Cooperate to Produce the Largest High-Precision Machine Tool in the World

• 1
  Why two builders had to cooperate to produce the largest high-precision machine tool
• 2
  Heat is the first enemy when you produce the largest high-precision machine tool
• 3
  Hand scraping sets the geometry that a control cannot fix
• 4
  Where 114 kW goes, and why stiffness beats raw power
• 5
  What this machine-tool story means for your own parts
• 6
  Dixi 210 P class machine vs typical large 5-axis work

A universal machining center with a 1,800 × 2,100 × 1,450 mm work envelope cannot be built by one team working alone. The Dixi 210 P came out of a joint effort between Mori Seiki and Deckel Maho, and the reason is geometric. Every extra millimeter of travel multiplies the alignment problem. A builder who owns the spindle and the builder who owns the bed each carry half the error budget, and neither can absorb the other's stack-up.

The numbers matter here. The machine holds spatial accuracy under 35 μm across the full cube, which is roughly one third of a human hair over a 2-meter diagonal. That figure is not a specification pulled from a catalog. It is the sum of squareness, straightness, and thermal drift measured under running conditions. To reach it, the two companies had to share not just drawings but measurement data, and agree on which side corrects what.

There is a commercial logic too. Building a machine this size in a single factory means one company carries the full capital risk for a tool that may sell a handful of units. Splitting the design lets each partner invest in the part it already makes well. The collaboration is not a marketing story. It is risk sharing on a scale that a single mid-size machine builder could not underwrite.

For anyone reading this who runs a shop, the takeaway is simple. When a part gets large and tight at the same time, no single process owns the result. The same logic shows up in how we quote large work.

• 1
  Shared error budgetEach partner owns the tolerances it can control and measure.
• 2
  One datum chainBoth sides align to the same reference before assembly.
• 3
  Split capital riskNo single builder funds the full development cost.

Cast iron grows about 11 μm per meter for every 1 °C rise. On a 2.1 m column, a 5 °C swing moves the spindle nose roughly 115 μm. That is more than three times the machine's total spatial accuracy. No control system can compensate its way out of a drift it cannot predict, so the design has to remove the heat at the source.

The Dixi 210 P cools every significant heat generator: spindle bearings, drive motors, ballscrew nuts, and the hydraulic unit. Coolant is circulated and temperature-controlled, not just pumped. The goal is not to keep the machine cold. It is to keep the whole structure within a narrow band, so the thermal gradient across the column stays small and predictable.

This is why a machine of this class is usually run for hours before a critical cut. The structure has to reach steady state. Cold starts are the worst case for accuracy, not the best. An operator who takes a finishing pass ten minutes after power-up is fighting physics.

The same principle applies at a smaller scale. On any job where the tolerance is a few micrometers and the part is longer than 500 mm, let the machine idle to temperature first. Measure a warm part, not a cold one.

Grinding gives you a surface. Scraping gives you a relationship. On a large machine, the slideways and mounting faces are hand-scraped and ground to create bearing contact over the full length of travel. The published figure for the Dixi 210 P is 500 hours of hand-scraped and ground contact surfaces. That work is done to control straightness, squareness, and the way the column leans as it moves.

Why hand work instead of a grinder? Because a large casting deflects under its own weight as it sits on the grinder. Scraping is done with the part in its final support condition, so the geometry is true in the position it will actually work in. It also lets a fitter correct local contact rather than chase a single number.

The practical limit shows up in the part, not the machine spec sheet. A machine that is square to 10 μm over 2 m will cut a bore and a face that are square to each other. A machine that is only ground may pass a static check and still walk out of square after an hour of cutting.

If you are evaluating a supplier for large, tight parts, ask how they hold the relationship between features, not just how they hold a single dimension. That is the question that separates a real capability from a catalog claim.

• 1
  Scrape in final positionGeometry is set where the casting will actually sit.
• 2
  Contact over full travelBearing points are checked at both ends, not just center.
• 3
  Squareness over sizeFeature-to-feature relationship matters more than one dimension.

A 114 kW spindle sounds like a headline number. In practice, power only matters if the structure can absorb the cutting force without deflecting. A machine that flexes under load will chatter long before it runs out of torque. The Dixi 210 P pairs its spindle with a heavily ribbed cast structure and large bearing contact, so the force path stays short and stiff.

Stiffness decides the depth of cut you can take without a second pass. It also decides surface finish. Vibration that appears at the tool tip leaves marks that no polishing step fully removes. This is why two machines with the same spindle power can produce very different finishes on the same material.

For large parts, the loop closes differently. A long boring bar or an extended tool holder adds compliance that the machine itself does not have. Deep pockets, thin walls, and long reaches all reduce the effective stiffness of the system, regardless of what the spindle is rated for.

When a part is both large and thin-walled, we often reduce depth of cut and raise the number of passes. It is slower, but it holds the wall thickness and avoids the spring-back that shows up after unclamping.

You will probably never order a machine of this size. But the engineering constraints are identical at every scale. Long parts, tight tolerances, and thin sections all fail for the same three reasons: thermal drift, geometry error, and deflection under cutting load.

In our shop, we see this on 4,000 mm travel work. A long aluminum extrusion fixture that measures true on a cold machine can drift out of tolerance after two hours of cutting if the coolant temperature is not controlled. The fix is not a better probe. It is a stable thermal environment and a warm-up routine.

For parts under 500 mm, the picture is simpler. A 5-axis machine with a Ø400 mm rotary table and a controlled environment can hold ±0.005 mm on qualified features without the drama. The engineering effort scales with size, not with the number on the drawing alone.

This is the practical lesson from the largest machine tool programs. Accuracy is a system property. It comes from the structure, the thermal design, the metrology, and the process, in that order.