What's the maximum gain you can run an OA at?

[smoguzbenjamin]

I was wondering,

What is the maximum gain you can run an opamp at? Is it mentioned in datasheets and am I blind or is there an unspoken border? I guess it should change per opamp, but I ain't sure. Anyone have some numbers?

[Ansil]

from what i have heard it varies with your supply voltage and the ability of th oa to swing ..   ok dirty minded people here i am talking the rail to rail or if it has a fixed swing. of a few volts shy of its v+ personally iuse r to r opamps now for every opamp except clean things.. and i have found a interesting new chip from TI that closely mocks the lm386

[R.G.]

What is the maximum gain you can run an opamp at? Is it mentioned in datasheets and am I blind or is there an unspoken border?

This is another of those questions that ought to be simple, but isn't.

Opamps are intended to be used with negative feedback, and so they have to be stable (that is, not oscillate) when negative feedback is used. This requirement turns out to limit the available gain as a function of frequency.

Opamps have huge gains in general before feedback. Typical *DC* voltage gains are one hundred thousand to several million. However, to be stable with negative feedback, that gain has to decline to under 1 before the frequency a  which internal phase shifts add up to 180 degrees, making the chip oscillate.

This is different for every opamp, as you supposed. Most opamp data sheets specify open loop gain as "V/V" and the unity gain frequency as a single frequency. older bipolar "741" style opamps were typically 1MHz unity gain. TL07x opamps are about 3Mhz unity gain, and others vary. 13 MHz is common, and there are newer ones with much higher bandwidths.

The crux is this: The highest gain you can use is limited by this falloff of gain versus frequency. At DC, you can use all million or so. At 1kHz, you might be able to use several thousand. At a one MHz, you're lucky to get a gain over 1, even open loop.

None of these are practical limitations, though. What limits the usable gain is much more likely to be layout and parasitic coupling at high gains, not the capabilities of the opamp.

[Mark Hammer]

....which is why you'll see the term "gain bandwidth" used.  A "gain bandwidth" product of 10meg mean you can successfully aim for a 10khz bandwidth at a gain of 1000 (1000 * 10,000 = 10,000,000), a 20khz bandwidth with a gain of 500, etc.

Of course, the gain you apply is meaningless once the signal amplitude exceeeds the maximum swing for the supply voltage.

Start with a supply-voltage of +/-15v (that's pretty healthy), and a maximum swing to one volt less than each rail (that's not unreasonable).  That affords an absolute maximum pre-squishing ceiling of +/-14v peak-to-peak.  What's the maximum gain you could hope for at ANY bandwidth and using a 1V peak-to-peak input signal?  Twenty-eight (starting with a half-volt either side of the midpoint).

Guitar signals tend to produce 50-100mvAC for most types of picking and pickups (less if you play just so, more if you have heavy strings and slam).  The maximum gain of the *circuit* (which is, of course, not your question and not the same as the gain limits of an individual op-amp) prior to clipping, with our hypothetical supply and max swing would be about 280 for the most part.  That could be distributed multiple ways; e.g., one stage with a gain of 28 and another with a gain of 10.

Lucky for guitar players we *like* clipping.

[Paul Marossy]

I remember seeing on some data sheets something like max. voltage = 36V or something to that effect. That means you can have a bipolar power supply that swings +18/-18 volts. Using +15/-15 volts as Mark suggested gives you a good safety margin. Most guitar effects use 9V and a 4.5Vref, which fools the opamp into thinking that it is working on a bipolar power supply.