Between two stored points, a historian has to decide how to draw the signal, and the honest answer depends on what the tag represents. A continuous measurement really did pass smoothly through the values in between; a discrete state did not. The interpolation mode is a per-tag flag that tells the historian which assumption to make, and setting it wrong quietly misrepresents the data on every trend and every calculation.
Interpolated vs Stepped Tags in one line: A tag's interpolation mode determines how the historian fills in the space between two stored points. An interpolated, or continuous, tag is drawn as a straight sloped line, assuming the value moved smoothly between the points. A stepped, or discrete, tag holds the last stored value flat until the next point, assuming the value stayed constant and then jumped. The flag should match the physics of the signal, sloped for analog measurements and stepped for states and counters.
A historian stores discrete points in time, but a trend and most calculations need a value at moments where nothing was stored. To supply one, the historian looks at the two stored points that bracket the moment and applies the tag's interpolation mode. In interpolated mode it draws a straight line between the two points and reads the value off that line, so the trend slopes smoothly from one point to the next. In stepped mode it ignores the later point until its timestamp arrives and simply holds the earlier value flat, producing a horizontal segment that then jumps.
The choice is not cosmetic. It changes what value the historian reports for any time between stored points, which means it changes trend shapes, exported data, and any aggregate calculated over the interval. A value queried halfway between two stored points comes out as the midpoint of the slope on an interpolated tag and as the earlier point's value on a stepped tag, and those can be very different numbers.
The mode is set per tag precisely because different tags describe different kinds of quantities. It is a property of what the signal means, not of how it was collected, so it has to be configured deliberately when the tag is created. A historian cannot infer physics from timestamps alone, so if the flag is left at a default that does not match the signal, every retrieval of that tag inherits the mistake.
Continuous analog measurements should almost always be interpolated. A tank level, a line pressure, a temperature, or a flow rate really does move through the intermediate values on its way from one reading to the next, so a sloped line between stored points is the faithful reconstruction. Drawing these as steps would falsely imply the value teleported, which misreads ramps and makes smooth processes look jagged.
Discrete and state tags should be stepped. A pump that is off and then on did not pass through half-on; a valve is open or shut; a mode is in one of a few named states; an alarm is clear or active. For these, holding the last value flat until it genuinely changes is the truthful picture, and interpolating them would invent nonsensical fractional states along a slope that never existed. Digital and status tags almost always want stepped mode.
Counters and totalizers are the case that surprises people. A running total that only ever increases and jumps at each stored point is better represented stepped, because between reported values the counter held its last figure rather than gliding up a slope. Getting these wrong is a common source of confusing totalizer trends that appear to ramp continuously when the underlying count actually stair-stepped.
On a SCADA trend screen the interpolation mode is the difference between a chart that tells the truth and one that lies convincingly. A discrete pump-status tag accidentally set to interpolated will show diagonal ramps between off and on, implying the pump spent time in states it never occupied, and an operator reading that chart could badly misjudge what happened. A continuous pressure tag set to stepped will look like a series of sudden jumps that hide the real ramps and slopes.
The same flag feeds automated calculations, so a mis-set mode does not just mislead the eye, it corrupts numbers. A time-weighted average, a runtime total, or a duration-in-state calculation all depend on how the historian believes the value behaved between points. A state tag treated as interpolated will produce wrong runtime hours; a continuous tag treated as stepped will bias its average toward the earlier readings in each interval.
Merobix lets operators set the interpolation mode per tag so states and analogs each display honestly, whether a signal is trended live in the browser or exported for a report. A wellsite's run-status flags step cleanly between on and off while its pressures and flows slope smoothly, so the dashboard reflects what the equipment actually did rather than an artifact of how the historian chose to connect the dots.
Use stepped mode for discrete and state signals such as pump on or off, valve open or shut, operating modes, alarm status, and counters or totalizers. These values hold constant and then jump rather than sliding smoothly, so holding the last value flat is the truthful reconstruction. Interpolating them would invent fractional states that never existed.
Its trend will show sudden jumps at each stored point with flat segments in between, hiding the real ramps and slopes the measurement actually followed. Calculations over intervals will also be biased, because the historian assumes the value stayed at the earlier reading until the next point. Analog measurements like pressure, level, temperature, and flow should be interpolated instead.
It affects both. The same flag that shapes the trend also tells the historian what value existed between stored points, which drives time-weighted averages, runtime totals, and duration-in-state calculations. A mis-set mode therefore corrupts computed numbers, not just the appearance of the chart, which is why matching the mode to the signal's physics matters.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.