I use PID controllers in a different context: game development. You'll often have a variable you need to drive to a value, but need to do it in a kind of smooth or organic way. Instead of writing a new way to do it every single time I need to do this, I use a PID controller. This is especially useful when other things are trying to effect this value as well, as the PID controller can react and push harder toward the set point and generally won't overshoot the set point. It is of course also very useful in physics simulations. I've made a hovercar just by making 4 "thrusters" on the bottom of the car, using the distance to the ground at that location as my setpoint and adding forces at those locations. If I were to try to program something like that, first I think it would take a while, but second I think I would end up with something a lot like a PID controller. They're just useful all over the place in gamedev. I also like to think in simplified terms, usually I'm not doing equations and figuring things out exactly in gamedev, it's more of a "put values in until it looks right" kind of task. So imagine a cube that can only move up or down and the PID controller is controlling an upward force against gravity and we want it to hover at a certain height. P is the power, it's responsible for the majority of upward force. D is the damper, it wants to stop all movement, it's responsible for killing oscillations. I is intelligent, it'll make the small adjustments that P and D can't do, namely without I the cube will never quite get to the desired height. This is definitely not engineering, so definitely not an engineer's way of thinking about it but I think it's very useful.
I think you missed the main point of the D term. If your oven has a lot of thermal mass, *both* the P and the I terms will suffer from overshoot. The D term is very specifically the anti-overshoot term; because it's the only one "smart" enough to see "the oven is below the target temperature, but nevertheless I'm going to vote *against* putting in more effort because it is trending up so fast that I can see it might overshoot soon". On the flip side though, it's refreshing to see a video that doesn't claim/imply that PID controllers are perfect in all scenarios; thank you for mentioning bang-bang controllers.
TheHue's SciTech this is what I was thinking too. The D term would take the current position and its derivative and use that to calculate an approximation for a future position. If it's getting close to the set point but is still pushing hard towards that set point, the D term puts on the brakes to help prevent overshoot. Then in the next time step it calculates the approximation again and adjusts accordingly.
So refreshing to see something about PID controllers on TH-cam that isn't just tuning by guessing random numbers until the steady state looks OK and never checking the step response which always ends up massively overdamped (like orders of magnitude slower than it could have been).
Great video, and contrary to other comments; keep the math or even have more of it. But with some people talking about including electronics, maybe you could have used a small heater element and an Arduino to demonstrate real-time tuning? Or even auto-tuning like many 3d printers use for their heaters. In the end a strong foundation in the theory followed by a good practical example of the implimentation is a great way to drive things home.
Can you post links in YT comments? This might be a good simulation. robotic-controls.com/static/inverted-pendulum/ If the link does come through google robotic-controls inverted-pendulum. There is a link in the article for the simulator.
This was so cool and so enlightening. I'm just a hobbyist, but I love diving into the work of degreed professionals and seeing what kinds of things they learned in school, and these EEVacademy courses are like a treasure trove of direct info as well as info to look up later.
Very well done. Not an easy topic to explain but hugely useful and widely utilized. Few robotics or flight controllers out there not using PID control.
Excellent! Excellent! Excellent! I learned basic control theory in college over 40 years ago. It was hard to understand just the theory. I've worked with PID controllers ranging from mechanical analog to fully digital controls for the past few decades. The best way to understand tuning is to do it hands on. It gives you a feel for the dynamics of the process and the amount of hysteresis in the loop. I can't wait to see the inverted pendulum project. This is incredible content. Well done!
Nice PID video. I thought I already knew everything about PID but you taught me several things I hadn't realized or come across. It was worth the 27 minutes!
Years ago, I developed several user-content 'toys' in Linden Labs' SecondLife that use PID controllers to simulate 3-D tracking behaviors and in one case, a very realistic tire-swing within the environment. It was fun to see this video and revisit my own experience in developing and tuning a PID controller in that context.
I hesitated to click on this video because I usually watch Dave's EEVBlog videos to slowly fall asleep, but now that I have found some time to watch it out of my usual schedule I have to say I really liked it for the educational aspect. Please make more videos like this one! Thanks! Oh and nice touch with the picture at the end. That made me actually chuckle a little bit.
Hey! I'm working through my Electrical Engineering degree. I thought this was honestly pretty good. I appreciate the layman explanations of control theory. I am actually taking a controls class soon, so this makes me feel better about that class. There was a bit of rambling and bouncing around, but that's fine. I do that when I explain things to newer students, lol. I also appreciate the math. It's nice to see application, so to speak, rather than just concepts. Keep up the good work! I look forward to more videos on this topic!!!
Nice job, Controls was my elected concentration as an undergrad EE student until they introduced the State Space method at which time I changed to digital and computer design, a choice I regret. I am looking forward to this series, keep up the good work.
Awesome! I've taken a controls course before but I never learned the Ziegler-Nichols method. In fact, we never actually touched tuning other than using a given transfer function and finding the corresponding PID values by using a root locus. The major issue is that you have to get the appropriate model first then convert to a transfer function which is already a lot of math. Then you can finally get the transfer function to find the PID values which, in most cases only gets you in the ballpark of the system response you desire. Even then there is trial and error involved at the end to get it dialed in to that response.
Absolutely a nice video - it started a bit rough, but once David got into the flow, it became very clear and nice to follow. When explaining the I-part, you should perhaps also have shown the actual "temperature" or state of the oven instead of just the error. You could have combined the two in a single graph where you had shown a line for the setpoint and drawn the error between the setpoint and the actual state. YES, keep the math, it is necessary here!
Ever since I learned Control, I see it everywhere, we are all in harmony because of the feedback and control of our systems with Brain OS, mine currently have 1024 bit cpu, 4096 TeraByte of RAM, but need an upgrade to my soft drive
my vote: construct on a protoboard the simple RC feedback opamp, feed in a pulse to obtain a desired pulse integral with visible over/undershoot. Be literal with how you get Ku and do the 4 iterations to adjust "pole-zero". Great Start!
Good effort. You forgot the essential role of the delay in the control feedback loop, without this system delay there would be no need for PID controllers. I wish I could recall the website I found many years ago(~15) explaining PID control, it was very spot on, covered the math in a reasonable manor(especially the strengths of the three components at different levels of initial error) and it had some very nice example graphs.
Delay isn't the only reason for PID controllers, infact in many system models it can be ignored. Delay is very significant in systems involving heating and cooling and I suppose I did talk about that so maybe should have mentioned it.
Wow that is late. I had it in the 3rd and 4th semester. But yeah its a really interesting topic, especially with the more advanced controllers and fuzzy control. Its a good choice :)
Thanks and I hope the other parts are better. As a matter of fact you presented more information than the others who talked about PID. PID is vague subject presented by books and many instructors.
Great video. I hardly understood how these kind of controllers work, like ambient room temperature controllers or a cruise control system, and how they deal with overshoot. Until now.
Thanks for your efforts. It would be cool to see this PID info from the point of view of model quadcopters, as PIDs, and tuning PIDs are a huge thing with getting rc model quadcopters working properly.
Excellent video. I develop software for autonomous vehicles in the agricultural industry. Our controls engineers use simulink to develop their "models" and I develop the wrapper code as we call it to feed the model with data. Our controls engineers are always tuning the system. Our systems controls a steering valve for an auto guidance application. This video filled in some gaps for me. Now I know what my controls engineers are really doing lol...
Just the other day i said to myself that i have to look at theory behind PID again since i need to do some regulation and here u come with this video - just perfect :D.
I don't know how this was boring lol. I guess I just like learning, so I found it interesting. I love learning how math works in the 'real world' - I die trying to learn math purely for math's sake. So something like a PID controller is perfect for me - shows how you can use math in the real world to achieve things you want. I'm glad you included the animation at the end - it definitely helps to see the effects of variable changes in real time like that. I hope you guys do a tear-down or something of a typical PID controller during this series. I would love to learn how the circuits / components work together to make this math happen.
feedback: Screech your voice like Dave, it throws my thinking off. Seriously, Dave, that was an excellent vid on digital data logic sheets. I went to college for Avionics tech and that was above and beyond. That is the kind of content I really enjoy watching from you. Something that furthers my understanding of a topic I already love.
Hi. Perhaps the spring analogy for the P component was really evident for people who already knew the theory, but I didn't get it at all. There was no working example. The I component explanation was very clear, and although the D component explanation was a little rushed, I think I've got it as well.
I know this is an old video, but it would be really great to see the math fully worked out. I've been dorking around writing out diff eqs for the response got a basic heater, but the coefficients turn into a mess pretty quickly.
I would like to use PID on a fresh water pump for water pressure. 45psi would be the ideal pressure. Opening lines for showers, washer, etc should raise the frequency of the pump. Ideally utilizing a second pump when water demand is high. Doesn't this sound like a fun challenge?
That's an interesting idea, but I think this problem can be solved with a diaphragm regulator. The pump should already have a high pressure cutoff so it doesn't overheat or waste power. Combine that with a standard off the shelf pressure regulator and you should have your whole system, just set the desired pressure on the regulator and forget about PID man. Unless you're doing this purely for the challenge, in that case, have it lol.
Isn't an ideal spring more like an integral controller because the total force based on hookes law, F=kx, changes based on distance, and distance changes based on time. So when you let go of a compressed spring, it oscillates back and forth based on time, thus changing the effort, until is reaches equilibrium.
Some items I would like to see include the effects of signal delays, and an analysis of real world PID controllers. Temperature controllers like a Solo SL4848 and pretty much any other process controller, use a specific PID algorithm. However the manuals don't tell you what the algorithm does inside. Which means that when you are tuning these kinds of controllers in a system with a signal delay, it can be quite a bit of black magic which falls outside of the PID analysis as noted in this tutorial. For example, if you have a water pump with an electric heater, which is heating some device. Your temperature probe is in the device. However lets assume the controller makes hot water and it takes 20 seconds until that water has reached the device. This causes all sorts of problems. If you do not have time base control, you're going to be forced to jack up the the integral term, which is plagued with integral windup, slow reaction times, etc. However standard process controllers allow you to set the time base. Which if set properly, will reduce your integral terms effects greatly. Think of it like Z domain vs S domain. The PID you talk about is more like S domain, but the real world controllers and real world systems are effected by delays which makes them more like a Z domain device. It would be great if you made a video about the effects of the time base in an PID algorithm.
Really good video! Interested in the inverted pendulum robot design, component selection, and build. Also excited for tuning the little guy. I've watched several people in my university try to build these for fun or for a class, but they never preformed very well.
The missile knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation. The guidance subsystem uses deviations to generate corrective commands to drive the missile from a position where it is to a position where it isn't, and arriving at a position where it wasn't, it now is. Consequently, the position where it is, is now the position that it wasn't, and it follows that the position that it was, is now the position that it isn't. In the event that the position that it is in is not the position that it wasn't, the system has acquired a variation, the variation being the difference between where the missile is, and where it wasn't. If variation is considered to be a significant factor, it too may be corrected by the GEA. However, the missile must also know where it was. The missile guidance computer scenario works as follows. Because a variation has modified some of the information the missile has obtained, it is not sure just where it is. However, it is sure where it isn't, within reason, and it knows where it was. It now subtracts where it should be from where it wasn't, or vice-versa, and by differentiating this from the algebraic sum of where it shouldn't be, and where it was, it is able to obtain the deviation and its variation, which is called error.
I need to do videos on neural networks, kalman filters, adaptive controllers, lidars, SLAM, and a few other concepts first though, maybe there is a way to simplify it but I haven't figured it out yet :(.
Thanks.one of the best explanation .I like the way you combine mathematics together with the actual thing. It really make sense to me, better understanding since I am only good at math but not so to relate it with the actual thing. Once again thank you so much for your fantastic effort to teach us. God bless you Sir.
WHERE is the NEXT VIDEO. I LOVE your method of teaching. pls do more. i have done courses in control but i feel enlighted each time i watch great videos as this really show how simple PID control theory is. i still am not very confided on the propotional part but i have a good idea as its just an offset constant but still i would like to improve my understanding of it furthur and thus seeing ure example with an actual system will be veyr useful. i am very greatful for this. MANY THANKS
it's just like pumping your tires, at first it checks the pressure, then it increases or decreases the pressure and checks again, repeat until set pressure.
this is the first pid controller video i have watched that has mentioned filtering. my pid controller has a filter on it but i have it turned off and have been fine tuning the i and d and can't get it quite perfect, at least for my autistic brain. i have not tried the filtering because it's labelled as digital filtering, but I'm going to give it a whirl
What's next a discussion of root locus? 😏 Great video, my EE specialty was control theory out of a very very plain navy book with no graphics, circa 2005. Used PID controllers for temperature control of large incinerators, >100 MBTU thermal oxidizers, down to small ovens for refractory tile, only 4 pallets or around 2000lb to dry them out before installation, they still had to be cured in the furnace/thermal oxidizers/flare.
This is an excellent starting video... I think you need to show the math a long side an application which I think will home in the concepts better... I like the way you presented this please do more on control theory...
You're a fantastic presenter! So detailed and engaging!
3 ปีที่แล้ว
Coefficient calculation: Can we just adjust the coefficient separately? First, we run the system only with P-controller, then only with I-controller, then only with D-controller. In the end, we collect them and put them together in a control algorithm. Can it work like this?
That was much better paced than your first videos. The next videos with real examples should be fun. I remember you talking to Dave about the balancing robots on The Amphour.
Easier voice to follow than Dave (ok I admit I'm used to Dave now). Needs a little polishing on the presentation but I think its a good effort. Not too much maths or too complex. As you continue the series consider doing something in mathlab or similar software to show the effects of tuning. The last graph was small and unclear.
It boggles the mind that a minor difference in pitch and intonation provokes such a reaction. Never go to Italy, by the way, you probably would not survive it. ;)
If you had a differential equation for the issuance of an impact recall, how to compute the PID parameters mathematically? It is interesting to see such readings for example for an oven.
To people who thinks there is too much math involved, you havent seen the tip of it, there isn't almost any math involved here. He just explained how it works, that's it. No s-domain representation, no formal PID design techniques, no bode plots for frequency domain analysis, not much talk about stability, and a risky method for designing a PID controller. This video isn't math intensive, it is just a tutorial. Btw, you do not use ziegler nichols technique for linear systems, there are indeed better and easier ways of designing PID controllers for linear systems, you use ziegler-nichols if the system is somewhat nonlinear (meaning that the system can not be linearized well enough), and you need a starting point for a PID design, that is it. Don't use ziegler nichols for linear systems. It is most likely that you will end up with much better PID controllers if you just use any of the PID design techniques for linear systems.
17:50 that graph is the step response of the system, you put error on your axis I think that is a mistake, it's also not the effort, it would be the feedback converted to the setpoint units, setpoint is a step like having the heater unplugged from AC and putting the dial to 10, then connecting it to AC, that would be the response to a 10 unit step
Great video, thanks! I first saw them in flight controllers for racing drones. There you have to tune your PID's and see the results immediately! Now I really understand how and why. With the drones the P is the responsiveness of the sticks in relation to the drones movement. You want it as high as possible, the I is the drones ability to hold the angle you set it at while flying, and the D is used to dampen oscillations from too high P, however it does make the drones more sluggish. As you said, D resists change, makes sense. Tuning those quadracopters is an art, a blend of snappy controls and smooth video. I really enjoyed this!
![[LIVE] : ONE ลุมพินี 88 | คู่เอก "ป้อมเพชร vs อัสลามจอน"](http://i.ytimg.com/vi/1ZqTVVbMgbo/mqdefault.jpg)
"control theory, feedback appreciated" 😊😊😂
Negative feedback is good 😆
I use PID controllers in a different context: game development. You'll often have a variable you need to drive to a value, but need to do it in a kind of smooth or organic way. Instead of writing a new way to do it every single time I need to do this, I use a PID controller. This is especially useful when other things are trying to effect this value as well, as the PID controller can react and push harder toward the set point and generally won't overshoot the set point.
It is of course also very useful in physics simulations. I've made a hovercar just by making 4 "thrusters" on the bottom of the car, using the distance to the ground at that location as my setpoint and adding forces at those locations. If I were to try to program something like that, first I think it would take a while, but second I think I would end up with something a lot like a PID controller. They're just useful all over the place in gamedev.
I also like to think in simplified terms, usually I'm not doing equations and figuring things out exactly in gamedev, it's more of a "put values in until it looks right" kind of task. So imagine a cube that can only move up or down and the PID controller is controlling an upward force against gravity and we want it to hover at a certain height. P is the power, it's responsible for the majority of upward force. D is the damper, it wants to stop all movement, it's responsible for killing oscillations. I is intelligent, it'll make the small adjustments that P and D can't do, namely without I the cube will never quite get to the desired height. This is definitely not engineering, so definitely not an engineer's way of thinking about it but I think it's very useful.
Exactly what I saw in watching this video, cheers.
Yeah I used it in software a few times as well.
uzimonkey I use this for AI controllers. :)
i love principles that can be applied to mass amounts of different applications
absolutely fabulous
I think you missed the main point of the D term. If your oven has a lot of thermal mass, *both* the P and the I terms will suffer from overshoot. The D term is very specifically the anti-overshoot term; because it's the only one "smart" enough to see "the oven is below the target temperature, but nevertheless I'm going to vote *against* putting in more effort because it is trending up so fast that I can see it might overshoot soon". On the flip side though, it's refreshing to see a video that doesn't claim/imply that PID controllers are perfect in all scenarios; thank you for mentioning bang-bang controllers.
That's a good point. To go futher PID terms could be considered the present, past and future terms, respectively.
ramueller: Thanks, that is a great way to look at it
TheHue's SciTech this is what I was thinking too. The D term would take the current position and its derivative and use that to calculate an approximation for a future position. If it's getting close to the set point but is still pushing hard towards that set point, the D term puts on the brakes to help prevent overshoot. Then in the next time step it calculates the approximation again and adjusts accordingly.
00:17 "any feedback would be appreciated". Aha feedback: control theory. I see what you're doing there ;-)
So refreshing to see something about PID controllers on TH-cam that isn't just tuning by guessing random numbers until the steady state looks OK and never checking the step response which always ends up massively overdamped (like orders of magnitude slower than it could have been).
Great video, and contrary to other comments; keep the math or even have more of it. But with some people talking about including electronics, maybe you could have used a small heater element and an Arduino to demonstrate real-time tuning? Or even auto-tuning like many 3d printers use for their heaters.
In the end a strong foundation in the theory followed by a good practical example of the implimentation is a great way to drive things home.
maybe steal a vid of an inverted double pendulum
I second that, esp using an Arduino. Great vid, many thanks
Does anyone know of a good simulator of an over or inverted pendulum in which you could try applying a PID controller to?
Can you post links in YT comments? This might be a good simulation. robotic-controls.com/static/inverted-pendulum/
If the link does come through google robotic-controls inverted-pendulum. There is a link in the article for the simulator.
why not?YT=Google and links are not copyright protected,soooo....
This was so cool and so enlightening. I'm just a hobbyist, but I love diving into the work of degreed professionals and seeing what kinds of things they learned in school, and these EEVacademy courses are like a treasure trove of direct info as well as info to look up later.
Very easy to follow, clear and fun explanation of what PID Theory is. Thank you David. Looking forward for the continuation of this topic. Good job!
Truly brilliant presentation. Relating the equations/terms to the graphs made it understandable to people like me.
I look forward to more control theory videos.
What a nice way to explain this, i wish my teachers at college taught me like this, keep up the good work!
Holy crap this is the boring part of the series?!? I can't wait for the rest this was incredibly interesting
Maniacal Cactus my thoughts exactly
Have a look at "Teaching old motors new tricks" for some real juicy control system nerdiness
I read this topic before, but this video makes it much easier.
Very well done. Not an easy topic to explain but hugely useful and widely utilized. Few robotics or flight controllers out there not using PID control.
Thanks, it is a very fresh way to explain the concepts. I am less lost with this explanation I think I am going though a good path
Excellent! Excellent! Excellent! I learned basic control theory in college over 40 years ago. It was hard to understand just the theory. I've worked with PID controllers ranging from mechanical analog to fully digital controls for the past few decades. The best way to understand tuning is to do it hands on. It gives you a feel for the dynamics of the process and the amount of hysteresis in the loop. I can't wait to see the inverted pendulum project.
This is incredible content. Well done!
Nice PID video. I thought I already knew everything about PID but you taught me several things I hadn't realized or come across. It was worth the 27 minutes!
Years ago, I developed several user-content 'toys' in Linden Labs' SecondLife that use PID controllers to simulate 3-D tracking behaviors and in one case, a very realistic tire-swing within the environment. It was fun to see this video and revisit my own experience in developing and tuning a PID controller in that context.
I hesitated to click on this video because I usually watch Dave's EEVBlog videos to slowly fall asleep, but now that I have found some time to watch it out of my usual schedule I have to say I really liked it for the educational aspect. Please make more videos like this one! Thanks!
Oh and nice touch with the picture at the end. That made me actually chuckle a little bit.
Hey! I'm working through my Electrical Engineering degree. I thought this was honestly pretty good. I appreciate the layman explanations of control theory. I am actually taking a controls class soon, so this makes me feel better about that class. There was a bit of rambling and bouncing around, but that's fine. I do that when I explain things to newer students, lol. I also appreciate the math. It's nice to see application, so to speak, rather than just concepts. Keep up the good work! I look forward to more videos on this topic!!!
Nice job, Controls was my elected concentration as an undergrad EE student until they introduced the State Space method at which time I changed to digital and computer design, a choice I regret. I am looking forward to this series, keep up the good work.
I was like in a total dark room, and now there is a ray of light. Thanks a lot Bro.
Awesome! I've taken a controls course before but I never learned the Ziegler-Nichols method. In fact, we never actually touched tuning other than using a given transfer function and finding the corresponding PID values by using a root locus. The major issue is that you have to get the appropriate model first then convert to a transfer function which is already a lot of math. Then you can finally get the transfer function to find the PID values which, in most cases only gets you in the ballpark of the system response you desire. Even then there is trial and error involved at the end to get it dialed in to that response.
Absolutely a nice video - it started a bit rough, but once David got into the flow, it became very clear and nice to follow. When explaining the I-part, you should perhaps also have shown the actual "temperature" or state of the oven instead of just the error. You could have combined the two in a single graph where you had shown a line for the setpoint and drawn the error between the setpoint and the actual state.
YES, keep the math, it is necessary here!
Thank you! I took control theory in school but this really helped to understand the PID controller much better.
Ever since I learned Control, I see it everywhere, we are all in harmony because of the feedback and control of our systems with Brain OS, mine currently have 1024 bit cpu, 4096 TeraByte of RAM, but need an upgrade to my soft drive
I spent alot of my career coding PID loops in PLCs.
Good video David :)
my vote: construct on a protoboard the simple RC feedback opamp, feed in a pulse to obtain a desired pulse integral with visible over/undershoot. Be literal with how you get Ku and do the 4 iterations to adjust "pole-zero". Great Start!
Great video and one of most clear explanation of PID control i have seen.
Good effort.
You forgot the essential role of the delay in the control feedback loop, without this system delay there would be no need for PID controllers.
I wish I could recall the website I found many years ago(~15) explaining PID control, it was very spot on, covered the math in a reasonable manor(especially the strengths of the three components at different levels of initial error) and it had some very nice example graphs.
Delay isn't the only reason for PID controllers, infact in many system models it can be ignored. Delay is very significant in systems involving heating and cooling and I suppose I did talk about that so maybe should have mentioned it.
Loved it, would like to see more of the math, and maybe a demostration implementing the control theory.
3rd year EE
woot and then you havent had a lecture about that? After 3 years?
nukularpictures taking it next semester actually.
Wow that is late. I had it in the 3rd and 4th semester. But yeah its a really interesting topic, especially with the more advanced controllers and fuzzy control.
Its a good choice :)
Thanks and I hope the other parts are better. As a matter of fact you presented more information than the others who talked about PID. PID is vague subject presented by books and many instructors.
Great video. I hardly understood how these kind of controllers work, like ambient room temperature controllers or a cruise control system, and how they deal with overshoot. Until now.
Thanks for your efforts. It would be cool to see this PID info from the point of view of model quadcopters, as PIDs, and tuning PIDs are a huge thing with getting rc model quadcopters working properly.
Excellent video. I develop software for autonomous vehicles in the agricultural industry. Our controls engineers use simulink to develop their "models" and I develop the wrapper code as we call it to feed the model with data. Our controls engineers are always tuning the system. Our systems controls a steering valve for an auto guidance application. This video filled in some gaps for me. Now I know what my controls engineers are really doing lol...
Glad to help :)
Excellent presentation that gives you a good "feel" without maths. Look forward to next viseo
Theory was great to follow alongside, looking forward to the practical.
Just the other day i said to myself that i have to look at theory behind PID again since i need to do some regulation and here u come with this video - just perfect :D.
The best video of the series so far, good job David!
Thankyou!
I don't know how this was boring lol. I guess I just like learning, so I found it interesting. I love learning how math works in the 'real world' - I die trying to learn math purely for math's sake. So something like a PID controller is perfect for me - shows how you can use math in the real world to achieve things you want. I'm glad you included the animation at the end - it definitely helps to see the effects of variable changes in real time like that.
I hope you guys do a tear-down or something of a typical PID controller during this series. I would love to learn how the circuits / components work together to make this math happen.
feedback: Screech your voice like Dave, it throws my thinking off. Seriously, Dave, that was an excellent vid on digital data logic sheets. I went to college for Avionics tech and that was above and beyond. That is the kind of content I really enjoy watching from you. Something that furthers my understanding of a topic I already love.
An actual good video on PID controllers, FINALLY! I have been trying to learn these for years!
Hi. Perhaps the spring analogy for the P component was really evident for people who already knew the theory, but I didn't get it at all. There was no working example. The I component explanation was very clear, and although the D component explanation was a little rushed, I think I've got it as well.
If I can't get this explanation I can't get any other. Thank you very much!
I know this is an old video, but it would be really great to see the math fully worked out. I've been dorking around writing out diff eqs for the response got a basic heater, but the coefficients turn into a mess pretty quickly.
DR RORPOPOR HERBAL on TH-cam changed my entire life with his herbal medicine. I appreciate you sir, for taken away my PID 🌿🌿🌿🌿
I would like to use PID on a fresh water pump for water pressure. 45psi would be the ideal pressure. Opening lines for showers, washer, etc should raise the frequency of the pump. Ideally utilizing a second pump when water demand is high.
Doesn't this sound like a fun challenge?
Sounds an appropriate use case :)
Skyler Peters this is something that I could use for future projects. I'd love to learn how to do it!
That's an interesting idea, but I think this problem can be solved with a diaphragm regulator. The pump should already have a high pressure cutoff so it doesn't overheat or waste power. Combine that with a standard off the shelf pressure regulator and you should have your whole system, just set the desired pressure on the regulator and forget about PID man. Unless you're doing this purely for the challenge, in that case, have it lol.
Jesse W, It would not work. Pump would be constantly turning off and on at a pretty high rate.
David Ledger 4:24 I think arrows belong only to the positive half of the axis, not the negative
More like this please, perhaps adding an actual device/program to demonstrate the concepts in context. Very good, clear explanations. Thanks!
Isn't an ideal spring more like an integral controller because the total force based on hookes law, F=kx, changes based on distance, and distance changes based on time. So when you let go of a compressed spring, it oscillates back and forth based on time, thus changing the effort, until is reaches equilibrium.
No, integrated position with respect to time would have a time component.
David, nice job balancing examples and theory.
loved it, simple, clear, and to the point
Some items I would like to see include the effects of signal delays, and an analysis of real world PID controllers. Temperature controllers like a Solo SL4848 and pretty much any other process controller, use a specific PID algorithm. However the manuals don't tell you what the algorithm does inside. Which means that when you are tuning these kinds of controllers in a system with a signal delay, it can be quite a bit of black magic which falls outside of the PID analysis as noted in this tutorial. For example, if you have a water pump with an electric heater, which is heating some device. Your temperature probe is in the device. However lets assume the controller makes hot water and it takes 20 seconds until that water has reached the device. This causes all sorts of problems.
If you do not have time base control, you're going to be forced to jack up the the integral term, which is plagued with integral windup, slow reaction times, etc. However standard process controllers allow you to set the time base. Which if set properly, will reduce your integral terms effects greatly. Think of it like Z domain vs S domain. The PID you talk about is more like S domain, but the real world controllers and real world systems are effected by delays which makes them more like a Z domain device.
It would be great if you made a video about the effects of the time base in an PID algorithm.
Really good video! Interested in the inverted pendulum robot design, component selection, and build. Also excited for tuning the little guy. I've watched several people in my university try to build these for fun or for a class, but they never preformed very well.
The missile knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation. The guidance subsystem uses deviations to generate corrective commands to drive the missile from a position where it is to a position where it isn't, and arriving at a position where it wasn't, it now is. Consequently, the position where it is, is now the position that it wasn't, and it follows that the position that it was, is now the position that it isn't.
In the event that the position that it is in is not the position that it wasn't, the system has acquired a variation, the variation being the difference between where the missile is, and where it wasn't. If variation is considered to be a significant factor, it too may be corrected by the GEA. However, the missile must also know where it was.
The missile guidance computer scenario works as follows. Because a variation has modified some of the information the missile has obtained, it is not sure just where it is. However, it is sure where it isn't, within reason, and it knows where it was. It now subtracts where it should be from where it wasn't, or vice-versa, and by differentiating this from the algebraic sum of where it shouldn't be, and where it was, it is able to obtain the deviation and its variation, which is called error.
Damn. You beat me to posting this. Here's the video:
th-cam.com/video/bZe5J8SVCYQ/w-d-xo.html
I like it. Would love to see you dig into how the self driving cars stay in one lane without swerving and navigate safely.
I need to do videos on neural networks, kalman filters, adaptive controllers, lidars, SLAM, and a few other concepts first though, maybe there is a way to simplify it but I haven't figured it out yet :(.
Thanks.one of the best explanation .I like the way you combine mathematics together with the actual thing. It really make sense to me, better understanding since I am only good at math but not so to relate it with the actual thing. Once again thank you so much for your fantastic effort to teach us. God bless you Sir.
Excellent explanation, much better than Dave, who can make the simplest concept incomprehensible by over-explaining it to death.
Excellent explanation of PID control. Thanks.
Thankyou!
Great video, very informative and well done 🤗 looking forward for the rest of the series!
David2 brilliant video, just the right balance of explanation in each area for me :)
Looking forward to the physical example.
More videos like this please !
WHERE is the NEXT VIDEO. I LOVE your method of teaching. pls do more. i have done courses in control but i feel enlighted each time i watch great videos as this really show how simple PID control theory is.
i still am not very confided on the propotional part but i have a good idea as its just an offset constant but still i would like to improve my understanding of it furthur and thus seeing ure example with an actual system will be veyr useful.
i am very greatful for this.
MANY THANKS
it's just like pumping your tires, at first it checks the pressure, then it increases or decreases the pressure and checks again, repeat until set pressure.
Yes, although that is a bang bang controller :)
I liked this explanation. Good job David.
Nice one David - clear and concise..
I wish I had resources like this when I was a student - maybe I would have learned more and played less poker :/
Absolutely great presentation and gace me a further understanding
Typo at 24:17. KsubD should equal TsubD times KsubP, not what is written: KsubD = TsubD X KsubD.
love the video dave2 as i am studting control engiuneering in uni and you also covered the topic and no mistakes to spot
This is my line of work :) Measuring instruments, controllers and process valves :) and PLC's naturally :)
Awesome Info about pid loops! I was just tuning my quadcopter and reading about pid's and now you uploaded more Info, cool thx =)
Valuable, informative and helpful content👍. Thank you.
Nice video, very well explained. Thank you David
this is the first pid controller video i have watched that has mentioned filtering. my pid controller has a filter on it but i have it turned off and have been fine tuning the i and d and can't get it quite perfect, at least for my autistic brain. i have not tried the filtering because it's labelled as digital filtering, but I'm going to give it a whirl
The only ERROR I found is the chicken would probably give you food poisoning....
Awesome video. I'm stoked to see the pendulum bot take shape! And yes, please keep the math in there.
What's next a discussion of root locus? 😏
Great video, my EE specialty was control theory out of a very very plain navy book with no graphics, circa 2005. Used PID controllers for temperature control of large incinerators, >100 MBTU thermal oxidizers, down to small ovens for refractory tile, only 4 pallets or around 2000lb to dry them out before installation, they still had to be cured in the furnace/thermal oxidizers/flare.
The program at the end was probably written in python using matplotlib. python is a good tool for studying control theory and doesn't cost anything.
What's the whiteboard software that you're using? Thanks!
Drawboard, from the windows store.
Thanks!
David Ledger On the Surface?
I find the undrawing of the pen stroke really distracting
Do more videos on control theory, that stuff is amazing.
Thanks, will be!
This is an excellent starting video... I think you need to show the math a long side an application which I think will home in the concepts better... I like the way you presented this please do more on control theory...
wow that was extremely interesting!, will look more into that, and looking forward for the next one, thank you!.
Great Video, David, much better presentation too, you are getting better every time, seems like you were tuning your constants...
The last graphic explains very well
You're a fantastic presenter! So detailed and engaging!
Coefficient calculation: Can we just adjust the coefficient separately? First, we run the system only with P-controller, then only with I-controller, then only with D-controller. In the end, we collect them and put them together in a control algorithm. Can it work like this?
DR RORPOPOR HERBAL on TH-cam changed my entire life with his herbal medicine. I appreciate you sir, for taken away my PID 🌿🌿🌿🌿
Great video! I was just thinking about learning about this and then this vid came out. Perfect timing!
That was much better paced than your first videos.
The next videos with real examples should be fun. I remember you talking to Dave about the balancing robots on The Amphour.
Thanks for the positive feedback!
dentakuweb Agree good pacing, getting really good now.
Easier voice to follow than Dave (ok I admit I'm used to Dave now). Needs a little polishing on the presentation but I think its a good effort. Not too much maths or too complex.
As you continue the series consider doing something in mathlab or similar software to show the effects of tuning. The last graph was small and unclear.
Can't stand Dave's voice. This was a welcome change.
It boggles the mind that a minor difference in pitch and intonation provokes such a reaction.
Never go to Italy, by the way, you probably would not survive it. ;)
Manick N i prefer daves voice, personally
Best explanation ever! Thank you!
If you had a differential equation for the issuance of an impact recall, how to compute the PID parameters mathematically? It is interesting to see such readings for example for an oven.
Many thanks for this amazing video. Best wishes!
To people who thinks there is too much math involved, you havent seen the tip of it, there isn't almost any math involved here. He just explained how it works, that's it. No s-domain representation, no formal PID design techniques, no bode plots for frequency domain analysis, not much talk about stability, and a risky method for designing a PID controller. This video isn't math intensive, it is just a tutorial.
Btw, you do not use ziegler nichols technique for linear systems, there are indeed better and easier ways of designing PID controllers for linear systems, you use ziegler-nichols if the system is somewhat nonlinear (meaning that the system can not be linearized well enough), and you need a starting point for a PID design, that is it. Don't use ziegler nichols for linear systems. It is most likely that you will end up with much better PID controllers if you just use any of the PID design techniques for linear systems.
17:50 that graph is the step response of the system, you put error on your axis I think that is a mistake, it's also not the effort, it would be the feedback converted to the setpoint units, setpoint is a step like having the heater unplugged from AC and putting the dial to 10, then connecting it to AC, that would be the response to a 10 unit step
Great explanation. Looking forward to the continuation of the series
"Feedback would be appreciated" ha. Good one
Excellent series, Keep it up, I am waiting for more of these EEVacademy videos
You changed microphone at 15:50 ? Also I think Ziegler should be pronounced "Zeegler".
I like this kind of videos. And you are getting better from video to video. Go on with the great work Dave 2!
Great mix of math and examples! Keep it up.
Great video, thanks! I first saw them in flight controllers for racing drones. There you have to tune your PID's and see the results immediately! Now I really understand how and why. With the drones the P is the responsiveness of the sticks in relation to the drones movement. You want it as high as possible, the I is the drones ability to hold the angle you set it at while flying, and the D is used to dampen oscillations from too high P, however it does make the drones more sluggish. As you said, D resists change, makes sense. Tuning those quadracopters is an art, a blend of snappy controls and smooth video. I really enjoyed this!
Great video, thanks! Looking forward to the next one.