Workholding Explained: Vises, Fixtures, Tombstones & Pallets
Every accurate part is the product of two quiet agreements. The first is that the cutting tool will be where the program thinks it is — the promise kept by the tool, the holder and the machine. The second is that the workpiece will be where the program thinks it is, held still while thousands of pounds of cutting force try to move it. That second promise is the job of workholding, and it is easy to underestimate because a good setup is invisible: the part simply stays put, cut after cut, and nobody notices the vise until the day a part shifts. But workholding is genuinely half of machining accuracy. The finest machine, the sharpest tool and the most perfect program are all wasted if the part moves, flexes or sits in a different place every time it is loaded. Shops that struggle with inconsistent parts, chatter they cannot explain, or tolerances that hold on the first piece and drift on the third are often fighting a workholding problem while looking everywhere else.
This guide is the reference base for that half of machining. It explains what workholding must do, the 3-2-1 locating principle and datum logic that every good fixture obeys, and the workholding catalog — the vises, soft jaws, clamps, fixtures, vacuum and magnetic tables, tombstones and pallet systems that fill a modern shop — with the logic for choosing among them. It sits at the heart of this library’s tooling topic as the workpiece-side complement to the toolholding guide, which covers the mirror-image problem of holding the tool: between the two, the entire chain from spindle to workpiece is covered. The workholding’s quality shows up where all accuracy does — in the tolerance the machine can hold and in the surface finish the part gets. Terms like workholding, fixture, pallet, chuck, datum and axis are in the glossary.
What workholding must do
A free workpiece has complete freedom of motion — six ways it can move and six ways it can rotate relative to the tool. Workholding exists to take that freedom away, and it does so with three distinct jobs that are easy to blur and important to separate:
Locating sets where the part is. Before anything is clamped, the part must be placed in a definite, known position and orientation relative to the machine — the position that makes the program’s coordinates true. Supporting holds the part against the forces it will meet — the cutting forces that try to push it down and aside, and the vibration those forces create — so it does not deflect, flex or chatter under load. Clamping holds the located and supported part in place against all of those forces, applying enough force to keep it there without so much force that it distorts the part. The order matters: a part must be located before it is clamped, and a clamp that fights the locators — squeezing the part off its locating surfaces instead of onto them — is a setup that will never repeat.
Accuracy, in workholding, is mostly a matter of repeatability: the part must sit in the same place every time it is loaded, because the program does not change between parts. A setup that locates the first part perfectly but lets the second sit a few hundredths different has already lost the tolerance game regardless of how good the machine is. And rigidity is the partner of repeatability — a workpiece that flexes under load machines differently at the start of the cut than at the end, and a workpiece that vibrates transfers that energy into the tool as the chatter that ruins finish and shortens tool life. The workholding is part of the machine’s stiffness loop, which is why a weak setup shows up not only as moved parts but as chatter and poor finish on parts that never visibly moved at all.
The 3-2-1 principle: how parts are located
The foundation of every good fixture is a way of thinking about location called the 3-2-1 principle, and it deserves to be understood rather than memorised, because it explains both why fixtures work and why over-engineered ones fail.
A free part has six degrees of freedom: it can translate in three directions (X, Y and Z) and rotate about three axes. To locate it fully, the fixture must remove all six — and the 3-2-1 principle does exactly that with six points of contact, arranged in three groups. Three points on the part’s largest flat surface — the primary surface, usually its bottom — establish a plane and remove three degrees of freedom at once: the part cannot move up and down or rock in two directions. Two points against one side of the part establish a second plane and remove two more: the part cannot slide sideways in that direction or rotate about the vertical. One point against an adjacent side removes the last degree: the part cannot slide along the remaining direction. With three, then two, then one contact points, the part is deterministically located — it sits in exactly one place and orientation, every time.
The discipline hidden in the rule is what it excludes. Three points on the bottom are correct; a fourth point on the same flat bottom is wrong, because four points on a nominally flat surface cannot all touch a real (never perfectly flat) part at once — the fourth point makes the part rock on whichever three happen to touch, and the location becomes unstable instead of deterministic. Over-locating is the classic fixture mistake, and it is the reason the rule says exactly six points and no more. The practical expression of the same discipline is the pin pair used when a part is located by two holes: one round pin in one hole, and one diamond pin — a pin relieved on two sides so it contacts the hole only across a narrow band — in the other. Two full round pins would try to locate both holes at their exact centres, but real parts have hole spacing variation, so the two pins would fight; the round pin locates position while the diamond pin locates only rotation and tolerates the spacing variation. The lesson generalises: give every feature one clean job, never two features competing to locate the same direction.
Datums: locating from the part’s own reference frame
The 3-2-1 principle tells a fixture how many points to use; the part’s drawing tells it where. Machined parts are dimensioned from reference surfaces called datums — the faces, edges or bores that the designer chose as the origins of the part’s measurements. A good fixture locates the part from those same datums, because machining is only as consistent as the reference it is cut from: if the drawing dimensions the hole pattern from the bottom face and one edge, the setup that locates from that bottom face and that edge will machine the features where the drawing says they are, run after run.
The rule of thumb that follows is to support the part on its largest, flattest, most stable surface — usually its primary datum — and to prefer machined surfaces over raw stock where the choice exists, because a machined surface is flat and true while a casting skin or a sawed face may not be. Over-referencing is the mirror-image mistake to under-locating: clamping a part against more reference surfaces than the design uses locks in dimensional relationships the part does not actually need held, and as the stock, finish and temperature vary, the excess references fight the real ones and the part sits differently from load to load. Locate from the datums the drawing cares about — no more, no fewer — and the fixture becomes an extension of the part’s own coordinate system rather than an argument with it.
The workholding catalog
With the principles in place, the catalog of workholding becomes a set of tools matched to part shapes and production situations. The main families:
Vises — the everyday default. The plain machining vise, with its fixed and moving jaws driven by a screw, is the workhorse of prismatic machining: it holds square and rectangular stock fast, sets up in minutes, and repeats well enough for most work. Its real power is in its jaws. Hard jaws are durable but generic; soft jaws — blanks of aluminium or mild steel bolted to the vise — are machined in place to match the part’s exact contour, so a batch of parts seats into a perfectly shaped nest every time. Soft jaws are the bridge between a standard vise and a custom fixture: cheap to make, gentle on finished surfaces, and repeatable for second operations, which is why they are one of the most valuable techniques in the whole workholding toolbox. Vises also scale sideways — several bolted side by side, or one vise with multiple stations, cuts several parts per cycle.
Clamps — for parts too big or too odd for a vise. Strap clamps, toe clamps and edge clamps bolt directly to the table or a fixture plate and press the part down or against a stop. They suit large plates, castings and irregular shapes that no vise can swallow, and they can leave the top and sides clear for machining. Their cost is setup time and reach — a clamp in the wrong place blocks a toolpath, so clamping is planned against the toolpaths it must not obstruct.
Fixtures — precision for a specific part. A fixture is purpose-built workholding that locates and clamps one particular part or part family, assembled from a fixture body, locators, supports and clamps. A fixture pays for itself when a part runs often enough that its setup speed and repeatability outweigh the cost of building it — and it is the only way to hold complex shapes that no standard vise or clamp can. Fixtures range from dedicated steel and aluminium builds to modular fixtures assembled from standard components on a drilled or slotted plate, which capture much of a custom fixture’s rigidity while staying reconfigurable for new parts.
Vacuum and magnetic tables — the flat-part specialists. A vacuum table holds thin, flat work by suction, drawing the part down against a sealed surface with no clamps above it at all — ideal for sheets, thin plates and parts needing full top access, so long as the part is flat, non-porous and cut lightly enough not to break the seal. A magnetic chuck does the same job by magnetism for ferrous parts, holding them without clamps and releasing them instantly. Both trade holding force for unobstructed access, which makes them perfect for their niche and wrong for heavy cutting.
Chucks and collets — the rotating world. On the lathe and turning side, the workpiece rotates against a stationary tool, and workholding means gripping it on the spindle axis: the chuck that clamps by jaws for general work, the collet that grips round stock evenly around its full circumference for concentricity, and their power-driven variants for production. Where the mill holds a stationary part against moving tools, the lathe spins the held part into the tools — the same locating, supporting and clamping logic, applied to round work and judged by the concentricity it preserves.
| Workholding | Typical workpiece | Strength | Watch out for |
|---|---|---|---|
| Vise + soft jaws | Prismatic blocks, plates | Fast, flexible, repeatable | Size range; jaw contact |
| Clamps | Large plates, castings, odd shapes | Handles anything; clear top | Setup time; blocking toolpaths |
| Dedicated / modular fixture | One part or family, run after run | Precision, repeatability | Build cost; only pays with volume |
| Vacuum / magnetic table | Thin, flat, ferrous parts | Full access; no clamp marks | Light cuts only; flat clean parts |
| Chuck / collet (lathe) | Round stock and turned parts | Concentricity on the spindle axis | Round work only |
Multiplying setups: tombstones, pallets and quick change
The economics of workholding turn on a simple arithmetic: every minute the spindle spends waiting for a part to be loaded is a minute it is not cutting. The workholding systems that make a shop money are the ones that keep the spindle cutting while parts are loaded elsewhere — and they do it by multiplying how many parts a machine can reach in one setup.
Tombstones are the mill’s answer to that arithmetic — vertical, multi-sided columns of cast iron or aluminium that bolt to a machine table and carry parts on two, four or more faces. A tombstone on a horizontal machining centre, or on a rotary table, lets the machine reach several sides of several parts in a single setup, and index from face to face without unloading anything. The machine-types story of horizontal versus vertical machining is largely a workholding story: the horizontal machine exists to work the sides of a tombstone that the vertical machine cannot easily reach.
Pallets and zero-point systems attack the same arithmetic from the other end. A pallet is a removable workholding plate: the operator loads and clamps parts onto pallets at a bench while the machine cuts a different pallet, then swaps the finished pallet for the loaded one in seconds instead of tearing down and rebuilding the setup in the machine. Zero-point clamping — quick-change receivers with a repeatable, precision interface bolted to the table — extends the idea to the vises and fixtures themselves, so a whole setup can be lifted off and the next one dropped on with the location guaranteed to microns every time. The same pallet pools feed the lights-out machines that run unattended: a stack of pre-loaded pallets is exactly what an automated cell consumes through the night. None of this changes the physics of locating and clamping a single part; it changes how often the machine has to stop to let it happen.
Rigidity, clamping force and the silent failures
The principles and catalog above handle the visible job of workholding — holding the part in place. The silent failures live in the forces, and two of them deserve naming because they masquerade as other problems.
The first is insufficient rigidity: workholding that is stiff enough to hold a part still at light cuts but flexes under heavy ones. The fix is not always more clamping force — a thin-walled part clamped harder does not get stiffer, it gets distorted, and the part springs to a different shape the moment the clamps come off. Thin parts, delicate features and parts that must hold tolerance after release are exactly where clamping force has to be tuned to the part’s stiffness rather than to the cutting forces alone. The second silent failure is the one that produces chatter and unexplained finish problems: a workpiece that is held but not supported vibrates, and every vibration of the part is a vibration hammering the tool. The workholding is a link in the machine’s stiffness chain, and like the toolholder on the tool side, a weak link here shows up as problems everywhere else — which is why the diagnosis of a chatter or finish problem has to include the setup, not just the tool and the parameters.
The discipline that catches both is the same one that verifies every other promise in machining: prove the setup before trusting it. The probing and inspection that verifies offsets and tooling extends naturally to workholding — probe the located part to confirm it sits where the program expects, and prove the first article before the run. A shop that treats a clamp as a permanent fixture rather than as a part of the accuracy system will rediscover, part after part, that workholding is not a detail but a decision.
Choosing the right hold
With the catalog in front of you, selection is a short chain of reasoning that runs from the part outward:
- What shape is the part? Prismatic stock in a vise; flat thin material on vacuum or a magnetic table; round stock in a chuck or collet; complex or delicate shapes in a fixture or machined soft jaws.
- How many parts, how often? A one-off or a prototype is held with the fastest flexible setup — a vise, clamps, soft jaws. A part that returns in batches is worth a fixture or machined jaws that make every setup repeatable. A part that runs for weeks is worth a tombstone or pallet system that keeps the spindle cutting.
- What surfaces must the tool reach? The workholding must not block the toolpaths that make the part, which is why clamping is planned together with the toolpath strategy and why vacuum and magnetic holding exist for parts that must be machined all over the top.
- What accuracy does the job demand? The tighter the tolerance, the more the setup must repeat and the more carefully the part must be located from its datums rather than held “close enough.”
The guiding principle across all four questions is the same one that governs the rest of tooling: use the simplest workholding that does the job. Every component between the machine table and the part is a link in a chain, and each link adds its own possibility of error, flex or setup time. The simplest adequate setup — the fewest links, the most direct hold, the cleanest location from the part’s own datums — is almost always the most accurate and the fastest to change. Fancy workholding earns its place when the job’s shape, volume or accuracy demands it; until then, the plain vise with a good set of soft jaws is hard to beat.
Frequently asked questions
What is the 3-2-1 principle in workholding? It is the rule for locating a part with exactly six points of contact: three points on the part’s primary (usually bottom) surface establish a plane and remove three degrees of freedom, two points on one side remove two more, and one point on an adjacent side removes the last. Six points fully locate a part in a deterministic position. Adding more points than the rule calls for over-locates the part — a fourth point on the bottom makes it rock on an uneven surface — which is why the discipline is knowing what to leave out as much as what to put in.
What is the difference between a jig and a fixture? A fixture locates and clamps a workpiece for machining; a jig does the same but also guides the cutting tool — typically through drill bushings — so the tool is positioned by the jig rather than by the machine. In practice, fixtures dominate CNC machining because the machine positions the tool; jigs survive mainly in manual drilling where guiding the tool matters. The terms are often used loosely, but the functional distinction is whether the tool is guided or not.
When should I use soft jaws instead of hard jaws? When you want the repeatability of a custom nest without the cost of a fixture. Soft jaws are machined to match the part’s contour, so every part seats identically and is gripped evenly without marring finished surfaces. Use them for batches of parts that need consistent location for second operations, for delicate or finished parts that hard jaws would damage, and whenever a run is large enough to justify machining the jaws but not large enough to justify a dedicated fixture.
How do I stop a thin part from distorting when I clamp it? Clamp it with less force, spread over more area, and support it where it is weak. The distortion comes from clamping force concentrated on a thin, unsupported section — so the answer is machined soft jaws or a fixture that contacts the part across a broad area, supports it from below, and applies clamping force against the rigid sections rather than across the thin ones. If the part still distorts, the clamping force is too high for the part’s stiffness, regardless of how securely it feels clamped.
When is a tombstone or pallet system worth it? When the spindle would otherwise wait for setups. A tombstone earns its place when a part (or family) runs in enough volume and across enough faces that machining many parts per setup — on a horizontal machine or rotary table — keeps the spindle cutting instead of stopping for reloads. A pallet or zero-point system earns its place when setup changeover time is the bottleneck, letting one setup be loaded off-machine while another runs — and both are the natural workholding of automated and lights-out machining, where a stack of pre-loaded pallets is what keeps an unattended machine fed.
Bottom line
Workholding is the half of machining that holds the workpiece to its promise, and it is governed by a small set of ideas that scale from a bench vise to a pallet pool. A workpiece has six degrees of freedom, and good workholding removes them with three jobs — locating, supporting and clamping — applied in that order, with repeatability as the goal. Location follows the 3-2-1 principle: six contact points, three on a plane, two on a side, one on the end, and no more — because over-locating a part makes it rock instead of repeat — and the points go where the part’s own datums say they should. The catalog turns those principles into practice: vises and machined soft jaws for prismatic work, clamps for the large and irregular, fixtures for the part that runs again and again, vacuum and magnetic tables for the flat and ferrous, and chucks and collets for the rotating world of the lathe. Above them sit the systems that keep the spindle cutting — tombstones that multiply parts per setup, and pallets and zero-point clamping that move setup off the machine entirely and feed the lights-out shop. The failures are the silent ones: workholding too weak to resist heavy cuts vibrates into chatter, and workholding too strong for a thin part distorts it out of tolerance. Choose the simplest hold that locates from the part’s datums, supports it against the forces, and clamps it without distortion — and the second half of machining accuracy is already bought.
This guide is part of the CNC Media guides library — the workholding reference base of the tooling topic, deliberately free of prices and of any single fixture maker’s catalogue to promote.