Coordinating Tank Calibration, Sender Kinematics and Gauge Filtering

System boundary

A specified tank geometry, fluid state, float and arm, pivot, resistive sender card, excitation and receiving gauge or controller. Vehicle-level range prediction, warning strategy, safety and regulatory compliance remain with the system developer; the card supplies only a drawing-defined transfer element.

Create a five-column calibration chain

For every calibration point, record liquid volume, equilibrium height at the defined attitude, float or pivot angle, card resistance or ratio and displayed quantity. Include datum, direction of approach, temperature and stabilization rule. This table exposes a duplicated or reversed point before interpolation is implemented. Keep unusable volume and warning reserve as explicit system definitions rather than forcing them into the printed track geometry.

Check local slope through every conversion

The electrical sensitivity to volume is the product of successive local derivatives.

Locate a weak-resolution region by example

At one illustrative point, suppose tank geometry gives 0.8 mm height change per liter, linkage gives 0.30 degree per millimeter, the card changes 4 ohms per degree and electronics scale 0.5 display unit per ohm. The combined slope is 0.48 display unit per liter. A flat tank region or linkage near a geometric singularity would reduce it. The figures show multiplication only and are not data for any tank or sender.

Validate the static chain before adding filters

Fill and drain slowly through defined points, allow the agreed stabilization interval and record actual volume, attitude, mechanical position and raw electrical output. Run both directions to reveal float friction and backlash. Compare direct card measurement with controller acquisition to isolate harness and input effects. Do not tune a filter against an unverified static curve, because lag can make an offset appear smaller during one test while preserving the wrong endpoint.

Define what filtering should and should not remove

Slosh rejection, refueling response, hill hold and warning behavior are software-system decisions. Specify sample rate, time constants, rate limits and reset events together with the raw-signal validity range. Filtering should not conceal open circuits, contact loss or implausible dual-channel relationships. The passive card has no awareness of vehicle motion; its role ends at the bounded electrical transfer.

Keep ownership visible across the curve

One controlled matrix prevents a card adjustment from silently absorbing a tank or controller change.

Allocate error where it is physically generated

Separate volume reference uncertainty, tank dimensional variation, float immersion, pivot play, card transfer, contact resistance, harness resistance, acquisition error and interpolation. Some terms vary systematically with level; others are random or direction-dependent. A root-sum-square number is appropriate only for justified independent distributions. Preserve signed worst-case checks at warnings and stops where containment matters more than typical accuracy.

Validate reference states and representative transients

Use traceable volume addition or removal, controlled attitude, specified fluid condition and time-aligned raw signals. Include empty, reserve, midrange, full, fill/drain reversal and relevant thermal states. Dynamic trials should retain raw and filtered channels. The integrator defines accuracy, warning and safe-test requirements. ChipSimple can support review of the drawing-defined resistor pattern; complete gauge calibration remains customer-owned.

Version the curve as a multi-part configuration

A calibration belongs to a tank, float, arm, pivot, card, harness, electronics and software set. Record those identifiers with the point table and interpolation method. Reopen after changes to tank baffles, installation angle, fluid density, arm geometry, wiper radius, track, supply, input impedance or filter. Without configuration control, a good curve can be applied to the wrong physical system and fail without any defective component.

Interpolation between calibration points needs a declared method and monotonicity check. Piecewise linear, polynomial or table-based approaches can produce different behavior near sparse points and endpoints. Evaluate quantization from the ADC and display separately from physical curve error. If temperature compensation is used, retain the uncompensated channel and the temperature source during verification. A controller update that changes interpolation, debounce or warning hysteresis is a calibration revision even when the card and tank are unchanged. Archive the executable parameter set or its controlled identifier with the physical configuration and the volume reference record.

Tank reference measurements also need uncertainty and thermal-volume treatment. State whether delivered volume is corrected for temperature, how residual liquid is handled and where the empty reference is established. Repeat selected points after disassembly and reinstallation to reveal mounting sensitivity. A calibration curve should not claim better accuracy than the combined reference, mechanism, electrical and interpolation evidence supports.