In this guide 6 sections
A sensor and a gauge often disagree because they measure different points, respond at different speeds, or use different scaling. Check probe placement, response lag, 4-20 mA range settings, drift and calibration dates. Then run both side by side with a manual log for several days before trusting the dashboard.
Are they measuring the same spot?
Look at where each instrument touches the process. A dial thermometer in a pocket near the pipe wall reads differently from a probe pushed into the middle of the flow. A pressure gauge before a filter will not match a transmitter after it.
Check the mounting too. A temperature probe sitting in air instead of oil, or a level sensor beside a turbulent inlet, tells you about the wrong thing. Compare the two positions on the drawing. Do it first.
Move one instrument before blaming the other.
- Distance from the heat source or inlet
- Immersion depth of the probe
- Before or after a filter, valve or pump
- Ambient air versus the process itself
Does response lag explain the gap?
A thick thermowell slows a thermocouple. A dial gauge with a damped needle smooths out quick swings. The dashboard may report every ten seconds while the operator glances at the dial once a minute. Both look correct and still disagree.
Compare during a steady state, not during a start-up or a batch change. If the values converge after a few minutes of stable running, lag is the cause, not a fault.
Note the reporting interval and the filter setting on the transmitter. Write both beside the reading. Small details matter.
Is the 4-20 mA scaling set correctly?
Many industrial sensors send a current signal. Four milliamps means the bottom of the range and twenty means the top. If the gateway thinks the range is zero to 10 bar and the transmitter is set for zero to 16, every reading is off by a fixed ratio.
Check the transmitter label, the configured range and the engineering units entered in the gateway. Measure the loop current with a multimeter and work the reading out by hand.
A reading that is wrong by a constant factor points to scaling. A reading that is wrong at one end only points to offset.
- Transmitter range on the nameplate
- Low and high values entered in the gateway
- Loop current measured with a multimeter
- Engineering units: bar, kPa or degrees
Could drift or missed calibration be the cause?
Sensors age. It is normal. Probes foul with scale, diaphragms creep and electronics shift with heat. A device that matched the gauge at installation may drift slowly over months. The gauge itself may be the one that is out, especially if no one has calibrated it since the machine was bought.
Find the last calibration certificate for both instruments. Check one against a reference, such as a calibrated handheld or a test bath. Do not skip this.
If neither has a date, treat both as unproven.
Run a parallel check before you trust the dashboard
Keep the gauge and the manual log going for at least a week of normal production. At set times each shift, the operator writes the gauge value next to the dashboard value. Include a start-up, a steady run and a stop.
Review the pairs together. Look for patterns. Decide an acceptable difference for each measurement and record who approved it. Only then retire the paper log.
Keep one gauge on critical points even afterwards. It is the cheapest cross-check you will own.
Parallel run log sheet
A table for the operator, on one page. Columns hold date, time, shift, machine state, gauge reading, dashboard reading, difference, acceptable limit and initials. A second block lists calibration dates, transmitter range, gateway range and probe position for each point, with a line for the approver’s sign-off.
Open a blank worksheet to printQuestions owners ask
Which reading is right, the gauge or the sensor?
Neither is right by default. Check calibration dates on both and compare against a known reference. Then look at position, lag and scaling before blaming either.
How much difference is acceptable?
It depends on the process and the decision the reading drives. Agree a tolerance with the quality or maintenance head before the comparison starts.
How often should sensors be calibrated?
Follow the maker’s guidance and your own quality procedure. Critical points, such as food safety temperatures, need tighter schedules than general monitoring.
Can software correct the mismatch?
It can apply an offset or a scale factor once you know the cause. It should never hide a failing sensor. Log each correction with a date and a name.