Why Does a Servo Feeder Drift Sideways on Cambered Strip?
A coil with 3 mm of camber in 2 m will push the leading edge about 3 mm sideways for every metre you feed. No amount of guide pressure removes that; it only decides where the error shows up.
The complaint usually arrives as a part that shifts laterally by 0.4 mm halfway through a run, with the feeder itself reporting perfect length.
Length and lateral position are two separate systems, and this one is almost never the drive.
What Camber Does to a 3 m Feed Length
Camber is a curve in the plane of the strip. It is not a defect you can straighten out by tension, because the strip is being pushed, not pulled.
An NC servo feeder measures length at the roll face, so a lateral error never enters its feedback loop at all.
A 3 mm camber over a 2 m chord is a radius of roughly 165 m. Feed one metre and the leading edge moves 3 mm off the line it started on.
Feed three metres, which is a normal run between two coil changes on a small die, and the leading edge is 9 mm off. The guide cannot absorb that, so something has to give.
What gives is either the strip, which bows elastically and springs back at the die, or the guide, which wears on one face and then stops correcting anything.
That is the mechanism behind a lateral error that grows through a run and resets when a new coil is loaded.
There is a second effect that confuses the diagnosis. A cambered strip is also slightly longer on one edge, so one side of the part carries more stretch.
On a wide die that shows as a length difference across the strip, which the feeder cannot see because it measures the strip as a whole.
The Four Places Lateral Error Enters
Before touching the feeder, separate the four sources. They look identical at the part and they are fixed in different places.
| Source | How it shows on the part | Where it costs you |
|---|---|---|
| Camber in the coil | Drift grows through the run, resets at coil change | Nothing to adjust. It is a material purchase problem |
| Guide clearance or worn guide faces | Steady offset, worse on narrow strip | 0.3 mm of clearance becomes 0.6 mm of part error |
| Coil not centred, or mandrel worn | Drift in one direction only, per coil | Every coil needs a re-set, which costs a minute each |
| Feed roll axis not square to the line | Drift proportional to feed length, constant per coil | 0.05 mm of skew over 200 mm is 0.5 mm at the die |
Two of those four are the feeder's problem. Two are not, and adjusting the feeder for them makes the next coil worse.
In our own service calls the order is usually camber, then guide clearance, then roll axis, with centring last.
The quickest separator is a coil change. If the error resets when the coil changes, it is camber or centring. If it persists across coils, it is the guide or the roll axis.
Test the roll axis with a dial indicator on a bar clamped in the feed rolls, reading against the guide face at both ends.
More than 0.05 mm difference over 300 mm of roll face is enough to steer a strip, and it is invisible from the operator panel.

How to Measure Camber in Ten Minutes
You do not need a shadowgraph. A chalk line on the floor and a tape is enough to settle the argument.
Cut or shear a 2 m length and lay it on a flat floor, concave side up. Pull a string tight between the two ends and measure the largest gap at the middle.
That gap is the camber. Under 1.5 mm in 2 m is normal mill tolerance and rarely causes trouble. Above 4 mm, no guide setting will hold a 0.2 mm lateral tolerance.
Do it on three coils from three different deliveries before blaming the machine. Coil-to-coil variation is larger than most buyers expect.
If the camber is fine, repeat the same string test on the strip after the roll stack ahead of the feeder. A camber that appears there is being introduced by roll misalignment, not by the mill.
Do the test with the strip at room temperature and lying free. A strip held under tension reads straighter than it is.
Photograph the string and the gap together, because a supplier dispute over camber is settled by a picture with a tape in it.
Which Correction Holds and Which One Only Masks
Four corrections are available. They are listed in order of how much they actually change the outcome.
- Buy camber-controlled coil. A 1.5 mm in 2 m clause costs 2 to 4 percent on the coil price and removes the root cause.
- Close the guide clearance to 0.3 mm total on the strip width, and replace guide faces rather than shimming them.
- Square the feed roll axis to the line within 0.05 mm over the roll face, checked with a dial indicator on a mandrel bar.
- Let the die pilot correct the last 0.1 mm, by releasing the pilot before the strip is clamped.
The first two change the geometry. The last two change how much of the remaining error reaches the part.
Adding guide pressure is the one correction that looks effective and is not. It reduces the visible offset by forcing the strip against one face, then wears that face and returns the error within a month.
Replacing guide faces is a two-hour job and costs USD 300 to 700 a pair. Shimmed faces look cheaper until a shim works loose mid-run.
What buyers and setup engineers ask about lateral drift.
How much camber is acceptable for a 0.2 mm lateral tolerance?
About 1.5 mm in 2 m, and only with guide clearance under 0.3 mm. At 4 mm in 2 m the strip moves 4 mm per metre fed, which no guide can hold.
Does a 0.05 mm roll skew really move the part 0.5 mm?
Yes, at a 200 mm feed length. The error scales with feed length, so a long feed on a wide die turns a small skew into a visible offset at the far edge.
Can a servo feeder correct lateral position on its own?
No. It controls length. Lateral position needs a guide, a pilot in the die, or a strip steering unit, and only the pilot acts at the point where the error matters.
How often should guide faces be replaced on a 300 mm wide line?
Every 12 to 18 months on two shifts. Measure the gap at both ends; more than 0.15 mm difference means the faces are worn unevenly and are steering the strip.
At what camber does a coil become unusable rather than inconvenient?
Above 8 mm in 2 m on strip narrower than 150 mm. At that point the strip walks out of the guide within one feed length and the coil is better returned.
Does oil on the strip make lateral drift worse?
It makes it less predictable. Oil film of 2 to 5 g/m² lets the strip slide sideways under guide pressure, so the same camber produces a different error each coil.
Where Guide Adjustment Stops Working
It stops at the point where the strip buckles instead of sliding. A guide tight enough to steer a cambered strip also creates an edge compression that shows as a wave along one edge.
That wave then reaches the die and becomes a flatness defect, so the cure has moved the problem rather than removed it.
It also stops on very thin strip. Below 0.5 mm, guide pressure deforms the edge before it steers the strip, and the coil is being damaged by the correction.
And it stops when the die has no pilot. Without a pilot to take the last correction, whatever lateral error survives the guide goes straight into the part, and no guide setting is accurate enough to replace it.
On a line where those three limits apply, the answer is coil quality, not machine adjustment.
Our coil line knowledge base puts a number on each contribution, including the lateral term, so you can see what a guide change is worth against a coil clause.
One more boundary deserves stating: a steering unit can correct lateral position, but it needs 1.5 to 2 m of strip to act on.
On a short line where the guide sits 600 mm from the die, there is no room for the correction to take effect before the pilot engages.
What to Put on the Coil Order
Three clauses in the purchase order remove most of this problem before the coil arrives.
- Camber limit stated as a number over a stated chord, for example 2 mm in 2 m, measured at the mill.
- Width tolerance at plus or minus 0.1 mm, since a 0.4 mm width swing changes the guide clearance you set.
- Coil weight range that suits your mandrel, so a 9 tonne coil does not arrive for a 6 tonne machine.
A supplier who will not quote a camber figure is quoting an average, and the average is not what reaches your guide.
Ask for the camber figure on the mill certificate rather than in an email. A certificate number is what makes a claim possible.
Then measure one coil from each delivery for the first three months. That is enough to know whether the clause is being held.
The feeder data library lists guide widths and roll face options by model, which is the sheet to open before assuming a wider guide will solve the drift.
Where to go next
Lateral position is one term in the error budget that runs from the coil to the die. The cluster's accuracy pillar adds up every term and shows which one is worth buying down first.
See the full feed accuracy budget



