Single-Supply Instrumentation Amplifier Circuit

The instrumentation amplifier amplifies the difference between two signals. Typical differential signals come from sensor devices such as resistor bridges or thermocouples. Figure 1 shows a typical application of an instrumentation amplifier, where the differential voltage from a resistor bridge is amplified by an AD620 (a low-power, low-cost, integrated instrumentation amplifier). In thermocouple and resistor bridge applications, the differential voltage is always quite small (a few millivolts to more than ten millivolts). The voltage of the same polarity and amplitude input to the two input terminals is about 2.5V, and there is also a common-mode component that is useless for measurement, so the ideal instrumentation amplifier should amplify the difference between the two input signals, and any common-mode component must be suppressed. In fact, suppressing the common-mode component is the only reason to use an instrumentation amplifier. In practice, instrumentation amplifiers never completely suppress common-mode signals, and there will always be some residual components at the output.

Common-mode rejection ratio (CMRR) is a comprehensive indicator used to measure the degree to which the common-mode signal is suppressed by the amplifier. It is defined by the following formula:

Figure 1 In a typical instrumentation amplifier application, the input common-mode voltage consists of the DC bias voltage (VS/2) from the bridge and any common-mode noise picked up in the input lines. Some portion of the common-mode voltage will always appear at the output of the instrumentation amplifier.

Where Gain is the differential mode gain of the amplifier, Vcm is the common mode voltage present at the input, and Vout is the result of the input common mode voltage at the output.

Substituting specific values, such as the AD620 integrated instrumentation amplifier, when the gain is set to 10, the CMRR is 100dB, and the common mode voltage in Figure 1 is 2.5V, the voltage at the output is 250mV as calculated by equation (1). With the above settings, it is noted that the output voltage caused by the input and output offset voltages is about 1.5mV, which shows that as an error source, CMRR is not as important as the offset voltage. So far, only the common mode rejection ratio of DC signals has been discussed.

AC and DC common mode rejection ratio

In Figure 1, the common-mode signal can be a steady-state DC voltage (such as the 2.5V voltage from the bridge) or from external interference. In industrial applications, the most common external interference is picked up from the 50Hz/60Hz power mains (for example, from lighting, motors, or any equipment running on the power mains). In different measurement applications, the interference at the input of the instrument amplifier is basically equal, so the interference signal is also regarded as a common-mode signal here, which is superimposed on the input DC common-mode voltage. What is obtained at the output is the attenuated form of this input common-mode signal, and the degree of attenuation depends on the CMRR at that frequency.

While DC offset voltages can be easily removed by trimming and calibration, AC errors at the output are troublesome. For example, if the input loop picks up 50Hz or 60Hz interference from the mains, the AC voltage at the output will degrade the resolution of the entire application. Filtering out interference is expensive and is only feasible in applications where speed is not a high priority. Obviously, high common-mode rejection over the entire frequency range helps reduce the impact of external common-mode interference.

Therefore, in practice, it is more meaningful to discuss CMRR over the entire frequency range than to discuss it at DC. The data sheet of the integrated instrumentation amplifier lists the CMRR at 50Hz/60Hz, and the graph section shows the curve of CMRR changing with frequency (see Figure 2).

Figure 2 shows how the CMRR of the AD623 (a low-cost integrated instrumentation amplifier) ​​varies over frequency. It remains flat before 100Hz, and then begins to decline (greater than 100Hz). It can be seen that 50Hz/60Hz grid interference is well suppressed. Also pay attention to the harmonic interference of the grid frequency. In an industrial environment, the grid frequency harmonics can reach the seventh harmonic (350Hz/420Hz). At this point, the CMRR drops to about 90dB (gain of 10). This makes the common-mode gain of -70dB still sufficient to suppress most common-mode interference.

Instrumentation amplifiers of different structures

Now let’s examine the different structures of instrumentation amplifiers. The choice of structure and the accuracy of the passive components will affect the CMRR of AC and DC. 3.1 Two-op-amp instrumentation amplifier

Figure 3 is a basic two-op-amp instrumentation amplifier circuit diagram. The differential mode gain can be given by equation (2):

(2)

Here R1=R4, R2=R3. If R1=10kΩ, R2=1kΩ, the differential mode gain is 11. From equation (2), it is impossible to make the programmed gain 1.

3.1.1 Common-Mode Gain of a Two-Op-Amp Instrumentation Amplifier

The output voltage caused by the DC common-mode voltage is given by equation (3):

Using formula (1), the expression of the circuit's CMRR can be obtained as

Because the resistor ratio in the denominator is always close to 1, we do not need to consider the gain of the instrumentation amplifier, and we can see that the CMRR of a two-op-amp instrumentation amplifier increases with the increase of differential-mode gain.

In the above resistor network, due to the existence of errors, the actual resistance value cannot be completely equal to the nominal value, that is, there is a mismatch. The percentage of the actual value of R1R3 to the difference between it and R2R4 can be defined as the mismatch. Formula (4) can be rewritten as

Where Mismatch is the mismatch rate.

Any mismatch between the four resistors that program the gain will directly affect the CMRR. Precision resistor networks are fine-tuned to achieve maximum accuracy at ambient temperature. Any mismatch caused by temperature drift of the resistors will further degrade the CMRR.

Obviously, the key to high common-mode rejection is the resistor network, so both the resistor ratio and the corresponding drift must be well matched, while the absolute values ​​of the resistors and their absolute drift are not important, the key is matching.

Integrated instrumentation amplifiers are particularly suitable for ratio matching and temperature tracking of gain programming resistors. The initial tolerance of thin film resistors made on silicon wafers reaches ± 20%, and laser trimming during the manufacturing process reduces the ratio error between resistors to 0.01%. In addition, the correlation between the value and temperature coefficient of each thin film resistor is very small, usually less than 3×10-6/℃.

Figure 4 illustrates the practical results of resistor mismatch at ambient temperature. In Figure 3, the measurement of the circuit CMRR (gain of 11) uses four resistors with a mismatch of about 0.1% (R1=9999.5Ω, R2=999.76Ω, R3=1000.2Ω, R4=9997.7Ω). The DC CMRR value is about 84dB (theoretical value is 85dB), and the CMRR drops rapidly when the frequency increases. Figure 4 also shows the oscilloscope waveform of the output voltage of the power grid interference. The output voltage caused by the 200mV (peak-to-peak) harmonic at 180Hz is about 800mV. Based on the above settings, the 1sb weight of a 12-bit data acquisition system with an input range of 0-2.5V is 610mV.

The phase shift or delay of the Vin- signal at the in-phase end of A1 after passing through A1 will cause a vector error between Vin- and the output signal of A1, causing a reduction in CMRR over the entire frequency range. To ensure a certain CMRR, the common-mode signal at Vin- and the output of A1 should have the same phase and amplitude, which is only possible when A1 has no delay. Choosing a matched high-speed dual op amp can extend the frequency range, thereby keeping the CMRR flat, but on the other hand, high-speed op amps will pick up external high-frequency interference. Another solution is to connect a trimming capacitor between the inverting input of A1 and the ground terminal, but the disadvantage is that it must be manually trimmed.

So the CMRR (over frequency) of Figure 4 is affected by two distinct parameters. At low frequencies, the CMRR is directly related to the mismatch in the programmed gain resistors, and at high frequencies, the differential-mode closed-loop gain of the op amp causes a reduction in the CMRR.

3.1.2 Common-Mode Range of a Two-Op-Amp Instrumentation Amplifier

The input common-mode range of a two-op-amp instrumentation amplifier is affected by the programmed gain. In Figure 3, when A1 operates at a closed-loop gain of 1.1, any common-mode voltage at the input is amplified (i.e., the input common-mode voltage appears at the output of A1 after being amplified by 1.1).

Now consider the case where the instrumentation amplifier has a programmable gain of 1.1 (R1 = 1kΩ, R2 = 10kΩ, R3 = 10kΩ, R4 = 1kΩ). A1's closed-loop gain is 11, and because the common-mode voltage is amplified, the input common-mode range is strictly limited by the swing of A1's output. This is particularly problematic in applications where forced low voltages are required, and in these cases, using a full-scale amplifier adds some swing to alleviate the problem.

Three-op-amp instrumentation amplifier

Figure 5 shows the structure of a three-op-amp instrumentation amplifier, which is the most commonly used structure for discrete and integrated instrumentation amplifiers. The transfer function of the entire gain is very complex. When R1=R2=R3=R4, the transfer function can be simplified to

(6)

R5 and R6 are set to the same value (usually between 10 and 50 kΩ). The overall gain of the circuit can be adjusted from unity to any high value by simply adjusting the value of RG.

3.2.1 Common-Mode Gain of a Three-Op-Amp Instrumentation Amplifier

As expected, the theoretical common-mode gain of the instrumentation amplifier is zero. To calculate the common-mode gain, assume that there is only a common-mode voltage of Vcm at the input (that is, Vin+=Vin-=Vcm). There is no voltage drop across RG, and the output voltage of A1 and A2 is also equal to Vcm. Assuming that A1 and A2 are ideally matched, the first approximation is that the first-stage common-mode gain is equal to unity and is independent of the programmed gain.

Assuming that op amp A3 is ideal, the second-stage common-mode gain is obtained from equation (7):

Substituting into equation (1), the common mode rejection ratio becomes equation (8):

The denominator in the formula is much more complicated than that of the two-op-amp instrumentation amplifier. As shown in formula (4), the denominator can be expressed as the mismatch percentage of the resistors, that is,

In equation (8), if all four resistors are equal (or R1 = R3, R2 = R4), the denominator becomes zero, and any mismatch in these resistors will cause a portion of the common-mode voltage to appear at the output. Similar to a two-op-amp instrumentation amplifier: any mismatch in temperature drift between resistors will reduce the CMRR.

3.2.2 AC CMRR of a Three-Op-Amp Instrumentation Amplifier

If A1 and A2 are well matched (i.e., have the same closed-loop bandwidth), the CMRR will not drop as quickly as the two-op-amp instrumentation amplifier. Comparing Figures 2 and 4, the CMRR of the three-op-amp instrumentation amplifier is relatively flat before 100Hz, while the CMRR of the two-op-amp instrumentation amplifier starts to drop at about 10Hz.

3.2.3 Common-Mode Range of a Three-Op-Amp Instrumentation Amplifier

The first stage of the three-op-amp instrumentation amplifier has a unity common-mode gain. The common-mode voltage appears intact at the outputs of A1 and A2 in Figure 5, while the differential-mode input voltage (Vdiff) drops across the gain resistors, resulting in current flowing through R5 and R6. This means that as the input differential-mode voltage increases, the voltage at A1 will be above Vcm and the voltage at A2 will be below Vcm. Therefore, as the gain and/or input signal increase, the voltage range of A1 and A2 will also increase, and will eventually be limited by the range of the power supply voltage. It can be seen that the range of common-mode voltage that can be achieved, the differential-mode input voltage, and the gain are interrelated. For example, increasing the gain will reduce the common-mode range and input voltage range. Similarly, increasing the common-mode voltage will limit the differential-mode input range and limit the maximum value that the gain can achieve. If the output swing of the input stage op amp is known, then the relationship between the input range, common-mode range, and gain can be well represented to serve the specific three-op-amp instrumentation amplifier.

When low supply voltages are used in industrial applications, the available swing range is increasingly limited. As with the two-op-amp in-amp, full-swing op amps can solve this problem, but in the three-op-amp in-amp, the full-swing output stage (A3) does not help at all because excessive input voltage, common-mode voltage, or gain will reduce the output voltage of the input stage (A1, A2).

3.2.4 Single-Supply Three-Op-Amp Instrumentation Amplifier Optimized for Low Common-Mode Applications

Figure 6 is a simplified diagram of the AD623 (a low-power, single-supply, full-range instrumentation amplifier), which uses the traditional three-op-amp instrumentation amplifier structure. Before being used as an input stage op amp, the positive and negative input voltages pass through a PNP tube, and the voltage is biased by 0.6V.

To understand the importance of level offset, we must first consider the normal conditions under which the instrumentation amplifier operates. Figure 7 shows a typical application of the AD623. The signal amplified by the instrumentation amplifier comes from a J-type thermocouple. The instrumentation amplifier and the A/D converter are powered by a single +5V power supply. In this application, the measured temperature range is from -200 to +200°C, and the corresponding thermocouple voltage range is -7.890 to 10.777mV.

As usual, one end of the thermocouple is grounded, allowing bias current to flow into the instrumentation amplifier. Therefore, the common-mode voltage between the noninverting and inverting input voltages is very close to ground. In fact, as the voltage from the thermocouple starts to go negative, the effective common-mode voltage also goes negative.

In a traditional three-op-amp instrumentation amplifier, when the thermocouple voltage begins to exceed zero, the voltage spreading effect of the input stage causes the output voltage of one of the input stage op amps to go to ground. The level-biasing structure of Figure 6 avoids this problem by effectively adding 0.6V to the common-mode voltage, thereby having more swing range to ground and allowing the output voltage of the full-scale op amps A1 and A2 to be in the linear region, even when the input voltage and common-mode voltage are below ground. The input voltage can be as negative as 150mV, which is controlled by the programmed gain and common-mode voltage.

In this example, the instrumentation amplifier is set to a gain of 91.9 (RG = 1.1 kΩ), and the voltage of the reference pin is set to 2 V. As long as the thermocouple voltage varies between -200 and +200°C, the output voltage range of the instrumentation amplifier is 1.274 to 2.990 V (relative to ground). This voltage swing range is well suited to the input voltage range of the A/D converter (2 V ± 1 V).

3.2.5 Single-supply two-op-amp instrumentation amplifier for low common-mode voltage applications

The method of adding a Vbe voltage drop to increase the common-mode voltage can be applied to two-op-amp instrumentation amplifiers. Figure 8 is a simplified diagram of the AD627, which is an integrated two-op-amp instrumentation amplifier that uses special techniques to obtain high CMRR over the entire frequency range. It must be pointed out that for three-op-amp instrumentation amplifiers, care must be taken to compensate for internal node voltages to avoid signal saturation, which is particularly critical in single-supply applications. Generally speaking, the maximum gain is determined by the range of the output valid signal (greater than 50mV for the inverting channel and within 100mV for the non-inverting channel). In single-supply applications where the input common-mode voltage is close to or equal to zero, there are certain limitations on the programmed gain. When the voltage ranges of the input, output, and reference pins (REF) are specified by the technical specifications, the voltage ranges of these pins affect each other. In Figure 8, driven by a differential voltage Vdiff with a common-mode component Vcm, the voltage at the output of op amp A1 is a function of the voltages of the Vdiff, Vcm, and Vref pins and the programmed gain:

VA1=1.25(Vcm +0.5V)- 0.25Vref -Vdiff(25kΩ/RG-0.625)

It can also be expressed by the actual voltage on the -IN and +IN (V- and V+) pins:

VA1=1.25(V-+0.5V)- 0.25Vref -(V+ -V-)25kΩ/RG

The output voltage of A1 swings within 50mV for the inverting channel and 200mV for the non-inverting channel, and the above equations can be used to verify that the voltage of A1 is within this range. From any of the above equations, it can be seen that when Vref increases as the positive bias of the output (A2) of the AD627 is increased, the output voltage of A1 will decrease. In addition, increasing the input common-mode voltage will increase the output voltage of A1. In single-supply applications with low common-mode voltage, the differential input voltage or the voltage on REF is too high, which will cause the output of A1 to go to ground level. An effective upward bias of 0.5V of the input voltage (such as the Vbe of T1 and T2) can increase some swing range.

Table 1 shows the maximum gain of the AD627 under different single-supply input conditions. The output swing is obtained according to the voltage on the REF pin. The voltage on REF has been set to 2V or 1V to maximize the gain and output swing range. Note that in many cases, there is no benefit in making the single-supply voltage greater than 5V (except when the input range is 0V to 1V).

Table 1. Maximum gain of AD627 for low common-mode single-supply applications

4 Filter out high frequency common mode signals

All instrumentation amplifiers can correct for out-of-band signals at high frequencies. Once corrected, these signals appear as DC offset errors at the output. The circuit of Figure 9 provides excellent RFI suppression without degrading performance within the instrumentation amplifier's passband. Resistor R1 and capacitor C1 (similarly R2 and C2) form a low-pass RC filter with a -3dB bandwidth of F=1/(2πR1C1). Substituting the component values, this filter has a -3dB bandwidth of approximately 40kHz. Resistors R1 and R2 should be selected to be large enough to isolate the circuit input from the capacitors, but not so large that they increase the circuit noise. To maintain common-mode rejection in the amplifier's passband, capacitors C1 and C2 must be ±5% or better components, or low-cost components that have been tested and can provide good matching.

Capacitor C3 is necessary to maintain common-mode rejection at low frequencies. R1, R2 and C1, C2 form a bridge circuit, and the output of the bridge circuit is connected to the input of the instrumentation amplifier. Any mismatch in C1, C2 will cause the bridge circuit to be unbalanced and reduce common-mode rejection. C3 ensures that any RF signal is common-mode (appears at both inputs of the instrumentation amplifier with the same polarity and amplitude) and does not differentially input. The - 3dB bandwidth of the second stage low-pass network (R1 + R2 and C3) is 1/[2π(R2 + R1)C3], and substituting C3 = 0.047mF, the -3dB signal bandwidth of this circuit is about 400Hz. The typical DC offset (over the entire frequency range) is less than 1.5mV, and the circuit rejects RF signals by more than 71dB. By reducing R1, R2 to 2.2kΩ, the - 3dB signal bandwidth of the circuit can be increased to 900Hz. The performance is similar to when using 4kΩ, except that the circuit before the instrumentation amplifier must drive a low impedance load.

The circuit of Figure 9 can be built using a PCB board. The component leads must be as short as possible. Resistors R1 and R2 can be 1% metal film resistors, while capacitors C1 and C2 must be ±5% tolerance components to avoid reducing the common-mode rejection of the circuit. 5% silver mica capacitors or ±2% PPS film capacitors from Panasonic are recommended.

Figure 9: Attenuation circuits for normal mode and common mode RF interference suppression

Figure 6 The AD623 uses a typical three-op-amp instrumentation amplifier configuration. By biasing both inputs 0.6V, it can operate from a single supply even at very low common-mode voltages.

Figure 7 The input stage level biasing of the AD623 is ideal for single-supply, low common-mode applications.

The temperature range is -200～+200℃, and the voltage range of the J-type thermocouple is from -7.890～10.777mV. The gain of 91.9 makes the output voltage range of the instrumentation amplifier 1 to 3V (i.e. 2V±1V), and the output end is connected to the AD7776A/D converter powered by a single power supply.

Figure 8: An integrated two-op-amp instrumentation amplifier AD627, also biased at the Vbe level to allow low input common-mode voltage to operate on a single supply.

Figure 4. A 0.1% mismatch between the four resistors of the programmable gain determines the CMRR of the two-op-amp instrumentation amplifier at low frequencies. The difference in closed-loop gain between the two op-amps results in a reduction in CMRR across the entire frequency band. At 180Hz, a 200mV line harmonic generates an 800μV voltage at the op-amp output.

Figure 5. The structure of a three-op-amp instrumentation amplifier. A 0.1% mismatch between R1, R2, R3, and R4 results in a worst-case CMRR of 60dB (gain = 1). Drift mismatch exacerbates the degradation of CMRR.

Figure 3 The input common-mode range of a two-op-amp instrumentation amplifier decreases as the differential-mode gain decreases (unity gain is not possible). The resistor mismatch determines the CMRR at DC and low frequencies, while the high-frequency CMRR depends on the phase shift of Vin- through A1.