Op-Amp Gain: The Feedback Ratio That Fails When Bandwidth Runs Out

Op-amp gain looks like one of the cleanest calculations in electronics.

For an inverting amplifier:

Av = −Rf / Rin

For a non-inverting amplifier:

Av = 1 + Rf / Rg

Pick the resistors, calculate the ratio, and the gain is done.

At least, that is the way it looks in a textbook.

In real analogue design, the resistor ratio is only the first layer. The circuit can have the correct gain on paper and still fail because of bandwidth, noise gain, output swing, input common-mode range, resistor value selection, bias current, or a simple misunderstanding of what the op-amp is actually being asked to do.

The trap is that op-amp gain is easy to calculate but easy to over-trust.

The basic inverting amplifier formula

The classic inverting op-amp gain is:

Av = −Rf / Rin

Where:

Av = closed-loop signal gain
Rf = feedback resistor
Rin = input resistor

The negative sign matters.

It tells you that the output is inverted relative to the input.

If:

Rf = 100 kΩ
Rin = 10 kΩ

Then:

Av = −100,000 / 10,000
Av = −10

So a 100 mV input signal becomes:

Vout = −10 × 0.1 V
Vout = −1.0 V

That is simple and useful.

But it is not the full design check.

Gain in dB

Voltage gain is often expressed in decibels:

Gain dB = 20 × log10(|Av|)

For a gain magnitude of 10:

Gain dB = 20 × log10(10)
Gain dB = 20 dB

For a gain magnitude of 100:

Gain dB = 20 × log10(100)
Gain dB = 40 dB

For a gain magnitude of 1000:

Gain dB = 20 × log10(1000)
Gain dB = 60 dB

This is a good quick sanity check.

A single-stage gain of 10 or 20 may be routine.

A gain of 100 may be possible, but bandwidth, noise, and offset start to matter much more.

A gain of 1000 in one op-amp stage is usually a warning sign. It may work in a narrow low-frequency case, but it often deserves a two-stage design or an instrumentation amplifier instead.

Non-inverting amplifier gain

For a non-inverting amplifier, the gain is:

Av = 1 + Rf / Rg

Where:

Rf = feedback resistor
Rg = resistor from inverting input to ground/reference

If:

Rf = 99 kΩ
Rg = 1 kΩ

Then:

Av = 1 + 99,000 / 1,000
Av = 100

The gain is positive, so the output is not inverted.

The important detail is that the minimum gain of the standard non-inverting configuration is 1.

You cannot get a gain of 0.5 from the basic non-inverting formula. If you need attenuation, you usually need a divider, an inverting attenuator, a difference stage, or another topology.

This is a common mistake when someone tries to use a non-inverting op-amp as a “buffer with adjustable gain below unity.”

A voltage follower can buffer at unity gain.

It cannot attenuate by itself.

Signal gain vs noise gain

The most important op-amp gain mistake is confusing signal gain with noise gain.

For an inverting amplifier, signal gain is:

Av = −Rf / Rin

But the noise gain is:

Av_noise = 1 + Rf / Rin

These two numbers are not the same.

Example:

Rf = 100 kΩ
Rin = 10 kΩ

Signal gain:

Av = −100,000 / 10,000
Av = −10

Noise gain:

Av_noise = 1 + 100,000 / 10,000
Av_noise = 11

The signal is amplified by −10.

But the op-amp’s input-referred noise, offset, and bandwidth behavior are governed by a noise gain of 11.

That small difference may not matter in a rough low-frequency circuit.

But it matters when checking bandwidth and stability.

Gain-bandwidth product

For many voltage-feedback op-amps, a first-pass closed-loop bandwidth estimate is:

BW = GBW / Av_noise

Where:

BW = estimated closed-loop bandwidth
GBW = op-amp gain-bandwidth product
Av_noise = noise gain

This is a screening calculation, not a complete stability analysis. But it is extremely useful.

Suppose an op-amp has:

GBW = 1 MHz

And the inverting stage uses:

Rf = 100 kΩ
Rin = 10 kΩ

We already calculated:

Signal gain = −10
Noise gain = 11

So the estimated closed-loop bandwidth is:

BW = 1,000,000 / 11
BW ≈ 90,909 Hz
BW ≈ 90.9 kHz

That may be acceptable for a 10 kHz signal.

It may be marginal for a 50 kHz signal depending on amplitude accuracy and phase shift requirements.

It is not acceptable if someone expected a clean 100 kHz amplifier with no margin.

This is where the resistor-ratio calculation can mislead the design.

The gain can be mathematically correct while the selected op-amp is too slow.

Worked example: a sensor amplifier that looks right but fails at frequency

Suppose an engineer needs to amplify a sensor signal.

The requirement is:

Input signal amplitude = 50 mV
Target output amplitude = 2.5 V
Signal frequency = 20 kHz
Op-amp GBW = 1 MHz
Topology = non-inverting

The required gain is:

Av = 2.5 / 0.05
Av = 50

For a non-inverting amplifier:

Av = 1 + Rf / Rg

Choose:

Rg = 2 kΩ
Rf = 98 kΩ

Then:

Av = 1 + 98,000 / 2,000
Av = 50

So far, the resistor calculation looks perfect.

Now check bandwidth.

For a non-inverting amplifier:

Av_noise = Av
Av_noise = 50

Estimated closed-loop bandwidth:

BW = GBW / Av_noise
BW = 1,000,000 / 50
BW = 20,000 Hz
BW = 20 kHz

The required signal frequency is also 20 kHz.

That means the amplifier is being used right at the estimated closed-loop bandwidth.

That is not a comfortable design.

At this point, the gain may be down, the phase shift may be significant, and the waveform may not match the intended signal-chain requirement.

The resistor ratio gave the right gain.

The op-amp selection did not give enough bandwidth margin.

A better design might use:

A higher-GBW op-amp
Lower gain per stage across two stages
A different signal-chain architecture
A reduced target bandwidth

The engineering mistake was not the gain formula.

The mistake was stopping at the gain formula.

The output swing check

Another simple failure mode is ignoring the supply rails.

Suppose the same amplifier has:

Input amplitude = 100 mV
Gain = 50

The expected output amplitude is:

Vout = 0.1 × 50
Vout = 5 V

If the op-amp is powered from a single 5 V supply, that output is probably not possible.

Even a rail-to-rail op-amp cannot usually swing perfectly to both rails under load. A non-rail-to-rail op-amp may need much more headroom.

So a gain calculation can say:

Av = 50

But the real circuit says:

The output saturates.

This is why every gain check should be followed by a voltage swing check:

Expected output min/max
Supply voltage
Output swing limit from datasheet
Load resistance
Required signal headroom

The op-amp does not care that the resistor formula is correct if the output stage cannot physically deliver the voltage.

Resistor values matter

The ratio sets gain, but the absolute resistor values still matter.

For an inverting gain of −10, all of these ratios work mathematically:

Rf = 1 kΩ, Rin = 100 Ω
Rf = 10 kΩ, Rin = 1 kΩ
Rf = 100 kΩ, Rin = 10 kΩ
Rf = 1 MΩ, Rin = 100 kΩ

All give:

Av = −10

But they are not equivalent designs.

Very small resistor values can load the op-amp output and waste current.

Very large resistor values increase sensitivity to input bias current, leakage, noise, parasitic capacitance, and PCB contamination.

For many general-purpose analogue stages, the practical resistor range is often somewhere around:

1 kΩ to 100 kΩ

That is not a universal rule. Precision, low-noise, high-speed, low-power, and sensor front-end designs may need different choices.

But it is a useful first-pass sanity check.

If a gain stage uses 10 Ω feedback resistors, ask why.

If it uses 50 MΩ resistors, ask why.

The ratio alone is not the whole design.

Difference amplifier: resistor matching matters more than the ratio alone

For a simplified matched difference amplifier:

Av_diff = Rf / R1

This assumes matched resistor pairs:

R1 = R3
Rf = R2

The ideal formula is clean.

But real common-mode rejection depends strongly on resistor matching.

That means two difference amplifiers can have the same nominal gain and very different real-world performance.

For example, a difference amplifier may be intended to reject a large common-mode voltage while amplifying a small differential signal.

If the resistor ratios are not matched closely, part of the common-mode signal appears at the output as an error.

So the mistake is thinking:

The gain is correct, so the difference amplifier is correct.

A better check is:

Is the resistor network matched tightly enough for the required CMRR?
Is the input common-mode voltage inside the op-amp limit?
Is the output swing still inside the rails?
Is an instrumentation amplifier more appropriate?

A discrete four-resistor difference amplifier can work well.

But for precision measurement, matched resistor networks or instrumentation amplifier ICs are often the better choice.

Summing amplifier: each input has its own gain

For an inverting summing amplifier:

Av_n = −Rf / Rn

Each input channel has its own resistor and its own gain.

For example:

Rf = 10 kΩ
R1 = 10 kΩ
R2 = 5 kΩ
R3 = 2 kΩ

Per-channel gains:

Av1 = −10,000 / 10,000 = −1
Av2 = −10,000 / 5,000 = −2
Av3 = −10,000 / 2,000 = −5

The output is:

Vout = −(1 × V1 + 2 × V2 + 5 × V3)

The common mistake is checking only the highest per-channel gain and forgetting the summed output magnitude.

Even if each input is inside a reasonable range, several channels can add together and drive the output into saturation.

For example:

V1 = 1 V
V2 = 1 V
V3 = 1 V

Then:

Vout = −(1 + 2 + 5)
Vout = −8 V

That may be fine on a ±12 V supply.

It will not be fine on a single 5 V supply.

Summing amplifiers need both gain checks and worst-case output swing checks.

The common engineering mistake: using gain as the only design criterion

A very common op-amp design review looks like this:

Required gain = 100
Rf = 100 kΩ
Rin = 1 kΩ
Av = −100
Done

But a useful review should continue:

What is the noise gain?
What is the closed-loop bandwidth?
What is the input signal frequency?
What output swing is required?
Can the op-amp output reach that voltage on the selected supply?
Are the inputs inside the common-mode range?
Are the resistor values practical?
Will bias current create an offset error?
Is the single-stage gain too high?
Is the op-amp stable in this configuration?

The resistor ratio is the beginning of the design.

It is not the design.

Practical design takeaway

Use the gain formula to get the first answer.

Then immediately check the things the formula does not cover.

For an inverting amplifier:

Av = −Rf / Rin
Av_noise = 1 + Rf / Rin
BW = GBW / Av_noise

For a non-inverting amplifier:

Av = 1 + Rf / Rg
Av_noise = Av
BW = GBW / Av_noise

For gain in decibels:

Gain dB = 20 × log10(|Av|)

Then ask the practical questions:

Is the bandwidth high enough?
Is the output swing possible?
Are the resistor values reasonable?
Is the input common-mode range valid?
Is the op-amp stable at this noise gain?
Is the single-stage gain realistic?

That workflow catches many problems before simulation, PCB layout, or bench testing.

Final thought

Op-amp gain is simple when it is only a resistor ratio.

But real circuits do not fail only because the ratio is wrong.

They fail because the gain was correct at DC but not at the required frequency.

They fail because the output saturated.

They fail because the designer confused signal gain with noise gain.

They fail because a 1000× stage was forced into one op-amp when two moderate-gain stages would have worked better.

The resistor formula is still the right starting point.

It just should not be the stopping point.

For quick checks of inverting, non-inverting, voltage follower, difference, and summing amplifier gain — including gain in dB, noise gain, and closed-loop bandwidth screening — use the Op-Amp Gain Calculator on CalcEngineer.