Why Does a Servo Feeder Drift After a Die Height Change?

The Drift That Arrives With a New Die

The job was a bracket run on 1.6 mm SPHC, 300 mm feed, 45 strokes a minute, with an NC servo feeder mounted on a fixed base.

The previous die sat 25 mm lower, so the toolroom added a spacer plate under the new one and kept the same bolster clamps. The die face rose 25 mm, and the pass line rose with it.

The feeder stayed where it was. Its exit rolls were now 25 mm below the die entry, so the strip had to climb between the two.

The operator saw three things in the first hour: short parts, a bright rub mark on the die entry, and a strip that bowed upward between the feeder and the die.

The controller was blamed first, then the encoder, then the roll pressure. All three were fine, and each test cost half a shift.

How a 25 mm Spacer Becomes 0.65 mm of Error

Take the geometry in two steps: the angle, then the projection of the feed onto the horizontal.

With a 25 mm rise over a 380 mm span, the angle is the arctangent of 25 divided by 380, which is 3.77 degrees.

The feeder measures strip travel along the slope. The die needs that travel projected onto its own plane, and the projection is the feed multiplied by the cosine of the angle.

Cosine of 3.77 degrees is 0.99784. On a 300 mm feed the loss is 0.65 mm, which is close to the 0.6 mm the operator measured on the part.

That single number explains most of the complaint. It also explains why a 5 mm rise is invisible and a 40 mm rise is a scrap event.

The rub mark is the second half of the story. Strip climbing into a die entry wears the entry radius on one side, and the wear then acts as a guide that pushes the strip sideways.

Read the table as a tolerance budget rather than a failure list. A 10 mm rise does not stop the line; it consumes the accuracy you paid for.

The table assumes the feeder exit roll is the last contact point before the die. Add a hold-down or a guide block at the die entry and the effective span shortens, which makes the same rise worse.

Measure the Pass Line Before You Touch the Controller

This check takes twenty minutes and settles the argument before anyone opens a parameter screen.

• Stretch a line. Run a taut cord from the feeder exit roll to the die entry, and measure the gap at the die with a feeler gauge.
• Measure the span. Distance from the exit roll nip to the die entry, not to the press centreline. On most lines it is 300 to 450 mm.
• Check the strip, not the machine. Look for a bright rub mark on the upper die entry and a bow in the strip between feeder and die.
• Compare against the last die. The old setup sheet should carry a pass line height, and if it does not, that is the first thing to add.

Four Ways to Restore the Pass Line

Once the number is known, the fix is a mechanical decision, and each option carries a different cost.

The fourth option is what most plants do at 2 a.m., and it is defensible for one shift. It is not defensible as a setup.

Raising the feeder is the right answer in most cases, because the pass line belongs to the die stack, and the die stack is what changed.

Limits of the Cosine Arithmetic

Cosine error is real, and on this job it accounted for about 0.6 of the 0.6 mm. It is not the whole family of drift complaints.

Roll slip. A strip climbing a slope pulls down on the exit roll. Grip falls, and the loss adds to the geometric error rather than replacing it.

Pilot release timing. A die that grabs the strip early pulls it forward, which shows as a long feed. That is a timing fault, not a pass line fault.

Frame deflection. A heavier die deflects the press frame differently, and the shut height at the bottom of the stroke is not the shut height on the nameplate.

Strip bow. On thin material the strip bows between feeder and die instead of climbing, and the bow absorbs feed in a way no cosine corrects.

The honest limit: geometry tells you the size of the error you created. It cannot tell you whether a die that was already marginal will now crack pilots.

The controller holds a data trap of its own. The feed length on the screen is the command, not the result, and on a climbing strip the two differ by more than the tolerance band.

Why It Grows at 200 SPM and Not at 60

The same 25 mm error is nearly invisible during tryout and obvious in production, and speed is the reason.

At 60 strokes a minute there is time for the strip to settle onto the die face after the feed. The angle costs 0.65 mm and the pilot absorbs part of it.

At 200 strokes a minute the feed completes in 90 milliseconds. The strip arrives still climbing, the pilot enters a strip that is not seated, and the error compounds.

That is why a line that ran at ±0.05 mm for years starts showing 0.3 mm of scatter after a die change and a speed increase in the same quarter.

It is also why the fix has to be mechanical. A controller offset applied at 60 strokes a minute will be wrong at 200.

The Setup Sheet That Prevents It

One line on a setup sheet removes this entire failure mode, and it costs nothing to add.

Record the pass line height, measured from the floor to the top of the lower die face, for every die in the cell.

Record the feeder base height next to it. When the two numbers differ by more than 5 mm, the feeder moves before the die runs.

Add the spacer thickness to the die card as well, because a 25 mm plate is easy to forget once it is painted the same colour as the bolster.

FANTY has built feed lines for 200-plus press cells, and the setups that hold ±0.05 mm at 200 SPM all carry a pass line number rather than a feed length offset.

The coil feeding equipment we build carries a screw adjustment on the feeder base for exactly this reason, sized to cover the shut-height range of the dies it will run.

The full error budget, including the terms this article leaves out, is set out in the coil line guide library, which is the right next read if the geometry is now under control.