CNC robot bidding: what it does to your machine cycle

• 1
  How a robot tender actually moves a part
• 2
  The gripper decides what parts fit the cell
• 3
  When robot tending stops paying off
• 4
  Design features that make automation work
• 5
  What the cell needs beyond the arm
• 6
  Part and run profiles for robot tending

A robot tender is a handling loop bolted onto the machine cycle. The loop has four states: the robot waits at the door, the door opens, the finished part comes out, and a new blank goes in. Between those states the machine does the cutting. Everything that improves throughput comes from shrinking the four states or overlapping them with cutting time.

The robot itself is usually a six-axis articulated arm, sometimes a SCARA or a gantry for simple pick-and-place. The physical interface is the end-of-arm tool, or EOAT: a gripper, a magnetic handler, a vacuum pad, or a dual gripper that holds the finished part and the blank at the same time. A dual gripper is the single biggest time saver, because it removes one full arm travel from every cycle.

The arm talks to the machine through the CNC controller, not around it. Door open, chuck clamp, and cycle start are handshake signals, typically over digital I/O or a fieldbus. If the handshake is out of order, the machine will not start, and the robot will wait. That interlock is what makes a cell safe and repeatable, but it is also where most integration time is spent.

Cycle time is set by the slower of two clocks: cutting time and handling time. If cutting takes 90 seconds and handling takes 25, the arm is idle most of the shift. If handling takes 100 seconds, the spindle sits still and you have automated the wrong thing. The first job in any robot cell is to measure both clocks honestly.

• 1
  Handling timeDoor open to door close, including arm travel and grip
• 2
  Cutting timeSpindle in cut, plus any in-cycle probing or tool change
• 3
  HandshakeI/O or fieldbus signals that lock the two systems together

Most failed robot projects fail at the gripper, not at the arm. A gripper needs a stable locating feature on the part, enough clamping force to survive acceleration, and clearance to enter and leave the workholding without scraping. If the part is a thin walled housing with no flat datum, no gripper will hold it repeatably at speed.

Part weight and geometry set the practical range. A 0.5 kg aluminum bracket with two bored holes is easy. A 12 kg steel shaft with a rough cast surface is not, because the gripper has to fight both mass and surface variation. For parts above roughly 10 kg, a gantry or a dedicated loader usually beats an articulated arm on both cost and repeatability.

Dual grippers add value when the part can be gripped from the same side in both states. If loading and unloading require different orientations, the arm has to rotate between them, and the time saving disappears. Check the approach vectors before assuming a dual gripper pays for itself.

Gripper jaws wear. Soft jaws or urethane pads protect the finish, but they compress under load and need periodic replacement. In a lights-out cell, jaw wear is one of the few failure modes that a robot cannot detect on its own, so plan a check interval based on cycle count, not on calendar days.

• 1
  Good candidateStable datum, 0.2-10 kg, repeatable blank geometry
• 2
  Poor candidateThin walls, no flat face, cast skin, heavy shafts
• 3
  Wear pointJaw pads and vacuum cups, replace by cycle count

Robot tending pays off when the part runs long enough to amortize the cell. Short runs of 20 to 50 pieces rarely justify the setup, programming, and gripper build. Long runs, or a family of parts that share one gripper and one workholding setup, change the math completely.

Lights-out operation is the real prize, and it has hard limits. Tool wear, chip evacuation, and dimensional drift are the three things that stop an unattended cell. If a tool lasts 40 minutes and nobody is there to change it, the cell stops at 40 minutes. Tool life monitoring and a spare-tool strategy are prerequisites, not upgrades.

In-process probing extends unattended time on tight parts. A probe can confirm a bore or a face before the next cycle starts, and if the measurement drifts, the controller can offset the tool. Without probing, a lights-out cell on a ±0.005 mm feature is a gamble on thermal stability and tool wear.

Chip control matters more than most teams expect. Aluminum stringers wrap around a gripper and pull the part out of position. Deep holes in stainless pack chips into the workholding. The robot will keep loading parts into a dirty fixture until something breaks. Air blast, through-spindle coolant, and a chip conveyor are part of the automation scope, not the machine scope.

• 1
  Run lengthLong runs or a shared part family amortize the cell
• 2
  Tool lifeShorter than the unattended window means the cell stops
• 3
  Chip controlStringers and packed chips defeat an unattended cell

Automation rewards parts that are designed for it. A robot needs a place to hold, a place to sit, and a way to know the part is oriented correctly. Add a gripping land, a locating flat, or a chamfered lead-in, and the cell gets simpler and cheaper. These features cost almost nothing on a CNC part and save a lot on the gripper.

Tolerances should be split into two groups: the features that must be tight, and the features that only need to be functional. A robot cell magnifies the cost of a tight tolerance on a non-critical face, because every part now has to be inspected or probed. Mark the critical dimensions on the drawing and leave the rest at general tolerance.

Material choice affects cycle time and gripper design. Aluminum 6061 and 7075 cut fast and are light enough for a compact arm. Stainless 316L and 17-4PH cut slower and need more clamping force. Titanium TC4 sits in between but generates heat that changes dimensions over a long unattended run.

Deburring is often the hidden bottleneck. If a part needs manual deburring after machining, the cell still needs a person, and the lights-out benefit shrinks. Design in chamfers or specify a tumbling or brushing step so the part leaves the cell finished, not half-finished.

• 1
  Add a grip landA flat or a bore the gripper can repeat against
• 2
  Split tolerancesTight only where function demands it
• 3
  Plan deburringKeep the part complete when it leaves the cell

The arm is one line item. A working cell also needs a part presentation system, a workholding interface, a safety enclosure, and a control layer. Part presentation is often the most expensive piece, because blanks must arrive in a known position every time. Trays, magazines, and vibratory feeders all solve this, at different cost and flexibility.

Workholding has to be robot friendly. A standard three-jaw chuck works, but a quick-change fixture or a zero-point system lets the cell switch between part numbers without a full teardown. If the cell only ever runs one part, a dedicated fixture is cheaper. If it runs a family, the quick-change system pays back in setup time.

Safety enclosures and interlocks are non-negotiable. Light curtains, door switches, and a safety rated controller protect the operator and the robot. They also add cost and floor space. Plan the footprint early, because retrofitting a fence into a crowded bay is painful.

The control layer ties it together. A cell controller or a PLC sequences the robot, the machine, and the presentation system. It also logs cycle counts, tool changes, and faults. That log is what tells you whether the cell is running at the rate you calculated, or whether something is quietly eating the gain.

• 1
  PresentationTrays, magazines, or feeders that fix blank position
• 2
  WorkholdingQuick-change or zero-point for part families
• 3
  Control layerSequencing and logging across robot and machine