EEVblog 1609 - Composite Amplifier Tutorial + Practical Demo

[@digitalzoul57]

Just a moment for appreciation Dave I started watching you 10 yeas ago , now I have a Bsc in industrial electronics Master in electrical engineering, and I am the head of the firmware development in my company. Just I want to say thank you 😃

[@EEVblog]

Awesome to hear, that made my day, thanks.

[@Morbuto]

I’ll join in the chorus… I started watching around the time you were in double digit episodes, watched everything, and retrained myself from SE to also knowing electronic product design. You have been the single most important teacher I had in that aside from datasheets and app notes 😁 Now I’m Head of “Product Development” for an interesting company developing connected sensors for improving insulation. I’m really glad to see you’re doing more of these classic educational videos again, hopefully another generation will learn from them (and I sure still am too)!

[@ppdan]

Dave was one of the guys that revived my passion for electronics (most credits go to him). At 50 years of age I have decided to finally finish my bachelor in electronics and also switched jobs. Went from repairing trains (mainly electric traction) to tester/faultfinder in a company that produces secondary radar systems (civil and military) with possibility to become test/installations engineer when I finish my studies.

[@rimmersbryggeri]

This type of video is my absolute favourite on this channel. It's a great time to live in to have such a variety of highly skilled people teaching useful skills with no paywall.

[@christhirion9474]

Congratulations as a retired industrial electronics engineer that spen 36 years in industrial instrumentation, automation and control you got a great experience ahead of you

[@ashleyzinyk399]

I love the fact that the previous episode in the series was published 10 years ago.

[@EEVblog]

I honestly didn't know that when I shot this, could have sworn it was like 4-5 years at most.

[@TheBodgybrothers]

Interestingly, i unsubbed a few years ago due to the busted videos. And this popped into my feed. Perfect algorithm.

[@HazeAnderson]

@@TheBodgybrothers clearly this series was produced for you! 😄

[@flymypg]

I remember when my first analog electronics course in university covered Darlington pairs, then later covered ideal and real opamps. We were a little confused when a subsequent class started with a Darlington pair drawn on the board. Our instructor replaced the transistors with opamps, then asked us to "discuss the implications" (our professor's favorite phrase). We discussed not only buffer amps, then went on to add feedback to make logarithmic amps, followed by homework that had us modeling the whole thing in Spice, down to the internal transistors. In the lab that accompanied the course, the corresponding project started with a box of "random" opamps and transistors (along with a standard kit of resistors, caps and inductors), then tasked us to make circuits meeting particular specifications for gain, power, bandwidth and noise using only the parts in the box. That really opened our eyes to the power of working with "what's at hand" when meeting a need, and, more importantly, realizing when those parts CAN'T do the job. Many of the specs had multiple successful implementation paths using those parts, and the "winner" was based on minimal parts count. The only thing missing was having prices on the parts, to force us to decide which made more sense to take to production. But the professor did call it a "hands-on theory course", so prices would have been a "distraction" as they don't map to theory!

[@SoonerRoadie]

That sounds like a great course and a great professor.

[@johnwest7993]

@@SoonerRoadie, ditto. I wish I'd had that class and prof. But now I have Dave, and I can't argue about the cost of tuition.

[@urugulu1656]

that must have been donkeys years ago. hadnt heard or such classes any recent time and i HAD a seriously old fashioned prof at Electronics 😮

[@flymypg]

@@urugulu1656 Yup! It was 1982, at UC San Diego.

[@flymypg]

@@SoonerRoadie It was intended to be tough, to weed out folks lacking math skills and/or lacking the ability to rapidly grasp and apply theory. I would have totally failed had I not just come from being a US Navy technician, where I understood all the fundamentals and had good lab skills, needing only the theory behind them and the math.

[@burstofsanity]

Long, long ago, when I took electronics in college, we had to design a low noise, very high gain circuit for one of our last projects. Our teacher loved to make assumptions about ideal Op Amps when explaining things in class and encouraged us to use them when appropriate when designing circuits. I felt pretty smart coming up with this same Idea using 3 amplifier circuits in series treated as an ideal Op Amp to then use to get the results I was aiming for with a fraction of the math to get the design to perform as expected.

[@mortenhattesen]

Great video. It does often cause instability (oscillation) which has to be compensated externally. It would be great to see a video explaining different ways to stabilize amplifier circuits and how to calculate component values (resistors and capacitors) for the compensation network.

[@hugofrisk1889]

Yes please! I've never understood how to do the compensation

[@SianaGearz]

@@hugofrisk1889 Uhm by lowpassing the feedback to attenuate the oscillation frequency?

[@esven9263]

Old comment but putting a low pass filter in the feedback path would make the problem far worse. Often the solution is the opposite, you use dominant pole compensation on the inner Amp to artificially reduce its gbwp until the system has adequate phase margin. Miller compensation is probably the approach I see most often.

[@mortenhattesen]

@@esven9263 I didn't mention "low pass", I wrote "compensation". A lead ompensation circuit (capacitor in parallel with feedback resistor) will improve stability. www.ti.com/lit/an/sloa020a/sloa020a.pdf

[@worroSfOretsevraH]

Dave, you are amongst the very few who dont spread the ads plague. It's rotting the whole YT from the inside out. When I hear: "and our sponsor today PCB-YaaaY" I want to throw my PC out of the window. Huge respect!

[@EEVblog]

Thanks. Turned them all down for 14 years. Many requests a week.

[@davidrick959]

I was pleased to see a HP 41C calculator make a cameo in this video, since there's still one on my desk. (But then it turned into a Sharp 😯Sacrilege!) This composite amplifier trick was taught to me by a Hewlett-Packard colleague about 35 years ago. There are excellent op amps today with both precision and a decent amount of bandwidth; back then, one often had to be more clever to achieve one's performance goals. Sometimes it takes frequency-dependent feedback paths to keep everything stable.

[@tomasbergh]

Finally Dave found the way back to the topic that everyone loves! Dave you are so welcome back! Thanks!

[@johncoops6897]

Somehow I feel that he will do an "Oh A Squirrel" 🐿 and jump back to his recent (lost) ways.

[@briantb5550]

One of the best electronics channel on KZbin, thanks Dave

[@adam3141]

I love all the different calculators on the white board between the edits. Great video. Bug thumbs up from me

[@guillaume8437]

It's so great to have this channel that teaches you things no one else talks about!❤

[@will742]

Thank you for bringing this series back. Commenting for the algorithm.

[@PrincipalAudio]

Dave, these fundamentals videos are the absolute best! So much better than trying to decipher textbooks. Thanks so much for uploading these. I'm learning tonnes of things I didn't know, and filling in gaps in things I did know. Really appreciate your time.

[@ViniciusMiguel1988]

That’s the kind of video I mostly miss from this channel! I hope you keep doing them Dave!

[@Klemmi.]

Thanks for those type of videos. I know, they don't create as much revenue as other formats, but they are pure gold for the really interested viewers! Love the increasing calculator size as well! :-D

[@KeritechElectronics]

A cool and interesting lesson, and it's nice to see the simple-and-elegant breadboard from the mailbag in action :).

[@EEVblog]

The fanboys wanted the jumperless breadboard!

[@KeritechElectronics]

@@EEVblog kids these days, always wanna have the newfangled stuff... LOL

[@matteo234321]

Loved the calculator debuts on the whiteboard! Also, at 21:40 the THD was terrible without the composite configuration, you could even see it on the scope!

[@DrFrank-xj9bc]

Thank you so much for this 'basics' video. Great, that you did it on that abstract / practical level, not too much details or variants, and not too high flying, so that everybody could catch the idea and advantages. Myself, I was very much fascinated of those amplifiers, when I did my first job @19 in the Airforce lab in 1980/1982.. most hp, FLUKE, Philips multimeters had one or several composite amplifiers of all kinds (in constrast to only a linear OpAmp). At that time, they consisted of discrete components, like vaccum tubes in combination with optical choppers, or a double FET and a 741 for the AC path, plus a capacitor coupled FET chopper path for the DC path. In the volt-nuts section, we discussed the classic null detectors, like the Fluke 845AB/AR, which also has this composite design.. If I find the PCBs, I'll post pictures of the PM2524 discrete chopper and A/D converter, which I had designed from the schematics. Great stuff. Let's see if it still works and how it performs after 40 years.

[@theondono]

Watched this just in case there were some gold nuggets, and it immediately reminded me of one of my first “deep dive” technical discussions at my first job. My boss could not see the difference between cascaded and compound amplifiers (we needed something similar to your DC precision example). In the end I sneakily added additional footprints to support both configurations and assembled both options. The guy could not believe the difference when we tested them, he even accused me of cheating to make him look bad. Luckily it took all of 5min for our in-company “guru” to side with me!

[@SeanBZA]

A very common amplifier, yet often totally hidden inside a package as well, with most opamps being made up of them as well internally. A good explanation of why you want the gains, and the closing of feedback loops as well, so as to get the desired performance, plus that you can mix different types of amplifier as well.

[@EEVblog]

Yes, totally forgot to mention that in the video. I knew there was something I wanted to add.

[@michaelmolter8828]

As a young designer starting my career during the pandemic, I learned to use the most common, universal components whenever possible. Never use a specific bespoke part when a generic one will do. If Digikiey stocks less than 5,000, stay away!

[@nameredacted1242]

This is really all that you should be doing on YT, given complete lack of such material, at this high quality, on YT.

[@Mr.Leeroy]

eevblog YT now is all about fear of staying relevant and twisted viewership statistics.

[@johncoops6897]

Yep, 100% agreed. Dave has lost his way, especially over the past few years. Didn't help that he started multiple channels and then puts the same kind of content randomly on them to ensure that viewers have no way of knowing where to look.

[@nameredacted1242]

@@johncoops6897 Have you considered the possibility that he is tired of saying the same thing over and over again to us morons, and not making much money from that??? Have you watched all of his older videos, as well?

[@johncoops6897]

@@nameredacted1242 - of course, and it's Dave's choice if he chases the algorithm instead of producing the quality content thet made him successful in the first place. I'm a long term subscriber end have watched his old videos. I use them as reference material. I don't bother with his boring newer stuff, or his other channels with duplicated end diluted content. Take a read down the comments here and see how many people are congratulating him for publishing something decent after so long. I am one of hundreds...

[@nameredacted1242]

@@johncoops6897 Have you considered that YT makes it impossible for honest and intelligent content producers to stay on top of the algorithm? If SSSl🌶t gets ten billion views for free for her "reaction videos", and Dave has to take a chunk out of his life to work for ten cents, it very quickly gets tiring and pointless to try to stay ahead of Shorts sl🌶ts on this platform.

[@johnwest7993]

Thanks Dave. This is a very useful short class in practical op-amp circuit designs that was never well covered in my text books.

[@klazzera]

Dave, your teaching skills and the information you give is so invaluable that I wouldn't be able to repay you🙏🏻 Not that I'd be willing to pay you money but I wouldn't be able to repay you🙏🏻

[@ppdan]

Great video. You covered something very interesting and useful that will certainly be of use. We need more of videos like this, were you not only explain how it works but also build and show it on a simple breadboard that any electronics enthousiast can reproduce at home.

[@drinductor8150]

Negative feedback has got to be my favorite principle in engineering. Like magic.

[@petersage5157]

Same basic principle as using an op amp to take a diode's forward voltage drop out of the equation. I always enjoy the little Easter eggs on your whiteboard tray.

[@LawpickingLocksmith]

Neat! I think his t shirts also have compound ultra low fading. 20 years ago perfect black, now uniform grey....

[@CatcatcatElectronics]

*_Hello from Ukraine. Live and learn. This video was very interesting for me. Thank you, you are doing a great job!_*

[@fredflickinger643]

Always my standby circuit for the difficult input specifications! Good explanation Dave.

[@PixelSchnitzel]

This is the kind of content I love from you! That was VERY clear and well presented (loved the changing calculators too). I've gotten away from watching this channel for a long time because the signal-to-noise ratio had declined a lot (in my view anyway). But stuff like this will draw me in every time. I gotta go look through your catalog and see what I've missed now.

[@HazeAnderson]

I was thinking "How did I miss this blast from the past" when I saw the thumbnail ... new content! And on OP AMPS! Thank you!

[@mrVetz]

I found a national semiconductor paper when I was doing a small headphone amp. Not knowing the. what Dave is talking about here I replicated the circuit and it was awsome. No DC drift and good low noise application. A good input amp and an higher current output buffer in this configuration, two resistors to set the gain and a decent power supply you get a really good amp. Thanks Dave

[@GooogleGoglee]

Damn.. those CALCULATORS on the board got a quick evolution or devolution along the video! 😂

[@JTordur]

Excellent video, I can never get enough of your opamp videos!

[@electroquests]

I really appreciate you making this video Dave, I've learnt so much from your channel over the years!!!

[@SeanB88]

Great video, Dave! Always love these whiteboard explanations of circuits!

[@Keethraxmn]

This was super helpful! Would have been more helpful a few months ago, but never too late to learn. THis is one of those situations where knowing the terminology to search for more details was all I really needed. Thanks!

[@AIM54A]

I love how the calculators keep changing on the white board.. Great video too.

[@CarlosChavez-gs1ld]

Well done. I have designed some analog circuits in my life, including some that went into consumer products, and my only real experience with "Composite" amplifiers is in analog filters. I found this video interesting because I have "Cascaded" (put in series) Op Amps on many occasions- usually in an effort to get good gain without adding any more noise than nesc., and also- obviously, sending the output signals through some kind of buffer/line driver- like everyone else does. I have not given building some of these designs as "Composite" before... but I will do so for the next time.... see what I come up with. Thanks again.

[@JonathanDFielding]

Fantastic gems of wisdom! Thank you. One thing that composite amplifiers cannot fix is quiescent current draw. I'm sure there are others but that gives to mind

[@JFirn86Q]

Wow, what a cool trick, worked perfectly in the real example too! Thanks - loved the calculators in the shots :)

[@AlessandroAllegretti]

This is the first time for me about composite amplifiers. Very intriguing! Thank you very much!

[@todirbg]

10x Dave! Seeing you in front of that board makes me click immediately! I love your tutorial series, please keep them coming :)

[@aturegano87]

Awesome explanation. I love this series of videos about electronic theory and how to use it in many applications.

[@andymouse]

Awesome, this is so refreshing and from a personal point of view couldn't be better timed as I'm over microcontrollers and coming home to where my heart is Analogue Electronics! proper electronics none of that Arduino rubbish. However there's a problem and thats where you come in Dave as it turns out I have forgotten more than care to admit here so any analogue stuff I will gobble up but keep the math to an appropriate (low) level as I know that I am still shit. I might just start you opamp stuff from the beginning if I remember rightly it should be required reading. Another good teacher like yourself IMHO is Prof Fiore anyhow cheers.

[@cornwallonline]

Your video and comments about the lack of info in books reminded me of something I read years ago. Yes, The Art of Electronics has a fair bit on the subject...I just checked!! Mind you the trilogy of Art of books are the absolute best on the subject of electronics that ever has been! Great video, by the way.

[@EEVblog]

The main Art of Electronics only has rather small mention of composite amps. The X chapters has a bit more.

[@GiovaniCauzzi]

The white board vídeos are great! Cheers, Dave!

[@leplayz4499]

right when I needed it. greetings from Germany

[@lis0x90d]

Big thank you, Dave! Very intelligible and clear

[@ElektronikLabor]

I like such tutorials abouz analogue topics. Pleas keep them up 👍

[@mcconkeyb]

It nice that you are giving basic tutorials, but if our system of education is graduating engineers and designers and they don't know about this, we are all in very big trouble!

[@TheDefpom]

Nice example and explanation.

[@gerryjamesedwards1227]

Hi, Dave. Great tutorial, I really apprecaite you making these! You mention the possibility of having to add some compensation to the loop. I've tried to follow Bob Pease's chapter on feedback loop compensation but it was aimed at more advanced EE's than me and most of it went over my head, but I'm fascinated by the subject. I know it would be a bugger trying to put together a practical example circuit that needs compensating, but I would love to see you do a tute on this subject.

[@hedleyfurio]

These 'tutorial' type videos are the reason I started watching your channel many years ago with the 34063 . 👍One day if you up to it please do one explaining where the energy goes if you have an off grid inverter MPPT controller that is fed by a solar panel and the output ( load ) is far smaller than the input power + the battery is fully charged . The panel is in direct sunlight - All I can think is that the output which is ' throttled by the MPPT' must then be dissipated as heat by the panel ? due to conservation of energy .

[@EEVblog]

Ask yourself "where does the heat go" if you have a 400V DC battery just sitting there...

[@hedleyfurio]

@@EEVblog I would guess nowhere , as it is just stored chemical potential energy which is only released when connected to a load and current flows.

[@kellen_mcsmellin]

Love the tutorial videos Dave, very enlightening

[@electroquests]

Hello Dave, can you please make more videos on composite amplifiers? Thanks for this one, love your videos! Learnt alot over the years from them!!!

[@McTroyd]

Seems like this is the fundamental building block for an audio channel strip on a sound console. High-precision preamp input, high-power bus drive output, with a bunch of adjustable filtering intentionally mangling the link between the two. Neat.

[@cyberphox1]

Excellent tutorial video. Fantastic channel

[@aaronk2242]

Thanks for sharing! I love learning clever circuit design tricks like this =)

[@kellen_mcsmellin]

Liked because of the Pilot G2, my favorite pen of all time

[@cremvustila]

Where's the two thumbs up button!? Thanks Dave, I learned something new, awesome!

[@NoLandMandi]

Geeee! I feel bad for that A2, you call that poor thing horrible so many times I wonder if it even bothers to operate at all! :))

[@SeanBZA]

It knows it is a brute force opamp, just cannot do fine work, so needs the first one to help with that.

[@EEVblog]

Kinda like asking Andre the Giant to repair your Rolex.

[@jcugnoni]

This circuit topology reminds me of tube driver & power amp with global negative feedback. Really cool.

[@tracynorton6410]

Excellent video. btw, I have that Casio fx calc that appears at 10:22...my wife's grandfather gave it to me back in '86 when I went back to college. He flew dive bomber in the Battle of Coral Sea. I still use it. Daily.

[@jannb.6811]

Great video. Have to check if there were changing calculators in other white board sessions as well. Counted 5 in this one.

[@LavalBolduc]

Absolutely great video, thank you !

[@martinvollderpro]

I recommend looking at the Application Note of Jim Williams on Composite Amplifiers. There are number of great circuits Dave hasen't covered in this video

[@alirezaeskandari2287]

Dave!!! We need a video on star ground too. Thank you very much.

[@p_mouse8676]

These are often also called nested amplifiers. This term is often used in more scientific papers.

[@TheHuesSciTech]

Very loud noise at 14:28, might be a good idea to just mute for a fraction of a second there to protect viewer's ears.

[@EEVblog]

I think that was the video card file size limit switchover. Too late now, can't edit audio after upload.

[@Ma_X64]

Looks like a good idea for a constant gain overdrive guitar FX. Internal FB can be non-linear and the external one would set the overall gain.

[@kensmith5694]

I have done the composite amplifier thing 3 op-amps deep to get high gain at high frequencies and a low offset. Opamp 1 = low offset and low noise, over compensated and inverting pin to (-) input of compensated via resistor.. Opamp 2 = darn fast opamp (+) input goes to the (-) input of composite and (-) input controlled by 1st stage's output Opamp 3 = Extremely fast inverting output of opamp2 to make the final output. For a step input it got settled in a big hurry but the offset was under 100uV

[@AlessandroAllegretti]

I like your calculators collection 😂👍

[@brouwereric644]

Awesome, thank you!

[@jackevans2386]

Nice job Dave !

[@WacKEDmaN]

Dave needed half a dozen calculators to work out his R1s + R2s :P ...thanks Dave... i had no idea about composite amps...

[@Bestcuriosity_1]

Awesome explanation

[@GaborGubicza]

BUF634.... Been there... Done that😂 exact same V drop under load conditions. I also solved it with compound amplification. :D Take care everyone.

[@EEVblog]

It's a little bit whimpy on its own.

[@ghlscitel6714]

In Former times we used the NatSemi LH0002 buffer amplifier.

[@kensmith5694]

I remember those and also the LH0033. It was the first time I saw the need for a heat sink on an op-amp.

[@zoltandiveki5233]

Ah, that Casio fx-50F! My first serious calculator. I bought mine in 1990 in Bremen, then West Germany...

[@EEVblog]

Nice.

[@devsysnet]

I also had a Casio FX-770P in the 80s. A wonderful programmable calculator. Unfortunately the folding interconnection foil broke in a short time.

[@RPBCACUEAIIBH]

I think we ware taught something like this in uni, here in Romania I don't remember well, because back then I didn't really understood it. May have been a language barrier, cause I was always interested in electronics, but Romanian isn't my native language, but now that I watched this I start to remember...

[@אביעדבגנו]

Wonderful.

[@roscozone8092]

Nowhere near a breadboard to test with ATM, and wondering whether reactive components can be used in the feedback loop to create a filter...? Would that still work? Would the equations need any tweaking in a composite configuration?

[@VandalIO]

I love video like these :D

[@Asdayasman]

Hello there Dave, this video was great. It got me thinking. If you put out a video series that was like "the beginner idiot's (yes I mean you Asday) practical guide to electronics" with each video being your answer to last month's project, some concept introduced, MAYBE some theory, then a description of a project relying on that concept, and a shopping list of parts, I would... Well I'm not going to finish that sentence. Even better if you stocked the BoM on your website with shipping to England and maybe a 50-100% markup in price. You could even follow the HelloFresh model and make it a subscription service thing so each month a bundle of components turn up like it's Christmas and then you go watch the video. Kinda like Ben Eater's stuff, but with more of a focus on "here is a specific concept that you will be able to take elsewhere and use in your circuit design" and less on "here is one specific project that needs a disparate multitude of concepts even if it is kinda cool". Maybe even a 300% markup.

[@stamasd8500]

This is a nice segue to delve into instrumentation amplifiers (aka composite amplifiers on steroids) :)

[@electroquests]

Hello, say I wanted to combine the precision of OP07 and TL081's slew rate, how would I do that, like which amplifier would be used for the input?

[@hcpcb]

If I remember correctly, you used 2 cascaded MAX opamps in the uCurrent design. Could you explain why not using 1 in front and a cheaper one as the second in a composite configuration?

[@mikeh7704]

I think the noise performance of the composite amplifier failed momentarily just after 14:28 🤔. Great educational video as always!

[@willeyex]

Would you Dave consider making and example use of this for say a headphone amplifier or some other audio circuit .

[@MartinBeret]

Like you calculators selection

[@lasersbee]

Excellent... I missed watching these Tutorials. I always learn something Exciting and New.

[@garygranato9164]

good tutorial thank you

[@SianaGearz]

1. I need to know why amp B needs to have a higher gbw than amp A. I might have an intuitive feel why it may be so but i'm not comfortable with the gap in my knowledge. 2. I know how to add a discrete output stage of one or several transistor pairs to an opamp, and indeed it works fantastically. Classic headphone amplifier schematic, but indeed can have versatile use. How does one add a discrete JFET input stage, and when might one want to do that vs. choosing an opamp with a different type of input stage in it?

[@ve2zzz]

10:20 Take care when choosing the amplifier gains, otherwise your Hewlett-Packard HP-41 will transform itself into a Sharp calculator !

[@andrewkissel3807]

So many calculator's 🙂 Love it...

[@VandalIO]

what happens if you combine different slew rate op amps ? in a composite amplifier configuration ?

[@Scrogan]

Wonder if this will reduce the THD of an LM386? It’s probably a lot slower than whatever op-amp you’d put in front of it, but you can probably make it stable by putting a cap in the feedback loop. Now I’m wondering how to do this with an instrumentation amp circuit.

[@maudiojunky]

It's basically the same if you mean using a fully-differential amplifier as the outer amplifier in the composite amplifier, you just have to mirror everything inside the two inner loops. You can use a FDA like this to create a bridge-tied output with basically zero DC offset and no even-order distortion. In either case, single-ended or bridge-tied, you can forego the LM386 entirely if you want more than 1W of power. It's the same process for a discrete output stage, you close the feedback loop around a complementary class B stage with power transistors, which can be of whatever size you like if you use them in darlington configuration or add a buffer amplifier in the loop to provide sufficient base current. With a fast enough opamp you won't measure any appreciable crossover distortion even with a very cold bias running like this, no need to waste current for class AB bias, which means a simple diode bias is fine without worrying about thermal compensation. Just ensure the opamp has sufficient voltage headroom and its overload and recovery are well-behaved for good large signal behavior.

[@Scrogan]

@@maudiojunky done the class-B stuff before, it takes up more board real estate and you need a pretty fast op-amp to avoid crossover distortion.

[@maudiojunky]

@@Scrogan I have to say my experience differs. If you bias the transistors just below turn-on with a simple diode pair the crossover residual can be almost immeasurable. Audio just isn't very fast, even with no biasing at all a 5V/µS opamp will pass through the crossover region in well under 300ns. Regarding the IC amp solution, you may have trouble wrapping an integrated audio amplifier inside the feedback loop because it may already be compensated too slow to work in this configuration since the inner loop must be much faster than outer loop. LM386 has a rated bandwidth of 300kHz, so it's quite slow compared to opamps and buffers. Even if the composite amp isn't quite stable at 20kHz bandwidth you might still find use for it powering a woofer paired with a comparatively sensitive tweeter driven directly by the other half of a dual opamp in a portable speaker.