being a perfectionist is a disease and once you start suffering from this disease your life is a disaster. And the best part of your lecture is that stop when you feel system is behaving good. And your emphasize on "stop when done once" is a life changer. really appreciate that.
I found this very helpful, as I have been trying to tune my home made sailing boat autopilot, and hadn't come across the differential band limit, which, based on experience of my systems behaviour, is possibly something that I need to programme in. Many thanks
Start out by demonstrating what your device is supposed to do. Preferably by setting it to work well and show us how it performs. It is hard for us to recognize poor performance when we don't know what it does.
Excellent comment. However, unless I'm horribly wrong, RUclips doesn't let you edit video once it's posted -- you can add annotations, or you can take it down, but you can't change the actual video. About the closest you can come is to take it down and post another one -- but the new one will be under a different URL.
I can give a verbal description: I want the thing to go to a target position promptly, without overshoot. Because of the nature of the tuning method, one has to settle for "as promptly as possible" and then find out what "possible" is after the fact, rather than being able to directly predict performance beforehand. One can have a rough idea if one has done something similar beforehand, but that's about it.
Lo, I was trying to check out what books you had on your shelf rather than hear what you were saying. I wish I had this video available when I was constructing the timing and phase loops for a project i was working on. Always appreciate your comment on comp.dsp
Bottom Shelf, right to left: "The Annotated C++ Reference Manual" (Stroustrop) "Universal Serial Bus System Architecture" (Mindshare) "Nonlinear Dynamical Systems" (Cook) "C: A Reference Manual" (Harbison & Steel) "Linux In a Nutshell" (Siever, Figgens & Weber) "Running Linux" (Dalhemer & Welsh) "HP 32SII Owner's Manual" (HP) "Enercell Battery Guidebook" (Radio Shack) "Design of Feedback Control Systems" (Hostetter, Savant, Stefani) "HTML & XHTML: The Definitive Guide" (Musciano & Kennedy) "Linux Kernel in a Nutshell" (Kroah & Hartman) "ISE 5.2i Reference Manual" (Xilinx) various bills Top shelf, left to right: Various CDs a very dusty Rolodex Two MSP430 Ultra-Low Power MCU devolopment kits Two Stellaris LM3S811 evaluation kits (the trainer uses an LM3S811) "Switching Power Supply Design" (Pressman) "Guide to GPS Positioning" (Canadian GPS Associates) "Optimal State Estimation" (Dan Simon -- THE book if you need to understand Kalman filtering) "The Verilog Hardware Description Language" (Thomas & Moorby) "Electric Machines: Steady State Theory & Dynamic Performance" (Sarma) "Digital Control Systems: Theory, Hardware, Software" (Houpis & Lamont) "The Debian System" (Krafft) "Detection Estimation & Modulation Theory, part 1" (Van Trees -- the hardest and most useful class I ever took) "Signals and Systems" (Oppenheim & Willsky with Young) "Digital Communications: Fundamentals and Applications" (Sklar -- shared with a customer so we have a common language) "The Sports Car: It's Design and Performance" (Cambell -- from the library of Rolla Volsteadt) "Directional Statistics" (Mardia & Jupp -- mind bending math) "Engineering and Scientific Computing with Scilab" (Gomez)
Amazing Work hope you keep posting videos like this, and why not some tuning with modeling. I'm fascinated with control theory. Greetings from Ecuador.
Well, videos provided by u r very knowledgeable content wise. Now, I have understood the PID and trying to make a temperature controller. After deriving all parameters P, I, D. How it it is used to control microcontroller pin? Means it provides on/ off time, but how to correlate.
Great video! I'm just wondering, what is the benefit of adding the "integral" control to this particular system? It seemed to work much better with just D & P (i.e. no visible overshoot at all). Is there a condition under which the integral control would be of benefit here?
Good question. Because the fan trainer has very free bearings, is well balanced, and has an actuator that develops a force, it already acts like an integrator by itself -- this is why it behaves so well with just proportional and derivative control. Where integral control would really come in handy with this thing (assuming that such a gizmo ever had a practical use), would be when there is a disturbance force to be overcome. Such a force could come from an imbalance in the mechanical arrangement, if you were using the thing to pull or push on something in the real world, or if you were using the thing outside on a windy day. If you look at this video, ruclips.net/video/2elEXcv0AV8/видео.html, I show the behavior when you have a disturbance caused by imbalance with just proportional + derivative control (starting at around 8:45, here: ruclips.net/video/2elEXcv0AV8/видео.html), then later I show the behavior with the same disturbance, but with full proportional + integral + derivative control (starting at around 11:45, here: ruclips.net/video/2elEXcv0AV8/видео.html).
Business got slow, so I had to get a corporate job. That job is in the middle of a big push right now, so I basically have time for eating, sleeping, and working. I've got some videos I want to make as things calm down, though.
I'm not familiar with the Matlab tools, so I can't answer. The limitation with Matlab is that you need to get measurements of the system behavior into the thing to do system identification on, or you need to model the system -- this method arose out of a community of millwrights who learned how to do the tuning by the seat of their pants. It's not the best, but it's often plenty good enough.
Thank you very much for taking the time to post this video. It was very useful to me. I would like to replicate your setup. Could you guide me on the kind of software you're using? Thanks.
It's just a cheap little "Kaizen" lapel microphone from Amazon. No longer available -- they probably realized they'd made something decent and expired from embarrassment. www.amazon.com/gp/product/B016OKE154/ref=oh_aui_search_detailpage?ie=UTF8&psc=1
It would take a few videos to work up to that, and I have them on the list. In the mean time: www.wescottdesign.com/articles/zTransform/z-transforms.html
being a perfectionist is a disease and once you start suffering from this disease your life is a disaster. And the best part of your lecture is that stop when you feel system is behaving good. And your emphasize on "stop when done once" is a life changer. really appreciate that.
I found this very helpful, as I have been trying to tune my home made sailing boat autopilot, and hadn't come across the differential band limit, which, based on experience of my systems behaviour, is possibly something that I need to programme in. Many thanks
I'm glad I helped. There's more than just differential band limiting that can screw you up -- but it's something to try.
Wescott Design Services yes, still suffering from differential kick and my programmed range for the I term is too high...
Start out by demonstrating what your device is supposed to do. Preferably by setting it to work well and show us how it performs. It is hard for us to recognize poor performance when we don't know what it does.
Excellent comment. However, unless I'm horribly wrong, RUclips doesn't let you edit video once it's posted -- you can add annotations, or you can take it down, but you can't change the actual video. About the closest you can come is to take it down and post another one -- but the new one will be under a different URL.
I can give a verbal description: I want the thing to go to a target position promptly, without overshoot.
Because of the nature of the tuning method, one has to settle for "as promptly as possible" and then find out what "possible" is after the fact, rather than being able to directly predict performance beforehand. One can have a rough idea if one has done something similar beforehand, but that's about it.
I've added an annotation to the video to clarify this, as well as a link to one of my other videos that goes a bit deeper into desired behavior.
Lo, I was trying to check out what books you had on your shelf rather than hear what you were saying. I wish I had this video available when I was constructing the timing and phase loops for a project i was working on. Always appreciate your comment on comp.dsp
Bottom Shelf, right to left:
"The Annotated C++ Reference Manual" (Stroustrop)
"Universal Serial Bus System Architecture" (Mindshare)
"Nonlinear Dynamical Systems" (Cook)
"C: A Reference Manual" (Harbison & Steel)
"Linux In a Nutshell" (Siever, Figgens & Weber)
"Running Linux" (Dalhemer & Welsh)
"HP 32SII Owner's Manual" (HP)
"Enercell Battery Guidebook" (Radio Shack)
"Design of Feedback Control Systems" (Hostetter, Savant, Stefani)
"HTML & XHTML: The Definitive Guide" (Musciano & Kennedy)
"Linux Kernel in a Nutshell" (Kroah & Hartman)
"ISE 5.2i Reference Manual" (Xilinx)
various bills
Top shelf, left to right:
Various CDs
a very dusty Rolodex
Two MSP430 Ultra-Low Power MCU devolopment kits
Two Stellaris LM3S811 evaluation kits (the trainer uses an LM3S811)
"Switching Power Supply Design" (Pressman)
"Guide to GPS Positioning" (Canadian GPS Associates)
"Optimal State Estimation" (Dan Simon -- THE book if you need to understand Kalman filtering)
"The Verilog Hardware Description Language" (Thomas & Moorby)
"Electric Machines: Steady State Theory & Dynamic Performance" (Sarma)
"Digital Control Systems: Theory, Hardware, Software" (Houpis & Lamont)
"The Debian System" (Krafft)
"Detection Estimation & Modulation Theory, part 1" (Van Trees -- the hardest and most useful class I ever took)
"Signals and Systems" (Oppenheim & Willsky with Young)
"Digital Communications: Fundamentals and Applications" (Sklar -- shared with a customer so we have a common language)
"The Sports Car: It's Design and Performance" (Cambell -- from the library of Rolla Volsteadt)
"Directional Statistics" (Mardia & Jupp -- mind bending math)
"Engineering and Scientific Computing with Scilab" (Gomez)
Amazing Work hope you keep posting videos like this, and why not some tuning with modeling. I'm fascinated with control theory.
Greetings from Ecuador.
Modeling will come. Next is the first video on difference equations.
Well, videos provided by u r very knowledgeable content wise. Now, I have understood the PID and trying to make a temperature controller. After deriving all parameters P, I, D. How it it is used to control microcontroller pin? Means it provides on/ off time, but how to correlate.
This was good, concise and enjoyable. Thank you.
Thanks for the feedback!
Great video! I'm just wondering, what is the benefit of adding the "integral" control to this particular system? It seemed to work much better with just D & P (i.e. no visible overshoot at all). Is there a condition under which the integral control would be of benefit here?
Good question. Because the fan trainer has very free bearings, is well balanced, and has an actuator that develops a force, it already acts like an integrator by itself -- this is why it behaves so well with just proportional and derivative control. Where integral control would really come in handy with this thing (assuming that such a gizmo ever had a practical use), would be when there is a disturbance force to be overcome. Such a force could come from an imbalance in the mechanical arrangement, if you were using the thing to pull or push on something in the real world, or if you were using the thing outside on a windy day.
If you look at this video, ruclips.net/video/2elEXcv0AV8/видео.html, I show the behavior when you have a disturbance caused by imbalance with just proportional + derivative control (starting at around 8:45, here: ruclips.net/video/2elEXcv0AV8/видео.html), then later I show the behavior with the same disturbance, but with full proportional + integral + derivative control (starting at around 11:45, here: ruclips.net/video/2elEXcv0AV8/видео.html).
Ah, many thanks! Good, concise explanation.
Hey, thank you for your content, it really helped me with my Stewart platform. But what happened to you ? Why aren't you uploading anymore ?
Business got slow, so I had to get a corporate job. That job is in the middle of a big push right now, so I basically have time for eating, sleeping, and working. I've got some videos I want to make as things calm down, though.
Sad to hear ;/ Wish you the best, I think that your channel got much potential. Merry Christmas from Poland ;]
You said set P at 1% of the D. Your D is set at 2000 But, you set the P at 2. 1% of 2000 is 20 not 2... Am i missing something...?
matlab tuning tools is good enough? Is there any problem?
I'm not familiar with the Matlab tools, so I can't answer. The limitation with Matlab is that you need to get measurements of the system behavior into the thing to do system identification on, or you need to model the system -- this method arose out of a community of millwrights who learned how to do the tuning by the seat of their pants. It's not the best, but it's often plenty good enough.
Thank you very much for taking the time to post this video. It was very useful to me. I would like to replicate your setup. Could you guide me on the kind of software you're using? Thanks.
Nevermind.... Just found the other video.
Thanks.
What Mic. do you use? Voice sounds crisp af
It's just a cheap little "Kaizen" lapel microphone from Amazon. No longer available -- they probably realized they'd made something decent and expired from embarrassment. www.amazon.com/gp/product/B016OKE154/ref=oh_aui_search_detailpage?ie=UTF8&psc=1
Wescott Design Services thanks for the response! Yeah, they probably did haha
how do you tune with control theory and math
It would take a few videos to work up to that, and I have them on the list. In the mean time:
www.wescottdesign.com/articles/zTransform/z-transforms.html
25:30... See what you did there :D