MIG versus robotic MIG welding for OEM parts

A 5 mm fillet laid at 300 mm/min puts down a metre of weld in 3.3 minutes of arc time. On a manual bench that metre costs around 8 minutes of paid labour once tacking, repositioning, spatter removal and crane time are added. In a robotic cell it costs around 4 minutes, load and unload included. That gap is the real case for one route over the other, and it is narrower than most quotations assume.

The choice between MIG versus robotic MIG welding is not a contest of welding speed. Both torches travel at the same 300–400 mm/min for a 5–6 mm leg. What changes is how much of the paid hour has the arc lit, and what has to be built and maintained to keep it lit on every part.

The decision rule we apply is a band, not a preference. Below roughly 150 parts per release, manual MIG wins. Above roughly 250 parts per release, a cell usually repays its fixture inside the first release. Between the two, a fixture cost review decides. The crossover arithmetic, with the fixture hours and the rework rate, sits in the robotic versus manual welding cost comparison.

What decides the route before quantity does: access, fit-up, presentation

Three properties of the part decide whether a cell is even possible: whether the torch can reach every seam, whether the parts arrive identical, and whether the whole assembly can be built in one fixture. A part that fails any of them is a bench job at any volume, because the cell will spend its arc-on advantage waiting for a fitter.

Joint access has a number

A MIG torch with a 16 mm nozzle and a 45° neck needs the shroud 10–15 mm from the joint with the arm behind it. In practice that means an approach within ±30° of the joint normal, 80 mm of clear space behind the nozzle along that approach line, and 150 mm of free run-in and run-out at each end of the seam.

A person breaks those rules routinely. The welder tilts the torch, lets the shroud ride over the rib and drags the seam from an awkward angle. A six-axis robot with a fixed program cannot, and a collision between the torch body and a stiffener is a crash, not a compromise.

This is the most common reason an existing part cannot move to a cell. The design was made around a hand, and the hand is more adaptable than the arm.

Torch angle and stick-out are fixed in a cell

A welder varies two things without thinking: the work angle, and the wire stick-out, meaning the length of wire between the contact tip and the arc. Both change penetration, bead shape and spatter.

A push angle of 10–15° gives a flatter bead and less spatter; a drag angle of the same amount gives deeper penetration and a narrower bead on 5 mm and above. Stick-out of 10–15 mm is typical for short-arc; beyond 20 mm the current drops, the arc becomes unstable and the joint loses fusion at the root. A hand corrects for both through a seam that moves. A program holds whatever it was taught, which is an advantage only if the taught value was right for the whole seam.

Fit-up must be repeatable, not merely acceptable

A manual welder absorbs variation. When the gap opens to 2 mm, the welder weaves, adds wire, slows down, and the joint is filled. The robot runs the program it was taught.

On 2–3 mm sheet, a gap beyond about 1.0 mm causes burn-through at normal parameters. A mismatch beyond about 1.0 mm produces a short leg on one side and undercut on the other, which a visual inspection may pass and a fatigue or leak test will not.

Bend angle is where that number comes from. A flange angled 1° off nominal moves its edge by 1.75 mm at 100 mm from the bend, because tan 1° × 100 mm = 1.75 mm. A fixture that locates on the cut edge therefore inherits the full bend error, while one that locates on the formed face and the flat halves it. Fixture datum choice, covered in the fixture-controlled welding guide, matters more to a cell than to a bench.

Presentation in one fixture

Every re-clamp is a new datum and a new chance to move the part. A repeat assembly should be built in one fixture with 3-2-1 location: three points on the primary face, two on the secondary, one on the tertiary. Clamps belong at least 60 mm from the seam so the shroud and the gas flow are not disturbed, and copper or ceramic backing belongs under any seam that runs over an open gap.

The numbers that separate MIG versus robotic MIG welding

Read the table left to right and the pattern is clear. The cell wins on consistency and on arc-on time; it loses on changeover and on tolerance to variation. A part with a stable geometry and a generous run is a cell part. A part with six variants at 120 pieces each is a bench part even if the annual total is large.

What the first article has to prove

A cell is only approved when the first article proves the process, not the part. That means a record taken on the fixture, with the part still clamped, of four things: gap and mismatch at every numbered seam, the leg size and length across the weld, the distortion of the assembly against the drawing diagonal, and the spatter and undercut condition on the visible face.

Practically, five parts are run before the route is released. The first is measured fully; parts two to five are measured at the three seams that vary most. If a seam moves beyond 0.5 mm across those five parts, the fixture or the fit-up is the cause, and no amount of program tuning will remove it.

Three failure modes account for most of what an audit finds on a cell part. Porosity from a draught across the shielding gas, usually because a shutter or a fan is open near the cell. Spatter build-up on the fixture stops, which changes the part height by 0.2–0.4 mm over a shift and silently opens the joint. And wire feed variation from a kinked liner, which shows as intermittent lack of fusion on a seam that passed at first article.

ISO 5817 sets quality levels for steel fusion welds and defines what counts as an imperfection and at what level. Its scope covers steel from 0.5 mm thickness upward, with quality levels B to D; it does not cover aluminium, which sits under ISO 10042, nor does it set acceptance for a specific load case — that stays with the drawing.

Weld symbols and the numbered weld map

ISO 2553 governs how a weld is called out on the drawing: symbol, leg length, seam length, pitch and whether the weld runs all round. Its scope stops at representation; it does not tell anyone what leg size is adequate for a load.

For a cell, the drawing must also carry a weld map — every seam numbered, with its own symbol and its own length. The robot program is then written seam by seam, and the inspector checks the same seams in the same order. Without numbering, a 14-seam frame becomes a discussion about which 6 mm fillet was undercut.

Two details on the map change the cost more than the leg size does. An intermittent stitch pattern, typically 60 mm of weld on a 120 mm pitch, halves the arc time and reduces heat input into thin panels. A weld-all-round callout on a closed box forces the part to be turned, which usually means a second fixture.

Part mix, changeover and the run that pays

A cell running one part number for a week is close to its best case. A cell changing part every two hours is a manual shop with extra capital in it, because 20–40 minutes of program recall and fixture swap on a 25-minute cycle removes most of the arc-on gain.

The test is simple arithmetic. Multiply the changeover time by the number of variants per release, and compare it with the cell hours available. If changeover consumes more than about 10% of the cell's planned hours, split the family: run the two highest-volume variants in the cell and keep the rest on a bench with a dedicated checking fixture.

Variant design also matters. If six canopy sizes share the same joint layout and differ only in length, one program with a shifted frame will often cover all six. If each variant has a different door cut-out, each needs its own fixture, and the family belongs on the bench.

How Ganesh routes a repeat weldment

Ganesh runs three CLOOS robotic MIG cells and more than 30 manual MIG/MAG, spot and SPM stations, with fixtures designed and built in-house; the equipment is listed under welded assemblies. That mix is deliberate: it lets a family of parts sit where it belongs rather than forcing every weldment into the cell.

Upstream of welding, the 12 kW fibre laser cuts on a 40 ft × 16 ft bed, so a 4,000 mm panel arrives with a straight, repeatable edge for the fixture to locate against, and the 200 T, 5-axis brake holds flange angle over a 4,450 mm bed up to 10 mm. Downstream, welded assemblies go through seven-tank pretreatment and the 18 ft × 14 ft powder booth. Customer work in this shape includes elevator frames for TKE and Eros Elevators, genset base frames for Mahindra Powerol and railway assemblies for Indian Railways.

Ganesh has manufactured sheet metal since 1992 across five units in Mumbai (Vasai) and Nagpur, holds ISO 9001:2015 and ISO 14001:2015, and offers PPAP and first article inspection. For a cell part, the first article record includes the weld map with leg size, the fixture gap check and the diagonal of the assembly, because those three catch most of what goes wrong later.

What to put on the drawing

• Material grade and thickness per part, to IS 2062 or IS 513, with the thickness tolerance.
• Weld map: numbered seams, ISO 2553 symbols, leg size, length and pitch.
• Which welds are structural, which are cosmetic, and which are sealing.
• Joint gap and mismatch the design assumes, and the fit-up the fixture must hold.
• Fixture locating faces, marked as machined or as-formed.
• Any seam that cannot be reached with an 80 mm clearance behind the nozzle, flagged on the drawing.
• Distortion allowance and the diagonal or flatness limits on the finished assembly.
• Surfaces that must stay weld- and spatter-free, and the coating that follows.

The welded assemblies hub collects the rest of this cluster, including distortion control and fixture design.

When the cell is the wrong answer

A cell is wrong for a one-off prototype, for a part whose geometry is still moving, and for an assembly with a seam on the inside of a closed box that can only be reached before the last panel goes on. It is also wrong for heavy multi-pass work in 10 mm plate, where the arc time is long, the distortion is large and a positioner with a manual welder is the more controllable route.

Ganesh does not offer laser welding, does not weld on site, and does not treat robotic welding as a fix for a part that cannot hold its own fit-up. If the drawing cannot hold 1.0 mm on gap and mismatch, the honest answer is a bench route with a checking fixture, not a program that fights the part.

Send the assembly drawing with the weld map, the quantity per release and the variants in the family. The engineering team reviews joint access, fixture strategy and the route before any price is put on the part, and the RFQ checklist lists what to attach.