Robot Cell Or Standalone Machine: Where The Line Actually Is

Robot Cell or Standalone Machine: Where the Line Actually Is

The photograph above was taken in a customer's plant, and it happens to contain both answers to the question this article is about. In front, a special-purpose machine built around one part: one welding head, one fixture, two buttons. Behind the fence, a robot cell. They are doing the same kind of work a few metres apart, in the same building, for the same company.

Which one a buyer should order is the question we are asked most often at the quotation stage, and the honest answer is that "automation" is not the decision. The decision is about part mix and handling, and it usually has very little to do with the price of the robot.

Both answers in one photograph, forty metres apart: a special-purpose machine built around one part in front, and a fenced robot cell behind it.

Two Different Answers to the Same Sentence

A special-purpose machine is a fixture with welding heads fixed around it. The part is located, the heads close, and every weld in that stroke is made at once. There is no program to select and no axis to teach; the machine knows one part, and it knows it very well.

A robot cell is a robot, a gun, a fixture and a safety system. The fixture still does the locating; the robot only carries the gun from one stored position to the next. Change the part and you change the fixture and the program. The robot is the flexible part of a system whose accuracy still comes from the fixture.

That last sentence is where most quotations go wrong, in both directions. A robot does not make a bad fixture good, and a special machine is not automatically faster than a robot that has been laid out properly.

Where the Robot Is Genuinely the Right Answer

The robot wins when the welds cannot all be reached from one direction. A box, a bracket with returns on three sides, a part welded on both faces: a fixed head cannot get there, and a robot with a compact gun can.

It wins again when the part mix is wide and the volume per variant is small. One fixture family and twenty programs is a normal robot cell. Twenty fixtures for twenty parts is what the special-machine route would cost.

It wins when the part is heavy. A robot moves a 60 kg gun to a 300 kg panel far more easily than an operator moves the panel to the gun, and that is a handling argument before it is a welding one. And it wins when the product does not exist yet: a cell can be given a new fixture and a new program; a dedicated machine is scrap the day the part changes.

A robot line at a customer, with two arms working over a common fixture. The fixture does the locating; the robot only carries the gun to where the fixture says.

Where the Special Machine Is the Right Answer

The special wins on one part at high volume, because all the welds happen in a single stroke. Four welds on a bracket take six seconds including loading, against twelve to fourteen for a robot doing the same four welds one at a time. At four thousand parts a day that difference is the whole project.

It also wins on people. A special machine is run by whoever is standing there; there is no program to lose, no teach pendant to be dropped, and no argument about who is allowed to change a position. In plants where the maintenance team is two people and the line runs three shifts, that matters more than flexibility on paper.

And it wins on repeatability that does not depend on the robot's absolute accuracy. Every weld is made at a hard stop that was machined once.

The Arithmetic, Because It Is Counter-Intuitive

Take a bracket with four welds. Six hundred a day, one shift. On a special machine that is about an hour of welding; on a robot cell it is a little under two. Neither number is the constraint on that line, and buying the robot to save forty minutes will not pay for the fence, the risk assessment and the programmer.

Now take the same bracket at four thousand a day, two shifts. The special machine needs under seven hours. The robot cell needs fourteen, which means two cells, or overtime, or a fixture that welds two parts at a time. Volume, in other words, pushes a buyer towards the dedicated machine, not towards the robot, which is the opposite of what most people expect.

The robot is justified by variation and by handling, not by volume. When a customer tells us they need a robot because their volumes are growing, we usually ask them to look at their variant count instead.

There Is Something in Between

The choice is not binary, and most of our better projects sit in the middle. A special machine can be built with quick-change nests for a family of parts, two heads whose spacing is adjustable in ten minutes, and a controller holding a schedule per variant. That covers a part mix of maybe six variants for the price of a machine and a half, and it needs no programmer.

The limit of that approach is the geometry. Modular tooling handles a family that shares weld positions; it cannot handle a part whose welds have moved to a face the heads cannot reach. When a buyer tells us the product will change every year, that sentence is worth more than the volume forecast.

The Costs That Never Make It Into the Comparison

A robot cell carries a list of costs that a special machine does not: fencing and light curtains, a risk assessment, integration and commissioning time measured in weeks, a spare gun, clean dry air, floor space around the cell for access, and a person who can own the program after we leave. None of those are optional and none of them appear on the quotation for the robot.

The special machine carries a different list: a fixture that only fits one part, and an obsolete machine if the part is redesigned. We have seen both failures. A cell sold for a two-hundred-a-day job because the customer wanted a man out of a hot spot ran about twenty minutes in every hour for two years, and then paid for itself three times over when the part changed twice in one season. A dedicated machine that welded a bracket beautifully became a problem the year the bracket gained a fifth weld on a face the heads could not reach.

What We Ask Before Quoting Either

Four answers decide the layout, and a buyer who has them ready gets a quotation that still fits two years later.

• How many parts a shift at peak, and how many variants of the part exist today?
• What does the part weigh and measure, and how many people does it take to lift it?
• Can every weld be reached from one direction, or does the part have to be turned?
• Who will own the program, the fixture and the maintenance in year three?

Neither column is better. They fail in opposite directions, which is exactly why the question is worth answering before the order rather than after the installation.

PDKJ continues to provide customized laser welding and resistance welding solutions based on different materials, workpiece structures, welding requirements, and production applications.

Related Products

• Robot Automated Systems
• Spot Welding Machines

Machines waiting on our floor in Dongguan. Which of them is right for a job is decided before the steel is cut, not after.

Contact PDKJ

Send us your part drawing or a sample, and our engineers will run a welding feasibility trial on your own material and recommend the equipment configured for your application.