A radar level error is easier to investigate when its pattern is recorded rather than described simply as "inaccurate". A nearly constant difference, a difference that grows with level and an abrupt jump suggest different checks. None identifies a failed transmitter by itself, and one comparison point cannot reliably distinguish them.
Collect comparable readings before adjusting anything
Use the same datum and the same physical quantity for every comparison. Radar distance from its reference face is not the same as liquid height above the tank bottom. Record the independent reference method and its limitations, then capture local distance, calculated level and the PLC value as closely together as practical.
Take observations at several safely available, sufficiently stable levels. Avoid comparing a moving surface with an old manual reading. The following patterns are diagnostic prompts, not acceptance tolerances or measured METRAVON results.
Read the radar level error pattern
| Pattern | First check | Avoid |
|---|---|---|
| Fixed gap | Reference planes, empty-distance setting and any applied offset | Correcting the sensor before confirming the same datum |
| Growing gap | Units, span and conversions between transmitter and host | Using one offset to hide a scaling error |
| Step change | Echo selection and process or installation changes at that time | Averaging the jump into a calibration correction |
| Host only | PLC scaling, data interpretation and stale-value behaviour | Replacing a transmitter whose local result remains consistent |

Use simple numbers to separate offset from slope
For illustration, suppose reference levels of 1, 2 and 3 m correspond to indications of 1.1, 2.1 and 3.1 m. The repeated 0.1 m difference suggests checking a reference or offset. If the indications are instead 1.1, 2.2 and 3.3 m, the proportional difference suggests a scale-factor check. These hypothetical values illustrate offset and scale error; they are not test results or product performance data.
Real data may contain both effects and measurement noise. Do not fit a correction to three points without investigating the underlying cause. A bad reference measurement or changing surface condition can also create a misleading pattern.
Keep the model specification separate from the diagnosis
METRAVON MV-RD12 lists +/-2 mm nominal accuracy. That value is not permission to adjust any installed discrepancy until it fits a desired result. The tank reference, installation and output chain must support the comparison. After an authorized correction, repeat the observations and retain both datasets.
Questions during fault finding
Q1. Can one full-tank comparison establish accuracy?
It checks only that condition. A span error may remain hidden elsewhere, and a high-level echo can behave differently from the low-level measurement.
Q2. Should I adjust the radar if only the PLC is wrong?
First investigate the receiving chain. Changing a correct local measurement to compensate for incorrect host scaling makes the instrument record misleading and may affect other users of the signal.
Q3. What evidence makes a support request useful?
Provide paired readings with units, timestamps, reference location and process state, plus the current settings. A screenshot of one unexplained number rarely separates the possible causes.
Summary
Classify the discrepancy before choosing a correction. Send METRAVON the paired measurement record so the review can address the suspected offset, scaling or echo issue without guessing from a single reading.
