Excelent descriptions and detail Martin and not too complex to lose people, I think this will prove to be very usefull to your viewers and I cant wait till the next video Peter
@ItistDe
9 жыл бұрын
Thank you for this good explanation
@rotlerin
9 жыл бұрын
Peter Oakes I agree Peter. Well explained that man!
@sdgelectronics
9 жыл бұрын
Hey Martin, this is a great video for beginners to start to get their head around stability. Well done.
@stephendixon8575
4 ай бұрын
Truly exceptional explanation here. I have come across one of your videos before, but had quite forgotten until I stumbled across this - Yet again Martin, I’m astounded at your ability to convey really quite complex topics with brilliant analogies and clear, visual explanations. I wish I had teachers and lecturers with your skills for passing on knowledge at college and University 🙌 🙌 🙌. Glad you skipped revisiting complex numbers - Have vague recollections of the anxiety they first induced when badly taught to me in my first year at Uni. The other thing that really tickled me was that I’d been reading some standard text online (trying to refresh my memory/understanding) about op-amps, covering differential amplifiers and feedback loops, when I got distracted, thinking why the voltage output ‘very nearly’ follows the input (with feedback) without oscillating out of control, unless the op-amp gain was unity (and not 200k or so as is common). Spent a few minutes scribbling down some ideas before deciding to just leave it alone, as the text didn’t cover it and move on - Then later the text then came up with an explained jump from the formula Vout = Gain (V+ less V-) and (magically) “solving for the output voltage we get the following:” - but without any mention of complex numbers or how they ‘solved’ the math 🤣🤣🤣 - Vout = Vin / 1 + (1/G). Anyway, somewhat serendipitous and bizarre that hours later I found your video which totally nailed the subject and filled in all the gaps others had conveniently left out. The World needs more teachers like you. Well done 👍
@WestCoastMole
8 жыл бұрын
WELL DONE Martin ! You have made an excellent presentation of a very complex topic. This is a subject tackled in most Electrical Engineering curricula in the 3rd year of study. It's typically entitled "Feedback Control Systems" or something of the sort. In a formal setting the Mathematics is quite intricate and involved. Your treatment reducing it to some simple Math and a Bode Plot does the viewer a great service. The plain language example you use also goes a longs ways to imparting an understanding to the uninitiated who don't have an engineering background with the adequate prerequisites. BTW your second pronunciation of Bode (with the long e) is the correct one. Thanks again
@mjlorton
8 жыл бұрын
+Bob Marston Thanks very much for the post Bob.
@mike-barber
9 жыл бұрын
Pretty good intro to a rather tricky subject. For me, the key is always that loop equation and thinking about when you're turning your nice negative feedback into positive feedback. It might be nice to show how an oscillator works too, actually! But a couple of thoughts on Bode plots... People can get a darn good understanding of this stuff playing around in LTSpice (or equivalent). I don't have the equipment for measuring this stuff either, so I usually land up simulating it first. Then just measure the circuit the a scope at a few different frequencies and check it agrees. I was thinking that it might be quite useful to show how to measure gain and phase with a scope and a sinewave generator, and do the plot manually. For me, I studied this stuff, knew all the maths, and never really understood it until I got my hands dirty and did some experimenting. It's super important, and I'm really glad you're doing a set of videos on it.
@power-max
9 жыл бұрын
This is great, I am in the same situation as you, and not understanding closed feedback control systems is proving to be a challenge. Hopefully I can bet my LTspice 0 through +-15V 0-5A linear supply working at all voltages and current now understanding how latency and gain change at different frequencies! Thanks!
7 жыл бұрын
Very nice explanation! I had to learn these with the math behind, but I must say, it's much easier to understand in the way you explained. And even with that, if someone is interested more, then it's more simple to learn the math later as well.
@edwaggonersr.7446
9 жыл бұрын
I don't understand much of what you say, but I thoroughly enjoy watching your videos. Your enthusiasm is contagious. Will it be possible for me to build this power supply without test equipment and extremely limited knowledge of electronics? I look forward to #14 in the power supply video series.
@joelrebaone9437
8 жыл бұрын
Martin your videos are really helpfull to my proffesion keep it up man we need people like you in EEE Engineering.
@YashKudale
9 жыл бұрын
Requested this kind of opamp stability video all over the internet thanxs for making it and in advance for the next video which I am eagerly waiting for !! Including a real life example of unstable system which then you compensate step by step, this will be really helpful for many guys like me!! ones again Thanx A Lot !! and ya two thumbs up !!
@DeeegerD
9 жыл бұрын
This has taken so long from the start that although I like where this project is going I ended up buying a Kit which reminds me of your circuit. This has been educational though, but some of us need a current regulated power supply before we go in the box ;)
@PapasDino
9 жыл бұрын
Nice lay down Martin, ancient university thoughts of poles, zeroes and transfer functions came rushing back as I watched! 73 - Dino KL0S
@КириллРагузин-р9в
9 жыл бұрын
I personally prefer the Laplace transform approach for I think it is easier to understand correctly that way. It is not hard to figure things out but it can be hard to make sure that you really do understand what you think you understand. The "not applying enough force" analogy is completely misleading here because in a closed loop with an aperiodic chain for a feedback and an amplified error output, gain is definitely not your friend. Low gain only means slow response and higher dynamic error, but it is making the gain too high what makes the system unstable for a given delay. So there is a trade-off between performance and stability. It is important to undarstand that they tend to work against each other and trying to "apply enough force" only makes things worse.
@bobdux
6 жыл бұрын
Do you hapve any plans to finish this project? I just got about half way through the series before realising it isn't complete!
@KX36
9 жыл бұрын
Very good video! Far too many textbooks dive straight into transfer functions, laplace notation, finding roots on Nyquist plots etc. One thing that might be a bit better explained is around 24:00 when you're talking about being too slow and not having enough force causing oscillation, but adding force could potentially change the oscillation to ringing. That seems to be backwards. Being too slow would mean no phase margin, therefore positive feedback, but that wouldn't be a problem if the gain is less than 0dB, depending on how much less than 0dB the gain is (gain margin), that might ring to a degree. Adding more force would give it what it needs to oscillate (have no phase margin or gain margin, therefore oscillate). Another thing you mentioned in the text is that ideally you want a ring of 2 cycles, less than this is overcompensated. Actually, any ringing at all is underdamped. Critical damping has no overshoot or ringing and settles faster than an underdamped or overdamped system and occurs at a phase margin around 76 degrees. The typical target phase margins of "aim for 60 degrees, but make sure its definitely more than 45 degrees" is underdamped, but it's a practical way of working when things like load resistance, input and output voltage, part tolerances etc all affect the loop.
@deanneumann8594
7 жыл бұрын
thank you for this series Martin. This series earned you my patreon support :)
@michaelo2l
8 жыл бұрын
If memory servers, the datasheet use to recommend adding a 10uF tantalum bead between the base and the emitter (directly on the 3055 device) when used in this type of application (power supplies)... Additionally the resistors should be mounted as close to the transistor as possible... even short cables (as you have currently have) may introduce instability... It's worth a try but I had to follow this recommendation on a previous build albeit a very long time ago ;) As regards the diode connected to the zenier, this was used to compensate for temperature drift (of reference voltage), as each diode has a different temperature coefficient, positive and negative ... Mike
@M07RealKing
8 жыл бұрын
there is another important thing , RF interference . and the way to reduce it is to put the circuit board in the project box (which is made out of aluminum or any conductive material ) and connecting the box to the circuit ground . also, EMI filter on the mains input will make a difference .
@BobMellor1954
9 жыл бұрын
Hi Martin, thank you, easy to understand; looking forward to the next video. Regards Bob
@BeetleJuice1980
4 жыл бұрын
i really wanted this project to finish..even without solving the oscillation issue.
@KX36
9 жыл бұрын
By the way, here's a potential use for your Red Pitaya. Get a signal injection transformer (not cheap, but still less than much of your test gear) and use it to make a Bode plot for your power supply under different loads, output voltages, input voltages. Depending on where you probe the 2 relative voltage signals you can measure the whole loop, just the error amp or just the power stage (plant). Do remember that input voltage, load resistance etc will be in the feedback loop, so the Bode plot, step response, stability etc. does change as they do. Unfortunately, I think the Red Pitaya only plots up to about 60kHz and a linear power supply may have more bandwidth than this. You have to have a stable system to measure a Bode plot, but often if it isn't already you can do that by putting a large capacitance across the error amplifier to reduce the bandwidth of the system so low that any poles and zeros are put under 0dB and so the whole system acts as an almost ideal integrator. The step response is aweful in this situation but its only temporary during testing. Once you've measured the plant Bode plots, you can design frequency compensation and then demonstrate the step response to us under different situations.
@RupertHandford
9 жыл бұрын
Great video and clear explanation. Thanks Martin.
@bassreaper7779
8 жыл бұрын
Sure would be nice if this series were to be completed!!! Will that ever happen?
@Haamedtm
5 жыл бұрын
Hi, why did you stop continuing this series?
@AliensInc.
8 жыл бұрын
I wonder if this series is ended? I so was looking forward to this but it feels there are some episodes left.
@Invisible_Tri
5 жыл бұрын
Great explanation..... worth it
@pomptang
6 жыл бұрын
Brilliant explanation! love your videos
@rowifi
9 жыл бұрын
At 15.00, you did some confusing edits. if ab goes to -1 you divide by zero and you get instability. then you state that the minus 1 gives a system gain of 1. I've watched this bit a few times and think there's sometime missing to make this clear.
@robertcalkjr.8325
9 жыл бұрын
Thanks Martin. That is very helpful.
@mrlazda
9 жыл бұрын
Bode plot can be obtained mathematically by plotting transfer function in frequency domain, or experimental as you described.
@mrlazda
9 жыл бұрын
Miloš Lazović here is simple demonstration of op amp loop stability, it could be easily replicated with real components www.linear.com/solutions/4449 you can do bode plot with just signal generator and any oscilloscope and paper (and calculator) but it is hard way
@yaghiyahbrenner8902
9 жыл бұрын
Miloš Lazović thanks, these will come in handy.
@mrlazda
9 жыл бұрын
***** That example is just special case, for more general usage you should use Middlebrook's or Tian's method (as I remember there is LTSpice example for Middlebrook's method).
@nickwagner9339
8 жыл бұрын
learnt lots from these videos, thanks!! but where is #15, dead project? :(
@SatyajitRoy2048
9 жыл бұрын
This would have been more interesting to watch if it was made with some real life testing. You could have made the explanation even simpler. Its more beneficial to have a reference to the real life opamp datasheet and their performances...nowadays datasheets are available for free.
@ThePokercoder
9 жыл бұрын
Hello Martin, I really appreciate your videos and your passion for electronics. keep going. I just wanted to express a doubt on the circuit: the voltage drop on the feedback resistor for the current is added to the zener voltage making the reference voltage unstable but dependent on the output current. Could it be a problem? Would not be better, therefore, that the anode of the zener diode is connected to ground instead of to the non-inverting input of the op-amp? Thank you for your feedback. Forgive my poor English (thanks google translator)
@power-max
7 жыл бұрын
Where is video 15?? I'm really needing to know how to do these measurements! From what I read I stick a 20 ohm resistor and use a isolation transformer, but I don't know what the signal I should be measuring is. Please help?
@aminzafar7171
8 жыл бұрын
hello when you add 15th part sir?
@zero_kelvin_
3 жыл бұрын
very good video
@AxelWerner
9 жыл бұрын
ok.. having a big phase shift in the regulation/feedback loop is a bad thing when its too large and all. but isnt it also a question of time? reaction time i mean. i mean, does it still matters if you just try to "suppress" fast swings/changes (frequencies) within the feedback-loop ? so lets say anything that goes quicker than 10 or 25ms is supposed to be ignored (suppressed by caps , RC or LC filters or somesort in the feedback look), wouldnt that be a good enough regulation if its not "lightning fast" ? big ass caps on the output maybe could help smoothing out the "over and undershoot" too , idnt it ?
@peterpv0001
9 жыл бұрын
Axel Werner Actually phase shift is usually expressed as the time delay a sine-wave of a certain frequency experiences. For example, a 1 kHz sine has a wave of 1 ms, 90 degrees phaseshift would then correspond to 90/360 * 1ms = 0.25 ms. So yes, that is a different way of expressing time delay ! Yes you can insert a very large capacitor in your loop to make it stable, BUT your feedback loop will get much slower ! So it will respond slowly to any changes in the system. So if you connect a load to the powersupply , it will take some time for the output voltage to reach the intended value. Indeed, loop stability is a compromise between required speed aan stability under all circumstances.
@rolfts5762
8 жыл бұрын
(Question: "rai" -going rai,..is it same as 'going wrong' ? ) Thanks for very nice videoes! br, Rolf
@ransomware7964
6 жыл бұрын
Thank you very much
@power-max
7 жыл бұрын
how do you get those bode plots you used for phase margin and gain margin in simulation? If I break the loop and inject a signal, op amps have insane gain (10^6) particulary below 10KHz?
@322doug
8 жыл бұрын
where is #15 ??? Please tell me there is a number 15 , i have been watching and building along with you , somebody , anybody , where is 15 !!!! lol help!!!
@AliensInc.
8 жыл бұрын
+Hammer USN Agree, I too need that rest of this series. It's so perfectly described and so easy to understand. Please let there be more to finish this project.
@mzaphod64
8 жыл бұрын
I don't think he intends to finish it. I might be very wrong but I wouldn't be surprised if he have been compensated for stopping the series.. His Most viewed videos are "DIY" and "How to.." made for noobs like myself. They gather a lot of views and I don't see the reason to stop making it. -On the other hand he majors in solar power gadgets which are very expensive and not many people are into it for practical reasons hence low view count on those I guess and low money in them. But he still makes those?? Sponsored content?
@khangowhar
9 жыл бұрын
Dear Sir, Can you lend me complete schematics of a versatile digital variable power supply 30v 5amp with an input from 90 to 220v @ 50hz
@PhillipS85
8 жыл бұрын
Great video thanks.
@alperenalperen2458
9 жыл бұрын
I like your videos in general. One suggestion though, I think you should stop finding analogy for every case. Go for calculus it is really easy
@amtpdb1
9 жыл бұрын
It would be better if you had examples you could show while you are talking about it. If you had examples it would keep you interested in what you are saying. This could be caused by.... or ........ it might do this..... or that.......! You loose me! Thanks for the video.
@Kangsteri
7 жыл бұрын
Well this series was a good waste of time..:D You didnt even get the current button cleared. Or how to put the limiting potentiometers to series. If you want clean good signal you need to do something to that AC transformer and bridge rectifier. To get the cleanest signal you need to use toroidal transformer and ceramic caps in the bridge rectifier (they need to be with every diode, 103, 203 and 102 ones). They also help if you add them between +/- DC rails. You might also want to test some differend values with the 2200uf ones. This same issue has existed for a very long time in 8bit Nintendos.
@dm-ot1st
7 жыл бұрын
There is something about solving this stability problem posted in an online forum. You may have to register to view the diagrams. www.edaboard.com/thread354838.html
@odddavelars1003
8 жыл бұрын
Most of his current videos seem to be reviews. Could be that he is bored with us newbees. It is kind of thankless job going over the basics. I was looking to a finish, as well as more beginning videos. Maybe I am missing something though.
@MrStricklen
9 жыл бұрын
most amps need tunes don.t make to much out of the small thing just show the fix move on you can't in the real would take in all veryables
Пікірлер: 60