3. Polarity verification methods
Polarity verification answers a single, deceptively simple question: when a load is applied in the intended direction, does the instrument produce a positive or negative reading, and is that sign consistent with the wiring convention the system expects? A load cell is a passive bridge; its output sign is determined by the interaction of four things — the strain gauge placement on the spring element, the bridge wiring (which arm is excited positive, which is signal positive), the mechanical mounting orientation, and the sign convention configured in the indicator or PLC. Any one of these can invert the result. Because a reversed channel often still reads plausibly (a compression load shown as tension is a perfectly stable, linear, repeatable number), polarity errors survive casual commissioning and only surface later as wrong totalization, runaway control loops, or safety interlocks that trigger in the wrong direction. Verification therefore needs to be deliberate, layered, and documented.
Establish the reference convention before testing anything
Before any measurement, fix the sign convention in writing for the whole project. A practical and widely used convention is: positive output = load acting in the cell's primary sensing direction as marked on the body, where a tension cell reads positive in tension and a compression cell reads positive in compression. State the unit (mV/V, kg, N, or engineering units after scaling), the excitation voltage, and the indicator's decimal and sign display. Every subsequent check — visual, electrical, and live-load — is judged against this written reference. If the site instead uses "positive = increasing weight on the platform," say so explicitly, because for a compression cell mounted under a hopper that is the same physical direction but a different label than "positive = compression." Ambiguity here is the root cause of most polarity disputes.
Visual inspection: reading the hardware before powering it
Visual inspection is the cheapest and most frequently skipped layer. It has three targets.
1. Cable and conductor identification. Confirm the standard four-wire color code actually used by the manufacturer — common conventions are red (excitation +, EXC+), black (excitation −, EXC−), green (signal +, SIG+), and white (signal −, SIG−), but this is not universal; some vendors use yellow or blue for signal, and six-wire cells add sense leads. Read the calibration certificate or the label on the cable jacket rather than assuming. A swapped SIG+ / SIG− pair produces a clean sign inversion with no other symptom, which is exactly the failure mode that visual inspection is meant to catch.
2. Arrow and marking orientation. Most cells carry a load-direction arrow or a "LOAD" face. Verify that the arrow points in the direction the mechanical structure will actually push or pull. On a tension link in a guy wire, the arrow should point along the wire axis away from the anchor; on a compression canister under a tank leg, it should point down into the cell. If the cell is installed upside down relative to its arrow, the electrical output will be inverted relative to the drawing even though the wiring is correct.
3. Junction box and terminal torque. Inspect the summing junction or terminal block for conductor strands bridging adjacent terminals, loose screws, and moisture. A single stray strand between EXC− and SIG− creates an offset that can mimic a polarity fault. Record the as-found wiring in a photo or sketch; this becomes the baseline for the electrical check.
Electrical check with a known load: the definitive test
Visual inspection narrows the possibilities but cannot prove sign. The definitive method is to apply a known, measurable load and observe the raw bridge output with a millivolt meter or a calibrated indicator in mV/V display mode. The procedure below is written for a compression cell rated 2.0 mV/V at 5,000 kg full scale, excited at 10 V DC — a typical example, with numbers chosen for illustration.
- Zero the meter with no load on the cell and the mechanical structure settled. Record the no-load output; a healthy cell should sit within roughly ±0.5 % of full-scale output, i.e. within about ±0.01 mV/V for this example.
- Apply a known load in the intended positive direction. A calibrated test weight, a dead-weight stack, or a verified hydraulic press with a reference gauge all work; the load must be at least 20 % of full scale to make the sign unambiguous against noise. For this example, apply 1,000 kg.
- Read the raw output. Expected magnitude is 1,000 / 5,000 × 2.0 mV/V = 0.40 mV/V, which at 10 V excitation is 4.00 mV. The sign is what matters: it must be positive under the stated convention.
- Remove the load and confirm the reading returns to the no-load baseline within the cell's specified return-to-zero tolerance.
- If the sign is wrong, do not "fix" it in software first. Correct the physical or wiring cause, then repeat. Software inversion is acceptable only when the hardware is proven correct and the indicator's convention genuinely differs from the project reference — and that decision must be recorded.
Expected output direction for tension versus compression
The table below states the expected raw-output sign for the two common cell types under the convention "positive = primary sensing direction." Use it as the acceptance criterion for step 3 above.
| Cell type | Applied load | Expected raw output sign | Typical symptom if inverted |
|---|---|---|---|
| Tension (S-type, link) | Pulling along axis, away from anchors | Positive | Weight reads negative; totalizer subtracts |
| Tension (S-type, link) | Pushing / compressing the link | Negative | Overload trip on release instead of on pull |
| Compression (canister, button) | Downward force into the cell | Positive | Hopper weight reads negative; batch control runs backward |
| Compression (canister, button) | Upward force / lift-off | Negative | Empty-vessel zero drifts positive |
| Bending beam (platform) | Downward load on platform | Positive (per arrow) | Platform reads negative; tare behaves inversely |
Two subtleties deserve attention. First, a tension cell and a compression cell of the same bridge wiring will produce opposite signs for the same physical direction of force, because the strain gauges are bonded to opposite faces of the spring element. This is normal and is the reason the cell type must be known before judging a sign. Second, multi-cell systems summed in a junction box can have one cell inverted relative to its neighbors; the summed output may still look reasonable at light load but will be non-linear or exhibit corner errors under eccentric loading. Verify each cell individually at the junction box before trusting the sum.
Documenting the result
Record, for each channel: cell type and serial, arrow orientation as installed, as-found wiring colors, no-load output in mV/V, applied test load, observed output in mV/V and its sign, and pass/fail against the reference convention. A channel that passes all three layers — visual, known-load sign, and return-to-zero — is verified. A channel that passes visual but fails sign should be corrected at the wiring or mounting level and retested; a channel that passes sign but drifts on return-to-zero has a mechanical or moisture problem that polarity testing alone will not resolve.
The three layers are complementary rather than redundant. Visual inspection catches the errors that are visible before power is applied — wrong color code, reversed arrow, bridged strands. The known-load test catches everything else, because it measures the actual electrical response of the assembled system. The sign acceptance step converts that response into a pass/fail decision against a written convention, which is what makes the result auditable and repeatable across shifts and contractors. When all three agree, the channel is verified; when they disagree, the disagreement itself is the diagnostic clue, pointing to whether the fault lies in wiring, mounting, or configuration.