A four-bend box that fits for the first twenty pieces and then runs out of square is usually a backgauge problem, not a bending problem. In multi-bend work the gauge is repositioned many times per part, so any lost motion, datum error or thermal growth appears once per move and accumulates across the sequence.

1. The practical problem

The symptom is a flange that is consistently short or long, or a box whose closing side does not meet after assembly, while every individual bend angle checks out. It worsens through the shift, or appears only on programmes with many repositioning moves, or only when the gauge is extended far from the bed. Because the angle is correct, the operator blames the drawing, the material or the press tooling, and the real error stays in the positioning chain.

2. Failure mechanism

The commanded X value passes through the controller, the servo drive, the ball screw, the linear guide, the finger and finally the sheet edge. Each element contributes an error. A ball screw with residual backlash loses that amount every time the axis reverses, which in multi-bend work happens constantly. A long screw grows with temperature, and the growth is not uniform because the drive end runs warmer than the free end. Guide rail parallelism and finger squareness add an error that changes with gauge extension, and the R axis adds geometry: as the finger swings, the effective stop face rotates, so the point where the sheet touches moves even when the commanded position is unchanged. A final contributor is the sheet itself: a burred, cambered or badly sheared edge contacts the finger at a single point, so the effective stop position varies from part to part even when the gauge is perfectly repeatable.

3. Key parameters

The parameters that determine the result are screw lead and backlash allowance, scale or encoder resolution, approach direction, finger height and squareness, and the temperature of the machine during the run. Approach direction deserves attention because it is free to control and easy to lose: a controller that always approaches the target from the same side converts backlash into a fixed bias that can be compensated, while a programme that approaches from either side turns the same backlash into a random error.

4. Diagnosis and calculation method

• Mount a dial indicator on the bed referencing the stop face, then run a sequence with reversals: 100, 400, 250, 600, 150 mm.
• Record the indicated position at each arrival and compute the spread; repeat after two hours to separate backlash from thermal growth.
• Repeat with the R axis engaged at three finger extensions, since geometric error grows with extension.
• Verify finger squareness against the bed and set finger height for the material thickness before the test, not after.
• Establish cumulative error by summing the per-bend deviation across the sequence; in a four-bend part a small constant error appears at the closing flange multiplied by four.
• Dimension from a single feature where the drawing allows, and place a verification stop on a critical bend when a chain of dimensions is unavoidable.

5. Prevention

Two design features carry most of the benefit: a preloaded ball screw with a rigid thrust bearing, and a gauge supported across its width instead of cantilevered from a single carriage. In process terms, enforce a single approach direction in every programme, set finger height per thickness, and verify the first-off part at the last bend rather than the first. Our 160T/4000 synchronised press brake uses a multi-axis backgauge with height-adjustable fingers that can swing away for deep boxes, and the same architecture is available across our 160 tonne hydraulic press brake range.

6. FAQ

Is a backlash error visible on one part?

Not usually, because a single bend may not reverse the axis. It becomes visible in a sequence, or after manual jogging during setup, and that is why it is often described as intermittent.

Does the scale eliminate backlash?

It measures the carriage, so it removes the effect of screw error from the control loop, but mechanical lost motion downstream of the scale, at the finger or the guide, is still present and must be controlled mechanically.

Should the gauge be re-zeroed every shift?

Re-zeroing hides thermal growth rather than managing it. A better routine is to verify against a known reference at the start of the shift and record the deviation, so that growth is visible as a trend instead of being reset daily.