What does "frequency" mean when discussing op amps?

[anfang]

There are plenty of websites that explain how to get more frequency from an op amp, what are the best frequencies at which certain op amps work and so on. But I haven't found a single book or blog post that will tell me: what is this frequency? Where does it come from?

I understand that it is something that has to do with the feedback loop. I imagine that it takes a while for the feedback loop to stabilize to the expected value. Is the frequency just

Code: [Select]

                     1
--------------------------------------------
amount of time the op amp needs to stabilize

?

[danadak]

Generally speaking it is the GBW (Gain Bandwidth) of the Opamp. So
think of a signal, like a sine, fed into an OpAmp, sweep it up in
cycles / sec and plot the output of the amplifier.

Here is example of a single pole compensated OpAmp open loop,
no feedback, curve, attached.

https://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=0ahUKEwj8-Jqvip_PAhWIbSYKHTI-Bp8QjRwIBw&url=http%3A%2F%2Fwww.ecircuitcenter.com%2FOpModels%2F2nd_and_3rd_Poles%2F2nd_and_3rd_Poles.htm&psig=AFQjCNHfeE0M7sxyysRCRIJOiAosAhz3tA&ust=1474499911632283

A single pole compensated OpAmp has a fixed GBW, so if it was 1 Mhz as an example,
the amp would pass 1 Mhz at 0 db gain, if loop was closed for a G = 10 then the 0 db
output freq would be 100 Khz. G X BW = Constant - 1 Mhz in this example.

Note there are other considerations about BW, one is slew rate. The amp must slew the Comp Cap
and it only has finite current. So small signal BW is >> slew rate limited BW. For a sine thats


http://www.radio-electronics.com/info/circuits/opamp_basics/operational-amplifier-slew-rate.php

Google "opamp books free", more info than you want to know in these books.

Keep in mind there are Voltage mode and Current mode opamps, as well as switch mode,
all with specific and common capabilities.


Regards, Dana.

[T3sl4co1l]

For a small (~mV), sinusoidal waveform applied between the input pins, the output magnitude equals the input magnitude, at the frequency called the gain-bandwidth frequency (GBW).

In general, op-amp gain varies inversely with frequency; they are best described as integrators with limited DC gain.  (A typical analysis considers op-amps as infinite gain amplifiers, but this is not only impossible to realize, but also impossible to stabilize -- such a concept can only ever work at DC.  The real world has AC, no matter how slow you go, so it is impossible to ignore AC outright.)

Also, it's noteworthy that such op-amps are voltage mode, dominant pole compensated types.  There are several other types, which have different responses.

Tim

[CatalinaWOW]

Another simplified way to think of it is that the op-amp circuit will perform pretty close to the way the cookbooks tell you for simple circuits like low pass filters, band pass filters and high pass filters at frequencies about one tenth of the gain bandwidth product.  Audio applications will generally work fine with the typical 1 MHz GBW of jelly bean op amps.  It you are doing something at higher frequencies you will need a better op-amp.

Pretty close may or may not be good enough.  You will either have to dive deeper into the math and understanding of what all this is about, or build some and test them to see if you are happy with the results.  Or even better, do both and understand whether your test and theory match up.

[matseng]

Don't forget the slewrate.... It can pop up and bite you in the *ss even if the GWP is within limits for your application.

This short tutorial video explains it very well.
https://youtu.be/UooUGC7tNRg

[anfang]

So the frequency we are talking about is the AC frequency, right?

Up to now I have tried only DC circuits, so the only frequency I deal with are that of PWM square waves. I suppose I will not have to deal with bandwidth for quite a long time. 

[T3sl4co1l]

DC is zero frequency by definition; there's no such thing as "PWM DC".  That's AC, quite richly AC I might add -- with a tinge of DC superimposed upon it.

Tim

[Zero999]

Op-amp stages can be cascaded together for more bandwidth at higher gains but it won't overcome the slew rate limitation, nor will it increase the gain bandwidth product.

[Brumby]

Whenever the term 'Frequency' is used, especially in regards to bandwidth, filters and similar topics, it is usually referring to the sine waves that are present in the signal being discussed.

Typically, the figures being used are, in fact, specified for a single frequency with no harmonics - that is, a pure sine wave.  When bandwidth, gain, attenuation, etc. are talked about in relation to 'frequency' they are describing the changes observed when pure sine waves of different frequencies are put through the circuit, one at a time.

However, the real world is a little more complex...

DC is zero frequency by definition; there's no such thing as "PWM DC".  That's AC, quite richly AC I might add -- with a tinge of DC superimposed upon it.

Tim

Any signal of changing amplitude that is not a pure sine wave will have more than one frequency in it.  For example, a square wave has harmonics - and lots of them.  PWM is a particular variant of a square wave and, as indicated above, has a generous content of multiple frequencies.

However, it is often the case that the speeds at which PWM systems operate and the applications they are found in are not particularly sensitive to the true AC profile of the signal, so you don't need to be too concerned about them.  BUT, the multiple frequencies are still there and they will exert their influence.  You may have heard this when, say, a 1KHz PWM signal causes a 'buzzing' sound in a nearby audio circuit.  It is all those harmonics above the 1KHz that cause that.

[danadak]

A simple treatment of cascaded identical stages with dominant pole
individual stage compensation/frequency response.


http://forum.allaboutcircuits.com/threads/op-amps-and-gain-bandwidth-product.43799/


Regards, Dana.

[John_ITIC]

To make it even simpler: At a certain point, an op-amp is unable to swing its output (positive or negative) quickly enough to track its input signal. This results in a lower output amplitude than what is seen at lower frequency input sine waves. That would then be the "frequency" of the op-amp.