How to use a recorded step response
Use a time column and a measured output column from a CSV or spreadsheet. If your file has sensor readings and command signals, choose the sensor column for the response. Check the first three readings before calculating. Enter the original time unit and select the starting timestamp of the part you want to examine. Results use elapsed time from that point.
Enter the pre-step baseline as the initial output and the value you want to assess as the final reference. For example, when a process moves from 100 toward a target of 80, enter 100 and 80. If you choose the last sample as the reference, the results describe the response relative to that choice. The calculator always prints the final measured output separately.
How do you calculate overshoot on a falling step?
Divide each sample's change from the initial output by the signed difference between the final and initial references. That normalized response moves toward 1 for either an increase or a decrease. Find the largest recorded normalized value and subtract 1. Multiply the positive excess by 100.
z = (output − initial) ÷ (reference − initial). Overshoot = 100 × max(0, max(z) − 1).
For a step from 100 toward 80, a recorded low of 75 is 5 units beyond the reference on a 20-unit step. Its observed overshoot is 25%. The farthest sample in the step direction is a minimum for a falling step. It is a maximum for a rising step.
How are transition time and the settling band measured?
The 10–90% transition time is the difference between the first observed crossings of the two normalized thresholds after your analysis start. With adjacent readings on opposite sides of a threshold, the calculator uses straight-line interpolation. If a threshold crossing is already past, outside the recording or across a gap larger than your limit, the calculator explains why it cannot report a complete transition time.
The settling band is the final reference plus or minus the chosen percentage of the step size. The calculator finds the final out-of-band observation and checks the run of in-band samples that follows it. Enter a planned hold and a maximum sample gap to evaluate that run. A capture ending two seconds after first entering the band cannot demonstrate a three-second hold. Return to the acquisition and capture a longer window when the result is needed.
These are measurements of the supplied samples. A peak or out-of-band excursion could fall between observations. A short recording cannot establish what happens later. You can compare common thresholds with the MathWorks sampled step response reference. Its mode and reference choices may differ.
What can I import?
Use 4–2,000 sample rows with 2–32 columns and an optional header. Timestamps must rise strictly. The import accepts rectangular CSV or tab-separated text. For a file that includes a separate units row or equipment settings before the header, copy the time and response columns into the paste box and choose their time unit. This browser tab keeps your working values in memory. Clear or close the tab to discard them.