How Loading and Unloading Devices Work on Automatic Towers

• 1
  What Loading and Unloading Devices Actually Do
• 2
  Bar Feeders and the Geometry They Tolerate
• 3
  Grippers, Locating Surfaces and Repeatability
• 4
  Pallet Changers and Setup Time
• 5
  Tolerances and Surface Finish on Automated Cells
• 6
  Material Behaviour Inside an Automated Cell
• 7
  How We Take an Automated Job From Quote to Shipment
• 8
  Which Loading Method Fits Which Job

A loading and unloading device is the interface between a stack of raw stock and a spindle that never wants to stop. On an automatic tower, that interface usually takes one of three forms: a bar feeder pushing stock through the spindle bore, a gantry or robot gripper moving a blank between a magazine and the chuck, or a pallet changer swapping a fixture plate on a vertical mill. Each one removes a pair of hands from the cycle.

The mechanism matters more than the marketing. A bar feeder works by maintaining axial push while the guide channel holds the bar straight. If the bar whips, the finished diameter drifts. A gripper works by repeating the same approach vector thousands of times, so every locating surface that touches the blank is a wear item. A pallet changer works by guaranteeing that the fixture lands in the same position every swap.

That is the whole engineering story in one sentence: these devices convert setup time into repeatable positioning. When the positioning repeats, cycle time drops. When it drifts, scrap rises faster than any labor saving can cover.

So the question is never whether automation is better in the abstract. The question is whether your part geometry, batch size and locating features can hold the repeatability the device needs.

• 1
  Bar feederBest for round stock under Ø80 mm, long runs, low part variety.
• 2
  Gantry gripperBest for prismatic blanks with two clean locating faces.
• 3
  Pallet changerBest when fixture setup, not cutting, is the bottleneck.

A bar feeder holds a bundle of stock and pushes one bar at a time into the collet. The pusher stays behind the bar, so the unsupported length between the guide channel and the collet is where problems start. As spindle speed climbs, that free length vibrates. The vibration is not a nuisance; it shows up directly in the turned diameter.

The practical rule is to keep the ratio of unsupported length to bar diameter low. For Ø12 mm stock, a short remnant is fine; for Ø60 mm stock the same free length behaves very differently because mass rises with the cube of diameter while stiffness rises with the fourth power. This is why large-diameter bar feed work often runs at lower spindle speeds than the tooling could otherwise handle.

Stock straightness is the other limit. Cold-drawn or ground bar feeds cleanly. Hot-rolled bar with visible bow will bind in the guide channel and stall the pusher. If the pusher stalls mid-cycle, the machine may alarm out or, worse, finish the part short.

Bar feeders also set a bar-end remnant that cannot be machined. For a 3,000 mm bar, the last 150–300 mm is typically lost depending on the collet and pusher design. On a 10,000-piece order that remnant is negligible. On a 200-piece order it is not.

A gantry gripper is only as good as the surfaces it grabs. Two parallel faces give it a clean datum. A cast surface with draft and parting-line flash gives it nothing repeatable, and the operator ends up re-teaching the pick position every few hundred cycles.

When we quote a part for an automated cell, we look at the locating features before we look at the tolerance callouts. A machined pad, a reamed hole or a turned journal can be gripped and re-gripped within ±0.05 mm without much effort. A rough blank cannot. If the drawing shows only as-cast datums, the honest answer is that the first operation should stay manual or be done on a pallet.

Grip force is a trade-off too. Enough force to resist a milling cut can deform a thin wall. On a 1.5 mm aluminium wall, a soft jaw with a machined pocket spreads the load; a hard metal gripper point-loads it. We have seen 0.1 mm of bow introduced purely by handling.

So the design sequence is: define the datum, then the grip, then the gripper. Reversing that order is how automation projects get stuck in debug for weeks.

On a vertical mill running small parts, cutting may take 12 minutes and setup may take 20. A pallet changer does not make the cut faster. It makes the setup happen while the spindle is still turning.

The gain is proportional to the setup-to-cut ratio, not to the part size. If your setup is already under two minutes, a pallet system adds complexity for very little return. If you are changing soft jaws three times a shift, the arithmetic changes quickly.

Pallet repeatability depends on the receiver. Ball-lock and cone-lock receivers hold position well when the pallet underside and the receiver taper stay clean. A single chip trapped between them can shift the pallet by 0.05 mm or more, which is enough to scrap a tight bore. Air-blast cleaning at every swap is not optional.

We keep a Ø400 mm rotary table in the cell for parts that need angular indexing between pallets. That combination covers a lot of mid-size work without a second machine.

Automation does not loosen tolerances. If anything it tightens the demand on the first operation, because a bad first op propagates through every subsequent pick. On our 5-axis centers we hold ±0.005 mm (±0.0002 in) on critical features, and that number only holds when the fixture is rigid and the blank is seated fully.

Surface finish follows the same logic. Gripper pads and collets mark the surface they contact. If the drawing calls for Ra 0.2–0.8 μm on a turned journal, that journal should not be the grip surface. Move the grip to a shoulder or an extra stock allowance that gets machined off later.

For parts that will be anodized or plated, remember that the finish step adds or removes a few micrometres. A hardcoat anodize layer on aluminium can shift a diameter enough to matter at ±0.01 mm. Cutting to the middle of the tolerance band leaves room.

We inspect 100% of parts before shipment, with raw material checks, in-process monitoring and a final inspection. Reports are available on request. That is how we catch a drifting gripper before it becomes a customer problem.

• 1
  ±0.005 mmAchievable on rigid setups with a fully seated blank.
• 2
  Ra 0.8–1.6 μmStandard machined finish for most gripped surfaces.
• 3
  Ra 0.2–0.8 μmFine finish — keep it away from grip and clamp zones.

Different materials behave differently when a machine handles them unattended. Aluminium 6061 and 7075 feed and grip predictably and are the easiest to automate. Stainless 303 machines well but work-hardens if a tool rubs, which matters more when nobody is watching the cut.

Titanium TC4 (Ti-6Al-4V) and Inconel generate heat at the edge and transfer very little of it to the chip. In a lights-out cell that heat builds in the tool and the part. Coolant delivery and dwell time have to be programmed in, not left to the operator's judgement.

Plastics are the opposite problem. POM and PEEK grip well but deform under clamp force, and ABS or PP can be marked by a gripper pad. Soft jaws with a machined profile solve most of it. Carbon fibre adds abrasion, so grip pads wear faster and need a replacement interval.

The practical takeaway: match the handling method to the material, not the other way round. A gripper tuned for aluminium will mark a PP housing on the first shift.