Signal Jump Troubleshooting Flow: Isolating Electrical, Mechanical and Fieldbus Causes Layer by Layer (Save This One)

A jumping signal is the most frequent class of magnetostrictive displacement-sensor site complaint, and the one most easily fobbed off with "try another one". The premise of effective troubleshooting is classify first, then act: jumps divide by cause into electrical-interference, mechanical/magnet and fieldbus-communication classes, and the three have completely different symptom features, trigger conditions and remedy directions. This article gives a layered troubleshooting flow that can be executed in order. For the fault overview see Troubleshooting Magnetostrictive Displacement Sensors; for the earthing topic see Shield Grounding at One End or Both.

Step 0: record the symptom; do not rush to change things

Before acting, obtain four pieces of information; they decide which branch follows:

• Jump amplitude: small last-digit jitter, or a leap to a large number or even an extreme?
• Jump location: random over the full stroke, or concentrated on a fixed stretch of stroke?
• Trigger conditions: does it also jump with the machine at rest? Does it always jump when a particular high-power device starts or stops?
• Time pattern: has it always been so, or only recently? What was changed recently?

Of these, "does it also jump with the machine at rest" is the most effective single criterion: still jumps at rest, look at electrical first; does not jump at rest, jumps as soon as it moves, look at mechanics and the magnet first.

Features of the three classes of jump

Branch A: still jumps at rest → the electrical-interference path

Execute in the following order; verify immediately after each step, and do not change several places at once.

1. Confirm supply quality: measure voltage at the sensor connector with a multimeter and watch whether there is a clear dip when high-power equipment starts or stops; on a long-cable supply, voltage drop must also be budgeted, see Long Cable Runs;
2. Check shield earthing: whether there is hidden both-end earthing (the shield landing on the frame through a metal connector at the sensor end); whether the cabinet side is a 360° clamp circumferential termination;
3. Check routing: whether the signal cable runs in the same trunk, in parallel, with inverter output cables and solenoid-valve wires; a contrast test can be made by temporarily pulling the cable out of the trunk and running a section in free air;
4. Treat the interference source: a freewheel diode or RC snubber on solenoid-valve coils, a reactor on the inverter output;
5. Confirm the signal type: 0-10 V voltage output is inherently weaker than a current loop in a long-cable, strong-interference environment; consider changing to a 4-20 mA type, see the 191 analog series.

Branch B: jumps only when it moves → the mechanical and magnet path

First determine whether the stroke location of the jump is fixed. A fixed location means a physical defect in that stretch:

1. A slow full-stroke pass by hand: take off automatic control, move by hand to either side of the jumping stretch and confirm that the jump is repeatable;
2. Check magnet coaxiality and clearance: whether clearance in that stretch differs from other stretches, whether there are rod-rub marks; see Five Magnet Installation Pitfalls;
3. Check bracket and carriage: on an external type, focus on bracket stiffness, whether the carriage binds, and whether the connection has floating freedom; see External Installation;
4. Check cylinder-integrated hole machining: hole eccentricity will produce abnormal clearance at a particular depth; see Cylinder Bore Machining Requirements;
5. Check ferromagnetic foreign matter: swarf in the oil held at the magnet will cause progressive degradation; see Hydraulic Oil Cleanliness;
6. Check whether a dead zone is being approached: if the jumping stretch is at the end of the stroke, the magnet may have entered a dead zone outside the effective measuring range; see Dead Zones at Both Ends of the Effective Range.

Branch C: fieldbus-type jumps → the communication path

A "jump" on a fieldbus type is often not a wrong measured value, but data that did not arrive. The discriminant is to read the master's communication diagnostic counters (error-frame count, dropped-station count); if the count keeps rising, the problem is at the communication layer rather than the measurement layer. Three first checks: whether there is one 120 Ω terminating resistor at each end of the bus (too many or too few will both cause errors), whether node addresses conflict, and whether connectors have taken water or damp. A CANopen type (such as the 194 CANopen series, baud rate up to 1 Mbps) must also confirm that baud rate and bus length match; the higher the baud rate the shorter the allowed bus length. See Three Causes of Fieldbus Data Loss and Terminating Resistors and Bus Length.

When a body fault of the sensor should be judged

There is reason to suspect the body only when the following are met: all three branches have been checked, the sensor has been taken off on its own, and a test magnet moved by hand over the full stroke in an interference-free environment still jumps. This step is the only means, besides "substitution", that can give a definite conclusion; for the method see Commissioning Tool: Use a Test Magnet to Separate Body Faults from Installation Problems in 5 Minutes. A multi-sensor system can also locate by mutual comparison; see Synchronizing Multiple Sensors.