Load Cell Reading Low on a Long Four-Wire Cable
A 50 kN load cell rated at 2.0 mV/V and 350 Ω is bolted under a batching hopper and connected to a weigh indicator 60 m away through 0.1 mm² copper cores (0.17 Ω/m per conductor, two conductors carrying excitation current). The indicator supplies 10 V DC excitation and uses the same supply as its ADC reference, so the measurement is ratiometric. The cable runs in a sealed conduit along an outdoor pipe rack whose temperature swings from 5 °C to 45 °C, and pulling a new cable would need a plant shutdown. At commissioning, with a certified 40.00 kN test weight on the hopper, the indicator reads 37.90 kN. The plant requires ±0.2 % of full scale over the whole ambient range. The load cell brings out only four wires and the indicator has unused remote-sense terminals. Zero has been re-checked and is correct, and repeatability at constant temperature is better than 0.05 %.
1. Account for the 37.90 kN reading with a calculation. Is the load cell mis-calibrated?
Read the worked answer
Full-scale output at 10 V is 2.0 mV/V × 10 V = 20 mV, so 40.00 kN should give 16.00 mV. The excitation loop resistance is the bridge plus both leads: 350 + 2 × (60 m × 0.17 Ω/m) = 350 + 20.4 = 370.4 Ω. Excitation current is 10 V / 370.4 Ω = 27.0 mA, and the lead drop is 27.0 mA × 20.4 Ω = 0.55 V. The bridge therefore receives 9.45 V and produces 16.00 × 9.45/10 = 15.12 mV. Because the ADC reference is still the 10 V supply, the indicator reports 40.00 × 15.12/16.00 = 37.80 kN, within 0.1 kN of the observed value. The load cell is not mis-calibrated; it is behaving correctly at the voltage it actually receives. The missing 5.5 % is entirely the excitation lead drop, and because the reference is taken at the supply rather than at the bridge, the ratiometric connection cannot cancel it.
2. Compare three options: re-span the indicator, pull a six-core cable for remote sensing, or mount a digitizer at the hopper. Recommend one.
Read the worked answer
Re-spanning only corrects the error at calibration temperature. Copper rises about 0.4 %/°C, so over the 5–45 °C swing the 20.4 Ω of lead grows by roughly 20.4 × 0.004 × 40 ≈ 3.3 Ω. The drop rises from 0.55 V to about 0.64 V, a change of 0.09 V or 0.9 % of excitation, which is four to five times the 0.2 % budget; a one-point re-span would pass in the workshop and fail in the field. Six-wire remote sensing is the textbook answer, but the conduit is sealed and only four cores exist, so it means a shutdown. Mounting a small digitizer or transmitter at the hopper reduces the excitation loop to about 2 m: lead resistance is 2 × 2 × 0.17 = 0.68 Ω, giving a drop near 0.02 V, roughly 0.2 % of span, which can be calibrated out with a thermal drift of only about 0.03 % over the full range. That meets the ±0.2 % requirement without pulling cable, provided the enclosure is rated for the area and the digital link is immune to the VFD environment. Recommend the local digitizer, with six-wire as the fallback at the next shutdown.
3. Before committing to either option, what verification measurements would you make at the installation?
Read the worked answer
Measure, do not assume. First, put a DMM across the load cell excitation terminals with the hopper loaded and compare with the indicator's supply terminals: the difference should be about 0.55 V and confirms the lead-drop model. Second, read the load cell's actual zero-balance and rated output from its calibration certificate rather than from the nameplate, since a 1 % certificate tolerance would consume a large part of the budget. Third, check the shield: it should be grounded at one end only, otherwise the 60 m run in a VFD-rich area will inject noise that no amount of span correction will remove. Fourth, verify the indicator's reference really is the excitation supply; if it is an internal reference, the ratiometric argument changes and supply drift must be added to the error budget. Finally, log a zero and a known load at two ambient temperatures to separate span error from zero drift before spending shutdown time.