Real Op-Amps: Limitations & Frequency Response Revision Notes | A Level AQA Physics

1. Why Real Op-Amps Are Not Ideal

Every op-amp circuit you have analysed so far has relied on a convenient fiction: the ideal op-amp. Infinite gain, infinite input impedance, zero output impedance, and perfect behaviour at all frequencies. Real devices — the 741, the LM358, the TL071 — fall short of every one of those assumptions.

That gap between ideal and real matters enormously in practice. A circuit that looks perfect on paper can produce a drifting output, distort high-frequency signals, or oscillate unexpectedly — all because of the limitations built into every real op-amp. Understanding those limitations is what separates a circuit that works in a textbook from one that works on a bench.

2. Limitations of Real Operational Amplifiers

The ideal model assumes infinite open-loop gain. In reality, the open-loop gain $A_{OL}$ of a typical op-amp is large — often around $10^5$ at DC — but it is finite. More critically, it is not constant: it falls steeply as frequency increases. At DC and very low frequencies the gain is at its maximum; by the time you reach a few hertz or tens of hertz for many general-purpose devices, the gain has already begun to drop.

A second real-world departure from the ideal is input offset voltage ($V_{OS}$). In a perfect op-amp, zero volts in gives zero volts out. In a real device, tiny mismatches in the internal transistor pairs mean a small differential voltage — typically a few millivolts — appears at the input even when both terminals are grounded. This gets amplified by the closed-loop gain and can produce a significant, unwanted DC offset at the output.

Related to this are input bias currents. The ideal model assumes no current flows into either input terminal. Real op-amps draw small currents — typically nanoamperes for bipolar devices, picoamperes for FET-input types — into both inputs. These bias currents flow through any resistance in the circuit and create additional voltage errors via Ohm's law.

Slew rate limiting is particularly destructive for large-amplitude, high-frequency signals. A 741 op-amp has a slew rate of roughly 0.5 V $\mu$s$^{-1}$. If you ask it to reproduce a 10 V peak sine wave at 100 kHz, the output cannot keep up — the waveform becomes severely distorted, approaching a triangular shape as the output is reduced to a series of rising and falling ramps. This is a hard physical limit, not something a feedback network can fix.

Finally, real op-amps have a finite output impedance (not zero as assumed in the ideal model) and a finite common-mode rejection ratio (CMRR). CMRR describes how well the op-amp rejects signals that appear equally on both inputs — a perfect device would reject them entirely; a real device amplifies them slightly. High CMRR is essential in instrumentation applications where the signal of interest sits on top of a large common-mode noise voltage.

3. The Frequency Response Curve

Now that the individual limitations are clear, the most important one to analyse in depth is the frequency dependence of gain. Picture a graph with frequency on the horizontal axis (plotted on a logarithmic scale, from 1 Hz to perhaps 10 MHz) and voltage gain on the vertical axis (also logarithmic, in decibels or as a plain ratio). This is the frequency response curve of the op-amp.

The curve has a characteristic shape. At low frequencies the gain is flat at its maximum value — this is the DC open-loop gain. At a certain frequency called the dominant pole frequency (or $-3$ dB frequency), the gain begins to fall. For most general-purpose op-amps this roll-off is at a surprisingly low frequency — often just a few hertz to a few tens of hertz. Beyond that point, the gain falls at a rate of $-20$ dB per decade (equivalently, $-6$ dB per octave), which means every time the frequency increases by a factor of ten, the gain drops by a factor of ten.

The roll-off continues until the gain reaches unity — a gain of exactly 1, or 0 dB. The frequency at which this occurs is called the unity-gain bandwidth, sometimes written $f_T$ or $f_{unity}$. For a 741 this is approximately 1 MHz; for faster op-amps it can be tens or hundreds of megahertz.

4. Gain-Bandwidth Product

The single most useful number to extract from the frequency response curve is the gain-bandwidth product (GBP). Because the open-loop gain rolls off at $-20$ dB per decade after the dominant pole, the product of gain and frequency is constant across that entire roll-off region. This means you can trade gain for bandwidth, or bandwidth for gain, but you cannot have more of both simultaneously.

Here is the practical consequence: if you configure an op-amp with a closed-loop gain of 100 and the GBP is 1 MHz, the usable bandwidth is only 1 MHz / 100 = 10 kHz. Reduce the gain to 10 and the bandwidth opens up to 100 kHz. This trade-off is not a design flaw — it is a fundamental property of the internal compensation that makes the op-amp stable. Knowing the GBP lets you predict, before building anything, whether your amplifier will work at the frequencies you need.

Worked example: Calculating Bandwidth from Gain-Bandwidth Product

An op-amp has a unity-gain bandwidth of 3 MHz. It is configured as a non-inverting amplifier with a closed-loop gain of 150. Calculate the $-3$ dB bandwidth of this amplifier circuit.

1. $\text{GBP} = f_T = 3 \text{ MHz} = 3 \times 10^6 \text{ Hz}$ — For a single-pole op-amp, the GBP equals the unity-gain bandwidth. State this clearly before substituting.
2. $\text{GBP} = A_{CL} \times f_{-3\text{dB}}$ — Rearranging for bandwidth: $f_{-3\text{dB}} = \frac{\text{GBP}}{A_{CL}}$
3. $f_{-3\text{dB}} = \frac{3 \times 10^6}{150}$ — Closed-loop gain = 150, GBP = 3 × 10^6 Hz. Units: Hz / (dimensionless) = Hz.
4. $f_{-3\text{dB}} = \frac{3 \times 10^6}{150} = \frac{3000000}{150} = 20000 \text{ Hz}$ — 3 000 000 ÷ 150: divide 3 000 000 by 15 to get 200 000, then divide by 10 to get 20 000. Check: 150 × 20 000 = 3 000 000. Correct.
5. $f_{-3\text{dB}} = 20 \text{ kHz}$ — 20 000 Hz = 20 kHz. This is the highest frequency the amplifier can handle before its gain drops by 3 dB (falls to $\frac{A_{CL}}{\sqrt{2}}$).

Answer: The $-3$ dB bandwidth is 20 kHz.

5. Reading and Interpreting the Frequency Response Curve

When you encounter a frequency response curve in an exam question, there are four features to identify immediately. First, read off the DC open-loop gain from the flat portion at the left of the graph. Second, locate the dominant pole frequency — the point where the curve begins to fall, which is also the $-3$ dB point of the open-loop response. Third, note the slope of the roll-off: a well-compensated op-amp rolls off at exactly $-20$ dB per decade. Fourth, read the unity-gain bandwidth where the curve crosses 0 dB.

A horizontal line drawn at the closed-loop gain value intersects the open-loop curve at a frequency — that intersection point is the $-3$ dB bandwidth of your closed-loop amplifier. This graphical method is a fast way to read bandwidth directly from the Bode plot without calculation, and examiners use it regularly in data-response questions.

6. Slew Rate and Its Effect on Output Waveforms

Gain-bandwidth product describes a small-signal limitation — it applies when signal amplitudes are small enough that the op-amp behaves linearly. Slew rate is a large-signal limitation that operates independently. Even if your signal frequency is well within the bandwidth, a large enough amplitude will cause slew-rate distortion.

The maximum undistorted output frequency for a given peak output voltage $V_p$ is found from the slew rate $SR$. The steepest point of a sine wave occurs at zero crossing, where the rate of change equals $2\pi f V_p$. For the output to follow faithfully, this must not exceed the slew rate.

Worked example: Checking for Slew-Rate Distortion

An op-amp has a slew rate of 0.5 V $\mu$s$^{-1}$. Determine whether it can faithfully reproduce a 10 V peak sine wave at 20 kHz.

1. $SR = 0.5 \text{ V } \mu\text{s}^{-1} = 0.5 \times 10^6 \text{ V s}^{-1} = 5 \times 10^5 \text{ V s}^{-1}$ — Always convert to V s^{-1} before substituting into the formula. 1 V per microsecond = 10^6 V per second.
2. $\left(\frac{dV}{dt}\right)_{max} = 2\pi f V_p = 2 \times \pi \times 20000 \times 10$ — This is the maximum slope of the desired output sine wave at its zero crossing.
3. $= 2 \times 3.1416 \times 20000 \times 10 = 6.2832 \times 200000 = 1.257 \times 10^6 \text{ V s}^{-1}$ — 2 × 3.1416 = 6.2832. Then 6.2832 × 20 000 = 125 664. Then 125 664 × 10 = 1 256 640 ≈ 1.26 × 10^6 V s^{-1}.
4. $1.26 \times 10^6 \text{ V s}^{-1} > 5 \times 10^5 \text{ V s}^{-1}$ — The required rate of change exceeds the slew rate by more than a factor of two.
5. Required slope $\approx 1.26 \times 10^6$ V s$^{-1}$; available slew rate $= 5 \times 10^5$ V s$^{-1}$ — The op-amp cannot keep up. The output will be distorted — the peaks of the sine wave will be clipped into a triangular shape. A faster op-amp or a smaller output amplitude is needed.

Answer: The op-amp cannot reproduce the signal without distortion. The required rate of change ($\approx 1.26 \times 10^6$ V s$^{-1}$) exceeds the slew rate ($5 \times 10^5$ V s$^{-1}$).

7. Exam Technique and Common Mistakes

Questions on op-amp limitations tend to fall into three patterns: describe a limitation, explain its consequence, or calculate bandwidth or slew-rate limits. Each demands a different response structure. For a describe question, state what the limitation is and give its direction — for example, 'the open-loop gain is finite and decreases with increasing frequency.' For an explain question, add the mechanism: 'because the internal compensation capacitor rolls off the gain at $-20$ dB per decade, the bandwidth available to a closed-loop amplifier is reduced as gain increases.'

For calculate questions involving GBP, the most common error is forgetting to check whether the given gain is open-loop or closed-loop. GBP applies to the closed-loop gain. A second frequent slip is mixing units for slew rate — always convert V $\mu$s$^{-1}$ to V s$^{-1}$ before substituting into $f_{max} = SR / (2\pi V_p)$, otherwise the answer will be out by a factor of $10^6$.