This is a very high quality piece of content. It is much harder to edit this type of video rather than doing a simple screen share and writing the code, but I think it has paid off. Big fan of showing the iterative procedure of proportional controller, what its limitations are, and moving to the PI controller. I hope you find the time to keep making videos like this.
Thanks so much Manuel. My main goal with this one was to explain all the details carefully. You basically have to understand how it works to have any hope of getting a decent response.
Excellent series of videos on dc motors and how to control them. Thank you for not shying away from a bit of theory and mathematics. I find it very 'digestible' and inspires deeper thinking through thorough study of the material you provide. You present a clear expression of your obvious command of the topic. Plus your voice is very pleasant to listen to. Keep up the good work.
The channel is brilliant indeed! The last part of the video blew my mind because it is so rare to see similar approach to learning and experimentation! Thank you.
From my humble point of view, your channel is brilliant and remarkable. Content very consistent with practical examples and perfectly done. We have to appreciate also you effort in preparation content. Your channel will grow. I keep finger for your goals!
@@curiores111 hello, I will design the simulink according to the label movement of the information coming from the video controlled with 3 motors, of course, first of all, it will get information from gprs, monitor the coordinates with the satellite camera, track the moving objects and turn in that direction in the motors.which type of block can we use in simulink
Thank you so much for this inspiring video! Finally seeing someone implementing PID control after visiting control engineering class and not seeing any practical examples warms my engineering heart ;D I don't have a motor+encoder flying around, but an led and light sensitive resistor. I guess pid brightness control is now a thing for me. I added a none feedback path to the feedback loop, to make the system more reactive/less prone to oscillating. I add u to a value x that is dependent on my target value. The relationship between the target and x is not perfect, but now the control loop only needs to compensate for the remaining deviation. Not useful in every situation, but worth it in some.
Excellent Video! The step-by-step approach coupled with the fluid animations and an amazingly soothing voice made this video an absolute treat watch and learn! Thanks!
Thanks Dustan! I find that arduino/this type of application is a great way to get better and more interested in programming, without an awful slog through a C++ class.
I agree with all of these reviews, this is quality information presented extremely well. Great job and thumbs up! Plus, your voice in these videos is so relaxing to listen to it almost puts me in a trance... Thank you so much for such a clear explanation of all of your topics, and please keep them coming 🙂
If been looking a while for dc motor speed control with arduino and this video is by far the best video i have found. Thank you very much for your effort!!!
I am so impressed by this. I wondered if you would talk about what issues you encountered before this was operational?! I am struggling with a project!
Hi Matthew, thank you! My biggest issue was the velocity measurement, which is why I spent so long talking about it in the video. Other than that the main issue I encountered was instability near zero velocity. What kind of issues are you seeing?
thanks for your video you saved my time. I am trying to controll a blcd motor but have diffuculties about low speed rpm control and your low-pass filter solved my problem.
Hello, great Video! I tryed to do it for 2 Motors like the dual Motor with encoder Video. But i cant do it something is missing. Someone got it for a dual setup ?
Literally one of the best videos Ive ever seen.. !! excellent explanation . the animations are really easy to understand... and the flow is great too..
Great Job! Terrific tutorial....easy to follow with great visual aids. Your logic flow is spot on as you develop the code....thanks for the tremendous job!
Thank you (...not sure how to read your name... let's say..) DJ Drover! ;) I've been struggling to decide on something so I appreciate the encouragement.
@@curiores111 ahah its DJ-red rover. Trust me you should keep at this, a few more of your top quality videos and companies are gonna be fighting to be your sponsor. Its always tough in the beginning when you put all this time and effort and seemingly nothing in return but before you know it, all this effort pays off in dt :D. Your mouse-trap (or what sets you apart) is the level of detail you go into and the completeness of the explanation in a very short amount of time. Other similar videos are either short and uninformative, or too long and over explained, you have hit the sweet spot with ur duration, detail, great graphics. Don’t give up!!
@@djredrover Ah DJ red rover! I see it now, lol, I can't believe I didn't get that. XD Well thanks so much, you are far too kind. I appreciate you pointing that out, I will keep that "mouse-trap" in mind and see if I can do the same in the next video.
@@curiores111 If you need some video ideas, here are a few that I personally would love to see your take on: -Kalman Filter -LQR Controller -AM Radio Receiver -Audio Synthesizer -Camera Gimbal (Investigating an IMU [MPU6050]) -SRAM & EEPROM Usage w/ Arduino -Implementing Multi-Class Software on Arduino -LiDAR Implementation Just some ideas to help you setup more videos, since most companies that sponsor tech videos/channels, are looking for at least 3.5 videos/month of output. So there needs to be a balance between putting time into making videos as good as possible (good for us) and speeding up production to increase output (good for you to attract sponsors). Maybe you should put a PID controller on it LOL. I am an Electrical/Computer Engineer as well so if there is anything I can do to help you out, by all means let me know and I will do what I can! Cheers.
It's a very good video! I advise you to make a video about position/velocity control of a motor with state feedback! I'm studying this things in the course of systems theory and maybe you can also create an Luenberger observator to obtain the state of the motor. Yes i know, there is a lot of linear algebra, matrices everywhere but nobody on youtube create a tutorial about this, it can be something good to bring on youtube!
Great video, I have suggestion. Can you create video that explain how to control DC motor at the same time by speed and position? I like you videos very much. It helped me allot as a student in industrial automation. Thanks in advance!
@@curiores111 Acceleration and Deceleration on a Garage door controller that I designed. Some mosfets (Same Part number and Manufacturer) would behave different to others. Using this(tweak here and there) I could get the Acceleration and Deceleration the same regardless of the irregularity in the mosfets.
Hello! Thank you very much for this high quality video. I was just wondering: why are we using an atomic block, or just stopping the interrupts, when fetching its value? What are the risks if not doing it?
Nice video. I am trying to implement such speed control on a 0.75kW motor using a VFD. The microcontroller will set the frequency, similar to how yours sets a PWM signal, which is essentially the same. The problem is that for my application, the motor speeds are relatively low (maximum 150 RPM). Initially, I wanted to use hall effect sensors that would trigger a pulse every time the motor passes a magnet over it. The problem is the low resolution: with my current setup, I am only able to measure 6 pulses per revolution. That, combined with the fact that the speeds are low, is making me concerned. Do you think the 6 pulses per revolution will be enough to have accurate PID control? Also, you mentioned there are some potential problems with method 2 of measuring the velocity. Would those issues be a problem in my low speed application? Which method did you end up using? I am guessing that method 2 will not be a good idea for me, because my speeds are low.
I ended up using method 1 (number of pulses per fixed time interval). The issue with method 2 near zero velocity was too much of a problem for me. 6 counts per rev is pretty low... it might be okay. Is the motor geared down? For low RPM applications usually using a geared down motor is better. You get more counts per rev, and have fewer issues low voltage required to keep the speed low (there's usually some stall for low voltages when using DC motors).
@@curiores111 yes, it is geared down. I meant that the actual output, the thing that is attached to the motor, will be maximum 150RPM. I am a bit concerned about my 6 counts per rev because compared to your 600, thats 100 times less resolution. To control near zero velocity, I can just turn off the system. For my application, that is sufficient. I don't need it to hold its position, just need to turn it off. But I do see that if I needed to hold it in one position, simply turning it off wouldn't work. Also, is it correct that method 2 won't be able to be filtered, because you need deltaT between sample, and the deltaT between samples would change depending on the speed
@@yengajaf I do also think along these lines - could you in some way make method 2 work. I made a couple of videos regarding speed control of sewing machines, and almost all of them got a problem with low speed control. Part of the problem is the static friction of the mechanics when you start sewing. Bad control means, that you try to make one stitch from start but you end up making 4 stitches. I have used another small DC motor as a Tachogenerator to make the feed back signal for speed. It works, but this DC-motor do also produce a ripple voltage, but it will also depend of motor quality. You need to apply a 1 uF capacitor on its terminals to filter out bouncing of the brushes at mechanical resonances. Finally I like to thank @curiores111 for an excellent video.
Great tutorial.. with method 2 couldn’t you implement a condition if speed set to 0 to skip pid calculation and set speed to 0 or a stall detection if no pulses read after set time.
A great video as always Curio Res, keep it up! ....I have a question as to how you are choosing your gain values. This has confused me a lot in the past and I wondered if you could give me some further insight...but how do you choose those gains appropriately? It just seems as though often it's trial and error? I've had a look at approaches and a lot of literature points to Ziegler Nichols. Many controllers do have automatic tuning functions, I'm also curious of its procedures which is a whole new side of the coin as to how they optimise there gain values, but for manual tuning surely there is more of a methodical approach? Let me know your thoughts. Thank you again ! :)
Hi Ronak, yes this is a very important question. There are several different approaches. Classically speaking I would say that the two basic approaches are 1. Manual tuning (based on trial and error or intuition) 2. Tuning using a mathematical model (you can use root-locus or other methods to decide the parameters that fit a set of requirements) Details: 1. For the first approach, you're right, you can use Ziegler Nichols. I prefer to play with the parameters and find a response that looks good. The thing I personally find the most helpful is this table that explains how changing the parameters affects the response: en.wikipedia.org/wiki/PID_controller#cite_ref-24 Once you understand that table, you can intuitively adjust the parameters one at a time for simple applications (like the DC motor) 2. In terms of mathematical models, there are several classical methods. It's generally fairly challenging, but Brian Douglas has a series of great explanatory videos: ua-cam.com/channels/q0imsn84ShAe9PBOFnoIrg.html Optimization is roughly a branch of (2), where you use an optimizer on the mathematical model to select parameters that improve one or more of the characteristics, like rise time and/or stability.
Oh by the way I loved your video I already downloaded the software for Arduino. I will hopefully use one for my project with a Pololu motor controller Jrk G2 18v27 USB Motor Controller with Feedback just not sure which Arduino unit to use yet any recommendations?
Thank you! The 25 Hz choice is pretty arbitrary. Basically, I chose the highest cutoff frequency that still removed most (~90%) of the oscillations created by the measurement method. You could look at the spectral content to decide (which I did afterward, and was happy enough with where it landed).
@@curiores111 Don't you think we can rely on the motor response time to pick a number. For example, a typical response time of a small dc motor is about 100ms, so a 10Hz cutoff frequency is a reasonable choice. Does such reasoning make sense or am I missing something? Thanks!
Certainly, using characteristics of the system would be a good way to choose a cutoff. I feel that a 10Hz cutoff might lead to some signal loss, but I'm not certain. Certainly worth doing some testing.
My first thought was that allowing the controller's output to go negative and cause it to try to drive the motor in reverse seems wrong for a setup like this. Maybe constrain the output to be non-negative? But I guess for a DC motor, that means letting it spin freely, which would rely on friction to slow it. And if there was an external force (for example, gravity if you are trying to raise or lower window blinds), I could see needing to apply a negative force to partially counteract that external force to control the speed. Next, I was thinking that if the current speed is below the target speed, you clearly need to increase the control output (and decrease the control output if current speed is faster than the target speed). Wouldn't that mean adding (some multiple of) the error to the current control output, instead of setting the output to be proportional to the error? But that's essentially what the integration term is doing (except that it is accumulating error from multiple iterations of the loop). Thanks for pointing out that velocity method two (timing the interval between pulses of the encoder) breaks down at zero velocity. That was something I hadn't considered. It seems like there ought to be a way to mitigate that. You know that an upper bound on the speed is determined by the time from the last pulse to current time, so you could constrain your last measured speed; eventually, you would reach the limits of your precision and get zero. I really appreciate your videos. The narration is clear and easy to understand (good pronunciation and enunciation). The pace is quick, but not too fast to follow. The graphics really help reinforce the narration.
Great to see you Mark! And I love your detailed thoughts about the controller. 1. Interestingly, the voltage switching happens fast enough that it's totally feasible to apply a negative voltage to slow the motor down. In fact, there's a lot of +/- switching that will happen in the controller that's unnecessary, but on **average** will produce the required speed. That's one of the unpleasant things caused by feedback control, but is a natural consequence. 2. Yes! This is why the integral term is essential in this case, because the friction (or other loading) is constantly trying to slow it down. So at steady-state you need to be applying additional voltage. What you're describing is a bit more like open-loop or model based control, where you use some information about the system to try to predict how much voltage to apply. 3. One way to mitigate this is to use open-loop control near zero velocity. Or an alternative strategy. I toyed with a couple alternatives. At the end I found that simply filtering method 1 was overall easier to implement and more consistent. It's the reason I made the video about low-pass filtering. 4. Thank you very much. 😄 I hope that you stop by at my next video.
Heelo, nice video ! There's one thing that I don't uderstand about PID controller: Imagine that you want your motor to spin a 100 RPM and for that speed the voltage should be around 6 V. If the motor is exactly at that speed the error will be 0 and then the pwm too so the motor's votlatge will be 0 rather than ~6V. The motor's speed will decrease the error go up and the pwm too to have the right speed then the error will be 0, this cycle will go indefinetly so my question is why the motor's speed does not oscillates like the cycle I mentionned earlier and why working purely with the error is best when you want to maintain a certain value?
Good observation Julien. This is actually exactly how proportional control works (neglecting for a moment, the integral and derivative terms). What ends up happening is that the rotation speed will drop slightly, and the control signal will immediately ramp up to keep it spinning. But overall that means the speed you reach will tend to be a bit lower than the desired speed on average. If this happens, you can add the integral term which will accumulate error over time. This will end up giving a little more power to the control signal than you would have with the proportional control alone.
And btw, if you choose the PID parameters poorly, you will definitely see the speed oscillate. Another thing to remember is that the physical characteristics of the motor tend to keep it spinning for a while, even if the voltage is turned off. So in other words, you wouldn't suddenly stop when the control signal goes to zero, you would just start to slow down.
@@curiores111 oh yeah, the inertial carateristics of the motor are useful here, it's like a bike where to mainain a certain speed you don't have to pedal continously but give "spikes" of power will keep you going at approximatively constant speed
Thank you so much.Very useful video. There is one thing I'm wondering to know. The input and the output of PID Controller is rpm, right? So how we define it to arduino pins as Voltage? Thanks
Hi there, great video! Learned alot from it. I just have one question. How do i replace the setPoint for a potentiometer? I cant figure that one out at all! Can you please help?
Thanks Kenan! There's more than one option. I think the most obvious answer to your question would be something like... (1) Read a value from the pot, let's call it potValue ( e.g.: www.arduino.cc/en/tutorial/potentiometer ) (2) setPoint = multiplier*potValue the value of "multiplier" should be based on the range of set-points that you are looking to achieve.
Hello, I’ve been search for a answer to a question for a long time and have found nothing that fits. Your videos are the closest to sorting out my problem. I have a BLDC Motor with Hall Outputs (x3) Not and encoder as I don’t need to run the motor in reverse. I do have a motor controller that has a PMW input. The motor controller also has a “signal output” which takes the 3 hall signals and amalgamates them to a single 12 pulses per rotation 2.65V Square wave. I can control the motor using the PMW output from both and Arduino Uno and an ATTiny 85. I can also read the RPM with the Arduino. My problem is that I would like to close the loop and control the RPM of the motor using the PMW signal which is being controlled by the RPM reader. I found your PID code but am not sure how to implement it correctly. My goal is to run the motor at either exactly 6250 RPM (75000 pulse per minute) or alternatively 6000 RPM (72000 pulse per minute) The question is; Can I do this with and Arduino/ATtiny 85? If I can, what to I need to change in this code to make it work? Here's a video of my motor running: ua-cam.com/video/GRr_g87BeAo/v-deo.html
Based on your description, sure, you can do this. The measured signal would be your RPM reading (this is the input to the controller), and you would use the control signal (the output of the control algorithm) to set the PWM. The target would be set to the RPM you want, such as 6250 RPM. In terms of how to adjust the code, I guess it depends on how you're writing it. There shouldn't be too much to adjust in terms of the PID controller, since that's pretty standard. It's more a matter of mapping your system to the interface and the controller. You'll also need to rescale the PID parameters, in all likelihood.
How you are measuring the rpm of the motor in the display? Please show the connection diagram of the LCD display. Is there is any different code to show the rpm ?
I had an idea to fix the issues of methods 1 and 2 for velocity measurement. It's basically a hybrid of the two: Regularly check how many pulses there have been since the last check, but instead of just assuming the check interval as time, always record the last pulse timestamp and use it to get a more accurate time between checks Would that work?
I have a question about the conversion from rpm to voltage. Have you supposed that 1 rpm equal to 1 V? Despite my question, this is one of the greatest videos on DC motor PID speed control I have ever seen.
@@curiores111 when the graphic is shown, it is calculated the error of rpm (y axis) and then it is said that if the error is 24, meaning 24 rpm, and Kp is 0.5 the tension given to the motor es 24 * 0.5... this is the part where i get confused. I want to excuse myself for the bothering
Thank you :) The encoder CPR = 64 is rated with the expectation that you will count all four events: trigger A rising, trigger B rising, trigger A falling, trigger B falling. I have only written an interrupt that triggers when A rises. So only one out of every four events is captured. If you add interrupts for the other three events, you will have the higher resolution of 64 CPR.
@Curio Res I am trying to implement this PID speed control but my motor driver board only has the INA and INB, not a PWM input like your Pololu has. I am using a Cytron MDD3A. I'm trying to figure out if the code can be made to work with this driver board or if I need to have that 3rd input? Any thoughts
Jeffery, it looks like on your motor driver both inputs are PWM. It seems like if you increase/decrease the PWM signals you should see speed changes. I'm not sure what combination you should use to produce the speeds required... the data sheet doesn't make it obvious. www.robotshop.com/media/files/content/c/cyt/pdf/cytron-3a-4-16v-dual-channel-dc-motor-driver-datasheet.pdf I think they're expecting you to use their library. Have you looked at their example? github.com/CytronTechnologies/CytronMotorDriver/blob/master/examples/PWM_DIR_DUAL/PWM_DIR_DUAL.ino If I wanted to figure out how to do it without their library, I would look at their arduino code (linked above) and determine what combination of PWM signals they use to generate the forward/reverse speeds, and wrap that in my own function.
Please reply ... What is the reason behind measuring only one out of every 64 events of encoder per revolution? and how we are measuring only one .?would you please explain it?
It's just an on/off function to switch the speeds from 0 to 100. Switches every 3.14 seconds. Nothing special about that number, just something simple.
Looking to use a tach to control motor speed, but the tach is on the device, not the motor, and will receive 4 magnetic pulses per revolution. It will need to stay at 1800 rpm but the motor will be running at a different speed due to voltage changes the volts will start at 12-13volts but go up to 14.5 to 15 volts how do you think I should do it?
Hi! Thank you for your great video! I have one question tho. How can i find the cut off frequency, or how did you perform the dft to get the plot in frequency domain? Thank you!
sure, atomic.h is for the atomic block. The atomic block prevents the interrupt code from running while in the atomic block. If it were to run at the same time, you could potentially read the value incorrectly, since it might be partially updated when the value is read. You can see an example at the bottom of this page: www.arduino.cc/reference/en/language/variables/variable-scope-qualifiers/volatile/ An alternative is to use nointerrupts() www.arduino.cc/reference/en/language/functions/interrupts/nointerrupts/
Thank you for the very informative video. It's somewhat unintuitive to me why method 2 exists if we won't be able to measure zero velocity. It also seems that you've used the LPF for it as well for some reason.
It's definitely the most intuitive method. There could be ways to workaround the issue at zero velocity. I'm not sure what you mean by "why it exists", do you mean "why did you put it in the video?"
@@curiores111 It's intuitive yes, but its existence provided that we'll get oscillations when setting v to zero is what isn't. It's only intuitive now because there could be ways to workaround the issue like you said.
Why can't we model the motor by giving step input to it from the Arduino and seeing the response in the plot..It's like we may get FOPDT System I guess... then we can design our PID controller in a very easy way right?
CRYSTAL CLEAR. But I have a question about one of your prior videos on PID-controlled DC motor position control. My main issue is that my motor continues to oscillate at the desired position even when I change the gain levels. Please share your explanation. These are the specs for my motor ( gear ratio 30: 1, 600 rpm and 4092 CPR)
thanks! Hmmm.. are you actually reaching the target and then oscillating? If so, I would try reducing your proportional gain down to 1e-8 and then multiplying it by a power of ten until you start to get a response (kd = ki = 0 too). There might be some other issue in your code, but if you copied the code I posted then it shouldn't be a code issue.
@@curiores111 I had copied same to same your code. Yeah I am reaching my target position then afterwards it ossicilates at target position. I will try thanks for replying me.
Hi, great video. One question ( 4:05 ), why does the motor increase speed for 3s and than stop and repeat it? If I print a variable "pwr" in serial, it constantly rise. I dont get it.
@@curiores111 at 4:05, line 27, int pwr. You set int pwr as 100/3*micros/1.e6. If I print a variable "pwr" in serial comunication, it constantly rise. It do what i expected. But when the hardware is connected (motor, driver, source, arduino), motor increasing speed for few second and then stop and repeat it in loop. I still dont get it. Have you some idea why this happening? regards
Just can't believe that this high quality content is available for free. God bless you.
I'm studying engineer for 3 years and your videos are pure gold, thank you very much!
Thank you Pavel - and good luck with your studies!
This is a very high quality piece of content. It is much harder to edit this type of video rather than doing a simple screen share and writing the code, but I think it has paid off. Big fan of showing the iterative procedure of proportional controller, what its limitations are, and moving to the PI controller. I hope you find the time to keep making videos like this.
This may well be one of the most clearly explained videos on UA-cam. I'm amazed.
Thanks so much Manuel. My main goal with this one was to explain all the details carefully. You basically have to understand how it works to have any hope of getting a decent response.
Excellent series of videos on dc motors and how to control them. Thank you for not shying away from a bit of theory and mathematics. I find it very 'digestible' and inspires deeper thinking through thorough study of the material you provide. You present a clear expression of your obvious command of the topic. Plus your voice is very pleasant to listen to. Keep up the good work.
If someone wants to know or is interested: there are graphic methods to get the ki, ke and even the kd constants. Amazing vid, cheers.
The channel is brilliant indeed! The last part of the video blew my mind because it is so rare to see similar approach to learning and experimentation! Thank you.
Well thank you Noxafurry. That's very kind :)
Thank you very much. Your way of explaining things are much better than the most professors in my university.
From my humble point of view, your channel is brilliant and remarkable. Content very consistent with practical examples and perfectly done. We have to appreciate also you effort in preparation content. Your channel will grow. I keep finger for your goals!
Thank you Markus, you are too kind 😊.
@@curiores111 hello, I will design the simulink according to the label movement of the information coming from the video controlled with 3 motors, of course, first of all, it will get information from gprs, monitor the coordinates with the satellite camera, track the moving objects and turn in that direction in the motors.which type of block can we use in simulink
@@curiores111 hey can you help me with implementing this project
Qjwwwww
Hands down best Arduino-related video I've ever seen. Great quality, in-depth explanations, just excellent.
You are too kind, thank you
This was exactly the video I needed to remind myself how to build good speed control, thank you!
Thank you so much for this inspiring video! Finally seeing someone implementing PID control after visiting control engineering class and not seeing any practical examples warms my engineering heart ;D I don't have a motor+encoder flying around, but an led and light sensitive resistor. I guess pid brightness control is now a thing for me. I added a none feedback path to the feedback loop, to make the system more reactive/less prone to oscillating. I add u to a value x that is dependent on my target value. The relationship between the target and x is not perfect, but now the control loop only needs to compensate for the remaining deviation. Not useful in every situation, but worth it in some.
Excellent Video! The step-by-step approach coupled with the fluid animations and an amazingly soothing voice made this video an absolute treat watch and learn! Thanks!
Amazing explanation, thank you! I liked how you compare the methods.
Glad to hear that 😄
I'm really enjoying these videos. So clearly explained and easy to follow especially for an unskilled programmer like myself
Thanks Dustan! I find that arduino/this type of application is a great way to get better and more interested in programming, without an awful slog through a C++ class.
Nice video. You can actually solve 'zero velocity' issues by sending a stop to the motor any time the reference is 0 instead of having it use the PID
100% better than my old control teacher.
I am building an arduino controller for a 4hp dc motor and this video is invaluable!
For me it's hard to put formulas like these into code so your video is very helpful, thank you!!
Thank you, friend. Happy coding.
Im doing this exact experiment but with state space. Can’t believe i just found your channel, loved your video
Awesome explanation, this is exactly what I subbed for! And those nice graphics are the icing on the cake. Thanks for sharing!
Thanks for the kind words and for stopping by Lucca 😊
I agree with all of these reviews, this is quality information presented extremely well. Great job and thumbs up!
Plus, your voice in these videos is so relaxing to listen to it almost puts me in a trance...
Thank you so much for such a clear explanation of all of your topics, and please keep them coming 🙂
If been looking a while for dc motor speed control with arduino and this video is by far the best video i have found. Thank you very much for your effort!!!
This channel is a hidden GEM!
Clearly explanation of how to control motor with pid. Thank for this tutorial. I'm looking for next video
Brilliant explanation - wish I had this explanation when I was at college. 👍🙏🙂
thanks for video series. this is the 3th time i am watching. They are full of information.
The Best Tutorial Video about PID I have seen! Great Job
I am so impressed by this. I wondered if you would talk about what issues you encountered before this was operational?! I am struggling with a project!
Hi Matthew, thank you! My biggest issue was the velocity measurement, which is why I spent so long talking about it in the video. Other than that the main issue I encountered was instability near zero velocity. What kind of issues are you seeing?
The "speed" of the video is a bit fast BUT it's a very very good explaination of PID control! thanks tons for this :))
thanks for your video you saved my time. I am trying to controll a blcd motor but have diffuculties about low speed rpm control and your low-pass filter solved my problem.
This is going to save me a lot of time and headache. Your channel is going to blow up soon. I know it!!!
Let's hope so! (that it saves you time and headache, not that the channel blows up ;)
This video was just what I was looking for and you explained everything so well. Thank you so much!
Hello, great Video!
I tryed to do it for 2 Motors like the dual Motor with encoder Video. But i cant do it something is missing.
Someone got it for a dual setup ?
Literally one of the best videos Ive ever seen.. !! excellent explanation . the animations are really easy to understand... and the flow is great too..
You are too kind Will, thank you
Thanks for your excellent video and explanation..Hope you will make many such engineering videos.
Exquisitely presented; so clear!
Great Video, Wish if you have shown the filtered speed using method 2 (the less noisy method)
Great Job! Terrific tutorial....easy to follow with great visual aids. Your logic flow is spot on as you develop the code....thanks for the tremendous job!
I freaking LOVE your channel. Your videos are well thought-out, useful and informative. Please keep them coming.
Thank you (...not sure how to read your name... let's say..) DJ Drover! ;) I've been struggling to decide on something so I appreciate the encouragement.
@@curiores111 ahah its DJ-red rover. Trust me you should keep at this, a few more of your top quality videos and companies are gonna be fighting to be your sponsor. Its always tough in the beginning when you put all this time and effort and seemingly nothing in return but before you know it, all this effort pays off in dt :D. Your mouse-trap (or what sets you apart) is the level of detail you go into and the completeness of the explanation in a very short amount of time. Other similar videos are either short and uninformative, or too long and over explained, you have hit the sweet spot with ur duration, detail, great graphics. Don’t give up!!
@@djredrover Ah DJ red rover! I see it now, lol, I can't believe I didn't get that. XD Well thanks so much, you are far too kind. I appreciate you pointing that out, I will keep that "mouse-trap" in mind and see if I can do the same in the next video.
@@curiores111 If you need some video ideas, here are a few that I personally would love to see your take on:
-Kalman Filter
-LQR Controller
-AM Radio Receiver
-Audio Synthesizer
-Camera Gimbal (Investigating an IMU [MPU6050])
-SRAM & EEPROM Usage w/ Arduino
-Implementing Multi-Class Software on Arduino
-LiDAR Implementation
Just some ideas to help you setup more videos, since most companies that sponsor tech videos/channels, are looking for at least 3.5 videos/month of output. So there needs to be a balance between putting time into making videos as good as possible (good for us) and speeding up production to increase output (good for you to attract sponsors). Maybe you should put a PID controller on it LOL.
I am an Electrical/Computer Engineer as well so if there is anything I can do to help you out, by all means let me know and I will do what I can! Cheers.
It's a very good video! I advise you to make a video about position/velocity control of a motor with state feedback! I'm studying this things in the course of systems theory and maybe you can also create an Luenberger observator to obtain the state of the motor. Yes i know, there is a lot of linear algebra, matrices everywhere but nobody on youtube create a tutorial about this, it can be something good to bring on youtube!
Thank you and I appreciate the suggestion. I don't mind linear algebra and matrices. ;)
Great video, I have suggestion. Can you create video that explain how to control DC motor at the same time by speed and position? I like you videos very much. It helped me allot as a student in industrial automation. Thanks in advance!
Excelente video, editing and explanation! Thank you!
Thank you so much, your explanation and your code really helps me to shorten the low level control in my mobile robot!! Liked!
Exactly what I like to hear! Thanks Abdul 😊
Thanks. Was able to implement this in my code and it helped solve the problem I had.
Good to hear that. What was the problem that you had?
@@curiores111 Acceleration and Deceleration on a Garage door controller that I designed. Some mosfets (Same Part number and Manufacturer) would behave different to others. Using this(tweak here and there) I could get the Acceleration and Deceleration the same regardless of the irregularity in the mosfets.
@@dannyoneill4888 Interesting. Thank you for sharing.
Hello! Thank you very much for this high quality video.
I was just wondering: why are we using an atomic block, or just stopping the interrupts, when fetching its value? What are the risks if not doing it?
The best channel that link theory with practice!!
You are awesome 👍👍
your voice is extremely relaxing :)
Nice video. I am trying to implement such speed control on a 0.75kW motor using a VFD. The microcontroller will set the frequency, similar to how yours sets a PWM signal, which is essentially the same.
The problem is that for my application, the motor speeds are relatively low (maximum 150 RPM). Initially, I wanted to use hall effect sensors that would trigger a pulse every time the motor passes a magnet over it. The problem is the low resolution: with my current setup, I am only able to measure 6 pulses per revolution. That, combined with the fact that the speeds are low, is making me concerned. Do you think the 6 pulses per revolution will be enough to have accurate PID control?
Also, you mentioned there are some potential problems with method 2 of measuring the velocity. Would those issues be a problem in my low speed application? Which method did you end up using? I am guessing that method 2 will not be a good idea for me, because my speeds are low.
I ended up using method 1 (number of pulses per fixed time interval). The issue with method 2 near zero velocity was too much of a problem for me.
6 counts per rev is pretty low... it might be okay. Is the motor geared down? For low RPM applications usually using a geared down motor is better. You get more counts per rev, and have fewer issues low voltage required to keep the speed low (there's usually some stall for low voltages when using DC motors).
@@curiores111 yes, it is geared down. I meant that the actual output, the thing that is attached to the motor, will be maximum 150RPM.
I am a bit concerned about my 6 counts per rev because compared to your 600, thats 100 times less resolution.
To control near zero velocity, I can just turn off the system. For my application, that is sufficient. I don't need it to hold its position, just need to turn it off. But I do see that if I needed to hold it in one position, simply turning it off wouldn't work.
Also, is it correct that method 2 won't be able to be filtered, because you need deltaT between sample, and the deltaT between samples would change depending on the speed
@@yengajaf I do also think along these lines - could you in some way make method 2 work. I made a couple of videos regarding speed control of sewing machines, and almost all of them got a problem with low speed control. Part of the problem is the static friction of the mechanics when you start sewing. Bad control means, that you try to make one stitch from start but you end up making 4 stitches. I have used another small DC motor as a Tachogenerator to make the feed back signal for speed. It works, but this DC-motor do also produce a ripple voltage, but it will also depend of motor quality. You need to apply a 1 uF capacitor on its terminals to filter out bouncing of the brushes at mechanical resonances.
Finally I like to thank @curiores111 for an excellent video.
Great tutorial.. with method 2 couldn’t you implement a condition if speed set to 0 to skip pid calculation and set speed to 0 or a stall detection if no pulses read after set time.
Hello. Your videos are fantastic. Can you tell me what software you use to produce this tutorial?
A great video as always Curio Res, keep it up! ....I have a question as to how you are choosing your gain values. This has confused me a lot in the past and I wondered if you could give me some further insight...but how do you choose those gains appropriately? It just seems as though often it's trial and error? I've had a look at approaches and a lot of literature points to Ziegler Nichols. Many controllers do have automatic tuning functions, I'm also curious of its procedures which is a whole new side of the coin as to how they optimise there gain values, but for manual tuning surely there is more of a methodical approach? Let me know your thoughts. Thank you again ! :)
Hi Ronak, yes this is a very important question. There are several different approaches. Classically speaking I would say that the two basic approaches are
1. Manual tuning (based on trial and error or intuition)
2. Tuning using a mathematical model (you can use root-locus or other methods to decide the parameters that fit a set of requirements)
Details:
1. For the first approach, you're right, you can use Ziegler Nichols. I prefer to play with the parameters and find a response that looks good. The thing I personally find the most helpful is this table that explains how changing the parameters affects the response:
en.wikipedia.org/wiki/PID_controller#cite_ref-24
Once you understand that table, you can intuitively adjust the parameters one at a time for simple applications (like the DC motor)
2. In terms of mathematical models, there are several classical methods. It's generally fairly challenging, but Brian Douglas has a series of great explanatory videos:
ua-cam.com/channels/q0imsn84ShAe9PBOFnoIrg.html
Optimization is roughly a branch of (2), where you use an optimizer on the mathematical model to select parameters that improve one or more of the characteristics, like rise time and/or stability.
Wow great explaination with Animations
Oh by the way I loved your video I already downloaded the software for Arduino. I will hopefully use one for my project with a Pololu motor controller Jrk G2 18v27 USB Motor Controller with Feedback just not sure which Arduino unit to use yet any recommendations?
Fabolous explained. Thank you a lot for your sharing. The tutorial are well useful and informative
Thank you for the kind words Le Dinh 😊 I am very glad you found them useful.
holy moly PID makes sense now
Can you explain desired position controll process using encoder, like providing user input & getting that position.
This video is very high quality content. Thank you.
How nice 😄. Glad you stopped by mist!
Great video!
Why did you choose a cutoff frequency of 25Hz? What is the reasoning behind this choice? Thank you very much for your great work!
Thank you! The 25 Hz choice is pretty arbitrary. Basically, I chose the highest cutoff frequency that still removed most (~90%) of the oscillations created by the measurement method. You could look at the spectral content to decide (which I did afterward, and was happy enough with where it landed).
@@curiores111 Don't you think we can rely on the motor response time to pick a number. For example, a typical response time of a small dc motor is about 100ms, so a 10Hz cutoff frequency is a reasonable choice. Does such reasoning make sense or am I missing something? Thanks!
Certainly, using characteristics of the system would be a good way to choose a cutoff. I feel that a 10Hz cutoff might lead to some signal loss, but I'm not certain. Certainly worth doing some testing.
BRILLIANT VIDEO.
My first thought was that allowing the controller's output to go negative and cause it to try to drive the motor in reverse seems wrong for a setup like this. Maybe constrain the output to be non-negative? But I guess for a DC motor, that means letting it spin freely, which would rely on friction to slow it. And if there was an external force (for example, gravity if you are trying to raise or lower window blinds), I could see needing to apply a negative force to partially counteract that external force to control the speed.
Next, I was thinking that if the current speed is below the target speed, you clearly need to increase the control output (and decrease the control output if current speed is faster than the target speed). Wouldn't that mean adding (some multiple of) the error to the current control output, instead of setting the output to be proportional to the error? But that's essentially what the integration term is doing (except that it is accumulating error from multiple iterations of the loop).
Thanks for pointing out that velocity method two (timing the interval between pulses of the encoder) breaks down at zero velocity. That was something I hadn't considered. It seems like there ought to be a way to mitigate that. You know that an upper bound on the speed is determined by the time from the last pulse to current time, so you could constrain your last measured speed; eventually, you would reach the limits of your precision and get zero.
I really appreciate your videos. The narration is clear and easy to understand (good pronunciation and enunciation). The pace is quick, but not too fast to follow. The graphics really help reinforce the narration.
Great to see you Mark! And I love your detailed thoughts about the controller.
1. Interestingly, the voltage switching happens fast enough that it's totally feasible to apply a negative voltage to slow the motor down. In fact, there's a lot of +/- switching that will happen in the controller that's unnecessary, but on **average** will produce the required speed. That's one of the unpleasant things caused by feedback control, but is a natural consequence.
2. Yes! This is why the integral term is essential in this case, because the friction (or other loading) is constantly trying to slow it down. So at steady-state you need to be applying additional voltage. What you're describing is a bit more like open-loop or model based control, where you use some information about the system to try to predict how much voltage to apply.
3. One way to mitigate this is to use open-loop control near zero velocity. Or an alternative strategy. I toyed with a couple alternatives. At the end I found that simply filtering method 1 was overall easier to implement and more consistent. It's the reason I made the video about low-pass filtering.
4. Thank you very much. 😄 I hope that you stop by at my next video.
Heelo, nice video ! There's one thing that I don't uderstand about PID controller:
Imagine that you want your motor to spin a 100 RPM and for that speed the voltage should be around 6 V.
If the motor is exactly at that speed the error will be 0 and then the pwm too so the motor's votlatge will be 0 rather than ~6V.
The motor's speed will decrease the error go up and the pwm too to have the right speed then the error will be 0, this cycle will go indefinetly so my question is why the motor's speed does not oscillates like the cycle I mentionned earlier and why working purely with the error is best when you want to maintain a certain value?
Good observation Julien. This is actually exactly how proportional control works (neglecting for a moment, the integral and derivative terms). What ends up happening is that the rotation speed will drop slightly, and the control signal will immediately ramp up to keep it spinning. But overall that means the speed you reach will tend to be a bit lower than the desired speed on average.
If this happens, you can add the integral term which will accumulate error over time. This will end up giving a little more power to the control signal than you would have with the proportional control alone.
And btw, if you choose the PID parameters poorly, you will definitely see the speed oscillate. Another thing to remember is that the physical characteristics of the motor tend to keep it spinning for a while, even if the voltage is turned off. So in other words, you wouldn't suddenly stop when the control signal goes to zero, you would just start to slow down.
@@curiores111 oh yeah, the inertial carateristics of the motor are useful here, it's like a bike where to mainain a certain speed you don't have to pedal continously but give "spikes" of power will keep you going at approximatively constant speed
Thank you so much.Very useful video.
There is one thing I'm wondering to know.
The input and the output of PID Controller is rpm, right?
So how we define it to arduino pins as Voltage?
Thanks
Hi there, great video! Learned alot from it.
I just have one question. How do i replace the setPoint for a potentiometer? I cant figure that one out at all! Can you please help?
Thanks Kenan!
There's more than one option. I think the most obvious answer to your question would be something like...
(1) Read a value from the pot, let's call it potValue ( e.g.: www.arduino.cc/en/tutorial/potentiometer )
(2) setPoint = multiplier*potValue
the value of "multiplier" should be based on the range of set-points that you are looking to achieve.
Hello,
I’ve been search for a answer to a question for a long time and have found nothing that fits. Your videos are the closest to sorting out my problem.
I have a BLDC Motor with Hall Outputs (x3) Not and encoder as I don’t need to run the motor in reverse. I do have a motor controller that has a PMW input. The motor controller also has a “signal output” which takes the 3 hall signals and amalgamates them to a single 12 pulses per rotation 2.65V Square wave.
I can control the motor using the PMW output from both and Arduino Uno and an ATTiny 85.
I can also read the RPM with the Arduino.
My problem is that I would like to close the loop and control the RPM of the motor using the PMW signal which is being controlled by the RPM reader.
I found your PID code but am not sure how to implement it correctly.
My goal is to run the motor at either exactly 6250 RPM (75000 pulse per minute) or alternatively 6000 RPM (72000 pulse per minute)
The question is; Can I do this with and Arduino/ATtiny 85?
If I can, what to I need to change in this code to make it work?
Here's a video of my motor running: ua-cam.com/video/GRr_g87BeAo/v-deo.html
Based on your description, sure, you can do this. The measured signal would be your RPM reading (this is the input to the controller), and you would use the control signal (the output of the control algorithm) to set the PWM. The target would be set to the RPM you want, such as 6250 RPM. In terms of how to adjust the code, I guess it depends on how you're writing it. There shouldn't be too much to adjust in terms of the PID controller, since that's pretty standard. It's more a matter of mapping your system to the interface and the controller. You'll also need to rescale the PID parameters, in all likelihood.
How you are measuring the rpm of the motor in the display?
Please show the connection diagram of the LCD display.
Is there is any different code to show the rpm ?
Tutorial bellissimo. Complimenti.
High quality video, many thanks, subbed
Thank you, and welcome! :)
the best video ever ,thanks
I wonder if this could be improved by defining a transfer function for the motor and doing the root locus to find an optimal gain?
amazing video and amazing chanal ,we are waiting more videos
I had an idea to fix the issues of methods 1 and 2 for velocity measurement.
It's basically a hybrid of the two:
Regularly check how many pulses there have been since the last check, but instead of just assuming the check interval as time, always record the last pulse timestamp and use it to get a more accurate time between checks
Would that work?
my head hurts but this is gold!
I hope to discuss how to control stepper motor with constant acceleration model, there is no resources on UA-cam .please....
excellent video production, outstanding didactics. subscribed.
How kind and concisely put! Thank you Copernico.
I have a question about the conversion from rpm to voltage. Have you supposed that 1 rpm equal to 1 V? Despite my question, this is one of the greatest videos on DC motor PID speed control I have ever seen.
1 rpm definitely does *not* equal 1V. What part of the video suggested that to you?
@@curiores111 when the graphic is shown, it is calculated the error of rpm (y axis) and then it is said that if the error is 24, meaning 24 rpm, and Kp is 0.5 the tension given to the motor es 24 * 0.5... this is the part where i get confused. I want to excuse myself for the bothering
Amazing video. I had a question too,
Why do we take 16 counts, even thought the motor gives 64 counts per relovution?
Thank you :) The encoder CPR = 64 is rated with the expectation that you will count all four events: trigger A rising, trigger B rising, trigger A falling, trigger B falling. I have only written an interrupt that triggers when A rises. So only one out of every four events is captured. If you add interrupts for the other three events, you will have the higher resolution of 64 CPR.
wonderful explanation
Is it possible that we take analog reading of the motor rotation and use ADC to create PID.
Is there any video tutorial?
@Curio Res I am trying to implement this PID speed control but my motor driver board only has the INA and INB, not a PWM input like your Pololu has. I am using a Cytron MDD3A. I'm trying to figure out if the code can be made to work with this driver board or if I need to have that 3rd input? Any thoughts
Jeffery, it looks like on your motor driver both inputs are PWM. It seems like if you increase/decrease the PWM signals you should see speed changes. I'm not sure what combination you should use to produce the speeds required... the data sheet doesn't make it obvious.
www.robotshop.com/media/files/content/c/cyt/pdf/cytron-3a-4-16v-dual-channel-dc-motor-driver-datasheet.pdf
I think they're expecting you to use their library. Have you looked at their example?
github.com/CytronTechnologies/CytronMotorDriver/blob/master/examples/PWM_DIR_DUAL/PWM_DIR_DUAL.ino
If I wanted to figure out how to do it without their library, I would look at their arduino code (linked above) and determine what combination of PWM signals they use to generate the forward/reverse speeds, and wrap that in my own function.
Please reply ...
What is the reason behind measuring only one out of every 64 events of encoder per revolution?
and how we are measuring only one .?would you please explain it?
At 11:22 in the video, line 63 in the code, the speed is using vt=100*(sin(currT/1e6)>0); how did you get that equation?
It's just an on/off function to switch the speeds from 0 to 100. Switches every 3.14 seconds. Nothing special about that number, just something simple.
Love you! You saved my life.
Great stuff, very helpful, I really enjoyed this.
wow thank you, I'm glad you liked it 😊
Looking to use a tach to control motor speed, but the tach is on the device, not the motor, and will receive 4 magnetic pulses per revolution. It will need to stay at 1800 rpm but the motor will be running at a different speed due to voltage changes the volts will start at 12-13volts but go up to 14.5 to 15 volts how do you think I should do it?
Hi! Thank you for your great video! I have one question tho. How can i find the cut off frequency, or how did you perform the dft to get the plot in frequency domain? Thank you!
Oh nevermind, just saw your other video 😂.
Those videos were made just for you 😉😄
Is it possible for you to explain "util/atomic.h"? Or any other resource you think is relevant. Thanks in advance.
sure, atomic.h is for the atomic block. The atomic block prevents the interrupt code from running while in the atomic block. If it were to run at the same time, you could potentially read the value incorrectly, since it might be partially updated when the value is read. You can see an example at the bottom of this page: www.arduino.cc/reference/en/language/variables/variable-scope-qualifiers/volatile/
An alternative is to use nointerrupts()
www.arduino.cc/reference/en/language/functions/interrupts/nointerrupts/
@@curiores111 Thanks for the lightning response, Really appreciated.
I am very grateful to you. Thank you so much.
Quick question if I want to include the 64 counts of the encoder what part of the code that is going to change in order to achieve this request
I need to come back here
Thank you for the very informative video. It's somewhat unintuitive to me why method 2 exists if we won't be able to measure zero velocity. It also seems that you've used the LPF for it as well for some reason.
It's definitely the most intuitive method. There could be ways to workaround the issue at zero velocity. I'm not sure what you mean by "why it exists", do you mean "why did you put it in the video?"
@@curiores111 It's intuitive yes, but its existence provided that we'll get oscillations when setting v to zero is what isn't. It's only intuitive now because there could be ways to workaround the issue like you said.
@@essamwisamfouad9938 I guess I see it from a slightly different perspective... thank you for sharing your thoughts. :)
you are pretty brilliant, thank you so much .
What additions need to be made when performing PID control with 2 motors?
Thank you very much. I'm appreciated that!
How did you manage to Plot the Data this Consitently in the Serial Plotter ?
Can you do that for 4 DC motor please it is so important for me.
Why can't we model the motor by giving step input to it from the Arduino and seeing the response in the plot..It's like we may get FOPDT System I guess... then we can design our PID controller in a very easy way right?
CRYSTAL CLEAR. But I have a question about one of your prior videos on PID-controlled DC motor position control. My main issue is that my motor continues to oscillate at the desired position even when I change the gain levels. Please share your explanation. These are the specs for my motor ( gear ratio 30: 1, 600 rpm and 4092 CPR)
thanks! Hmmm.. are you actually reaching the target and then oscillating? If so, I would try reducing your proportional gain down to 1e-8 and then multiplying it by a power of ten until you start to get a response (kd = ki = 0 too). There might be some other issue in your code, but if you copied the code I posted then it shouldn't be a code issue.
@@curiores111 I had copied same to same your code. Yeah I am reaching my target position then afterwards it ossicilates at target position. I will try thanks for replying me.
Shouldn't using a kd and D component smooth the oscillation?
Hi, great video. One question ( 4:05 ), why does the motor increase speed for 3s and than stop and repeat it? If I print a variable "pwr" in serial, it constantly rise. I dont get it.
Thanks. :)
I'm not sure I see where the question is pointing to. Could you specify which lines of code are suspect (or which output)?
@@curiores111 at 4:05, line 27, int pwr. You set int pwr as 100/3*micros/1.e6. If I print a variable "pwr" in serial comunication, it constantly rise. It do what i expected. But when the hardware is connected (motor, driver, source, arduino), motor increasing speed for few second and then stop and repeat it in loop. I still dont get it. Have you some idea why this happening?
regards