How to Set Up Engineering Unit Scaling for a Tag
Scaling turns a raw signal - a milliamp current, a count, a register value - into the pressure or flow an operator actually reads, and getting the endpoints wrong makes every value on that tag quietly wrong. This guide is for the engineer configuring the raw-to-engineering-units conversion on a new point who wants the number on the screen to match the transmitter. It covers matching the transmitter range exactly, handling the 4-20 mA live zero, applying square root only where it belongs, and proving the scale at both ends.
Set Up Engineering Unit Scaling in one line: To set up engineering unit scaling, take the transmitter's configured range as the truth: map the raw low value (4 mA or its equivalent count) to the range's low engineering value and the raw high (20 mA) to the high. Respect the 4-20 mA live zero so 4 mA reads the bottom of range, not zero signal, apply square-root extraction only for differential-pressure flow and only once in the chain, then verify the reading at both endpoints against a simulated signal before trusting the middle.
Match the Transmitter Range Exactly
Scaling has exactly two anchor points and they must match the transmitter, not a guess. Whatever range the transmitter is configured for - say a pressure transmitter set for 0 to 250 psi - is the range your scaling must reproduce: raw-low maps to 0 psi, raw-high maps to 250 psi. If the SCADA scaling says 0 to 300 while the transmitter is 0 to 250, every reading is off by the ratio, and it will look plausible enough that nobody notices until a calibration check. The transmitter's configured range is the single source of truth for the endpoints.
Read the actual transmitter configuration rather than assuming the nameplate or the loop sheet. Transmitters get re-ranged in the field, and a scale built from an outdated drawing reproduces the old range against a device now set to a new one. Confirm the live range from the transmitter itself, because the two anchor points are the entire scale, and if either is wrong the whole tag is wrong. This is foundational to any raw SCADA tag that carries an analog value.
Keep the engineering units and the range together as one intentional choice. The units on the tag must be the units the range is expressed in, and mixing them - a range entered in one unit while the tag is labeled in another - is a classic silent error. Decide the unit, enter the range in that unit, label the tag in that unit, and the reading is coherent end to end.
Handle Live Zero and Square Root Correctly
The 4-20 mA standard uses a live zero, meaning the bottom of range is 4 mA, not 0 mA, and this is the point most likely to be scaled wrong. A signal at 4 mA should read the low end of the engineering range, and 0 mA should read as a fault, not as the bottom of range. Scale from 4 to 20, treat below-4 and above-20 as out-of-range signals worth flagging, and you get both a correct reading and a free broken-wire detector, since a severed loop drops toward 0 mA and lands clearly outside the live band.
Apply square-root extraction only where the physics demands it and only once. A differential-pressure flow measurement has a square-law relationship between DP and flow, so somewhere in the chain the square root must be taken to linearize it - but only once, and in one place. If both the transmitter and the SCADA scaling apply it, the value is wrong twice over; if neither does, the flow reads badly nonlinear. Decide where the root is taken and confirm it happens exactly once, a decision closely tied to where to apply square-root extraction.
Do not scale signals that are not linear-to-range as if they were. A level from a differential-pressure transmitter on a non-uniform vessel, or a flow with a nonlinear element, needs the right transform, not a straight two-point scale. Straight linear scaling is correct for most 4-20 mA transmitters because the transmitter already linearized the measurement, but confirm that assumption rather than applying it blindly to a signal that carries a nonlinear relationship the transmitter did not remove.
Verify the Scale at Both Endpoints
Prove the scale by injecting known raw signals at the endpoints and reading the result. Feed the equivalent of 4 mA and confirm the tag reads the bottom of range; feed 20 mA and confirm it reads the top. A decade box, a loop calibrator, or a signal simulator drives the raw input to known values, and the two endpoints together catch both an offset error (wrong zero) and a span error (wrong high point). If both ends read correctly, the linear middle follows.
Add a midpoint check when the transform is not a straight line. For a square-root flow or any nonlinear scale, the two endpoints can be perfect while the middle is wrong, because a nonlinearity error hides at the ends. Inject a midscale raw value and confirm the reading matches the expected transformed value, which is the only way to catch a square root applied twice or not at all. This endpoint-plus-midpoint check is the kind of step that belongs in a point commissioning checklist so no tag goes live unscaled-tested.
Common Mistakes to Avoid
The most common mistake is a scaling range that does not match the transmitter's configured range, usually because the scale was built from an outdated drawing after the transmitter was re-ranged. Read the live range from the device. The second mistake is mishandling the 4-20 mA live zero - scaling from 0 mA instead of 4 mA - which offsets every reading and throws away the broken-wire detection a correct live-zero scale gives you.
The third mistake is applying square-root extraction in two places or none, giving a flow that is wrong twice over or badly nonlinear. Take the root exactly once and know where. The fourth is verifying only at one endpoint, which catches a zero error but misses a span error or a hidden nonlinearity in the middle. Check both ends, and add a midpoint for any nonlinear scale, before the tag is trusted.
Frequently Asked Questions
How do I scale a 4-20 mA signal to engineering units?
Map 4 mA to the low end of the transmitter's configured engineering range and 20 mA to the high end, using the transmitter's live range as the source of truth. Respect the live zero: 4 mA is the bottom of range, and 0 mA should read as a fault, not zero signal, which gives you broken-wire detection for free. Then inject 4 mA and 20 mA to verify both endpoints before trusting the linear middle.
Where should square-root extraction be applied when scaling flow?
Exactly once, in one place - either the transmitter or the SCADA scaling, never both and never neither. A differential-pressure flow has a square-law relationship, so the root must be taken to linearize it, but applying it twice makes the value wrong twice over and omitting it leaves the flow badly nonlinear. Decide where the root lives, confirm it happens once, and verify with a midscale injection since the error hides at the endpoints.
Why does my scaled tag read wrong even though the endpoints look right?
Because a nonlinearity error hides at the endpoints. On a square-root flow or any nonlinear scale, both ends can read perfectly while the middle is wrong from a square root applied twice, omitted, or a wrong transform. Inject a midscale raw value and compare the reading to the expected transformed value. A straight linear scale will not have this problem, but any nonlinear transform needs the midpoint check to catch it.
Automation services
Need help turning this into a working system?
Merobix integrates SCADA, programs Allen-Bradley and Siemens PLCs, and designs and fabricates industrial control panels.
Meeting requests are reviewed before confirmation.