Synchronizing Multiple Sensors: A Common Reference, a Unified Range and Time Alignment, All Together

On multi-axis synchronous machines (four-column presses, multi-cylinder levelling, gantry beams), even a highly accurate single sensor is not enough — what determines synchronisation accuracy is consistency among the sensors. Calibrating multiple sensors must solve two things at once: spatially, bring every axis to the same physical reference, and in time, align the sampling instants of every axis. The former is a calibration procedure; the latter is a communication architecture. This article gives the practical methods for both. For the meaning of accuracy figures, see the three accuracy terms; for the calibration principle, see factory calibration and linearisation; for multi-axis selection, see multi-axis synchronisation selection.

Step 1: establish a common mechanical reference

Multi-axis calibration starts with the mechanics, not the sensors. A single physical datum face for the whole machine must first be defined (for example the upper face of the table, the lower die face, or a machined datum on the frame). Every axis zero is defined relative to that face, not relative to each cylinder's own retracted position.

Method: use a spirit level, gauge blocks or a laser levelling tool to bring the moving beam parallel to the datum (or to the same set height). In that state, read the raw value of every sensor at the same time and take that set of values as the zero offset of each axis. Then, whenever the axis readings are the same, the beam is necessarily level — which is exactly the semantics that synchronous control needs.

A common mistake is to zero each axis independently against its own mechanical limit. Because cylinder assembly dimensions differ, after calibration "every reading is 0" but the beam is not level, and the synchronising controller works to hold a skewed state.

Step 2: unify the span coefficients

Calibrate the span of each axis by the two-point method (see zero calibration step by step), but the calibration stroke must fall in the same physical interval. If axis A is calibrated over 0–500 mm and axis B over 50–550 mm, each coefficient may be correct on its own, yet the synchronisation error will change with position. After the interval is unified, the reading difference among axes is comparable at any position over the full stroke.

Step 3: full-stroke consistency verification

After calibration, run a full-stroke synchronisation check: move the axes together, stop at several positions, and record each reading against the actual mechanical position. There are two criteria:

• Inter-axis difference: the maximum difference among the axis readings should lie within the machine's synchronisation-accuracy requirement;
• Shape: if the inter-axis difference is constant over the full stroke, the zeros are not aligned — re-reference to the datum; if it grows with stroke, the span coefficients are inconsistent and must be re-calibrated; if it appears only locally, that axis has a mechanical or magnet problem — see tracking down reading errors.

Be realistic about the expected error magnitude: non-linearity is typically <0.02%FS and must be converted to millimetres for each axis range before being stacked; repeatability of the order of ±0.002 mm is a laboratory figure and will degrade on site with vibration and temperature drift. Setting a synchronisation-accuracy requirement tighter than the sum of these figures cannot be achieved by calibration.

Step 4: time alignment — the half that is often ignored

Once the spatial references are aligned, inconsistent sampling instants among axes still produce "false synchronisation error" during motion. The faster the speed, the larger the position difference caused by sampling time skew. Time consistency of different architectures differs as follows:

Therefore, for multi-axis systems with a demanding synchronisation-accuracy requirement, choose from the selection stage a fieldbus model that supports a synchronisation mechanism, such as the Series 197 EtherCAT, the 199PROFINET series or the Series 194 CANopen (used with SYNC; baud rate up to 1 Mbps). Fieldbus synchronisation mechanisms are covered in power-loss retention and synchronous refresh on fieldbus types; matching update rate to the control cycle is in response time and the control cycle.

When to re-calibrate in service

The following cases require a fresh synchronisation calibration; old parameters must not be retained:

• Replacement of any sensor or position magnet (the new part has a different zero offset; see spare parts replacement);
• Cylinder overhaul, seal replacement or adjustment of limit stops;
• A structural change to the frame, or an impact event;
• A sustained rise in synchronisation-error alarms, or a clear shift in the inter-axis-difference baseline.

Archive the "inter-axis difference versus position" curve from the first calibration as a health baseline. Later, comparing the shape of the curve quickly shows which axis has degraded. The idea is the same as comparison in a redundant system; see redundant installation comparison.