PID Tuning: The Ziegler Nichols Method Explained

In this short tutorial I will take you through the two Ziegler-Nichols tuning methods. This will let you tune the derivative, proportional and integral gains on your P, PI or PID controller. Thankfully these methods are extremely simple to understand and implement.

[erTurbo]

Correct PID controller formula: Kp(1 + 1/(Ti*s) + Td*s)
Ti and Td are time constants. K are gains. So therefore Ki gain is actually Kp/Ti and Kd gain is actually Kp*Td.

[kestans]

for anyone else read old comments. it is first method always, you dont want actual equipment to be oscillating and go damage. In HMI say it is pump running at 20% and PLC pid controls analog output. In HMI force analog output to be in Manual mode, so it pid no longer control output, than change eg. from 20% to 25% and see historian for flow feedback data at exact time of setpoint been changed. Most cases dont use PID, use PI control but instead D term apply software first order lag filter for the feedback signal. increasing Kp makes faster response but increase overshoot, Increasing Ki reduces error and damps Kp oscillation but too much will come back to as if it was only Kp control but even with increased overshoot and increased oscillations. If output out of bounds of physical equipment capability do not integrate output(integral antiwindup), when output is within the range continue integrating.

[laughy]

Cant believe it took me so long to find this video. Thanks so much for this!

[muhammadqaisarali]

One thing please.
The second method is used once the first is failed, and it fails whenever the plant response doesn't follow that S shape.(S shape will be there whenever there are no complex poles pair or zero at origion).
In that case 2nd method will be used..

[Bangalibabu]

When to use first method and also when to use second one?
I mean how to differentiate that in the given question we are required to apply first method or second one?

[TrickyNekro]

The second method should be easy to implement in an auto-tuning PID, apart from the 0.6 factor the others are easily divisible. Very nice and helpful video indeed! Thank you!

[PafLeRenard]

Thank you for this quick presentation. There all we need.
I think there is an error for the Kp value of the PID in the second method: this should be Kp = 0.6*K_crit and not Kp = 0.6*P_crit to be coherent with what is above (and with physical units)

[mohammedtaha]

For the second method it should be 0.6*Kcrit and not 0.6*Pcrit in your table.

[alexjc]

very clear, easier to understand compare to professor's one hour lecture. lol

[mortenlund]

Thank you so much. What a nice and cool explanation which is understandable and straight to the point.

[tebogolegoabe]

Thank you so much that was very useful!!!

[demoinja]

i think in the first method that your table is wrong
well not completely wrong but the values ain't for Ki and Kd
those r for Ti and Td
and to calculate Ki and Kd
u should use the formulas
Ki = Kp/Ti
Kd = Kp*Td

[EdgarPoe_Raven]

Thank you very much for the upload and recording!

[champcv]

@The Complete Guide to Everything
. I have an estimator chart from my tutor which states for Method 2, PID, Kp is 0.6KCrit, not 0.6Pcrit as per video. Is this correct?

[nagashreeneha]

How did you write that table? For P, Pi, PID?

[ryanho]

What was the process behind filling in that table?

[fethilakhdari]

I would like to suggest to define your response curve horizontal axis and vertical axis. The presentation will gain immensely in clarity.
Is the horizontal axis a representation of time and the vertical axis that of PV? or what is it?

[MORAL]

how can I apply a step function to a ball, on an inclined axis, to use the ziegler-Nichols method?

[kenmagalnik]

In the first method, does the TC not depend on the size of the step taken? Should the size of the step be included in the calculations somewhere?

[alsahhim]

At 1:00 he says to apply a unit step, will a value of 1 suffice?