EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Darkwing on April 27, 2019, 09:51:40 pm
-
Hey mates!
Well, I don't get it ... I've now read a lot of times about OpAmps, but I don't see the use of them, when I can do amplifications of all sorts with MOSFETs, BJTs and Darlingtons or even Optocouplers (which have an advantage of their own).
What is this mystic operation, that is best to be amplified via an Operational Amplifier? Can someone please point out? Do they have an advantage in audio in terms of noise or the like?
Side quest 1: What's this "rail-to-rail" thing?
Side quest 2: Can you name your favorite OpAmp? Should there be such a thimg, what could be the best go-to jelly bean I should have in my box?
Thanks y'all! :)
-
Sure you can make a basic amplifier out of a bunch of BJTs, but by the time you have designed it to be as quiet, stable, precise, efficient, small as a 20cent op amp, you would have effectively designed your own inferior op amp anyway. Opamps are for people who have more productive and interesting things to spend their design efforts on.
Of course there are applications where a custom amplifier design is necessary and an op amp won’t do, but for everything else there is an opamp.
-
Indeed, the selling point of an opamp is that it is (more or less, ok actually less then more) an ideal gain block, within certain limits you can assume open loop gain is large, input impedances are high and bias currents are small (but never zero).
This means that rather then winding up with circuits where beta matters I can assume (again within limits) that the closed loop gain is set almost solely by the feedback network, epic maths time saver.
Also often my other option is to build an inferior difference amp out of discrete components (that will not match near as well as the ones in the opamp), it will be bigger, probably noisier (unless I am good) and worse behaved, worse, it will take me a day two design it.
Regards, Dan.
-
The ideal op-amp has infinite gain and infinite input resistance. Then the gain of that stage is controlled by the feedback circuit. You can design an amp of any gain you want just by using a couple resistors for the feedback.
Try to do that with a transistor or two.
-
Another advantage of opamps over discrete amplifiers is matching of their components. Process variations mean you generally can't guarantee absolute values, but relative values on the same die can be amazingly close without trimming. And for even tighter matching things like laser trimming are a matter of automation. If matching is important for your homebrew discrete amplifier (think: differential input amplifier), you'll be mucking around with matched components or trimpots and the like just to achieve what a single monolithic opamp delivers for remarkably low cost.
Monolithic opamps are one of the modern miracles of electronics. The same three terminal device can yield hundreds of different results based solely on the feedback loop you put around it. Think of them as the analog equivalent of the microprocessor... both incredibly flexible building blocks, one in the analog domain, the other in the digital.
EDIT: As for "rail to rail", that refers to an opamp whose input and/or output voltage swings more closely approach the supply rails. Early opamps required some margin from the rails, but rail to rail has been available for literally decades and gives you more dynamic range within the supply rails.
-
Well, I don't get it ... I've now read a lot of times about OpAmps, but I don't see the use of them, when I can do amplifications of all sorts with MOSFETs, BJTs and Darlingtons or even Optocouplers (which have an advantage of their own).
What is this mystic operation, that is best to be amplified via an Operational Amplifier? Can someone please point out? Do they have an advantage in audio in terms of noise or the like?
1. The matching between the two sides of the symmetrical differential input stage reduces DC offset and offset drift by orders of magnitude allowing great precision. A bipolar transistor has an offset of 0.6 volts and drift of about -2 millivolts per degree C. But use that same transistor in a differential stage and it can be 0.6 millivolts and +/-2 microvolts per degree C or a thousand times better. Some parts are 10 times better even than that.
2. The differential input allows high common mode rejection. A singled ended amplifier relies on one of the supply leads for common.
Side quest 1: What's this "rail-to-rail" thing?
This refers to the allowable input and output voltage swing and specifically how close it can approach the supply voltages.
Side quest 2: Can you name your favorite OpAmp? Should there be such a thimg, what could be the best go-to jelly bean I should have in my box?
My favorite part is the LM301A which is one of the earliest operational amplifiers but still produced. Its input voltage range includes the positive supply and its compensation pin can be used as an input clamp or transconductance output.
-
Op Amps are very useful devices, & solve many problems in Electronics, but they do have "things you have to watch".
They are not at all fond of reactive loads, for instance, & many "young (& old) players" come to grief with this, although it is usually fairly easy to tame.
People do fall in love with them, & you will find "noobs" posting RF amps using "jellybean" Op Amps, with an LC circuit at the input, output or both, in puzzlement as to why the thing oscillates!
-
Maybe they're useful when you put a capacitor in the feedback path and create an integrator. Analog computing...
Wait! That's why they were invented in the first place! Operational Amplifier - performs 'operations' like differentiation and integration or addition and subtraction.
See attached photo of Comdyna GP-6 doing "Mass-Spring-Damper" problem and .pdf of MATLAB Simulink schematic.
The wiring on the GP-6 is overkill because 2 different problems are patched. The other simulates a swinging door like those between the kitchen and the dining area in a restaurant. Same equations, different constants...
One way or another, hardware or software, these simulations bring differential equations to life!
-
A neat historical note; before solid state op amps existed, there were vacuum tube op amps:
https://www.youtube.com/watch?v=mEFdADrU9MA (https://www.youtube.com/watch?v=mEFdADrU9MA)
-
And before tubes, there were (essentially) mechanical opamps. Some early military aircraft had scary-elaborate mechanisms that basically performed integration/differentiation/etc. operations that we take for granted today in our monolithic devices. I tried to find a photo to link but no luck.
-
Maybe they're useful when you put a capacitor in the feedback path and create an integrator. Analog computing...
Wait! That's why they were invented in the first place! Operational Amplifier - performs 'operations' like differentiation and integration or addition and subtraction.
Single transistors can do all of these things as well but with much less precision. The single transistor "Miller integrator" is included inside most operational (and audio) amplifiers where it provides the frequency compensation and usually most of the voltage gain.
A single transistor shunt feedback adder or inverter is also completely feasible but compared to the version using an operational amplifier, it has terrible offset and drift.
-
James Thomson first discussed solving differential equations with a mechanical ball-disk integrator back in 1876 but it was used earlier by Lord Kelvin in a tide prediction computer in 1872. Bright brothers!
https://en.wikipedia.org/wiki/Differential_analyser
Lord Kelvin came up with the integration method of modeling differential equations and the magic is based on the '=' sign.
Take an equation like Ay'' + By' + Cy + D = 0 - just a simple second order DE with constant coefficients. Lord Kelvin's contribution was to solve for y'' as y'' = (-By' - Cy - D) / A = (-1/A)*(By'+Cy+D). Pretend you actually have y'' and integrate once to get y' and again to get y then mix in all the coefficients creating the right hand side of the expression. Now, if we just HAD y'' we'd be all set. But we DO have it because of the '=' sign. All that mess we just created on the right hand side IS y'' so just feed to output back to the input, apply initial conditions and let the thing run.
Somehow I find this exciting! I really wish I had had access to an analog computer back in college. Not just for the Differential Equations course but also for Control Systems.
See that MATLAB Simulink schematic I posted earlier to see how this all plays out.
Oh, and accuracy is everything. We want to be able to read values off a strip chart recorder or oscilloscope. No drift, no effects from bias current (think 1 uA bias current with a 1 Mohm timing resistor - HUGE error!). We can overcome some of that by scaling time such that the timing resistor can be 1/10 or 1/100 as large. Of course, the timing capacitor has to be pretty high quality as well. The op amps are the least of the problems.
Sometimes we only want the shape of the result and the absolute values don't matter.
-
I really wish I had had access to an analog computer back in college. Not just for the Differential Equations course but also for Control Systems.
Oh, and accuracy is everything. We want to be able to read values off a strip chart recorder or oscilloscope. No drift, no effects from bias current (think 1 uA bias current with a 1 Mohm timing resistor - HUGE error!). We can overcome some of that by scaling time such that the timing resistor can be 1/10 or 1/100 as large. Of course, the timing capacitor has to be pretty high quality as well. The op amps are the least of the problems.
The analogue computer I saw in ~1973 at a local university when I was at school, had 100V supplies and a flying lead patchboard. Eventually people learned that one end of the patch cables bit.
My school math teacher had an interesting discussion with him about whether or not digital computers would supplant analogue and hybrid computers.
-
My school math teacher had an interesting discussion with him about whether or not digital computers would supplant analogue and hybrid computers.
There was a time when folks wondered if this newfangled "electronics" thing would supplant hydraulic circuitry. No, that is not a typo.
-
Well, I don't get it ... I've now read a lot of times about OpAmps, but I don't see the use of them
High amplification allowing high linearity with feedback, rail to rail operation, auto-zero operation, very low offset with very low drift, very low bias and leakage current.
Most of that require either 100s of transistors to equal an opamp, or might even require very small area monolithic pairs which simply don't exist as discretes.
Side quest 1: What's this "rail-to-rail" thing?
They generally automagically transition between P/N input pairs near the rails so they maintain amplification from rail-to-rail, again requiring tons of transistors.
-
My school math teacher had an interesting discussion with him about whether or not digital computers would supplant analogue and hybrid computers.
I remember when digital multipliers caught up with analog multipliers in precision and speed. Everybody who understood Moore's Law knew that it was only a matter of time before digital computing completely outperformed analog computing.
-
My school math teacher had an interesting discussion with him about whether or not digital computers would supplant analogue and hybrid computers.
There was a time when folks wondered if this newfangled "electronics" thing would supplant hydraulic circuitry. No, that is not a typo.
I once (1983) had a consultancy contract into exactly that.
My conclusion was that replacing the hydraulic logic was too difficult and with insufficient benefit. I included the possibility of remote measurement and control, well before the internet per se existed.
The application domain: off shore unmanned oil platforms where there was currently zero electricity, for obvious reasons.
-
My school math teacher had an interesting discussion with him about whether or not digital computers would supplant analogue and hybrid computers.
I remember when digital multipliers caught up with analog multipliers in precision and speed. Everybody who understood Moore's Law knew that it was only a matter of time before digital computing completely outperformed analog computing.
Indeed. That was my maths teacher's opinion, and the person at the university didn't have their heart in a counter-argument.
As a knowingly incompetent schoolkid, I found that interplay fascinating and revealing.
-
I remember when digital multipliers caught up with analog multipliers in precision and speed. Everybody who understood Moore's Law knew that it was only a matter of time before digital computing completely outperformed analog computing.
There's always the quantization error of digital as compared to analog. But that becomes an argument between "number of bits" versus "signal to noise ratio".
-
A neat historical note; before solid state op amps existed, there were vacuum tube op amps:
https://www.youtube.com/watch?v=mEFdADrU9MA (https://www.youtube.com/watch?v=mEFdADrU9MA)
Interestingly, "scale of 2" binary dividers ( plain old bi-stables) were made in an almost identical format.
Apparently they used them in tube type computers.
I never saw any "in the flesh", all the TV SPGs used individual 12AT7s or similar.
Some early SPGs used special "black magic" divide by 5 or 10 stages, but they had to be "babied" to get them to work properly.
-
Interestingly, "scale of 2" binary dividers ( plain old bi-stables) were made in an almost identical format.
Apparently they used them in tube type computers.
I never saw any "in the flesh", all the TV SPGs used individual 12AT7s or similar.
Some early SPGs used special "black magic" divide by 5 or 10 stages, but they had to be "babied" to get them to work properly.
I have a Tek 184 time mark generator. It uses nuvistor valves for the 10MHz crystal oscillator and multipliers up to 500MHz. It divides the 100ns crystal output down to 5s in divide by two and divide by 5 stages.
Each division is done with three transistors and some discrete components - including some tweakable components to "baby" the stage. The basic principle of operation is that each stage is an analogue divider: the input injects "glugs" of charge onto a capacitor, and when sufficient glugs have been received a pulse is output and the capacitor discharged.
-
Instead of having 20 transistors all over your PCB, you now have a neat little package that has those in silicon, and they are all connected nicely in the tiny little package.
What's not to like?
-
If you know what one is and what it can do then you wouldn't be asking the question, but making the presumption that you do not know what it is or what it does then it is a very valid question.
It's easy to say stuff like "why wouldn't you use one" but if you didn't know how they worked then you probably wouldn't? I know it took me a while to determine what the point of them are, but when you understand it "clicks" into place.
I think most people have answered your original question already, but here is a webpage with a small video that might help:
https://www.allaboutcircuits.com/video-lectures/op-amp-applications/ (https://www.allaboutcircuits.com/video-lectures/op-amp-applications/)
It tells you the applications for an op-amp.
There are also plenty of websites and videos that go into them.
-
Well, I don't get it ... I've now read a lot of times about OpAmps, but I don't see the use of them
It's in the name, "operational", the opamp can do mathematical operations among all your basic amplifier stuff. They also make it easier to implement active filters and other blocks.
-
To give the OP a direct answer to his question, I suggest he looks at the opamp applications in
https://www.analog.com/en/education/education-library/op-amp-applications-handbook.html (https://www.analog.com/en/education/education-library/op-amp-applications-handbook.html)
That should keep him going and answer his questions.
Op Amp Applications Handbook, Edited by Walt Jung, Published by Newnes/Elsevier, 2005, ISBN-0-7506-7844-5 (Also published as Op Amp Applications, Analog Devices, 2002, ISBN-0-916550-26-5).
This may well be the ultimate op amp book. It is brimming with application circuits, handy design tips, historical perspectives, and in-depth looks at the latest techniques to simplify designs and improve their performance. But this is more than just the last word on applications. A brief but fascinating History section outlines the early development of the feedback amplifier, starting with H. S. Black’s invention of seventy years ago—and provides priceless insights into the application needs, technological developments, and creative personalities that drove the many generations of op-amp designs.
The Op Amp Applications book is available for download:
Section H: Op Amp History (pdf)
Section 1: Op Amp Basics (pdf)
Section 2: Specialty Amplifiers (pdf)
Section 3: Using Op Amps with Data Converters (pdf)
Section 4: Sensor Signal Conditioning (pdf)
Sections 5-1 to 5-4: Analog Filters (pdf)
Sections 5-5 to 5-8: Analog Filters (pdf)
Section 6: Signal Amplifiers (pdf)
Section 7: Hardware and Housekeeping Techniques (pdf)
Index (pdf)
-
I still have some hardbound copies of books like that. National did a bunch of linear books that are near-priceless (like the original Audio Applications books, which I still have!). Analog Devices did a bunch on converter technology too. Over the years I've shed more and more of my books and other flotsam and jetsam, but those books are likely to pass on to my son (who leaves for CalPoly this fall to study EE).
In a tangental response to the OP: I have long emphasized to my son that he include a LOT of analog in his studies. We've had many a kitchen table discussion with pen and paper on the topic of opamps. Digital is nice, digital is important, but analog is the basis of everything and as digital frequencies keep going up a lot of the design rules are more analog than purely digital. I promise the more you study opamps, the more you will appreciate and respect them. It's stunning what a simple three terminal black box (OK, triangle! :)) can do.
-
I still have some hardbound copies of books like that. National did a bunch of linear books that are near-priceless (like the original Audio Applications books, which I still have!). Analog Devices did a bunch on converter technology too. Over the years I've shed more and more of my books and other flotsam and jetsam, but those books are likely to pass on to my son (who leaves for CalPoly this fall to study EE).
In a tangental response to the OP: I have long emphasized to my son that he include a LOT of analog in his studies. We've had many a kitchen table discussion with pen and paper on the topic of opamps. Digital is nice, digital is important, but analog is the basis of everything and as digital frequencies keep going up a lot of the design rules are more analog than purely digital. I promise the more you study opamps, the more you will appreciate and respect them. It's stunning what a simple three terminal black box (OK, triangle! :)) can do.
Yes to all of that, except everything is analogue except femtoamp circuits and photon counting applications. Think signal integrity, and understand that 10Gb/s signals are analogue :)
I too have saved the NS books, and many more
-
Over the years I've shed more and more of my books and other flotsam and jetsam, but those books are likely to pass on to my son (who leaves for CalPoly this fall to study EE).
Congratulations to you and your son!
-
Yes to all of that, except everything is analogue except femtoamp circuits and photon counting applications.
Well, if we dig deep enough everything goes digital at the quantum mechanics level, I suppose. {grin} And I'd argue that those femtoamp circuits (really, electron counters!) and photon counting applications had better adhere to proper analog design techniques too! {bigger grin}
The point I think we're both making is that folks ignore analog at their own peril, and these days opamps are a big part of the analog world.
-
Over the years I've shed more and more of my books and other flotsam and jetsam, but those books are likely to pass on to my son (who leaves for CalPoly this fall to study EE).
Congratulations to you and your son!
Thanks, very proud Daddy here. He still rolls his eyes at me when I drag him over to the oscilloscope for something, but I know data is seeping in through the cracks. ;D
-
Thanks, very proud Daddy here. He still rolls his eyes at me when I drag him over to the oscilloscope for something, but I know data is seeping in through the cracks. ;D
I know the feeling! My grandson is wrapping up what is effectively his 2d year of an ME program. It's a 5 year program so he has a long way to go. Then there is grad school... I told him that BS degrees are pretty common around Silicon Valley and if he wants to stand out, get an MS in something. Or an MBA if he would rather be a manager than an engineer (typically pays better). An MBA with an engineering undergrad should land him a job somewhere! Or even an MSES (Engineering Science - kind of a generalist program).
He has a little less than no interest in electronics or programming. The programming thing changed a little bit as he wandered through a MATLAB course this semester and I think he began to understand and appreciate the thought process behind problem solving in code. We'll see...
-
If he wants to be a manager at all, make sure he gets the technical skills too. Managers who have no clue about how things work "in the trenches" are the worst. IMO the ideal technical manager is someone you could replace one of the engineers with and not notice the difference. [edit] that's not to say that everyone with the technical background makes a great manager...that is certainly not true!
-
If he wants to be a manager at all, make sure he gets the technical skills too. Managers who have no clue about how things work "in the trenches" are the worst. IMO the ideal technical manager is someone you could replace one of the engineers with and not notice the difference. [edit] that's not to say that everyone with the technical background makes a great manager...that is certainly not true!
BSME
MSES
MBA
In that order - technical stuff first. There's a LONG way to go. We'll see...
-
Over the years I've shed more and more of my books and other flotsam and jetsam, but those books are likely to pass on to my son (who leaves for CalPoly this fall to study EE).
Congratulations to you and your son!
Thanks, very proud Daddy here. He still rolls his eyes at me when I drag him over to the oscilloscope for something, but I know data is seeping in through the cracks. ;D
Yeah! Deal with it.
Kids: who'd 'ave 'em :)
-
A super high gain function block
-
Apart from the hijackers: thank you very much guys! :) ;D
The linked PDF book is a very good read, I like that! Must read more. 8)
I understand for now, that such an OpAmp is a very flexible signal amplifier and in terms of "manufacturing tolerance" and reliability and signal quality to prefer, instead of simple BJTs.
But what I don't understand is: can they drive loads? Are they suitable for "the power stage" of a project? It doesn't seem so. I would be interested, if someone could point me to a part what I would then call an "OpAmp Driver", that can handle bigger loads.
(Maybe that's why I never saw a use for them in any of my projects so far, because I understand 'amplification' as being able to 'amplify voltage AND current'. [A MOSFET with 1.4mΩ Rdson is not even slightly impressed, when there's a current of 5A or so ... would an OpAmp be?])
However, you guys helped me a lot! Thanks! :-+
-
can they drive loads? Are they suitable for "the power stage" of a project? It doesn't seem so.
There used to be "power opamps" available. Apex made some that could drive multiple amps, came in suitable heaksink-able packages, etc. I don't know if Apex is even still around anymore.
Generally opamps are used for the small-signal part of the signal chain. Then it's common to see a discrete power transistor stage driven by an opamp. Many opamps have enough output current to drive power transistors, at least the first stage, and if you need even more power you cascade another discrete power amplifier stage after the first one, with the latter acting as a preamp for the former.
Stated differently: You'd use opamps to do small-signal amplification, filtering, and other signal conditioning. Then, once the signal has been manipulated the way you like, you use power transistors as a final to get the voltage and/or current you need to drive the load.
EDIT: Apex is still around! https://www.apexanalog.com/ (https://www.apexanalog.com/) I see they have output currents to 50A and voltages exceeding 350V. Most likely they are simply encapsulating my above description, a traditional opamp with a discrete power final. But it is nice that they make them available in a ready-to-go package.
-
Ok, thanks for clarification! :)
Btw: Apex PA75CD Power Op Amp (https://www.ebay.de/itm/Apex-PA75CD-Power-Op-Amp-1-4MHz-30-V-7-Pin-TO-220/163124059668?epid=27021016844&hash=item25faf39e14:g:J94AAOSwmHdbNlWE) 40 bucks!? :wtf:
Oh boy ... and that's their entry level ... This must be a nery niche product I promise to never use! :-DD
-
Btw: Apex PA75CD Power Op Amp (https://www.ebay.de/itm/Apex-PA75CD-Power-Op-Amp-1-4MHz-30-V-7-Pin-TO-220/163124059668?epid=27021016844&hash=item25faf39e14:g:J94AAOSwmHdbNlWE) 40 bucks!? :wtf: Oh boy ... and that's their entry level ... This must be a nery niche product I promise to never use! :-DD
Yep, they're very proud of them, but Apex has been around for a very long time so they're selling them to someone.
They'd be useful for one-offs where you didn't know how craft your own, or your focus was on some other part of the device and you just wanted a quick solution, etc. For example, I've always thought they'd be a quick fit for a galvanometer amplifier. I'm sure there are other applications, and their continued existence seems to confirm that.
-
there's many more power opamps. open any kind of surround sound amplifier these days and you will find plenty of TDA , STK and other series opamps.
i personally like the ICEPOWER opamps. :) no analog muckery.
-
Class D amps are great for audio applications. The Apex products are actual opamps, just with power finals. If all you're doing is audio, Class D (TDA style) is a good choice and will save a ton of money. But since the OP asked:
can they drive loads? Are they suitable for "the power stage" of a project? It doesn't seem so. I would be interested, if someone could point me to a part what I would then call an "OpAmp Driver", that can handle bigger loads.
...something like an actual linear opamp with power capability seems like a better fit.
Basically, figure out your application and choose the proper component.
-
Somebody must be buying/using op amps because the annual sales of analog ICs is expected to hit $69 billion by 2022. Considering the 'jelly bean' price of these devices in volume, there's a lot of analog going on somewhere.
I just left op amps lumped in 'analog ICs' because I didn't see a stand-alone figure.
https://www.globenewswire.com/news-release/2018/04/03/1458822/0/en/Global-Analog-Integrated-Circuits-ICs-Market-Will-Reach-USD-68-97-Billion-by-2022-Zion-Market-Research.html (https://www.globenewswire.com/news-release/2018/04/03/1458822/0/en/Global-Analog-Integrated-Circuits-ICs-Market-Will-Reach-USD-68-97-Billion-by-2022-Zion-Market-Research.html)
-
Perhaps a good way to learn the use of op-amps is to try building some circuits commonly using them without one and see what problems you run into. Try low pass, mid pass and high pass filters, try making a sine wave oscillator, an integrator, or circuits to do simple analog addition or multiplication. Convert a signal between single ended and differential and vice versa, then once you've done all this the hard and painful way, try doing the same stuff with some jellybean op amps and the advantages should become obvious.
-
Lots of the Apex stuff ends up in sonar applications, you don't really care about 40 bucks for a transmit gain block when the transducer array is costing £10k or so.
I would bet their stuff ended up in some old design that Thales wound up the design authority for by purchasing the original manufacturer, at which point ANY change becomes a sufficient paperwork dance to make it not worth it.
Regards, Dan.
-
Apart from the hijackers: thank you very much guys! :) ;D
The linked PDF book is a very good read, I like that! Must read more. 8)
I understand for now, that such an OpAmp is a very flexible signal amplifier and in terms of "manufacturing tolerance" and reliability and signal quality to prefer, instead of simple BJTs.
But what I don't understand is: can they drive loads? Are they suitable for "the power stage" of a project? It doesn't seem so. I would be interested, if someone could point me to a part what I would then call an "OpAmp Driver", that can handle bigger loads.
(Maybe that's why I never saw a use for them in any of my projects so far, because I understand 'amplification' as being able to 'amplify voltage AND current'. [A MOSFET with 1.4mΩ Rdson is not even slightly impressed, when there's a current of 5A or so ... would an OpAmp be?])
However, you guys helped me a lot! Thanks! :-+
I was glad to see you ask this question as I don't really see the day to day use for them either, but I think that is because my projects have not progressed that far. I was hoping for some sort of revelation here, and I agree, you did get a lot of op amp information that did not answer your question. I think it is a question that will get clearer once we get more involved in the proper type of circuits. I did download a copy of the Op Amp book by Walt Jung, but at 800 pages it is going to take a while to digest :) Tempted to buy a used copy as I find a printed book easier to work with, guess that's the printer in me :)
-
But what I don't understand is: can they drive loads? Are they suitable for "the power stage" of a project? It doesn't seem so. I would be interested, if someone could point me to a part what I would then call an "OpAmp Driver", that can handle bigger loads.
Operational amplifiers are designed two drive their expected load, which is often the input network to another operational amplifier, *and* their feedback network. So general purpose parts are usually specified for loads of 2 to 10 kilohms. Low voltage noise parts can drive much heavier loads down to hundreds of ohms because they must use low impedance network to preserve their low voltage noise. Audio parts might be expected to work with 32 ohm headphone loads.
Video parts can usually drive 150 ohms from a double terminated 75 ohm transmission line plus their own feedback network. RF parts of course will do the same for 100 ohms from double terminated 50 ohm transmission line. Some Video and RF parts can drive multiple double terminated transmission lines in parallel.
High output current comes at a price. The larger transistors take up more space making the die more expensive. The large transistors also will not fit in the smallest packages. Higher output current requires higher power dissipation. Higher power dissipation creates temperature gradients across the semiconductor die which compromises precision. Thermal feedback from the output transistors to the input transistors actually limits open loop gain among other things so high precision (and low distortion) designs benefit from using an external buffer if they must drive heavy loads like the low impedance feedback network required for low noise.
Back in the 70s, just about everybody who made a 741 had a power version of the 741 like the uA791.
(Maybe that's why I never saw a use for them in any of my projects so far, because I understand 'amplification' as being able to 'amplify voltage AND current'. [A MOSFET with 1.4mΩ Rdson is not even slightly impressed, when there's a current of 5A or so ... would an OpAmp be?])
A very common audio power amplifier topology is just an operational amplifier scaled up to provide hundreds of watts; it has the same 3 or sometimes 4 stages doing the same things.
-
It, of course, entirely possible to have an opamp that doesn't fit on a single die. The first opamps were all like that.
Compound opamps avid the weaknesses of integrated opamps, e.g. higher voltage or power etc.
-
You can also get opamp boosters that are typically more or less unity gain POWER stages that are fast enough to place within the feedback loop after the opamp proper and could provide significant output current. The trick is to use a booster that is MUCH faster then the opamp that it is boosting so it does not eat too far into available phase margin.
See for example the BUF634 from TI.
BTW, that 1.4mR mosfet will absolutely notice 5A is you run it in the linear region with say 5V across it (25W will be dissipated), those things are generally designed as switches not linear components. You can get mosfets for linear use, but they tend to be packaged to dissipate the inevitable heat and have much less emphasis on low RDS(on) as a figure of merit.
Regards, Dan.
-
Hello there,
The op amp is most basically a negative feedback amplifier with high gain. This combination means we can get very low error at the output.
The basic operation in most pure analog op amp circuits is the output of the op amp tries to force the two inputs to be the same value. As it does this, the side effect is that it provides some function like amplification of the input signal.
An interesting side effect is that the output is the inverse of whatever we put in the negative feedback path. For example if we put a high pass filter in the feedback path then the output becomes a low pass filter for the input signal. If we put a squaring circuit in the negative feedback path then we get the square root of the input at the output.
Probably the reason it is called an Operational Amplifier is because it is basically an amplifier that enjoys very widespread applications. From amplifiers to filters to math operations to even logic. The limiting factor comes from the specifications of the specific op amp being considered.
The LM358 is a very popular and low cost op amp very suitable for experimentation. It has limitations but the cost is very low.
By contrast the LM741 that used to be popular is much harder to use.
-
If you take rail to rail input and output opamp you can have voltages at input and/or output at the supply voltages, else you have a rail offset with normal opamps wich limits your dynamic range.
These TLV272IP are nice opamps ( someone gave me the hint in this forum and i tryd ), they have input to negative rail, cost around 1 euro, if you use only 5 volt those MCP are cheaper.
For the rest i go with transistors to make things smaller if possible and it wont sound bad, these opamps cost more power.
-
Common rail-to-rail parts are not universal improved replacements however.
Rail-to-rail input parts have compromised precision and common mode rejection over their input common mode range, usually close to the positive supply. Some parts do various things to improve this situation like external control of exactly where the input crossover point is so that it can be moved outside of the common mode input range but such a specialized part comes at a premium in cost and interoperability. Rail-to-rail and single supply inputs also inherently lack input bias current compensation because it requires compliance beyond the input range.
Rail-to-rail output parts only approach the supply rails so if operation all the way to the supply rails is required, they are not the solution. Their common emitter/source output stages are also more finicky about load impedance.