Practical Implementation Issues with a PID Controller 

These videos are so high quality. I really like the extended, classroom-lecture-style videos. There are a lot of short 10 minute lectures on RU-vid, and many of them are quite good but nothing beats in-depth lectures when you really want to learn a topic. Thanks!

Hey Christopher, I'm a control engineer and I must say it was an amazing lecture, thanks!

Отличный урок, спасибо!

Brilliant lecture.

I don't know how to thank you. You are one of the best teacher.

AE511: Great lecture Chris! I liked the story about the Aerosonde.

Hey professor. i couldn't thank more for your videos. Paying in millions at university but no one is there to teach like you and that too at free of cost. Once again thank you so much.LOVE FROM PAKISTAN

Great work. I truly enjoyed this video. Thank you.

Thank you Christopher it is necessary lecture

Please upload "Introduction to PID controller " video sir

Thanks for uploading these videos, it's amazing!!

As always, great video! I'm happy to be a patron. I thought I would comment that there is another reason to evaluate the integral in PID in the order Ki then the sum (discrete) / integral (continuous): (error -> Ki -> 1/s -> output). When changing the Ki coefficient, if you have Ki term preceding the sum, then any change to the Ki term will change the rate that the total integral term "changes". There will be no step discontinuities. If the sum precedes the Ki term, a change in the Ki term will result in a step change in the total integral term. This is almost never desired. When would you change the Ki term on a "hot" system? 1) When tuning on a test stand and 2) gain scheduling 3) when you switch your control law "off" temporarily. In addition, the Aerosonde example is a great example of how the whole problem could have been avoided by using this scheme. When switching from PID control to manual control, the Ki term could have been set to 0. If the Ki term precedes the integral sum, then the integral sum wouldn't build up and it's super easy to code this up (this does not eliminate the reason to limit the integral sum). Of course, this avoids the issue that a live control loop should never be left executing when not "hooked up" to anything. Leaving he forward path "broken" is the ultimate in non-linear control! :). Many thanks to Chris for making a video with such meaty topics!

hi Christopher ... Started following your lectures on control .... They are amazing and simple to understand

This is amazing, one of the best channels on RU-vid

I can't describe how great this video is ... thank you sir for Sharing your knowledge with us i am very glad

Stuck with the controller of the Atomic Force Microscopy device and this video is much better to understand the PI controller. I hope to solve the problem soon. Thanks, Professor for such amazing videos on control theory along with Simulink

[AE 511] This lecture helped me understand the integrator part of the controller much better. The accumulation of error or wind up is much more clear.

Great video sir

thanks for a great video sir...in India we use word "GURU" for great teacher....and you are guru!! thanx a lot guruji

Once more, another great vedio. Thanks a lot.

This is amazing

Thanks!

great lecture.....Thanks....keep doing......

AE511: This lecture going insane rn! 😎😎😎

Hi Chris, I loved the Aerosonde story and the way you tied it to the material. I've been a controls engineer for a while and I love the way you present the subject. I like how you highlight the physical meaning of theory-based insights, like the delay and disproportionate noise amplification inherent in the derivative or the subtlety the integral no longer commutes after saturation. Something I'd like to see is your take on how to go about discretizing a continuous-time design for the purpose of then implementing it on digital hardware. Considerations on forward/backward Euler vs Tustin vs others, etc. Keep it up!

Today is my exam on control theory paper thank youuu ❤

Great video

Your way of teaching is fantastic and very step-by-step. Let me request you to make a video on "Quadrotor control". PID is a linear controller but Quadrotor is a non-linear system. We know that linearization works in a limited domain. But if we control it via non-linear technique, its very good. Can you upload a video in which you 1st do modeling of a Quadrotor and then control that with a non-linear technique in MATLAB?

Hi Chris, thank you so much for these videos. Not sure if you know about Signal Viewer and how easy it is to use that (for viewing multiple signals together, separate, or whatever) instead of the scope blocks. All you'd need is just an output port for each signal you're trying to add to the Viewer. I discovered about this a few months back and it has made my life so much easier.

Hi Chris, thank you so much for these videos. Would you please share the matlab/simulink code for this video? I tried to make matlab/simulink code based on this video and previous, but designing the guidance module is not presented here.

when comparing two plots, u can put the arrows to same scope for ease of visualization, but ig u have figured it out by now, considering this video is 2 years old. Thank you for making such a crystal clear video

AE511: It seems like the integrator is required to guarantee no steady state error but also has the most tendency to make the controller behave adversely.

Hi Christopher, Thanks for the video, I really enjoyed it. I want to know why we don't consider a lead-lag compensator instead of a PID to avoid noise amplification problem for derivative and integral windup problem for the integrator. I mean does PID have a spectacular benefit over lead-lag compensator that makes it the main option?

AE511: Is there an analog to a pre filter and damping ratio? It seems like decreasing the coefficients you're effectively increasing the damping ratio of the system.

AE511: This helps clear up some of the adverse behavior I was seeing in my use of the integrator in the previous homework. Is the decision of which nonlinear integrator fix (solutions 2-4) to use largely made via trial-and-error?

Do you remember what signal you were using for the error term in the example you gave of a limited integral state vs limited output? When I set up a similar system and used a square wave as the input signal, I got the same output when limiting the integral block (what you called limiting the state of the integral) as when I put a saturation block on the output of the integrator. I think I'd understand the concept better if I knew what the input signal was, as I presume this is only an issue for certain types of functions/signals.

Sir, do you have any idea about the implementation of Nonlinear controllers with analog circuits? As for PID, we can construct analog PID controllers by resistors, capacitors, and opamp, because they involve simple calculus, but nonlinear controllers involve very difficult and complex mathematics, so how to implement these controllers in real-world systems?

AE 511: Looking at the list of anti-windup solutions, I had made the assumption that solution 2 of disabling the integrator would be the easiest approach. I don't understand how having the "if statement" on the integrator would make the system non-linear... How would the system become non-linear?

Sir pls about mathematical representation of rotary inverted pendulum pls pls

I can't find the introduction to PID video anywhere on your channel. Is it still available?

Would there be any risk to applying prefilters to only the I and D terms, but leaving the P term unfiltered?

You know you are speaking fast when you almost loose your breath while speaking

Does anyone know of a pid controller that will vary the voltage output rather than just cycle on and off. Thanks

Integrator does not have any accumulation until t-w, in option 3.

how can get your full lecture

In case it is helpful, here are all my Control Theory videos in a single playlist ru-vid.com/group/PLxdnSsBqCrrF9KOQRB9ByfB0EUMwnLO9o. You can support this channel via Patreon at www.patreon.com/christopherwlum. Please let me know what you think in the comments. Thanks for watching!

Good stuff sir, but pardon me. Audio is hard to follow, it sounds like your yelling even thou I have my sound at minimum. Maybe you have too much of gain or normalization. I guess you have some reason for over processing this audio. Thank you for great videos.

The problem I have with these videos is that too much time is wasted drawing or wiring the equations. They should be done ahead of time. The information density is low. Most 'noise" is not really noise. It is due to low resolution feedback. The feedback signals are "quantized" by the resolution of the feed back. A fast sample time helps but feedback resolution is more important. I didn't like the statement at the end about poles and zeros don't apply to a modified PID. The derivative low pass filter adds poles and the integrator pole is valid as long as the controller doesn't go into saturation. If the system does go into saturation then the control is basically open loop anyway.

I don't know how to thank you. You are one of the best teacher.