Electrical Engineering › Section Q · Thyristors, operational amplifiers, and power conversion › Topic Q.7

Differential and instrumentation amplifiers

Many sensors produce a tiny difference riding on a large shared voltage. A four-resistor difference amplifier extracts that difference with a set gain, and the three-op-amp instrumentation amplifier does it with high input impedance and a gain set by a single resistor.

Measuring a difference, not a voltage

A bare op-amp amplifies the difference between its inputs, but its gain is far too high to use. The feedback amplifiers of Negative feedback and divided feedback tame the gain by tying one input to ground or to the output, so they amplify only one signal measured from ground. Many measurements need both inputs. A strain-gauge bridge on a 5 V supply gives two outputs that both sit near 2.5 V and differ by a few millivolts. The few millivolts are the differential voltage \(V_d = V_2 - V_1\), the information wanted. The 2.5 V both share is the common-mode voltage, roughly \(V_{cm} = (V_1 + V_2)/2\), which should be ignored.

The four-resistor difference amplifier

Combine the inverting and noninverting circuits. \(V_1\) drives the inverting input through \(R_1\), with \(R_2\) from the output back to that input. \(V_2\) drives the noninverting input through a divider, \(R_3\) in series and \(R_4\) to ground. By superposition,

V_{out} = -\frac{R_2}{R_1}V_1 + \Big(1 + \frac{R_2}{R_1}\Big)\frac{R_4}{R_3 + R_4}V_2

When the ratios match, \(R_4/R_3 = R_2/R_1\), this collapses to

V_{out} = \frac{R_2}{R_1}(V_2 - V_1)

With all four resistors equal the gain is 1 and \(V_{out} = V_2 - V_1\). \(V_2\) acts as the noninverting input of the whole circuit and \(V_1\) as the inverting input.

Worked example: a difference amplifier

A difference amplifier has \(R_1 = R_3 = 10\,\text{k}\Omega\) and \(R_2 = R_4 = 47\,\text{k}\Omega\), with \(V_1 = 1.98\,\text{V}\) and \(V_2 = 2.03\,\text{V}\). 1. Matched ratios, so the gain is \(47/10 = 4.7\). 2. Output:

4.7 \times (2.03 - 1.98) = 4.7 \times 0.050 = 0.235\,\text{V}

3. Check by superposition: the \(V_1\) term is \(-4.7 \times 1.98 = -9.306\,\text{V}\); the \(V_2\) term is \(5.7 \times \frac{47}{57} \times 2.03 = 9.541\,\text{V}\); their sum is \(0.235\,\text{V}\). The common-mode level of about 2 V has vanished, and only the 50 mV difference, amplified 4.7 times, remains.

Common-mode gain and CMRR

An ideal difference amplifier has a common-mode gain \(A_{cm}\) of zero: tie both inputs together, sweep them up and down, and the output never moves. A real one amplifies the common-mode voltage a little. The figure of merit is the common-mode rejection ratio, the differential gain divided by the common-mode gain, usually quoted in decibels with the voltage form:

\text{CMRR} = 20\log_{10}\frac{A_d}{A_{cm}}

In the four-resistor circuit the main cause of common-mode gain is not the op-amp but resistor mismatch: if \(R_4/R_3\) differs even slightly from \(R_2/R_1\), part of \(V_{cm}\) leaks through. Precision difference amplifiers use closely matched or trimmed resistors.

Worked example: a 1% mismatch

In the previous amplifier, suppose \(R_4\) is 1% high, 47.47 kΩ, while the others are exact. 1. Common-mode gain: set \(V_1 = V_2 = 1\,\text{V}\) in the superposition formula:

A_{cm} = -4.7 + 5.7 \times \frac{47.47}{57.47} = 0.00818

2. Differential gain: set \(V_1 = -0.5\,\text{V}\), \(V_2 = +0.5\,\text{V}\): \(A_d = 4.704\). 3. CMRR: \(4.704/0.00818 = 575\), or \(20\log_{10}575 = 55.2\,\text{dB}\). 4. Effect: a 5 V common-mode voltage adds \(5 \times 0.00818 = 41\,\text{mV}\) of error to the output. Check: that error is a sixth of the 0.235 V signal in the earlier example, so one ordinary 1% resistor spoils a precision measurement.

Two practical weaknesses

The basic difference amplifier has modest, unequal input resistances. \(V_2\) sees \(R_3 + R_4\), while the current \(V_1\) must supply depends on both inputs. Any source resistance adds to \(R_1\) or \(R_3\), so a sensor with a few hundred ohms of resistance unbalances the ratios and creates common-mode gain. The cure is to buffer each input with a voltage follower, so the sources see only op-amp inputs and the followers drive the resistors. The second weakness is gain adjustment: changing the gain while keeping the ratios matched means changing two resistors together, \(R_2\) and \(R_4\), and keeping them matched.

The instrumentation amplifier

The instrumentation amplifier fixes both problems with three op-amps. \(V_1\) and \(V_2\) go straight to the noninverting inputs of two input op-amps, giving very high input impedance. Each input op-amp has a resistor \(R\) from its output to its inverting input, and one resistor \(R_{gain}\) joins the two inverting inputs. Their outputs feed a unity-gain difference amplifier. Feedback holds the ends of \(R_{gain}\) at \(V_1\) and \(V_2\), so a current \((V_2 - V_1)/R_{gain}\) flows through \(R\), \(R_{gain}\) and \(R\) in series. The output is

V_{out} = \Big(1 + \frac{2R}{R_{gain}}\Big)(V_2 - V_1)

One resistor sets the gain, and the minimum, with \(R_{gain}\) removed, is 1.

Worked example: a strain-gauge amplifier

A bridge on 5 V gives \(V_1 = 2.494\,\text{V}\) and \(V_2 = 2.506\,\text{V}\), a 12 mV difference. An instrumentation amplifier has \(R = 25\,\text{k}\Omega\), and a gain of about 100 is wanted. 1. Solve \(1 + 50{,}000/R_{gain} = 100\): \(R_{gain} = 505\,\Omega\); the nearest E24 value is 510 Ω. 2. Actual gain: \(1 + 50{,}000/510 = 99.0\). 3. Output: \(99.0 \times 12\,\text{mV} = 1.19\,\text{V}\). 4. Input stage: the current is \(12\,\text{mV}/510\,\Omega = 23.5\,\mu\text{A}\), dropping 0.588 V across each \(R\), so the input op-amp outputs sit at 1.91 V and 3.09 V. Check: their difference is 1.19 V, and both are well inside a ±12 V supply.

Why the design works so well

The input stage amplifies the difference by \(1 + 2R/R_{gain}\) but passes the common-mode voltage through at a gain of only 1: if \(V_1 = V_2\), no current flows in \(R_{gain}\) and both outputs simply follow the inputs. The final difference amplifier then rejects that common-mode level. So the signal is boosted before the stage whose resistor matching limits the rejection, and the overall CMRR improves roughly in proportion to the gain. Integrated instrumentation amplifiers put all the matched, trimmed resistors on one chip and leave only \(R_{gain}\) outside. They are standard for bridges, thermocouples and medical electrodes, where mains hum appears as a common-mode voltage.

Common mistakes to avoid

First, assuming the four-resistor circuit subtracts correctly with any resistors: the ratios \(R_2/R_1\) and \(R_4/R_3\) must match. Second, ignoring source resistance, which adds to the input resistors and upsets the match. Third, writing the instrumentation gain as \(1 + R/R_{gain}\): both resistors \(R\) carry the current, so it is \(1 + 2R/R_{gain}\). Fourth, confusing CMRR forms: it uses \(20\log_{10}\) because it is a voltage-gain ratio. Fifth, thinking the input stage amplifies the common-mode voltage: it passes it at a gain of 1.

Key takeaways

A differential measurement wants V_d = V₂ − V₁ and should ignore the common-mode voltage. The four-resistor difference amplifier gives V_out = (R₂/R₁)(V₂ − V₁) when R₄/R₃ = R₂/R₁; resistor mismatch creates common-mode gain, measured by CMRR = 20 log₁₀(A_d/A_cm). Its inputs have modest, unequal resistance, fixed by follower buffers. The three-op-amp instrumentation amplifier has very high input impedance and gain 1 + 2R/R_gain, set by one resistor, with a minimum of 1 and better CMRR at high gain.

Flashcards

Flashcard

Differential voltage

V_d = V₂ − V₁, the difference between the two inputs

The information a bridge sensor carries.

Test your understanding

Question 1 of 5

A difference amplifier has R₁ = R₃ = 22 kilohms and R₂ = R₄ = 220 kilohms. What is its output for V₁ = 3.10 V and V₂ = 3.05 V?

Question 2 of 5

A three-op-amp instrumentation amplifier has R = 10 kilohms for both input-stage feedback resistors and R_gain = 1.0 kilohm, with a unity-gain difference stage. What is its gain?

Question 3 of 5

A difference amplifier has a differential gain of 5.0 and a common-mode gain of 0.010. What is its CMRR?

Question 4 of 5

Why are voltage followers often added in front of each input of a four-resistor difference amplifier?

Question 5 of 5

What is the gain of a three-op-amp instrumentation amplifier (unity difference stage) when its gain resistor is removed entirely?

← Previous topicNext topic →