I knew this was a thing, and I knew vaguely how it was supposed to work, but this video took a fuzzy, nonspecific concept to perfectly understanding in seconds. Turning the pot and watching the distortion creep down the waveform and disappear brought it into perfect HD focus instantly for me. Simple, but still very impressive.
I had this exact issue today Alan - with an LM358 too. Thanks yet again for your simple explanations and better still your resolution methods! Very much appreciated.
I remember data books from the early 90s. Manufacturers were so much more prolific with info back then. Essentially forcing class A style operation to get over the crossover distortion associated with class B operation. Nice.
I saw a bench test that showed running this chip Class-A not only reduces the distortion drastically, but actually takes it lower than more favorable chips across a wide range of operating conditions. *With that said, the author concludes the article that there are still much better chips to select from no matter how you run it lol.*
Fantastic. I'm having some issues with this. Didn't know what it was, but now I do, and also how to eliminate it and all thanks to you. You're a star! 😊
I am always impressed by the way that you take the time to explain the small details that we see on your scopes. Very useful and informative. Thank you
Hi Alan Excellent material as always. The trick of applying an external load also works for regulator circuits where the current flow reverses itself. Without it the Regulator can enter regimes where it no longer functions and the output voltage soars.. This can easily damage components.
Thanks Alan, that was a great demonstration. When I have ever built a Push Pull audio o/p stage, I diminished this problem by placing 2 diodes in the collector circuit of the equivalent to Q12. but obviously that isn't possible here.
Sr Alan muito obrigado, estou aprendendo muito com suas informações, infelizmente aqui no Brasil não temos informações de qualidade a educação aqui é deficiente, depois de assistir vários vídeos do sr comecei a estudar o idioma inglês para de fato compreender seus vídeos, obrigado você me incentivou a estudar e fico muito grato, consegui comprar um osciloscópio, obrigado Sr Alan.
Wow! There I thought I knew at least the basic stuff to watch out for with opamps, but I never heard about crossover distortion before. And that is with the LM358 being the main opamp for most projects where you need "just any opamp".
It's a great general purpose op amp, crossover distortion is usually not a big deal unless it has to do with audio. Do not swap a modern opamp in an audio amplifier with a LM358 unless you like fuzzy distortion :P
I'm a new subscriber, and found this video was really excellent! I'm an EE myself (I still have some of those National Semi data books!), and appreciated your use of the scope and the data sheet schematic to clearly explain what can be a pretty subtle circuit problem. Looking forward to browsing through the rest of your postings.
I came across this statement in the book (I have the TI version of it), that I wish you would have put in the video. They specifically call for that resistor when you have more than 50pF capacitance on the output, because they built it to have a powerful mode, and an efficient mode, and that distortion occurs when it doesn't stay in one mode (like when a capacitor is at peak current draw, ie. near zero crossing). Here is an excerpt from the paper, and it serves as a powerful reminder that there is critical info that is not in the tables, charts, and drawings.: "To reduce the power supply current drain, the amplifiers have a class A output stage for small signal levels which converts to class B in a large signal mode. This allows the amplifiers to both source and sink large output currents. Therefore both NPN and PNP external current boost transistors can be used to extend the power capability of the basic amplifiers. The output voltage needs to raise approximately 1 diode drop above ground to bias the on-chip vertical PNP transistor for output current sinking applications. For ac applications, where the load is capacitively coupled to the output of the amplifier, a resistor should be used, from the output of the amplifier to ground to increase the class A bias current and prevent crossover distortion. Where the load is directly coupled, as in dc applications, there is no crossover distortion" ~~ SNOSBT3I Rev.: Dec 2014 Texas Instruments
As always, great video! I love to see when low level concepts like current mirrors can create higher level building blocks like op amps. Keep up the good work!
I just bought some LM358P opamps to experiment for making for rudimentary audio signal mixing and source following in a portable amplifier. It has 19 some odd volts on a single rail LiFe battery pack. I was hoping to just use the single rail power supply, and capacitively couple it to remove the DC bias. In talking with my roommate, he said that there might be issues with my scheme, and then I found your video, which was just the thing I was looking for, fixing a problem that I didn't even know I was going to have--before it happened. Thank!!!
Thanks, Alan; great explanation of crossover distortion! I always appreciate your well-presented videos. Your channel should be required material for any electronics hobbyist.
Wow. Great video. Exactly what I needed to know and well explained without any unnecessary and annoying diversions. High IQ plus good communication skills; a rare combination in online electronics vids.
Having problems with a transistor current source buffer that "kicks back" all the way to the buffered op amps. Seeing your videos helps me to try to learn what the problem is. Much appreciated! More videos! :-)
Thanks, I was baffled by this until I found this video! I was doing an LTSpice simulation and when I used the ideal opamp, all was well. But when I used an LM2902 spice model (from TI) I suddenly got these weird "ears" at the crossover point. I figured out that I needed a DC load to fix it, but I didn't know why, until now!
Dear Sir, First time viewer to your channel. This was an excellent video, I was trying to understand how X over distortion was caused and seeing a view on the scope I now have a understanding of why this happens, Thank you kindly for sharing this...have a good day!!
Hi Alan, I absolutely love your theory/tutorial videos! They are excellent! And thank you for putting the notes in the links, I like using them as I play in the lab trying to reproduce your steps. The notes on this one gives a '404' message... Cheers, Mark
This is great! Thanks! I have this problem with a TL071 op amp handling a square wave. The crossover distortion is short and subtle but I can see it on the oscilloscope. I guess it is short because of the TL071's high slew rate? I am going to try this solution!
+Chip Guy Vids I have about three shelves worth of old data books still and I use the PDFs online myself mostly today. I guess if you want to curl up to a good book at night though the data books are still nice to have? I just downloaded one of the paper books I have in PDF The Nat Semi Linear Applications Handbook.
***** I would not have found it myself except for looking for the book because of this video. I wanted to make sure I had the title right. Because the photo of my bookshelf is a little blurry. i.imgur.com/wFswkcH.jpg
+Paul Frederick I've got a lot of them too - especially the yellow TI books, the white Signetics books, and the blue (and older dark brown) Motorola books. I even have some of the same SAMs and TAB books. None of the microprocessor or Intel books though - I'm an analog guy... ;-)
I recall you mentioning this in an earlier video, I had no idea that this part did this. In examining my National Linear Data Book (purchased at Radio Shack! :-) I *do* see mention of this in the application notes for this part...
Very nice video, I was puzzled by this issue and now I know how to solve it. Another thing has to do with square waves, LM358 doesn't reproduce them as square but as trapezoid instead, I expected a more quick response from this chip!
this is very cool, but is there any way to implement a circuit for a DME pulse pair of. gausian pulse spaced lets say 12us apart. iam trying to design a simple dme. generator for aircraft dme transponder. i cansend you info theory about that i think it is a cool project.
It really depends on where your output voltage is going to go. You want the output of the op amp to always source current. So, if the output voltage is going to go below ground, then you want the pulldown to go to VEE.
Awesome vid as always. Just to clarify.. since the input is a full sine wave, the opamp ..internally.. still performs the full transition swing. however relative to the load, it only drives the positive swing and never sinks through it since the load will sink through the parallel load resisters. Since the opamp does not sink, does the internal transition still take take place ? The opamp sink is still part of the parallel circuit and once its resistance on negative swing becomes low enough it will conduct more than the load sink resisters, but by that time the noise will be a non factor ?
+Tim Copeland With the DC load resistor, the NPN output device operates as Class A - it is conducting through the entire cycle - the PNP device is never turned on. It is just that it is pushing less current into the DC load at the trough of the waveform, where the AC coupled load is contributing more current to the DC load. The PNP device never turns on because the collector voltage of Q12 is always biasing the NPN output devices.
+w2aew Now I think I see .. by loading the output just enough you change the feedback dynamics so as the only way the opamp can maintain equilibrium to its inputs, is by continuing to conduct through the entire swing.
I don't think you see. It's simpler than that. NPN can only push, PNP can only pull. When you have bare capacitor on the output and you are sending a push-pull signal (sine wave), NPN pushes, PNP pulls. With the distortion inbetween. So you put a resistor on the output to pull instead of the PNP. On opamp side of the capacitor, the voltage varies (push-pull) somewhere btween 0-5V. The other side of the cap is on 0 V without signal. Pushing current through the cap and load resistor makes load res. voltage positive, pulling the current makes load voltage negative. Capacitors can have different levels of DC voltage on sides but they pass vibrations of that voltage through as a current. And you can convert those current vibrations to your prefered voltage by resistor (a resistor is current to voltage converter and vice versa).
I was expecting to see a little crossover distortion, but the amount surprised me. This begs the question as to why bother having the PNP output transistor at all, if folk have to bias the output current to such an extent that the transistor is never used. Surely the designers must have foreseen this problem, yet they took no steps to avoid it. At first I thought that the load might be too high, but a 1k resistor is hardly excessive. I realise that the LM358 is not the newest kid on the block, but did the state of the art at the time prevent the incorporation of circuitry to circumvent this issue? (For example, as per your video on Vbe multipliers or a couple of diodes or some such alternative design to eliminate or at least reduce the dead band). I've used many OP amps in the past, but have never really paid more than a passing glimpse at what's inside them, but I'll pay more attention in the future. Your videos are excellent, by the way - keep up the good work.
Many applications operate at very low speeds, and the cross-over doesn't raise a problem. The dead band is only an issue when the output has to switch between sourcing and sinking current quickly.
Trying to do it the lazy way on NI Multisim, I get a little crossover but not as much. Probably a silly question, why did you use a analog scope instead of a digital. Great video's as always Alan, I will try it tomorrow in really life. Back to Multisim to try and replicate your results.
+T Komoski There was no particular reason that I used the analog scope, other than I love using them - and I didn't need any of the features of the digital for this video.
Generally, I like your videos very much - very informative and enlightening. On this specific topic, I get what it is, how it looks on the scope, and what is causing it. But as a total n00b I would greatly appreciate, if you could at least update the description and put in why it is bad (or at least undesirable), or any other effects on the rest of the circuit. Is it most common in RF circuits or power supplies, or where?
When using an amplifier on a signal, any type of distortion is undesirable, because it alters the shape (and thus, the frequency content, purity, etc.) of the signal.
My transistor guitar amp is hissing at every note played, but silence when im not playing anything, can anyone help suggest a solution? This is way too technical for me to understand, thank you
Is crossover distortion applicable to high speed current controlled op-amps in case when you use them as voltage buffer + 50 ohm / AC coupled output ? My educated guess is that line between output of op-amp and AC coupling cap is a transmission line, and it should have 50 ohm impedance, so we can say that we have 50 ohm load on output of op-amp, so crossover distortion should be a case there.
Devices designed to drive high speed 50 ohm loads will usually not have crossover distortion because the output stage would be specifically designed for those loads.
Alan, thanks for this video. Here is a (probably very dumb) beginner's question: why does the op amp need to sink current at all? Since the output can swing only between the supply rails (+5V and GND) and the load is connected to GND, the output never has a negative voltage, i.e. in my understanding, the op amp would always source current.
isn't that related to amplifier class? you had a class B output (one transistor on per half cycle) and you made it a class A by never opening the bottom half.
+nraynaud1 Yes, absolutely. Without the DC pulldown resistor, each of the two output transistors is operating Class B. As I added the output resistor and starting lowering it, the NPN went into Class AB, and the PNP into Class C, until eventually the NPN was operating Class A (and the PNP was off).
I found some very old KA3403 (Gold Star or Samsung brand, I think) op amps in a "surplus surprise box" from a well-known electronics supply store on the west coast of the USA. The KA3403 has four op amps on one chip. The outputs of all four op amps are class AB. No crossover distortion! Nice. I can't seem to find other op amps that have class AB outputs except for some expensive, "audio grade", Burr-Brown op amps. I wish more op amps had a class AB output.
@@w2aew Hi, again! I had the same problem with a TL072 op amp. I used your solution to get the crossover distortion below the bottom peak of the sine wave. Question: Is there any downside to this approach? Is forcing the upper, NPN, output transistor into class A mode bad in any way? Is it inefficient? Is too much heat generated?
It uses a little more current. But provided that you don’t exceed the maximum output current rating, and you can still drive your load and achieve the voltages you need at the output, all is well.
I've been thinking on this... Could we change the feedback resistor values, in an amplifying stage, to draw more current and have the same efect? I think that sounds right. In the follower example that you use the inverting input isn't connected to ground as it would be in an amplifying stage. (I'm still getting the hang of this stuff, so apologies if that's a daft question).
You just have to think about the *direction* of the current provided by the output. The distortion occurs when switching between sourcing and sinking current. If the feedback arrangement and signal levels will always keep the output sourcing current, then you'll have no crossover. Otherwise, a simply pulldown resistor at the output will suffice.
@@w2aew Just tested this in the lab. Works as you suggest. I tried it with a different op amp (5532), I get a x-over blip and can't move it, perhaps a different output stage.
Hi Mr. Cloudman! I love to watch your vids, your visualisations are always fun to see and easy to learn and remember. Nice Tek 485, it looks like new. I wanted to ask about delayed sweep functionality, but I just saw you already made a video about that. Would you consider tearing it down and explaining basic principles of operation, such as how is delayed sweep realized? Greetings, Nino
+nTOOBEa I've done a few videos that showed delayed sweep - the very first video, and the one that shows characterization of an RF power detector. An in depth video on how it works might be fun - I'll think about that.
By looking at the output, I understand that lm358 has a current limit when the output is sourcing current, but none when it is sinking current? In order to limit the current, it has to flow through Rsc, am I right? Sorry for my broken English :)
+MrJohhhnnnyyy Yes, for sourcing current, it goes through Rsc, and then when the voltage drop gets large enough, it turns on Q7 which starts to steal bias current away from Q5/6.
Thank you Alan! Have you ever thought about doing any videos on Delta-Sigma or Sigma-Delta? I know it's probably a little bit off base here but it is a very interesting topic.
Datasheet no longer has the nice photo's of the dies and a drawing of the terminals for the bare dies, as used in the heyday of hybrid manufacture. It also shows as an aside why the output stage is protected against shorts to ground, with the transistor and sense resistor, but that it will blow up immediately with a short to the supply rail. The current sources are also not shown, the sink and some of the sources are simply a pinched off FET, not too accurate but if you only need a low current very usable. A diffused resistor of high value would take up most of the die. the more precision ones use a few transistors, but still have the pinched off FET as a startup, and all the ones referenced from the positive rail actually are a single current mirror using scaled transistors to determine the current. That is why the simple opamp tracks well over temperature, as all currents inside are generally used in ratios and rarely as absolute values.
+SeanBZA Some of the other databooks I have do show the more detailed schematics, including the current source structures as you described. I chose to show this one, because that level of detail was irrelevant to the video topic so I didn't want to show the unnecessary complexity.
All designs have tradeoffs based upon the available technology and design goals. In this case, the output stage is low power, operates over a wide supply range, and achieves output swing to near ground (lower supply rail). For many applications, the tradeoff with having some crossover distortion is not an issue. The datasheet discusses this, and it is easily overcome for those applications where it is important.
It should be better titled as improper LM358 application specific biasing requirement. It is stated in the databook to install a resistor for capacitor coupled load so as to increase the internal bias requirement.
+Mike Shekhtman Thankfully, most op amps these days don't suffer from this - but the only way to be sure is to check the datasheet. The text description in the datasheet mentions this.
+w2aew You certainly stacked the deck with the 358. It is a low performing chip these days. A signal generator kit I got came with one and in the comments everyone said replace it with a TL072. Once I compared the data sheets of the two parts I did myself. A fun toy. Have you played with one of these? goo.gl/SJMCED I'd like to see you do a video about it.
I know these videos consume a lot of your available time, but you don't know how much we appreciate them Alan. Thanks! 73
+Don Coltrane It's the good, positive feedback that keeps me motivated.
+w2aew Alan outputs infinity gain
+Don Coltrane I Very Agree!
I knew this was a thing, and I knew vaguely how it was supposed to work, but this video took a fuzzy, nonspecific concept to perfectly understanding in seconds. Turning the pot and watching the distortion creep down the waveform and disappear brought it into perfect HD focus instantly for me. Simple, but still very impressive.
I had this exact issue today Alan - with an LM358 too. Thanks yet again for your simple explanations and better still your resolution methods! Very much appreciated.
I remember data books from the early 90s. Manufacturers were so much more prolific with info back then. Essentially forcing class A style operation to get over the crossover distortion associated with class B operation. Nice.
I saw a bench test that showed running this chip Class-A not only reduces the distortion drastically, but actually takes it lower than more favorable chips across a wide range of operating conditions.
*With that said, the author concludes the article that there are still much better chips to select from no matter how you run it lol.*
Fantastic. I'm having some issues with this. Didn't know what it was, but now I do, and also how to eliminate it and all thanks to you.
You're a star! 😊
Thanks!
Best guides/explanations on the internet, always easy to understand... nice!
I am always impressed by the way that you take the time to explain the small details that we see on your scopes. Very useful and informative. Thank you
Hi Alan Excellent material as always. The trick of applying an external load also works for regulator circuits where the current flow reverses itself. Without it the Regulator can enter regimes where it no longer functions and the output voltage soars.. This can easily damage components.
Yes, you're right. Some regulator designs even have a minimum load current requirement.
Simply amazing, Alan. You make electronics so simple and easy to understand much much better than teachers!
Thanks Alan, that was a great demonstration.
When I have ever built a Push Pull audio o/p stage, I diminished this problem by placing 2 diodes in the collector circuit of the equivalent to Q12. but obviously that isn't possible here.
Sr Alan muito obrigado, estou aprendendo muito com suas informações, infelizmente aqui no Brasil não temos informações de qualidade a educação aqui é deficiente, depois de assistir vários vídeos do sr comecei a estudar o idioma inglês para de fato compreender seus vídeos, obrigado você me incentivou a estudar e fico muito grato, consegui comprar um osciloscópio, obrigado Sr Alan.
Wow! There I thought I knew at least the basic stuff to watch out for with opamps, but I never heard about crossover distortion before. And that is with the LM358 being the main opamp for most projects where you need "just any opamp".
It's a great general purpose op amp, crossover distortion is usually not a big deal unless it has to do with audio. Do not swap a modern opamp in an audio amplifier with a LM358 unless you like fuzzy distortion :P
I'm a new subscriber, and found this video was really excellent! I'm an EE myself (I still have some of those National Semi data books!), and appreciated your use of the scope and the data sheet schematic to clearly explain what can be a pretty subtle circuit problem. Looking forward to browsing through the rest of your postings.
Thanks! I hope you enjoy some of the other videos. There is a complete index available, link on the title graphic.
I've been hunting for this video. I watched a good while ago and wanted to remove some xover from a circuit. Thank you.
I came across this statement in the book (I have the TI version of it), that I wish you would have put in the video. They specifically call for that resistor when you have more than 50pF capacitance on the output, because they built it to have a powerful mode, and an efficient mode, and that distortion occurs when it doesn't stay in one mode (like when a capacitor is at peak current draw, ie. near zero crossing). Here is an excerpt from the paper, and it serves as a powerful reminder that there is critical info that is not in the tables, charts, and drawings.:
"To reduce the power supply current drain, the amplifiers have a class A output stage for small signal levels which
converts to class B in a large signal mode. This allows the amplifiers to both source and sink large output
currents. Therefore both NPN and PNP external current boost transistors can be used to extend the power
capability of the basic amplifiers. The output voltage needs to raise approximately 1 diode drop above ground to
bias the on-chip vertical PNP transistor for output current sinking applications.
For ac applications, where the load is capacitively coupled to the output of the amplifier, a resistor should be
used, from the output of the amplifier to ground to increase the class A bias current and prevent crossover
distortion. Where the load is directly coupled, as in dc applications, there is no crossover distortion" ~~ SNOSBT3I Rev.: Dec 2014 Texas Instruments
Great video as usual. Please keep up the great work as the sponge upstairs wants to soak up some more.
This guy makes the best videos!! Thank you so much for taking time to make these!
As always, great video! I love to see when low level concepts like current mirrors can create higher level building blocks like op amps. Keep up the good work!
I just bought some LM358P opamps to experiment for making for rudimentary audio signal mixing and source following in a portable amplifier. It has 19 some odd volts on a single rail LiFe battery pack. I was hoping to just use the single rail power supply, and capacitively couple it to remove the DC bias.
In talking with my roommate, he said that there might be issues with my scheme, and then I found your video, which was just the thing I was looking for, fixing a problem that I didn't even know I was going to have--before it happened.
Thank!!!
Yes, ensuring a DC pulldown on the output, and AC coupling in/out for your audio signals, should all work well. Have fun!
The Most important electronics channel ever!
I know this video is old, but it was really useful understanding this issue. Thanks for uploading this!
Thanks Alan. Great lesson as always.
Thanks, Alan; great explanation of crossover distortion! I always appreciate your well-presented videos. Your channel should be required material for any electronics hobbyist.
Thank you, this was very much informative.
Wow. Great video. Exactly what I needed to know and well explained without any unnecessary and annoying diversions.
High IQ plus good communication skills; a rare combination in online electronics vids.
Having problems with a transistor current source buffer that "kicks back" all the way to the buffered op amps. Seeing your videos helps me to try to learn what the problem is. Much appreciated! More videos! :-)
Hi Alan. Op-amp crossover distortion very nicely explained. Thanks.
thank you for the very educative videos
What can I add that has not been said already? Thank You for the video and it was another great one, very informative.
Wonderfully done video and great explanation of cross over distortion Alan. I enjoyed it very much.
So that's what that was! I was beginning to think I had bad LM385 chips. Thanks.
Thanks, I was baffled by this until I found this video! I was doing an LTSpice simulation and when I used the ideal opamp, all was well. But when I used an LM2902 spice model (from TI) I suddenly got these weird "ears" at the crossover point. I figured out that I needed a DC load to fix it, but I didn't know why, until now!
Your videos are great...thanks for taking the time and always finding interesting subjects to explain.
Thanks a lot for this video. I'm not an engineering student, I'm studying physics in fact, but this helped me with my lab report so many thanks
I still have that exact data book on my shelf. Great video and explanation as always!
Another absolutely brilliant video. Huge fan.
Best guides on youtube. Thank you very much!
Wonderful Alan , great to learn this I am really thankful for your efforts to teach us with simple effective
practical set up,
Dear Sir, First time viewer to your channel. This was an excellent video, I was trying to understand how X over distortion was caused and seeing a view on the scope I now have a understanding of why this happens,
Thank you kindly for sharing this...have a good day!!
Hi Alan,
I absolutely love your theory/tutorial videos! They are excellent!
And thank you for putting the notes in the links, I like using them as I play in the lab trying to reproduce your steps.
The notes on this one gives a '404' message...
Cheers,
Mark
Should be fixed now. Here is the updated link:
www.qsl.net/w2aew/youtube/crossoverdistortion.pdf
Hence the difference between class A and class B output topology.
Very educational.
Thanks for yet another informative video. Love it when you use analog scopes, too!
Spectacular, very neat tutorial and example. Thank you.
Great video Alan. I wish they still printed ic data books, so handy. Thanks.
This is great! Thanks! I have this problem with a TL071 op amp handling a square wave. The crossover distortion is short and subtle but I can see it on the oscilloscope. I guess it is short because of the TL071's high slew rate? I am going to try this solution!
Thanks as usual Alan, great way to start the weekend by learning something new! 73 - Dino KL0S
Great video as always. As usual I learned something new and practical. Thanks so very much.
thanks for sharing. Do you know of any audio amplifier diagram, achievable by anyone, that has the best harmonic and crossover distortion values?
Nicely done. I like that you still have the real data BOOK, not just PDFs ;)
+Chip Guy Vids I still have a small selection of paper databooks - mostly analog stuff.
+Chip Guy Vids I have about three shelves worth of old data books still and I use the PDFs online myself mostly today. I guess if you want to curl up to a good book at night though the data books are still nice to have? I just downloaded one of the paper books I have in PDF The Nat Semi Linear Applications Handbook.
+Paul Frederick Thanks for the tip on the handbook PDF...
*****
I would not have found it myself except for looking for the book because of this video. I wanted to make sure I had the title right. Because the photo of my bookshelf is a little blurry. i.imgur.com/wFswkcH.jpg
+Paul Frederick I've got a lot of them too - especially the yellow TI books, the white Signetics books, and the blue (and older dark brown) Motorola books. I even have some of the same SAMs and TAB books. None of the microprocessor or Intel books though - I'm an analog guy... ;-)
Love your show...tech dragon here.
Nice video! It would be nice to see the effect of changes in gain and changes in frequency on the "fix" for crossover distortion.
As long as you keep the output stage sourcing current, the cross-over distortion won't be dependent on gain or frequency.
I recall you mentioning this in an earlier video, I had no idea that this part did this. In examining my National Linear Data Book (purchased at Radio Shack! :-) I *do* see mention of this in the application notes for this part...
Another excellent video Alan. Thanks.
This is great. I was stuck for the last so many days. excellent explanation. thank you so much.
Good explanation on this matter. I learn a lot from you. Thanks.
Alliteration: Awesome as always Alan.
Excellent video Alan,very educative and informative,straight to the point..
Do you plan to make tutorial about active filter design?
+alex nnn It's on my list of to-do's... Big topic, though....
Thank you! Very well explained.
Wow! Such a good demonstration. Thank you!
Thanks Allan for another great video. I liked your QSO Today podcast. Keep up the great work.
Thank you!
Very nice video, I was puzzled by this issue and now I know how to solve it. Another thing has to do with square waves, LM358 doesn't reproduce them as square but as trapezoid instead, I expected a more quick response from this chip!
You're probably seeing the "slew rate" limit of the chip. I have a video on that too...
ua-cam.com/video/UooUGC7tNRg/v-deo.html
***** I will check that... Thanks.
Please make more videos like this. Also how to analyse curcuits and read schematics
this is very cool, but is there any way to implement a circuit for a DME pulse pair of. gausian pulse spaced lets say 12us apart. iam trying to design a simple dme. generator for aircraft dme transponder. i cansend you info theory about that i think it is a cool project.
Would you sink the resistor to Vee when using a split rail supply? Or would it still be ground? Awesome video!
It really depends on where your output voltage is going to go. You want the output of the op amp to always source current. So, if the output voltage is going to go below ground, then you want the pulldown to go to VEE.
Love these videos. Keep em coming.. excellent explanation as always!
Awesome vid as always. Just to clarify.. since the input is a full sine wave, the opamp ..internally.. still performs the full transition swing. however relative to the load, it only drives the positive swing and never sinks through it since the load will sink through the parallel load resisters. Since the opamp does not sink, does the internal transition still take take place ? The opamp sink is still part of the parallel circuit and once its resistance on negative swing becomes low enough it will conduct more than the load sink resisters, but by that time the noise will be a non factor ?
+Tim Copeland With the DC load resistor, the NPN output device operates as Class A - it is conducting through the entire cycle - the PNP device is never turned on. It is just that it is pushing less current into the DC load at the trough of the waveform, where the AC coupled load is contributing more current to the DC load. The PNP device never turns on because the collector voltage of Q12 is always biasing the NPN output devices.
+w2aew Now I think I see .. by loading the output just enough you change the feedback dynamics so as the only way the opamp can maintain equilibrium to its inputs, is by continuing to conduct through the entire swing.
I don't think you see. It's simpler than that. NPN can only push, PNP can only pull. When you have bare capacitor on the output and you are sending a push-pull signal (sine wave), NPN pushes, PNP pulls. With the distortion inbetween. So you put a resistor on the output to pull instead of the PNP. On opamp side of the capacitor, the voltage varies (push-pull) somewhere btween 0-5V. The other side of the cap is on 0 V without signal. Pushing current through the cap and load resistor makes load res. voltage positive, pulling the current makes load voltage negative. Capacitors can have different levels of DC voltage on sides but they pass vibrations of that voltage through as a current. And you can convert those current vibrations to your prefered voltage by resistor (a resistor is current to voltage converter and vice versa).
Another GREAT video. Watched it twice. THANKS!!
I was expecting to see a little crossover distortion, but the amount surprised me. This begs the question as to why bother having the PNP output transistor at all, if folk have to bias the output current to such an extent that the transistor is never used. Surely the designers must have foreseen this problem, yet they took no steps to avoid it. At first I thought that the load might be too high, but a 1k resistor is hardly excessive. I realise that the LM358 is not the newest kid on the block, but did the state of the art at the time prevent the incorporation of circuitry to circumvent this issue? (For example, as per your video on Vbe multipliers or a couple of diodes or some such alternative design to eliminate or at least reduce the dead band).
I've used many OP amps in the past, but have never really paid more than a passing glimpse at what's inside them, but I'll pay more attention in the future. Your videos are excellent, by the way - keep up the good work.
Many applications operate at very low speeds, and the cross-over doesn't raise a problem. The dead band is only an issue when the output has to switch between sourcing and sinking current quickly.
you are a fantastic teacher, thank you!
Trying to do it the lazy way on NI Multisim, I get a little crossover but not as much. Probably a silly question, why did you use a analog scope instead of a digital. Great video's as always Alan, I will try it tomorrow in really life. Back to Multisim to try and replicate your results.
+T Komoski There was no particular reason that I used the analog scope, other than I love using them - and I didn't need any of the features of the digital for this video.
Really good video, Alan.
thanks Alan, another clear video explaining something that could have taken me all day to find out :) ps, love the analouge oscilloscope..
Generally, I like your videos very much - very informative and enlightening. On this specific topic, I get what it is, how it looks on the scope, and what is causing it. But as a total n00b I would greatly appreciate, if you could at least update the description and put in why it is bad (or at least undesirable), or any other effects on the rest of the circuit. Is it most common in RF circuits or power supplies, or where?
When using an amplifier on a signal, any type of distortion is undesirable, because it alters the shape (and thus, the frequency content, purity, etc.) of the signal.
My transistor guitar amp is hissing at every note played, but silence when im not playing anything, can anyone help suggest a solution? This is way too technical for me to understand, thank you
Is crossover distortion applicable to high speed current controlled op-amps in case when you use them as voltage buffer + 50 ohm / AC coupled output ?
My educated guess is that line between output of op-amp and AC coupling cap is a transmission line, and it should have 50 ohm impedance, so we can say that we have 50 ohm load on output of op-amp, so crossover distortion should be a case there.
Devices designed to drive high speed 50 ohm loads will usually not have crossover distortion because the output stage would be specifically designed for those loads.
Alan, thanks for this video. Here is a (probably very dumb) beginner's question: why does the op amp need to sink current at all? Since the output can swing only between the supply rails (+5V and GND) and the load is connected to GND, the output never has a negative voltage, i.e. in my understanding, the op amp would always source current.
The answer is, not all loads are DC coupled or are ground referenced. Each of these situations can lead to the need to source current.
isn't that related to amplifier class? you had a class B output (one transistor on per half cycle) and you made it a class A by never opening the bottom half.
+nraynaud1 Yes, absolutely. Without the DC pulldown resistor, each of the two output transistors is operating Class B. As I added the output resistor and starting lowering it, the NPN went into Class AB, and the PNP into Class C, until eventually the NPN was operating Class A (and the PNP was off).
I found some very old KA3403 (Gold Star or Samsung brand, I think) op amps in a "surplus surprise box" from a well-known electronics supply store on the west coast of the USA. The KA3403 has four op amps on one chip. The outputs of all four op amps are class AB. No crossover distortion! Nice. I can't seem to find other op amps that have class AB outputs except for some expensive, "audio grade", Burr-Brown op amps. I wish more op amps had a class AB output.
@@w2aew Hi, again! I had the same problem with a TL072 op amp. I used your solution to get the crossover distortion below the bottom peak of the sine wave. Question: Is there any downside to this approach? Is forcing the upper, NPN, output transistor into class A mode bad in any way? Is it inefficient? Is too much heat generated?
It uses a little more current. But provided that you don’t exceed the maximum output current rating, and you can still drive your load and achieve the voltages you need at the output, all is well.
@@brockwood5208 Try LMV324/358/321, or for slightly less noise LMV324A/358A/321A.
I've been thinking on this...
Could we change the feedback resistor values, in an amplifying stage, to draw more current and have the same efect? I think that sounds right. In the follower example that you use the inverting input isn't connected to ground as it would be in an amplifying stage. (I'm still getting the hang of this stuff, so apologies if that's a daft question).
You just have to think about the *direction* of the current provided by the output. The distortion occurs when switching between sourcing and sinking current. If the feedback arrangement and signal levels will always keep the output sourcing current, then you'll have no crossover. Otherwise, a simply pulldown resistor at the output will suffice.
@@w2aew Just tested this in the lab. Works as you suggest. I tried it with a different op amp (5532), I get a x-over blip and can't move it, perhaps a different output stage.
FANTASTIC STUFF KEEP THEM COMING!!
excellent video, more video about op-amp please. thanks.
Hi Mr. Cloudman!
I love to watch your vids, your visualisations are always fun to see and easy to learn and remember.
Nice Tek 485, it looks like new. I wanted to ask about delayed sweep functionality, but I just saw you already made a video about that.
Would you consider tearing it down and explaining basic principles of operation, such as how is delayed sweep realized?
Greetings,
Nino
+nTOOBEa I've done a few videos that showed delayed sweep - the very first video, and the one that shows characterization of an RF power detector. An in depth video on how it works might be fun - I'll think about that.
Woohoo, this is the fix I've been looking for! Thank you!
Awesome video. Thanks a lot. And that 485....
Thank you for the useful lecture .
By looking at the output, I understand that lm358 has a current limit when the output is sourcing current, but none when it is sinking current? In order to limit the current, it has to flow through Rsc, am I right? Sorry for my broken English :)
+MrJohhhnnnyyy Yes, for sourcing current, it goes through Rsc, and then when the voltage drop gets large enough, it turns on Q7 which starts to steal bias current away from Q5/6.
Thank you Alan! Have you ever thought about doing any videos on Delta-Sigma or Sigma-Delta? I know it's probably a little bit off base here but it is a very interesting topic.
+mopar3502001 These topics are on my long "to-do" list...
If I wanted to increase the crossover distortion how would I go about doing that? Add a second lm358 following the first?
Yes, if you operate at a lower amplitude, then the relative distortion will be worse.
Fantastic. Your videos are great.
Excellent video! Thanks.
Datasheet no longer has the nice photo's of the dies and a drawing of the terminals for the bare dies, as used in the heyday of hybrid manufacture. It also shows as an aside why the output stage is protected against shorts to ground, with the transistor and sense resistor, but that it will blow up immediately with a short to the supply rail.
The current sources are also not shown, the sink and some of the sources are simply a pinched off FET, not too accurate but if you only need a low current very usable. A diffused resistor of high value would take up most of the die. the more precision ones use a few transistors, but still have the pinched off FET as a startup, and all the ones referenced from the positive rail actually are a single current mirror using scaled transistors to determine the current. That is why the simple opamp tracks well over temperature, as all currents inside are generally used in ratios and rarely as absolute values.
+SeanBZA Some of the other databooks I have do show the more detailed schematics, including the current source structures as you described. I chose to show this one, because that level of detail was irrelevant to the video topic so I didn't want to show the unnecessary complexity.
Two questions come to mind. 1. Why bother designing an op amp like this and, 2. I assume not all are like this. Very enlightening.
All designs have tradeoffs based upon the available technology and design goals. In this case, the output stage is low power, operates over a wide supply range, and achieves output swing to near ground (lower supply rail). For many applications, the tradeoff with having some crossover distortion is not an issue. The datasheet discusses this, and it is easily overcome for those applications where it is important.
It should be better titled as improper LM358 application specific biasing requirement. It is stated in the databook to install a resistor for capacitor coupled load so as to increase the internal bias requirement.
wow! that was a fascinating and important. topic i didnt knew at all. is This important Trick mentioned in the datasheet/App Notes by the vendors?
It is discussed in the datasheet for the LM358
Very informative video.
Great video thanks
How can I tell which opamp is susceptible to crossover distortion?
+Mike Shekhtman Thankfully, most op amps these days don't suffer from this - but the only way to be sure is to check the datasheet. The text description in the datasheet mentions this.
+w2aew You certainly stacked the deck with the 358. It is a low performing chip these days. A signal generator kit I got came with one and in the comments everyone said replace it with a TL072. Once I compared the data sheets of the two parts I did myself. A fun toy. Have you played with one of these? goo.gl/SJMCED I'd like to see you do a video about it.
Would the distortion start to appear again if you were to increase the input frequency ?
As long as the output stage remains in class A mode, then no, the distortion from crossover would not reappear
Enjoyed that, thanks.