix. Run a three-inch fillet on the same joint and drag a thumbnail along both toes; it should catch nothing, and the toe line should read as a continuous shadow rather than a groove. If undercut survives the settings change, travel speed is the culprit, not the machine.
What you see. A bead that looks acceptable from above but sits on the plate rather than in it, with rounded, soft edges that never wet out and a clear line of demarcation at each toe. This is the defect that passes a visual check and fails a bend test.
Why it happens. Fusion requires plate and filler molten together and in contact long enough to mix, so anything that cuts heat into the base metal or moves the puddle away before mixing leaves cold lap: settings too low, travel too fast, a stickout crept past specification so arc energy is spent melting wire instead of plate, a push technique laying metal ahead of the fusion zone, paint or scale preventing wetting, or a heavy fixture sinking heat out faster than the arc supplies it.
The fix, in order.
Confirm the fix. Visual inspection is not enough: cut the coupon and bend it. A fused joint breaks through the weld metal; an unfused one opens at the toe like a hinge. Where cutting is impractical, arc across the suspect tie-in with corrected settings — a proper repair melts the old edge in, a cold one leaves it floating, and the load behavior of each is discussed in how strong a weld really is.
What you see. Frozen balls of metal thrown well beyond the weld zone, stuck to the plate, welded to the nozzle, embedded in adjacent surfaces. Flux core always throws more than gas-shielded MIG because the flux reaction is turbulent, so the line matters: normal is fine beads under a quarter inch from the toe that brush off, abnormal is marble-sized shots an inch or two out.
Why it happens. Spatter is metal expelled by a sudden gas release or an unstable short circuit. Voltage too low for the wire speed makes the wire stub and short before re-ignition, and each re-ignition pops a droplet out of the pool; wrong polarity magnifies this by enlarging the droplets. Contamination boils and ejects metal with it, a long stickout destabilizes the arc column, magnetic arc blow throws it in the direction of travel, and a partly plugged nozzle disturbs the atmosphere around the arc.
The fix, in order.
Confirm the fix. Weld two inches and measure the spatter field with a ruler — anything past a quarter inch means a cause is still active — and listen for a steady frying sizzle instead of sharp cracks. More detail is in reducing flux core spatter.
What you see. An arc that will not hold still: it wanders, sputters, cuts out, re-strikes, and sometimes dies for a fraction of a second before returning. Instead of a continuous sizzle you hear irregular popping, snapping, or a stuttering buzz that changes tempo as you travel.
Why it happens. Stability needs a consistent circuit and consistent consumable delivery, so anything intermittent in either shows up here first. Electrical suspects: voltage and wire speed that do not match, supply voltage sagging under load, a poor ground with variable resistance, wrong polarity, and magnetic arc blow. Mechanical suspects: a kinked or contaminated liner, a worn contact tip arcing internally, drive rolls slipping and delivering wire in pulses, a crushed section of wire in the gun, and a spool brake loose enough to let the wire overrun. The arc is the last link in that chain, so an unstable arc always points behind it, toward the supply described at breaker sizing for welders.
The fix, in order.
Confirm the fix. Run a six-inch bead and grade the sound before the appearance: a stable arc holds one note from start to finish. TWI’s technical notes on arc stability are worth a read when the cause is not obvious.
What you see. The arc dies with a click while the trigger still feels live, and the wire is fused inside the contact tip or the tip is glowing and eroded.
Why it happens. The tip is shorting to the puddle or to a stub welded to it: stickout so short the tip sits in the arc’s heat, a tip bored larger than the wire so current passes through a poor arcing interface, an eroded or spatter-filled tip, or a crater that froze around the wire end because the trigger was released while the tip was still buried in the pool.
The fix, in order.
Confirm the fix. Ten consecutive starts and stops without a melt-back is a pass, and the arc should strike immediately every time — a hesitant strike means the wire end is still attached to something, usually because tip and wire diameter do not match.
What you see. A tangle of wire exploding out in front of the inlet guides or wrapping the roll axle, with a grinding sound from the drive motor. Birdnesting is a blocked-output fault, not a wire fault: the rolls are pushing wire somewhere it cannot go.
Why it happens. Something downstream restricts flow while the rolls keep shoving: a kinked liner, a crushed gun neck, a spatter-filled tip, a tight bend radius, cross-wound or tangled wire on the spool, roll tension so high it deforms the tubular wire, or a groove that does not match the diameter so the rolls grip and release.
The fix, in order.
Confirm the fix. Wire should pass through the disconnected liner with light finger pressure and no catch points, and twelve inches of free feed should leave no slack behind the inlet guide. Routine care for these parts is covered in drive roll maintenance.
What you see. The trigger clicks and you hear the motor, but nothing moves at the gun, or the motor labors and stalls and the arc never establishes.
Why it happens. A total blockage or a total loss of traction: a liner the wrong size, missing, or badly kinked; inlet guides worn wide so the rolls spin on the wire; worn or wrong-groove rolls; tension too loose; a wire end bent at the spool or welded at the tip; or an electrical fault in the trigger circuit, which is a machine fault rather than a feed fault.
The fix, in order.
Confirm the fix. Free-feed twelve inches with the gun in the air and time it: the measured rate should match the display within about ten percent, and the feed must not lag once the arc is struck, as explained in wire feed fundamentals.
What you see. Holes melted clean through the plate, usually at the start where the metal was cold and you paused, or mid-seam where travel slowed.
Why it happens. Thin metal has almost no mass to sink heat, so the arc’s energy has nowhere to go but through. Voltage set for thicker material, wire speed too slow to add filler quickly, travel that drops while you negotiate a joint, a stickout too short and therefore too concentrated, a dwell at the start while the puddle builds, or a ground placed so the current path crosses the thin section all push local heat past what 20 or 16 gauge can carry.
The fix, in order.
Confirm the fix. Weld a coupon of identical thickness and inspect the back for a uniform heat tint and a slight penetration bead with no drop-through; heavy oxidation or lumps mean still too hot or too slow.
What you see. Two opposite failures sharing one root. A narrow, tall, humped bead sits up on the plate with steep sides, poor toe wetting, and a rough surface. A wide, flat, washed-out bead lies nearly flush with thin edges and sometimes undercut, because the arc is spreading rather than concentrating.
Why it happens. Profile is a direct readout of heat input against deposited metal. Too little voltage for the wire speed means the wire arrives faster than the arc can melt and spread it, so the puddle builds vertically and freezes with a high crown. Too much voltage spreads the arc wide and lets filler disperse across it, flattening the bead while the arc cuts the edges.
The fix, in order.
Confirm the fix. Measure rather than eyeball: a fillet should show equal leg lengths with a flat to slightly convex face, and a butt joint about an eighth of an inch of reinforcement across a width of roughly one and a half times the plate thickness.
What you see. The covering that should pop off in clean curls instead adheres: thin, glassy, and stubborn through the middle of the bead, or welded along the toes where it has become part of the surface.
Why it happens. The flux melts into a continuous film that floats on the puddle, then solidifies more brittlely than the steel so differential contraction cracks it loose on cooling. Fused slag means the film was disturbed, too thin, or chemically overwhelmed: welding over old slag without chipping, settings too hot so the film runs thin and breaks at the toes, travel fast enough to tear the cover before it forms, damp wire altering the flux chemistry, or an angle that aims the arc at the film behind the puddle.
The fix, in order.
Confirm the fix. A correct bead releases in long curls under light, sharp taps and leaves a metallic surface with visible ripples and no black film. Where film remains, note your settings at that moment and compare them with the sections that lifted clean — the difference between those two readings is the diagnosis.
What you see. Thin raised lines running along the bead surface or just under the slag, like something has dragged a worm across the weld.
Why it happens. The tubular wire deposits flux in a channel the arc is supposed to stir into the pool. When the pool is too shallow or too viscous — settings at the low end, stickout past specification, travel too fast, wrong polarity — the channel solidifies before it mixes, and unincorporated flux freezes as a raised ridge.
The fix, in order.
Confirm the fix. The next pass comes out with a smooth skin under the slag and no ridges; tracks that persist in one section usually mean a flattened or kinked section of self-shielded wire feeding through at that point.
What you see. The weld runs perfectly until you stop, and the restart is ugly: the arc hesitates, pops, or refuses to strike until you jab at the plate, and the next few inches are erratic.
Why it happens. At the stop the wire end freezes into the crater and the tip cools, so the next start is a re-strike through an oxidized, sometimes welded wire end. A liner with slack springs the lead forward after the trigger opens, a loose spool brake lets the spool overrun, an eroded tip arcs internally, and a crater left unfilled gives the wire a bad surface to start on.
The fix, in order.
Confirm the fix. Make five stop-and-start welds in a row: each restart strikes on the first try with the same sound as the original pass, and the wire does not move after you release the trigger.
What you see. The arc dies partway through a weld with no feed and no arc, usually after several minutes of continuous work.
Why it happens. Every welder has a duty cycle: the percentage of a ten-minute period it can run at a stated output before an internal protector trips. A small 120-volt machine may manage twenty or thirty percent at maximum, which means four minutes of arc and six of waiting, and pushing past that trips the switch by design. Restricted vents, a machine sitting on carpet, high ambient temperature, and low supply voltage shorten the time to trip.
The fix, in order.
Confirm the fix. Time a continuous bead at your working settings: if the machine holds it for a typical pass without tripping, you are inside the duty cycle. Miller’s technical resources publish duty cycle definitions and test conditions, and our duty cycle guide covers the shop-side fixes.
What you see. The arc is fine on the bench and weak at the far end of the shop, the machine seems to lose authority as the cord warms, the lights dim when you pull the trigger, and the feed motor sounds like it is climbing a hill.
Why it happens. Current through a conductor causes a voltage drop proportional to length and inversely proportional to cross-section, so a long thin cord steals volts before they reach the machine. A welder fed 105 volts produces noticeably less output than one fed 125, and because the control electronics and feed motor run off that same supply everything sags together: the arc softens and the motor slows.
The fix, in order.
Confirm the fix. Read the voltage at the machine while a helper strikes an arc: a drop of more than a few volts means the supply is still the limit. Generator users see identical symptoms for identical reasons, as explained in generator sizing for welders.
What you see. The first half inch of every weld is cold, lumpy, and poorly fused, sometimes with a stub of wire left on the plate, while the rest of the bead is fine. Restarts are worse: the arc spits, fails to establish, or melts a small crater before catching, leaving a visible hard spot at every restart.
Why it happens. The plate at the start is at ambient temperature, so the arc builds a puddle on a heat sink while the wire delivers cold metal into it. Add an oxidized wire end from the last stop, an unfilled crater, mill scale at the strike point, stickout that lengthened while you positioned the gun, or settings tuned for the middle of the pass, and the start will always be the weakest inch of the weld.
The fix, in order.
Confirm the fix. Run three starts and inspect the first half inch of each: full fusion, the same width as the middle of the pass, and no transition line where the arc struck. Starting discipline is covered in flux cored arc welding techniques.
What you see. A concave dish at the end of the bead, often with a star-shaped crack or a ring of fine cracks in the center, sometimes visible only after grinding the stop flat.
Why it happens. Molten steel shrinks as it solidifies and a crater is a pool with no incoming filler to compensate, so the last metal to freeze is pulled apart under tension. This is solidification cracking, not a shielding problem, which is why cleaning never fixes it.
The fix, in order.
Confirm the fix. Grind the stop flat: a filled crater shows solid metal with no star crack, and ten stops in a row without one means the technique has taken.
What you see. A lumpy overlap where a new pass begins, sometimes with a dark line of trapped slag between the old crater and the new bead, and a cold, unfused patch at the joint.
Why it happens. Starting directly on a crater piles fresh metal over an oxidized, possibly cracked surface that cannot fuse, and the geometry leaves a pocket that traps slag and gas.
The fix, in order.
Confirm the fix. Grind a section through the tie-in and look for a continuous bead with no dark seam. Restarts are where cracks start, so the check is worth the thirty seconds.
What you see. Panels that oil-can, frames twisted out of square, tube pulled toward the weld side, and gaps that were tight before welding and open afterward.
Why it happens. Weld metal contracts as it cools while the surrounding plate is restrained unevenly, so the part bends to accommodate the shrinkage. Heat input sets the size of the affected zone: slow travel, high voltage, large fillets, and long continuous seams all distort more. Sequence matters as much as energy — welding one side completely before the other concentrates shrinkage on one face — as do clamp position, fit-up gaps that close as you weld, and parts already stressed from cutting or forming.
The fix, in order.
Confirm the fix. Measure against your baseline before and after and record it: a within-tolerance result on a repeat part means the sequence works. Distortion that changes between identical parts usually means inconsistent travel speed or clamping, and panel welding practice covers the sheet-metal case.
What you see. A bead that swells and narrows along the seam, ripples that change spacing, and edges that wash in and out. It is the most common cosmetic complaint and it almost never means the machine is wrong.
Why it happens. Puddle width tracks three things the operator controls: travel speed, work angle, and stickout. Speeding up narrows the bead and tightens the ripples, slowing widens it; drifting the angle as your wrist rotates through a long seam moves the puddle sideways; letting the gun creep away lengthens stickout and shifts amperage mid-run.
The fix, in order.
Confirm the fix. Run a six-inch bead and measure width at three points: variation beyond a sixteenth of an inch means technique is still drifting, and the same drill applies whichever process you are practicing — deliberate practice cures it faster than new equipment.
Troubleshooting is what you do after something goes wrong, and most of the defects in this guide never appear if a handful of unglamorous habits are part of every setup. Run this list before the first weld of the day.
Run two identical beads on clean scrap, flip the polarity, and run a third beside them without changing anything else. If the reversed bead throws more spatter, sits proud with soft toes, and leaves patchy slag, you found it. Self-shielded wire belongs on electrode positive, and the polarity guide shows the switch on every machine type.
Most self-shielded wires on small machines want a contact tip to work distance of about half an inch to three quarters, with your wire’s specification sheet as the final word. Measure once with a ruler so you know what correct looks like from behind the hood.
Drag. The shielding gases and slag come off behind the wire, so pulling keeps the arc on clean metal and lets the cover follow it. Pushing lays flux ahead of the arc and produces lack of fusion, worm tracks, and an erratic arc. The trade-offs and exceptions are in push versus pull.
No. Porosity is a hole, and welding over it traps the defect under new metal while repeating the original cause. Grind or gouge back to sound metal, fix the root cause — wind, moisture, contamination, or settings — and re-weld. The reasoning is in why porosity forms.
Three usual suspects: the machine reaching its duty cycle, an extension cord heating and dropping voltage, or a breaker close to tripping. Time the failure — a consistent number of minutes points at thermal limits, a gradual decline at voltage sag. Check the numbers in our settings reference.
Because the tip is in the heat zone or the crater was not filled. Reset stickout to about five eighths, fill the crater before releasing the trigger, and fit a tip that matches the wire exactly — an oversized bore arcs internally and overheats, which is why wire and tip matching matters more than it looks.
Start mid-range for your thickness, weld a coupon, then adjust one variable at a time: raise voltage until the bead wets to the toes, balance wire speed until the profile is flat to slightly convex, and write the numbers down.
Undercut, lack of fusion, crater cracks, and porosity all reduce load-bearing section or introduce stress raisers, so yes, they are structural rather than cosmetic. If a weld carries real load, treat every defect in this article as a repair item before it goes into service.
Four things, weekly: clean the liner and nozzle, check drive roll groove and tension, blow out the machine’s vents, and inspect your power cord and plug. Add proper wire storage and the habit of trimming the wire before each start, and you will have removed the cause of most feed faults, porosity, and arc instability. Machine care overlaps with shop safety habits, so the routine protects quality and people both.
Ellis’s four-hour afternoon ended with a closed door and a dry spool, but the real fix came earlier than that: he changed five variables at once and learned nothing from any of them. Everything in this article exists to prevent that. Run the eight checks in order, keep settings honest against a chart, measure stickout and wire speed instead of guessing them, drag the gun, protect the shielding, clean and ground the joint properly, and change one thing per coupon with the result written down.
The symptom sections are for the days when the method alone does not close the gap: match what you see, read the physics, work the fix list from the top, and confirm with a test rather than a hope. Flux core punishes only one habit badly — guessing — and a disciplined setup turns a stubborn afternoon into eleven minutes of work. When you want the wider frame, Britannica’s overview of welding and Lincoln Electric’s technical library are good places to keep reading.