A deep pocket and a tall core are the same cut

A pocket sunk into a plate and a core standing proud of that plate look like opposite jobs, and on the machine they are the same cut. Both send most of the block to the chip conveyor; the only difference is whether the metal that leaves was inside the outline or outside it. They fail in the same ways, they are priced by the same rules, and they are decided by the same thing: the order the cuts are taken in.

The difficulty is not the cutting. It is what the cutting releases. A rolled steel block carries internal stress that stays balanced while the block is whole. Take a deep pocket out of one face and that face loses its share of the balance, so the plate bows towards the cut — slightly on an ordinary plate, plainly on a large one. Nothing has gone wrong; that is what steel does when it becomes unbalanced. The useful question is therefore not how to prevent the movement, but where in the sequence to let it happen.

The answer is before the finish cut, never after it. Rough the pocket leaving stock on the floor, on the walls and on the face that will sit on the table, then take the plate off the clamps, let it stand, and measure it free on a surface plate. A plate that bowed while clamped will spring back, or spring further, the moment the pressure comes off, and the free measurement is the only one worth acting on. The finish cut then comes off the corrected position rather than off the position the roughing left. The same interval brings the plate back to room temperature, which matters as much: roughing puts heat into thin metal, and a finish cut taken against grown steel arrives having shrunk out of tolerance.

Skip that interval and the pocket measures right on the machine and assembles wrong on the bench. With the plate held down at four points, the floor is flat because the clamps hold it flat. Free it and it dishes. The insert then bears in the middle of the pocket and rocks on its corners, or the reverse; the cure at the far end is shimming, which is a symptom of a sequence rather than of a tolerance. A tall core behaves the same way — flat while clamped, leaning once free. It shows up in three readings: a dial indicator across the floor, a wall measured at root, middle and top, and a core height checked at four corners — the things an inspection record should let you reconstruct.

A deep pocket and a high core are also usually the same plate seen from two sides. Sink a pocket on the cavity side and what is left standing around it becomes the core when the plate is turned. The second side is part of the job rather than a separate one, and the turn is where position is lost: machine and mark the reference edges first, then pick them up with an indicator after the flip — never assume the vice.

Underneath all of this sits a machine constraint. Our machining centres are three-axis, the ordinary configuration for mold base work: the spindle comes down vertically, and there is no way to angle the tool into the work. A pocket with vertical walls and a floor smaller than its opening can only be reached by a tool long enough to get there, and a tool that long deflects — the note on corner radii and reach sets out that arithmetic. The consequence is simple: a through pocket can be opened from both faces and the reach halved, where a blind one cannot. Through or blind is a price line, not a detail.

Depth is never deep in the abstract; it is deep relative to the floor left underneath. That floor is what resists injection pressure behind the insert, and a floor thin enough to deflect takes the pocket floor with it. A depth callout with no minimum floor thickness beneath it is half a specification, and the shop finishes the other half by guessing — usually conservatively, which costs the buyer either money or the depth they wanted.

On a high core the weak point is the wall rather than the floor. Two pockets cut either side of a rib leave a standing wall, often the thinnest section in the base, and it behaves like a spring: cut one side and it leans away from the cutter, cut the other and it comes back. Finishing one face completely and then the other leaves both cuts taken against different stress states, which is why such a wall measures one thickness at its root and another at its top. Leave stock on both faces, then alternate light passes and finish the two sides in one setup.

A tall core also needs support under it while it is being cut. Clamped only at the edges of the plate, a core standing well proud of the face is a cantilever, and a cantilever rings. The finishing cut then leaves chatter on the side walls and a top face that is flat one way and crowned the other. Blocking up beneath the core, or clamping on the core itself, is decided before the first cut; a spring pass at the end does not recover it.

What belongs on the drawing is more than an outline and a depth. The pocket wants its corner radii, a depth measured from a named face rather than from “top”, a minimum floor thickness, a minimum wall width, and a plain statement of through or blind. If the pocket floor is to be ground rather than milled, say so, because that is a second operation on a different machine. If the plate is a die-casting frame the walls need draft, and a vertical wall drawn today cannot acquire draft later.

That list explains why two pockets of similar size can be quoted far apart. Depth against width, floor thickness, wall thickness, a ground floor and whether the job can be opened from both sides each add operations rather than minutes: a pocket a fifth deeper can need a longer tool, lighter passes, a second setup and an extra interval. A drawing that says “pocket, depth as required” is not a specification, and a price quoted against it is a placeholder that gets revised — usually after the plate is cut.

None of this is exotic. It is the ordinary sequence of a shop that cuts its own plates: rough, release, measure free, semi-finish, finish late, and support what stands up. Work leaves the floor as custom and non-standard mold bases or, where the shape is deep and awkward, as a complex surface mold base, and as a precision ground mold base where the drawing asks for ground faces. The seven-day figure covers standard sizes and plain jobs, with deep work quoted a firm date before cutting starts; over 2,000 sets and pieces a month pass through our own machines; and there is no middleman between the drawing you approve and the machine that cuts it.