# Op Amp - Specifications are given for unity gain, how to choose another gain?

TL081 datasheet gives the following specifications:

• How can I find the open loop amplification?
• I know that the relationship between the differential voltage amplification and the (closed or opened?) amplification is A=Avd vid+Avc vic where Avc is much lower than Avd, but how can I find A?
• The bandwidth for unity gain is 3MHz. Assuming that I have a single pole, I will cross 20 kHz (I want to have an open loop gain of, minimum, 20kHz for audio applications) at an open loop amplification of approximately 40 dB. That's very bad. Does that mean that TL081 is not a good op amp for audio?

I need to design a preamplifier using that op amp, and the only information I have is the image I posted. This is an exam exercise, I don't have more information.

Thanks!

The open loop gain is the 'voltage amplification' line in the data sheet above. It specifies 15 or 25V/mV minimum (depending on temperature) and typically 200V/mV at room temp. Let's call that 100V/mV to make the sums easier, but keep in mind it could be worse.

The GBW is 3MHz typical (so it could be worse, an unspecified amount worse!). That means you can expect unity open loop gain at 3MHz, x10 at 300kHz, x100 at 30kHz etc.

If you make a controlled gain stage, using feedback resistors to set the closed loop gain (check the terminology you've used in your OP), then it will have completely run out of feedback gain at 30kHz, and won't have much at 20kHz. So, under the conditions that the gain is specified, it would not make a very good amplifier, at 20kHz, at 10v swing, if it was a typical gain amplifier. If it was a minimum gain amplifier, the 40dB bandwidth would be rather less than 20kHz.

The gain is specified as large signal gain. What's the expected small signal gain, if we wanted only a 1v line level out? We could expect a bit more, but how much? The data sheet remains silent on that.

If you want 40dB gain per stage, then as specified, it doesn't appear to be a good audio amplifier. Bear in mind that any distortion caused by inadequate feedback gain will not be audible if it occurs on single signals in the 10-20kHz region, the harmonics will fall out of band, and the content of audio signals tends to be dropping at high frequencies anyway. Intermodulation products will fall in band though. If you wanted 20dB gain per stage, and 1v output level, then the picture is rather better.

On a historical note, when the venerable 741 came out in 1900 and whatever, it was taken up enthusiastically by guitar and PA amplifier manufacturers. It wasn't ideal, low gain, high noise, low slew rate, but the products did sell. The TL081 is head and shoulders above the 741 for those three parameters.

If you were to build professional audio stuff, you may want to look at things like OP275 and the like. But use TL081 first, and see whether it sounds adequate.

• Perfect. I studied a lot this exercise and I got the same conclusiones that you posted :). But what can I do if I only have the information that I put in the main thread? How can I find the open loop amplification (I call it "a", and A=a/(1+af))? How can I "translate" from unity gain feedback to another gain of the feedback? – JFKalman Jul 8 '17 at 16:58

A picture helps (from the data sheet): -

As you can see (hopefully): -

• Open loop gain is typically about 2 x $10^5$ (or 200 V/mV)
• Open loop gain at 20 kHz is about 2 x $10^2$ (or 46 dB)

As to whether the TL081 is any good for audio that depends on how much amplification your circuit needs. For a unity gain amplifier, 46 dB open-loop gain is fine but, if you wanted a closed-loop gain of 100 (40 dB) you would begin to be a little wary about using the TL081 but, remembering that most people won't even hear 20 kHz and any harmonic distortion certainly won't be heard, I expect, in most applications, the TL081 will be good for closed-loop gains up to 100.

• Thanks for your answer. With that graphic I know the value of the Open Loop Gain, but what can I do if I only have the data that I put in the main post? – JFKalman Jul 8 '17 at 16:55
• @JFKalman the trick to this is recognising that the slope of the graph is 6 dB per octave or 20 dB per decade. That sloping line will intercept the dc open loop gain which is specified as 200 volts per milli volt and the same value i estimated from the graph. Knowledge and the two numbers paints the graph. Just look at the graph - at 3 MHz it's 0 dB and at 300 kHz it's 20 dB etc.. – Andy aka Jul 8 '17 at 19:46

The bandwidth for unity gain is 3MHz. Assuming that I have a single pole, I will cross 20 kHz (I want to have an open loop gain of, minimum, 20kHz for audio applications) at an open loop amplification of approximately 40 dB. That's very bad.

that's only applicable to VFB opamps. CFB opamps don't have that concept.

Does that mean that TL081 is not a good op amp for audio?

it depends on what you are using it for. gain is only one part of the equation when picking an opamp (or any active device), and in many cases it is a not-so-important part of that equation.

• Why do you say that Current Feedback op amps don't have that concept? – JFKalman Jul 8 '17 at 16:59

The opamp gives you 46dB open loop gain at 20kHz. You use some of that as actual gain in your circuit, and the rest is your feedback, whose purpose is to correct for open loop errors and distortion.

For example, if you want high DC accuracy at high gain for a sensor, then you need high open loop gain to minimize gain error. This isn't for audio.

Here, I assume you'd like your closed loop bandwidth to be at least 20kHz. But with your 40dB gain, you only have 6dB of extra OL gain left. That's assuming you get the 3MHz the datasheet says are "typical". If the weather is a bit cold, or too hot, and the opamp you get isn't the best in the wafer, maybe you won't have the typical 3MHz, you could have a bit less, and then your 6dB margin evaporates, and the actual closed loop bandwidth of your circuit will be reduced, and it will no longer pass 20kHz.

Also, 6dB of feedback is insufficient to correct for open loop nonlinearities. TL081 has a really wimpy output stage (notice its distortion is specified into a very high resistance load, not 600 ohms like beefier opamps). So you really need that extra feedback around the output stage.

I'd use another opamp, with more GBW... and maybe less noise also, because this one has 18nV/rtHz which is ENORMOUS! You can also split your 40dB gain into two 20dB stages, using two opamps.

Your single TL081 could well perform worse than a pair of crummy 4558's.

If this is DIY, do yourself a favor and don't penny-pinch, put an OPA1652 in there, its price is really reasonable (1.5€ qty 1) and it has excellent performance, although it too dislikes low impedance loads.

You can also use decompensated opamps which will only be stable at high gains. This gives you extra GBW, if you find the right one.

• Thanks!! This is a very helpful answer. But this in an exercise from an exam and I only have the data that I posted. I can't change the op amp, and I was told to build a preamplifier with that. – JFKalman Jul 8 '17 at 17:02
• In this case, you could measure distortion vs frequency (use a PC soundcard and some software like a freeware version of RMAA) and then make an interesting point about how distortion at high frequency will suck due to the lack of OL gain. – peufeu Jul 8 '17 at 21:56

TL081 has JFET input diffpair, so that source of distortion will be better than bipolars. Whether you have single tones, or standard music program material, the distortion will be inherently ----- even with little 1/GH ---- lower.

If you can tolerate inaccurate gains, use the TL081.

If you can tolerate poor output drive, use the TL081.