Stamping vs CNC machining: choosing for flat parts

A 5 mm steel plate with thirty holes and a rectangular cutout can be made three ways: pressed, milled or laser cut. All three produce a part that fits, and the quotations can differ by a factor of three. Stamping vs CNC machining is settled by arithmetic on four numbers — cycle time, tolerance, scrap and tooling — and the arithmetic is short enough to do before the RFQ goes out.

The usual mistake is to compare piece prices. A milled part has no tooling and a high piece price; a stamped part has a tool and a low piece price; the comparison only means something once the tool is amortised over the real annual volume. The second mistake is to assume the tighter tolerance decides it. On a bracket, the tolerance is rarely the constraint; on a dowel hole it always is.

What follows is the cycle-time comparison, the crossover calculation, the tolerance and scrap differences, and the cases where neither process is the right answer.

Stamping vs CNC machining: when does the press win?

Stamping wins when the annual volume is high enough to amortise the tool, the tolerance is no tighter than about ±0.1 mm, and the part is a flat or formed sheet between 0.8 mm and 6 mm. CNC machining wins when the tolerance is tighter than ±0.05 mm, the part needs a pocket or a machined face, or the volume is too low to justify a die.

The decision is a break-even, not a rule of thumb. Everything else — burr, edge quality, flatness after stress relief — is a tie-breaker on parts that sit near the crossover.

The crossover arithmetic, worked

Cycle time first, because it sets the size of the prize. Take the 5 mm plate: 30 holes of 8 mm, a 120 × 80 mm cutout and an outside profile. The holes need 30 × 25 mm of path, the cutout 400 mm and the profile 800 mm, and the finishing pass, lead-ins and tool repositions take the total to about 5,200 mm. At a 2,000 mm/min feed rate that is 2.6 minutes of cutting, four tool changes at 30 seconds add 2 minutes, and loading adds 1 minute. Six minutes per part.

The press does the same part in one stroke. At 30 strokes per minute the cycle is 2 seconds, and with manual loading and unloading the part takes about 6 seconds. The ratio is 60:1, and that ratio is the engine of the whole decision.

Now the break-even. Let the tooling cost be T, the machining cost per part be C, and the stamping cost per part be S, where S includes material and press time but not tooling. The break-even quantity is N = T / (C − S).

Put the numbers in machine-hour units so the result does not depend on a currency. Machining at 6 minutes per part uses 0.1 machine-hours. Stamping uses 6 seconds, or 0.0017 hours, plus material, which for the same plate is the same in both routes. If the tool is priced at the equivalent of 500 machine-hours, the break-even is 500 / (0.1 − 0.0017) = 5,090 parts.

That last number is the one to take to a quotation review. A tool priced at 500 machine-hours pays back at about 5,100 parts, and a tool priced at 1,000 machine-hours pays back at about 10,200. The arithmetic is identical in rupees or dollars as long as the rates are consistent.

What the operation sequence looks like on each route

The two routes diverge after the first cut. A stamped part starts as a coil, is fed through a progressive die that blanks, pierces, forms and often welds or taps in the same tool, and comes out as a finished part every 2 seconds. A machined part starts as a saw-cut plate, is fixtured once, faced, profiled, drilled and tapped, then deburred and inspected. The machining centre holds the part for the whole cycle, which is why the per-part cost tracks machine time so closely.

That sequence difference has a commercial consequence. Stamping costs are front-loaded into the tool, so the price per part falls sharply as volume rises. Machining costs are spread evenly, so the price per part is almost flat and falls only with better programs, better cutters or a second fixture that halves the load time.

It also changes how a design change is absorbed. A revised hole position in a stamping is a tool change with a lead time and a cost. The same revision in a machining programme is an edit and a re-post, usually within a day. Early in a programme, that flexibility is worth more than the piece price difference.

Where the tolerance actually decides

Stamped hole position holds about ±0.1 mm, and a checked feature with a fixture can hold ±0.05 mm. A machining centre holds ±0.02–0.05 mm on position without effort, and ±0.01 mm with a boring operation.

That difference matters on three features and almost nowhere else: bearing bores, dowel holes that locate two parts, and any dimension that sets a running clearance. On a bracket with 11 mm clearance holes for M10 bolts, a ±0.1 mm positional tolerance is already generous, and paying for machining to hold ±0.02 mm buys nothing the bolt notices. The progressive die tolerance guide sets out what a press can actually hold feature by feature.

There is a second tolerance trap that neither process controls: flatness. A 5 mm plate rolled at the mill carries residual stress, and removing material from one face — by machining or by blanking — releases it and the plate bows. A milled plate can bow after the last pass, and a stamped plate can bow after blanking. If flatness is the requirement, the answer is usually stress-relieved material, a thicker plate or a straightening operation, not a tighter cut.

Hole fit deserves its own calculation, because it is the place where a stamped part is asked to do a machined part's job. An 8 mm dowel at h7 runs 7.985–8.000 mm. A pierced 8 mm hole with a +0.1/0 tolerance and a slight taper can measure 8.00–8.10 mm at the entry, so the dowel floats by up to 0.115 mm before the taper grips. A reamed H7 hole closes that float to 0.015–0.035 mm. If the hole locates, ream it; if it clears a bolt, pierce it and stop worrying.

Where the drawing is silent, ISO 2768-1 supplies general tolerances for parts formed from sheet metal and for metal removal, which is why a part can pass the general class and still fail at assembly. Put the tolerance on the features that matter instead of relying on the class.

Scrap, burr and the second operation

Material utilisation favours machining on paper and stamping on labour. A progressive strip uses 60–70% of the blank area, so 30–40% of the steel becomes skeleton. A nested laser uses about 92%. Machining from plate uses 80–85% of the plate area, but the removed material becomes chips that carry the cutting fluid with them.

On a 1.19 kg bracket, that difference is 0.64 kg of scrap per part on the press against 0.2 kg on the machine. Multiply by 100,000 parts and the material line separates the two routes by 44 tonnes of steel. Scrap is sold back, but the credit rarely recovers more than about a third of the purchase rate, so the net material penalty stays close to 70% of the difference: 31 tonnes of bought-in steel that produces nothing.

The crossover therefore moves with the blank ratio, not with the press alone. On a heavy bracket in a high-cost steel grade, laser cutting with 92% nesting can beat a progressive die on total cost at volumes well above the usual threshold, because the press saves labour and loses material. A part with a compact outline behind a wide carrier is a worse press candidate than one nested tightly.

Burr is the hidden second operation. A sheared edge has four zones — rollover, burnish, fracture and burr — and the burr height on 5 mm CRCA is typically 5–10% of thickness, so 0.25–0.5 mm. A machined edge carries an exit burr that is smaller but present on every profile. Both need deburring if the part is handled, and deburring is a per-part labour cost that often equals the cutting cost at low volume.

Holes are the other second operation. A pierced hole in 5 mm CRCA has a slight taper and a fracture zone, so a hole that must take a dowel is drilled or reamed after piercing. That is fine in a progressive die — add a station — but it is a station you have to plan for at the tool design stage, not a detail to add after the first article fails.

What the press shop covers, and what we do not

Stamped parts run on 20 T to 300 T mechanical presses with 250 T punching, and the tooling is designed and built in the in-house tool room alongside the welding and checking fixtures. The blank-to-part route is press, weld on three CLOOS robotic cells or at 30-plus manual stations, then shot blasting, seven-tank pretreatment and powder coating in the 18 ft × 14 ft booth.

We do not run a CNC machining centre. When a part needs a milled pocket, a bored bearing seat or a turned feature, we say so rather than quoting a route we cannot hold, and the honest answer is that the part is either a machined part or a pressed blank that goes to a machinist for the critical features. The stamping capability page lists what the presses and the tool room cover, and the stamping and press tool hub has the tonnage and tooling articles behind this comparison.

What to put on the drawing

1. Process-neutral geometry, with the flat pattern and the formed dimensions stated separately.
2. Tolerance on the features that matter, with a datum, rather than a general class alone.
3. Hole function: clearance, tapped, dowel or bearing, because the function decides whether the hole is pierced, drilled or reamed.
4. Burr limit, stated as a height and a direction if the part is handled or stacked.
5. Flatness, with the measurement method and the state — after blanking, after forming, after coating.
6. Annual volume and the release pattern, because the tooling decision depends on both.
7. Material grade and thickness, with the standard, because formability and springback follow the grade.

The stamping capability page lists the press range, and the laser versus stamping comparison covers the third route when the volume sits between the two.

When neither process fits

Stamping is wrong when the volume cannot carry a tool, when the design is still moving, and when the part needs a machined feature that a die cannot produce. CNC machining is wrong when the volume is high, the part is thin, and the per-part machine time dominates the price.

Laser cutting plus forming is the third route and it wins a large middle band: it needs no tool, holds ±0.1 mm on position, uses about 92% of the sheet and runs at any volume from one to a few thousand. What it does not do is compete with a press on cycle time, so above roughly 5,000–10,000 parts a year the press usually takes the part back.

Fine blanking is the fourth route, and it is the answer when the edge must be square and smooth with a controlled shear zone. It needs a special press with a counter-punch and a die cushion, a tool built to a much tighter clearance than a conventional blanking tool, and a strip that is often annealed first. The result is an edge with 90% or more of its thickness cut clean, which removes the deburr and the fracture zone but adds tool cost and limits the material grades that run well.

Send the drawing with the annual volume and the tolerance on the critical features through the quotation route. If the part belongs on a machine we do not run, the reply will say so, and that answer is worth more than a quotation for a route we cannot hold.