Alongside the engineering value an operator reads, most control systems also carry a second, unitless version of every signal: where it sits between the bottom and top of its calibrated range, expressed as a number from zero to one hundred percent. This percent-of-scale value is what bargraphs fill to, what many controllers actually compute in, and what makes signals of wildly different units comparable at a glance. This guide explains what percent of scale means, how it is derived from the engineering value and the range, why so much of a control system runs in percent internally, and the important distinction between percent of span and percent of reading in accuracy specifications.
Percent of scale in one line: A percent-of-scale value is a dimensionless number from 0 to 100 percent expressing where a signal sits within its calibrated span, computed as the engineering value minus the range low, divided by the range span, times one hundred. HMIs, faceplates, and many controllers work in percent internally because it normalizes every signal to a common 0-to-100 scale regardless of units. It should not be confused with percent of reading, a different basis used in accuracy specifications.
Percent of scale answers a simple question: given where a signal's calibrated range starts and ends, how far up that range is the value right now? Zero percent is the bottom of the range, one hundred percent is the top, and fifty percent is exactly halfway, whatever the units happen to be. It strips away the units and leaves only the position within the span, which is why it is sometimes called percent of range or, loosely, a normalized value. A pressure at the middle of its range and a level at the middle of its range are both fifty percent, even though one is in psi and the other in feet.
The arithmetic is a small rescaling of the engineering value. You subtract the range low from the engineering value to find how far above the bottom it sits, divide that by the total span, the range high minus the range low, to get a fraction between zero and one, and multiply by one hundred to express it as a percent. A pressure ranged 0 to 150 psi reading 75 psi is 75 minus 0, over 150, times one hundred, which is fifty percent. A level ranged 0 to 20 feet reading 5 feet is 5 over 20, or twenty-five percent. The same formula converts any engineering value to its place in its span.
The conversion runs both ways. Given a percent and the range, you recover the engineering value by reversing the steps: take the percent as a fraction, multiply by the span, and add the range low. This matters because a faceplate might let an operator enter a setpoint as a percent while the controller stores it in engineering units, or vice versa. Percent of scale and the engineering value are two representations of the same point on the same range, and the range endpoints are the bridge between them.
Percent of scale is the natural internal language of a control system because it makes everything comparable. A bargraph or a trend needs to know how full to draw a bar or where to place a line, and it can only do that if the value is expressed as a fraction of its range; a raw psi number means nothing to a drawing routine without knowing the range, but a percent maps straight onto the height of a bar. So HMIs convert engineering values to percent to render them, and faceplates show a small percent-of-scale bar next to the engineering readout precisely so an operator can see at a glance whether a value is near the top or bottom of its range without doing the arithmetic in their head.
Many controllers, especially traditional PID loops, historically did their computation in percent as well. When a controller works in percent, the process variable, the setpoint, and the output are all on the same 0-to-100 scale, so the error between setpoint and process variable is a percentage and the tuning constants relate percentages to percentages rather than psi to valve position. This keeps the loop math independent of the particular units and ranges involved, so the same controller block behaves consistently whether it is regulating pressure, level, or temperature. Even where modern controllers work in engineering units, the percent-of-scale idea survives in how ranges, outputs, and tuning are reasoned about.
The practical upshot is that percent of scale and the engineering value coexist throughout the system, and both need their range to be correct. A percent value is only meaningful against the range it was computed from, so a bargraph that fills to the wrong level, or a percent readout that disagrees with the engineering value, usually traces back to a range mismatch between where the percent was calculated and the range the display assumes. Keeping the range consistent everywhere a signal is shown is what keeps the percent picture and the engineering number telling the same story.
There is an important distinction hiding behind the word percent when it appears in accuracy specifications, and it is easy to conflate with percent of scale. Percent of span, sometimes called percent of full scale, expresses an error as a fraction of the whole calibrated range, so a given percent-of-span error is the same absolute amount everywhere in the range, whether the signal is near the bottom or the top. Percent of reading expresses an error as a fraction of the current value, so the absolute error shrinks as the reading falls. The two describe very different error behavior even at the same headline percentage.
This matters when comparing instruments or judging how good a measurement is at the low end of its range. A device specified at a percent of span carries the same absolute uncertainty at ten percent of range as at ninety percent, which means the error is a large fraction of a small reading. A device specified at a percent of reading holds a tighter absolute error at low readings but the spec is quoted differently. Confusing the two, or assuming a percent-of-span figure behaves like percent of reading, leads to expecting an accuracy at low flow or low pressure that the instrument does not actually deliver. Percent of scale as a display value and percent as an accuracy basis are related ideas but must be kept separate in the head.
In a cloud SCADA architecture such as Merobix, percent of scale is what makes a dashboard of dissimilar signals legible: bargraphs, gauges, and normalized trends all lean on the value's position within its range, so the platform carries each tag's range and can present both the engineering value and its percent of scale. Keeping the range as explicit metadata means a bar always fills to the right level and a normalized comparison across many tags is honest. And when accuracy or alarm thresholds are reasoned about, being clear about whether a figure is percent of span or percent of reading keeps expectations of low-end precision realistic, which is exactly the sort of detail that having the range and the raw value visible end to end helps an analyst check.
Subtract the range low from the engineering value to find how far above the bottom of the range it sits, divide that by the span (range high minus range low) to get a fraction between zero and one, then multiply by one hundred. For example, 75 psi on a 0-to-150 psi range is 75 divided by 150 times one hundred, which is fifty percent. The same formula works for any signal once you know its range endpoints, and it reverses to recover the engineering value from a percent.
Because percent of scale normalizes every signal to a common 0-to-100 scale regardless of units, which makes signals comparable and easy to render. A bargraph or trend needs a fraction of range to know how full to draw, and a percent maps straight onto that. Many PID controllers historically computed in percent so the loop math and tuning were independent of the particular units and ranges, and that convention still shapes how outputs and tuning are reasoned about.
Percent of span, or percent of full scale, expresses an error as a fraction of the whole calibrated range, so the same percent is the same absolute amount everywhere in the range. Percent of reading expresses an error as a fraction of the current value, so the absolute error shrinks as the reading falls. They describe very different accuracy behavior at low readings, so confusing them leads to expecting a low-end precision the instrument may not actually provide.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.