Guide to CNC plasma cutting

• 1
  How the plasma arc forms and cuts metal
• 2
  What kerf, taper and the heat-affected zone do to a part
• 3
  Thickness limits and where the process stops working
• 4
  Consumables, gas and running cost per part
• 5
  What happens after the cut, and when to switch to milling
• 6
  Plasma cutting compared with other cutting methods

A plasma torch holds a copper nozzle with a small orifice and a tungsten or hafnium electrode behind it. Gas flows through the gap and a pilot arc jumps between electrode and nozzle. When the torch nears the workpiece, the arc transfers to the plate and the gas heats until it ionizes. That ionized gas is the plasma, and it leaves the nozzle at roughly 20,000 °C.

The constricted arc melts a narrow strip of metal. The high-velocity gas jet pushes the molten material out the bottom of the kerf. Nothing burns away in the chemical sense. This is a melting and blowing process, which is why it works on any conductive metal rather than only on steel.

Cut quality comes from three settings working together: current, gas flow and travel speed. Raise current and the arc widens, so kerf grows and the edge squares less. Raise speed and the cut narrows, but too fast leaves dross welded to the bottom edge. The window between those two failures is narrower on thick plate than on thin sheet.

Oxygen or air is the usual gas for carbon steel. Stainless and aluminum are cut with nitrogen, or with argon-hydrogen blends where the edge must stay clean. The gas choice changes the chemistry of the cut face more than it changes the geometry. On 304 stainless, an air plasma leaves an oxidized edge that must be ground before welding.

A CNC table adds nothing to the physics. It only holds the torch at a constant standoff, moves it at a programmed feed rate, and keeps the pierce points off the finished profile. The machine's job is repeatability: the same part at 8:00 and at 16:00. Hand cutting cannot hold that, which is why the process moved to gantry tables.

Kerf is the width of material the arc removes. On a 1.5 mm carbon steel sheet with a 45 A cut, expect roughly 1.0–1.5 mm. On 12 mm plate at 130 A, kerf runs closer to 2.0–3.0 mm. The kerf is not a fixed offset you can set once. It changes with current, speed, gas and electrode wear.

Taper is the angle of the cut face. A plasma kerf is wider at the top than at the bottom, because the arc loses energy as it travels down. On thin sheet taper may be 1–2°, which is invisible. On 20 mm plate it can reach 4–6°, and a 4° taper over 20 mm shifts the bottom edge by about 1.4 mm.

The heat-affected zone is the band of metal whose microstructure the arc has changed. It is narrow, but on carbon steel above roughly 0.3% carbon it can harden enough to crack during forming. Parts that will be bent or welded usually need the plasma edge dressed back before the next operation.

Hole quality is the weakest point of the process. A hole smaller than about 1.5 times the plate thickness will come out tapered and out of round, because the arc has no room to lead in. For holes that must hold a bolt or a bearing, drill or mill them after cutting.

Corner rounding is the other limit. The torch cannot reverse instantly, so every sharp internal corner gets a radius roughly equal to half the kerf. If the drawing calls for a true sharp corner, plasma is the wrong process for that feature.

Every one of these effects is a function of thickness. Thin sheet behaves almost like a shearing operation. Thick plate behaves like a thermal cutting operation with real distortion. Treat 3 mm and 25 mm as two different processes that happen to share a torch.

A typical 45 A air plasma on a hobby table will sever 10 mm carbon steel, but it will only produce a clean, weldable edge up to about 6 mm. The gap between sever and cut quality is where most shop arguments come from. A vendor saying a machine cuts 25 mm usually means it severs 25 mm.

Production machines at 130–200 A hold good edge quality on carbon steel to roughly 20–25 mm and can sever past 50 mm. Aluminum and stainless behave differently: they conduct heat away faster, so the practical clean-cut limit is lower than for carbon steel at the same current.

Above 50 mm, plasma stops being economical. The kerf widens, dross becomes hard to remove, and the heat input distorts the plate. At that thickness, waterjet or oxy-fuel usually wins on edge quality even if it is slower.

Material matters as much as thickness. Copper and brass conduct heat so well that they need high current and still cut with a rough edge. Titanium cuts cleanly but reacts with air, so it needs an inert gas shield. Coated or painted steel should be stripped before cutting, or the coating will contaminate the edge.

Stack cutting changes the numbers. Two 3 mm sheets clamped together can be cut as one 6 mm stack, which doubles throughput and reduces per-part taper. It only works if the sheets are flat, clean and firmly clamped, otherwise the arc wanders between them.

The process also has a pierce limit. Piercing thick plate blows molten metal upward and eats consumables. Many shops drill a start hole above 12 mm and edge-start from there. That single change can double nozzle life.

A plasma torch is a consumable stack: electrode, nozzle, swirl ring and shield. Electrode and nozzle wear together, and worn parts widen the kerf and increase taper before the cut visibly fails. Replace them on a count or on a kerf measurement, not on how the cut looks.

Gas cost is usually smaller than most shops assume. Electricity and consumables dominate on thin sheet, where the torch runs at high duty cycle and pierces often. On thick plate, gas and pierce time dominate. The crossover is roughly around 12 mm on carbon steel.

Pierce count is the hidden cost driver. A nest full of small parts may pierce 400 times per sheet. Each pierce consumes electrode life faster than the cut length does. Grouping parts and using common-line cutting reduces pierce count and cuts cost per part without changing any setting.

Cut speed matters for cost but also for quality. Running slower than the book speed raises heat input, widens the heat-affected zone and produces more dross. Running faster than the book speed leaves a ragged bottom edge. The book speed in the consumable chart is a starting point, not a target.

Air plasma is the cheapest gas option and works well on carbon steel. It is a poor choice for stainless or aluminum that will be welded, because the cut face picks up nitrogen and oxygen. That contamination shows up later as porosity in the weld.

A plasma-cut edge is a starting point, not a finished surface. Most parts get deburred, then either welded, formed or machined. The cut edge is square enough for a fillet weld in most cases, but a full-penetration weld on thick plate usually needs the edge ground back to sound metal.

When a drawing calls for a bore, a pocket, a thread or a flat face, plasma cannot deliver it. Those features need material removal with a controlled tool path. Our shop runs 127 high-precision CNC machines, including 16 simultaneous 5-axis machining centers and 16 mill-turn centers, and we hold ±0.005 mm on milled features.

A common production route is plasma for the blank and milling for the features. Cutting the outline on a plasma table and then fixturing the blank for a 3-axis mill is often cheaper than milling the whole profile from solid plate, especially on parts above 300 mm across.

For a mixed order, we quote both operations together. Quotation and free DFM analysis come back within 12 hours, production can start within 24 hours, and parts ship in 3–5 days. There is no minimum order quantity, from one prototype to 10,000+ part runs.

If the part is a thin bracket, a gusset or a mounting plate, plasma alone is usually enough. If it carries a bearing, seals a fluid or bolts to a machined mating face, plan the machining step before you release the drawing.