PID tuning calculator
A named tuning method turns a few measured numbers into three constants. This calculator applies the published correlations — Lambda/IMC (SIMC), Cohen-Coon, open-loop Ziegler-Nichols, and the closed-loop rules that start from ultimate gain — and converts reset and rate into the units on your faceplate.
It is a reference, not a recommendation. PIDSnap never originates a tuning value from a model, and it never asks you to find the ultimate gain on a running plant. Closed-loop Ziegler-Nichols and Tyreus-Luyben are here so you can see what those formulas produce. They are labelled unsafe for a reason: the experiment is a sustained oscillation.
The numbers assume a self-regulating first-order-plus-dead-time process and a bump you have not shown us. If the valve is sticking, the process is integrating, or the response is not first-order, the arithmetic is still tidy and still wrong.
Which method to pick
Lambda/IMC is the one worth using as a starting point on an operating plant. You choose how fast the loop should be — lambda — rather than pushing it as fast as the formula will go. A conservative default is the larger of the process time constant and three dead times.
Cohen-Coon and open-loop Ziegler-Nichols use the same three FOPDT numbers and aim closer to the stability boundary. They are useful for comparison. They are a poor default to type into a live controller.
Closed-loop Ziegler-Nichols and Tyreus-Luyben need the ultimate gain and period. Those two numbers are found by raising the gain until the loop cycles. Do not run that experiment on a production unit. If you already have Ku and Pu from a textbook or a simulator, the calculator will apply the correlations faithfully and tell you not to go looking for them on the plant.
What the units dropdowns are for
Kc is dimensionless controller gain in the ideal form. Ti and Td come out in the reset and rate units you pick, using the same conversion engine as the rest of the site. A minutes-per-repeat controller and a seconds-per-repeat controller are a factor of sixty apart. Getting that wrong is the same class of error as the reset-units converter exists to catch.
If your controller uses proportional band rather than gain, convert Kc with the band converter before you type it. If it solves the series or parallel equation, convert the three numbers for form.
What PIDSnap does instead
PIDSnap works from a photograph of your trend and faceplate. It does not fit an FOPDT model, does not invent a Kc, and does not tell you to cycle the loop. If the trend shows stiction, backlash or a saturated output, it says so rather than handing you a constant.
Questions that come up
Is this what PIDSnap recommends I type into the controller?
No. These are the classical correlations, applied as published, so you can see what a named method produces. PIDSnap's own guidance is computed from the values you confirmed off your faceplate and the response you photographed — never from a formula.
Why is closed-loop Ziegler-Nichols even on the page?
Because people search for it, and because hiding the numbers does not stop someone running the experiment. Showing the correlation next to a plain warning is more honest than omitting the method and leaving them on a site that will not warn them.
My process does not look first-order. What then?
Do not use the open-loop methods. Extra lags, an integrating character, or a strong nonlinearity mean the FOPDT numbers are a guess. Start with the structure of the loop, not its constants.
Related
Tuning guides
Free tools
These tools assume a self-regulating loop and a bump you have not shown us. Photograph the trend and the faceplate if you want a check against the loop in front of you, not a formula.
Last reviewed 2026-09-10.