Hollow Faces and Snipe After a Knife Change on the Surface Planer
A three-knife 260 mm surface planer in a workshop has just had a full set of freshly ground HSS knives fitted by a technician who set them level with the outfeed table, judged by eye against a steel rule. The first job is 2.1 m x 150 mm x 32 mm European oak boards for a bench top. After face-side planing, every board is 0.7-0.9 mm hollow along its length: a straightedge laid on the face rocks visibly in the middle, and the last 80-100 mm at each end is a further 0.4 mm low. Edge jointing against the fence produces convex edges, so the boards will not close up dry. The outfeed table was left 0.05 mm below knife throw at top dead centre; the infeed table reads true. The operator presses down hard on the outfeed side and lifts the board slightly at the end of the pass to avoid the cutter. Ambient 22 C, 55% RH; oak at 9% MC.
1. Explain, from the geometry of the machine, why the boards come out hollow with snipe at both ends, and why the edge-jointed faces come out convex rather than straight.
Read the worked answer
The finished surface is generated only where the knife circle is tangent to the outfeed table, so the outfeed table is the reference surface - not the infeed table and not the depth-of-cut scale.
1. Hollow along the length. The knives stand about 0.05 mm proud of the outfeed table plane. Once the trailing part of the board is running on the outfeed table, each knife passage removes slightly more than the infeed setting, so the cut deepens progressively and the face ends as a shallow concave curve. The error is small per pass but accumulates over 2.1 m as the board pivots on the outfeed lip, which is why 0.05 mm at the table shows up as 0.7-0.9 mm of hollow.
2. Snipe at both ends. As the leading end passes over the block, and again as the trailing end leaves the infeed table, the board loses support for 80-100 mm. Under the operator's heavy downward pressure it drops by the table error and the knives cut deeper, producing the step the straightedge finds at each end. Lifting the board at the end of the pass makes this worse, not better - it changes the feed geometry and can also let the board be dragged back over the block.
3. Convex edges. A hollow face cannot sit flat on the table for edge jointing; the board rocks on its belly and pressure at the ends tilts it, so more is removed at the ends of the edge than in the middle - a convex edge. The convexity is therefore mostly a consequence of the face fault, not an independent fence fault.
Do not correct technique first. Set the knives correctly, re-test, and only then check the fence for square and the tables for transverse co-planarity if convexity persists.
2. Set out the knife-setting and table-verification procedure you would use to bring the machine back into specification, including tolerances and how each check is made.
Read the worked answer
Work in a fixed order, because each check invalidates the next if done out of sequence.
- Isolate and lock out the machine, remove the guard, and confirm the block will not rotate.
- Set the outfeed table exactly level with knife throw at top dead centre, not 0.05 mm below. Use a dial indicator on a magnetic base running on the outfeed table, sweeping the knife edge at TDC; adjust the table until the reading is 0.00 mm. A setting jig resting on the outfeed table does the same job faster.
- Set every knife to the same projection, within 0.02 mm of each other and of the outfeed table, and check the projection at both ends of each knife across the 260 mm width. Knives that are parallel but unequal, or equal but skewed, both cut tapered or stepped work.
- Torque the gib screws to the manufacturer's figure in the specified sequence, then re-check projection, since tightening moves the knife.
- Check table co-planarity with a 1 m straightedge and feeler gauges laid across both tables at three fence positions; aim for 0.05 mm or less. This is where worn ways and a sagging outfeed table show up.
- Check the fence with an engineer's square off the tables and confirm it holds under pressure; a fence that deflects gives convex edges even on a good machine.
- Run a test piece - a 1 m board, light cuts, constant pressure, no lifting - and check with a straightedge and winding sticks. Target 0.15 mm or better over the metre, with no detectable snipe.
- Record the setting (date, knife set number, projections, who set it) and repeat the check after the first ten minutes of running, because knives settle.
The depth-of-cut scale and the operator's eye are not measuring instruments and should never be used for this.
3. An apprentice wants to run 300 mm offcuts through the same machine with the bridge guard swung aside because the pieces are too short to pass under it. The technician says the machine is now 'fine'. Give your judgement.
Read the worked answer
Refuse the operation as proposed. Three separate failures are being combined.
Guard. The bridge guard is not there to be got out of the way; it must be lowered to just above the timber so the unused part of the block stays covered. On a 150 mm wide job the guard covers the remaining 110 mm. Swinging it aside exposes a full-width cutter block running at several thousand rpm within reach of the hands, and the guard is interlocked or not - either way its removal is a management decision with a risk assessment behind it, not something an apprentice can decide at the machine.
Short stock. A 300 mm piece gives no support at entry and exit: the operator's hands pass directly over the block at both ends and the piece can tip into the throat. The practical minimum for hand feeding is around 300-400 mm and only with a push block and constant, controlled pressure - not with a swinging motion. If the offcuts are needed, they should be planed on a jig or sled that carries them past the block, or faced on the bandsaw and brought to thickness on the thicknesser.
'The machine is fine'. The technician's opinion is not verification. Until the knife projections have been measured, the tables checked and a test board produced a straight, snipe-free face, the machine is not released for use. The signed setting record is what makes the statement true.
A useful outcome: use the offcuts as the test pieces for the re-setting exercise, which removes the temptation to run them as production work.