By Wu Peng, Senior Process Instrumentation Engineer · Last reviewed July 27, 2026
An accuracy of percent of full scale (%FS) is a fixed error: the percentage times the instrument range, in engineering units, at every reading. An accuracy of percent of reading (%RD) scales with the measurement: the same percentage of whatever the instrument currently shows. A 0–100 psi gauge at 0.5 percent FS is ±0.5 psi everywhere; at 0.5 percent of reading it is ±0.5 psi at 100 psi but only ±0.125 psi at 25 psi.
The two conventions describe different-sized error bands from the same-looking number, and the gap grows the lower in the range you operate: a full scale spec read at 25 percent of range is already four times worse as a fraction of the measurement. This guide gives the conversion math, the span vs full scale distinction, combined specifications, the range-oversizing penalty, and how to compare two datasheets honestly before you buy.
Contents
- The two conventions
- Error math compared
- Span vs full scale
- Combined specifications
- The oversizing penalty
- Flow meters and turndown
- Reading a datasheet
- FAQ
The two conventions
Percent of full scale anchors the error to the top of the range:
Error (%FS) = spec × full-scale value
A 0–100 psi transmitter at ±0.5 percent FS may be off by ±0.5 psi at any reading. At 100 psi that is 0.5 percent of what you measure. At 25 psi the same ±0.5 psi is 2 percent of the measurement, and at 10 psi it is 5 percent. The absolute error stays constant; the relative error grows quickly as the reading falls.
Percent of reading anchors the error to the measurement itself:
Error (%RD) = spec × current reading
A ±0.5 percent of reading instrument is ±0.5 psi at 100 psi, ±0.125 psi at 25 psi, ±0.05 psi at 10 psi. The relative error stays constant across the range, which is why percent of reading is the convention for meters sold on wide rangeability.
Two conventions exist for historical and physical reasons. Percent of full scale comes from mechanical gauges, where the error was set by how finely the dial could be graduated. Analog panel meters still state their class as a percentage of full scale deflection (FSD).
Percent of reading became practical with digital sensors, but no sensor can hold a pure percent of reading spec all the way to zero; near the bottom of the range there is always a noise floor, which is why realistic percent of reading specs carry a floor or a stated valid range.
One convention trap to memorize: a spec quoted as a bare percentage with no basis, such as accuracy 0.5 percent, is understood as percent of full scale by industry convention.
Error math compared
To convert a full scale spec into percent of reading at any operating point, divide by the fraction of range in use:
Error as %RD = spec (%FS) × full scale / reading
Here is a ±0.25 percent FS transmitter on a 0–10 bar range, worked at five operating points:
| Reading | Absolute error | Error as percent of reading |
|---|---|---|
| 10 bar (100%) | ±0.025 bar | 0.25% |
| 7.5 bar (75%) | ±0.025 bar | 0.33% |
| 5 bar (50%) | ±0.025 bar | 0.5% |
| 2.5 bar (25%) | ±0.025 bar | 1.0% |
| 1 bar (10%) | ±0.025 bar | 2.5% |
Error percentages are expressed relative to the true reading. The pattern is general: at half range the relative error doubles, at a quarter range it quadruples.
When you compare a full scale spec against a reading spec, there is a crossover point below which the reading spec wins:
Crossover reading = (FS spec / RD spec) × full scale
Example: a ±0.05 percent FS calibrator against a ±0.1 percent RD calibrator, both 0–100 psi. Crossover = (0.05 / 0.1) × 100 = 50 psi. Above 50 psi the full scale unit gives the smaller error band; below 50 psi the reading unit wins, and it keeps winning by a wider margin the lower you go.
Span vs full scale
On transmitter datasheets the basis is often percent of span rather than percent of full scale, and the two are not always the same number. Four terms to keep straight:
| Term | Meaning |
|---|---|
| URL / LRL | Upper and lower range limits: the widest the sensor can ever be set. |
| URV / LRV | Upper and lower range values: where 20 mA and 4 mA actually sit after ranging. |
| Calibrated span | URV minus LRV. A percent of span spec is a percentage of this number. |
| Full scale | The top-of-range value. Equal to span only when the range starts at zero. |
The difference matters most on compound and elevated ranges. Take a gauge range of −1 to 5 bar: full scale is 5 bar, but the span is 6 bar. A ±0.05 percent of span spec allows ±3 mbar of error; the same number read as percent of full scale would be ±2.5 mbar. Same digits, 20 percent more error band.
Percent of span also moves when you re-range a smart transmitter: narrow the calibrated span and the absolute error band narrows with it, down to the turndown limit. Our differential pressure transmitters quote reference accuracy as 0.075 percent, and per the convention above that reads as percent of the calibrated span.
Combined specifications
Once you read enough datasheets you meet three spec grammars beyond the plain single number, all of them attempts to describe a real sensor that behaves like percent of reading in the middle of its range and hits a noise floor near the bottom:
Additive: ±(0.1% of reading + 0.02% FS). Work it at the operating point. On a 0–100 psi range at 20 psi: 0.1 percent of 20 psi is 0.02 psi, plus 0.02 percent of 100 psi is 0.02 psi, total ±0.04 psi, or 0.2 percent of the reading. A plain ±0.05 percent FS unit at the same point allows ±0.05 psi, or 0.25 percent of the reading; the combined spec is tighter here even though it prints two terms.
Whichever is greater: ±0.6% of reading or ±0.1% FS, whichever is greater. The FS term acts as a floor: below the crossover point the fixed floor governs, above it the reading term takes over.
Split range: ±1% of reading from 20 to 100 percent of range, ±0.2% FS below 20 percent. Same idea stated piecewise. Read the boundary carefully, because the spec you compare against a competitor depends on which segment your operating point falls in.
None of these grammars is deceptive by itself. Comparisons go wrong when a bare full scale number is set against a combined spec without converting both to engineering units at the actual operating point.
The oversizing penalty
Under the full scale convention, range oversizing costs accuracy directly, and oversizing is the most common specification mistake we see in requests for quote. The math from a real sizing review:
A process runs at 8 bar. A 0–40 bar transmitter at ±0.5 percent FS carries ±0.2 bar of error, which at 8 bar is 2.5 percent of the reading.
The same 0.5 percent FS class on a properly sized 0–10 bar range carries ±0.05 bar, which at 8 bar is 0.625 percent of the reading.
Same accuracy class, same process, four times the error, purely from range selection. The working rule: choose the range so the normal operating point sits between 60 and 80 percent of full scale, with enough headroom for the worst transient.
Smart transmitters soften the penalty because a percent of span spec tightens as you range down, but not indefinitely; past roughly 10:1 turndown many manufacturers publish a degradation term of the form ±(base + coefficient × turndown) percent, so the gain flattens out. If your operating point lives below 10 percent of any available range, that is usually a sign to change the measurement approach, not the spec sheet. Our guides on pressure transmitters and gauge pressure transmitters list the selectable range steps we build to.
Flow meters and turndown
The convention split runs straight through flow measurement, and it decides how much rangeability a stated accuracy really delivers:
| Technology | Usual accuracy basis | Typical class |
|---|---|---|
| Coriolis | Percent of reading | 0.1 to 0.5 |
| Turbine | Percent of reading | 0.2 to 1.0 |
| Vortex | Percent of reading | 1.0 to 1.5 |
| Variable area (rotameter) | Percent of full scale | 1.0 to 4.0 |
| Thermal mass | Percent of full scale, sometimes combined | 1.5 to 2.5 |
Classes are the ranges our own product pages state for each technology. Always confirm the basis line on the specific datasheet.
A 1.0 percent FS rotameter running at a quarter of its scale is a 4 percent instrument at that flow; a 1.0 percent of reading turbine meter at the same quarter flow is still a 1 percent instrument. That single line explains most of the price and selection logic between the two.
Differential pressure flow elements deserve a special warning, because flow follows the square root of the measured DP. Turn flow down to one third and the differential falls to one ninth of full scale. A DP transmitter at 0.075 percent of span then carries an error of 0.675 percent of the actual differential, which translates to about 0.34 percent of flow after the square root halves the sensitivity.
The penalty grows with the square of turndown, which is why DP flow installations run out of accuracy at 3:1 to 4:1 while a percent of reading meter keeps going. The worked turndown table in our flow rate and pressure relationship guide shows the full progression.
Reading a datasheet
When two quotes land on your desk with different accuracy grammars, do not compare the percentages directly. A fair comparison takes five minutes:
1. Fix the operating point. Normal reading, minimum reading you still care about, and the range each vendor proposes.
2. Convert every spec to engineering units at those points. Full scale specs multiply by range; reading specs multiply by the reading; combined specs get both terms added at each point.
3. Check the basis words. Full scale, span, calibrated span, reading, URL: each one changes the multiplication. A bare percentage means full scale by convention.
4. Add the adders. Reference accuracy is a bench number; temperature effect and long-term stability stack on top in the field, and they are often larger than the headline figure.
5. Compare the engineering units, not the percentages. The smaller band at your operating point wins, whatever the grammar. The same discipline separates accuracy from repeatability, which we cover in accuracy vs precision.
Application example
Fertilizer plant, Middle East. Challenge: an ammonia producer needed to monitor differential pressure across the catalyst beds of its primary and secondary reformers, where the useful signal is a small, slowly rising pressure drop. Solution: we specified differential pressure transmitters with 0.075 percent of span reference accuracy for the catalyst-bed DP measurement. Result: at that class the instrument error band stays small relative to the pressure-drop trend the plant watches, so a genuine rise in bed DP stands out from instrument error instead of hiding inside it.
FAQ
What is 0.1% of full scale accuracy?
An error band equal to 0.1 percent of the instrument range, at any reading. On a 0–100 psi gauge that is ±0.1 psi whether you measure 100 psi or 10 psi. As a fraction of the measurement it is 0.1 percent at full scale but 1 percent at 10 psi, which is why a full scale spec should always be judged at your operating point, not at the top of the range.
What does 1% of full scale mean?
The maximum error equals 1 percent of the full-scale value. A classic exam example: a 0–150 V voltmeter with 1 percent of full scale accuracy may be off by ±1.5 V at any reading. Measuring 75 V, that ±1.5 V is 2 percent of the reading; the relative error doubles because you are at half range.
What is the 4 to 1 calibration rule?
The reference standard used to calibrate an instrument should be at least four times more accurate than the instrument under test, a test accuracy ratio of 4:1. If a transmitter carries ±0.2 psi of allowed error, the calibrator should contribute no more than ±0.05 psi at the test points. Watch the ranges: a calibrator that meets 4:1 at full scale can fail it when the device under test uses only the bottom of the calibrator range.
How do you calculate reading accuracy percentage?
Multiply the percent of reading spec by the current reading. A flow meter at ±0.5 percent of reading showing 40 m³/h may be off by ±0.2 m³/h. To convert a full scale spec into the equivalent percent of reading, multiply the spec by full scale divided by the reading: 0.5 percent FS on a 100 m³/h meter at 40 m³/h is 0.5 × 100 / 40 = 1.25 percent of reading.
Request a quote
Send us the operating point, the minimum reading that still matters, and the error you can live with in engineering units. We will size the range, state the basis of every accuracy number, and quote the class the job actually needs. Tell us the application and we configure one unit, not a shelf part.
Written and technically reviewed by Wu Peng and the Instranova engineering team.