Thickness is a profile, not a number
Every sheet is sold with a nominal thickness and a tolerance, and both of them describe a single number. Neither describes the sheet. What matters to the thermoformer, the printer and the die-cutter is the profile: how the thickness varies across the width and along the length, and whether that variation is stable.
A 2.00 mm sheet that runs 2.01 at one edge, 2.16 in the middle and 2.02 at the other edge is inside a plus or minus 0.10 tolerance and is still a bad sheet. It will heat unevenly, form unevenly and trim unevenly, and the parts from the thin band will be different from the parts from the thick band.
That is why the number to specify is the cross-direction spread, not the thickness. Spread is the difference between the thickest and thinnest point at one station across the web, and it is the number the die bolts actually control.
What one turn of a die bolt does
A die bolt pulls or pushes the lip against melt pressure. The lip is a beam, so turning one bolt changes the gap most directly under that bolt and less at a distance, with the effect falling off over roughly two bolt spacings. That is why adjusting a bolt to fix a local thick band can create a thin band two bolts away.
The practical correction is a wave, not a set of independent points. A profile that runs thick, thin, thick, thin needs every other bolt moved slightly, not the thick ones moved a lot. Operators who correct each peak to nominal produce a profile with the same amplitude and a shorter wavelength, which is worse because it is harder to see on a hand gauge and harder to correct on the next pass.
The second thing one turn does is change the total output slightly. Tightening several bolts reduces the die opening and raises the pressure upstream, which changes the melt temperature and the die swell, so the profile shifts as well as scaling. This is why adjustment is iterative rather than calculated.
The passes converge, then stop
Cross-direction spread after each pass: 0.120 mm at the start, 0.060 after the first, 0.035 after the second, 0.025 after the third and 0.022 after the fourth. The first pass halves it, the second nearly halves it again, and then the curve flattens.
The flattening is the point at which the die bolts stop being the limit. Below about 0.025 mm on a 2 mm sheet the remaining variation comes from the lip's own flatness, the temperature uniformity across the die, the flow balance of the manifold and the roll stack. Adjusting bolts past that point just moves the error around.
Knowing where the curve flattens saves a lot of production time. A line that spends an hour chasing the last 0.01 mm is an hour of scrap that buys nothing, because the limit is no longer something a bolt can reach.
| Pass | Spread after | What changed |
|---|---|---|
| Start | 0.120 mm | Profile as found, eight peaks and troughs |
| Pass 1 | 0.060 mm | Every other bolt corrected towards the mean |
| Pass 2 | 0.035 mm | Second harmonic corrected |
| Pass 3 | 0.025 mm | Small local corrections at two bolts |
| Pass 4 | 0.022 mm | Barely measurable, and the limit is no longer the bolts |
Table 1 — Cross-direction spread after successive adjustment passes
The gauge decides the answer
A hand-held thickness gauge on a hot sheet reads low, because the sheet is soft and the foot sinks in. The same sheet measured cold reads higher, and the difference on a 2 mm sheet at 60 C can be 0.03 mm, which is larger than the tolerance being argued about.
Contact gauges also read low on a compressible sheet and high on a stiff one, and they read differently depending on how hard the operator presses. That is why a thickness argument between a supplier and a customer is usually a measurement argument, and why the measurement method has to be on the drawing.
An online scanner avoids most of this. It measures without contact, it measures continuously, and it reports the profile rather than a point. The trade is that a scanner needs calibrating against a known standard, and an uncalibrated scanner is a very confident source of the wrong number.
When the problem is not the die
A profile that wanders along the length rather than across the width is not a die problem. It is a line speed, melt pressure or cooling problem, and adjusting bolts will make it worse by adding a fixed error to a moving one.
The diagnostic is simple. Measure at a fixed position and watch the number over a few minutes. If it drifts, the cause is upstream of the die. If it is stable, measure across the width and the cause is in the die.
| What you see | Where the cause is | What to change |
|---|---|---|
| Stable across the width, drifting along the length | Extruder or line speed | Melt pump, screw speed, haul-off |
| Fixed wave across the width | Die lip and bolts | Adjust the bolts in a wave |
| One edge thick, the other thin | Die temperature gradient | Balance the die zones |
| Periodic bands at a fixed interval | Roll runout or a drive fault | Inspect the rolls and the gearbox |
| Thickness correct but haze varies | Cooling, not thickness | Roll temperature and flow |
Table 2 — Reading a thickness complaint back to its source
What to specify
A sheet specification should carry the profile, the method and the stability.
- Nominal thickness, and whether the tolerance applies to any point or to the mean.
- Cross-direction spread, which is the number the bolts control.
- The measurement method: instrument, contact pressure, sheet temperature at measurement.
- Longitudinal stability, as a limit on drift over a stated length.
- Where the measurement stations are, since the edge is not representative.
- Whether the tolerance applies before or after trimming.
The last item catches people regularly. An untrimmed sheet is wider and includes the thick edge bead that the die produces; a trimmed sheet excludes it. Quoting a tolerance without saying which one is being measured leaves the argument open every time.
What is cross-direction spread?
The difference between the thickest and thinnest point measured across the web at one station. It is the number the die bolts control, and it matters more than the thickness tolerance.
How much does one turn of a die bolt change?
It depends on the thread and the lip stiffness, typically a few hundredths of a millimetre under that bolt, falling off over about two bolt spacings.
Why does adjustment stop working after a few passes?
Below about 0.025 mm on a 2 mm sheet the limit moves to lip flatness, die temperature uniformity and manifold balance, none of which a bolt can correct.
Why does my hand gauge disagree with the supplier's?
Contact pressure, sheet temperature and instrument calibration. A hot sheet reads low because the foot sinks in. Put the method on the drawing.
How can I tell a die problem from a line problem?
Measure at a fixed position over a few minutes. Drifting means upstream; stable means the die.
Does the thickness tolerance apply to the trimmed sheet?
It should say so explicitly. An untrimmed sheet includes the edge bead, which is thicker, so the same sheet passes one specification and fails the other.
Sheet that measures right but forms wrong?
Send the thickness map, the width and what the sheet is formed into. We will tell you whether the spread is the problem and what it will take to hold it.
JA 
EN 
ES 
FR 
DE 
PT 
AR 
RU 
KO 
IT 
NL 
TR 
PL 
ID 