“As precision requirements increase, fully differential signal chain components stand out for their performance, and a key benefit of these components is the ability to pick up noise rejection through signal routing. Because the output picks up this noise, the output often experiences errors and thus further attenuation in the signal chain. In addition, differential signaling can achieve twice the signal range of single-ended signals on the same supply. Therefore, the signal-to-noise ratio (SNR) of the fully differential signal is higher.The classic three-op amp instrumentation amplifier has many advantages, including common-mode signal rejection, high input impedance, and precise (adjustable) gain; however, when a fully differential output signal is required, it
Q: Can we use an instrumentation amplifier to generate a differential output signal?
A: As the requirements for accuracy increase, fully differential signal chain components stand out for their excellent performance. One of the main advantages of these components is the ability to pick up noise suppression through signal routing. Because the output picks up this noise, the output often experiences errors and thus further attenuation in the signal chain. In addition, differential signaling can achieve twice the signal range of single-ended signals on the same supply. Therefore, the signal-to-noise ratio (SNR) of the fully differential signal is higher. The classic three-op amp instrumentation amplifier has many advantages, including common-mode signal rejection, high input impedance, and precise (adjustable) gain; however, it is powerless when a fully differential output signal is required. Several methods have been used to implement fully differential instrumentation amplifiers with standard components. However, they have their own shortcomings.
Figure 1. Classic instrumentation amplifier.
One technique is to use an op amp to drive the reference pin, with the positive input being common mode and the negative input being the center of two matched resistors that connect the outputs together. This configuration uses the in-amp output as the positive output and the op-amp output as the negative output. Since the two outputs are different amplifiers, the mismatch in dynamic performance between these amplifiers can greatly affect the overall performance of the circuit. Additionally, the matching of the two resistors causes the output common mode to move with the output signal, which can result in distortion. When designing this circuit, stability must be considered when choosing the amplifier, and a feedback capacitor may need to be placed on the op amp to limit the overall bandwidth of the circuit. Finally, the gain range of this circuit depends on the instrumentation amplifier. Therefore, it is not possible to achieve a gain of less than 1.
Figure 2. Using an external op amp to generate the inverting output.
Another technique is to put two instrumentation amplifiers in parallel with the input switch. This configuration has a better matched drive circuit and frequency response than the previous circuit. But it cannot achieve gain less than 2. The circuit also requires precisely matched gain resistors for a purely differential signal. A mismatch in these resistors causes a change in the output common-mode level with the same effect as in the previous architecture.
Figure 3. Using a second instrumentation amplifier to generate an inverting output.
Both approaches have limitations on the gain that can be achieved and on the requirements for matching components.
New cross-connect technology
By cross-connecting two instrumentation amplifiers, as shown in Figure 4, this new circuit uses a single gain resistor to provide a fully differential output with precise gain or attenuation. By connecting the two reference pins together, the user can adjust the output common mode as desired.
Figure 4. Cross-connect technique—a solution for generating the output of a differential in-amp.
The gain of In_A is derived from the following equation. Since the input voltage appears at the positive terminal of the input buffers of in-amp 2 and the voltage across resistors R2 and R3 is 0 V, the gain of these buffers follows the equations applicable to the non-inverting op amp configuration. Likewise, for the input buffer of in-amp 1, the gain follows the inverting op-amp configuration. Since all the resistors in the differential amplifier are matched, the gain of the buffer output is 1.
Figure 5. Matching resistors inside the instrumentation amplifier is the key to the cross-connect technique.
VOUT_A = CV1 × (R1/R3 C R1/R2)
VOUT_B = V1 × (R1/R3 C R1/R2)
According to the principle of symmetry, if a voltage V2 is applied at In_B and In_A is grounded, the result is as follows:
VOUT_A = V2 × (R1/R3 C R1/R2)
VOUT_B = CV2 × (R1/R3 C R1/R2)
Adding these two results gives the gain of the circuit.
VIN = In_A C In_B = V1 C V2
VOUT = VOUT_A CVOUT_B
Gain = 2 × (R1/R2 C R1/R3)
Gain resistors R3 and R2 set the gain of the circuit, and only one resistor is required to achieve a fully differential signal. Positive/negative output depends on installed resistors. Not installing R3 will cause the second term in the gain equation to go to zero. From this, the gain is 2×R1/R2. Not installing R2 causes the first term in the gain equation to go to zero. From this, the gain is -2×R1/R3. Another point to note is that the gain is purely a ratio, so a gain of less than 1 can be achieved. Remember, since R2 and R3 have opposite effects on gain, using two gain resistors will make the first stage gain higher than the output. If care is not taken when choosing the resistor values, the result will be increased bias due to the first stage op amp at the output.
To demonstrate this circuit in action, we hooked up two AD8221 instrumentation amplifiers. The data sheet lists R1 as 24.7kΩ, so a gain equal to 1 is achieved when R2 is 49.4kΩ.
CH1 is the input signal of In_A, CH2 is VOUT_A, CH3 is VOUT_B. Outputs A and B are matched and out of phase, and the difference is equal in magnitude to the input signal.
Figure 6. Using the cross-connect technique to generate a differential in-amp output signal, measured at gain=1.
Next, move the 49.4kΩ gain resistor from R2 to R3, and the new gain of the circuit is -1. Now Out_A is out of phase with the input and the difference between the outputs is equal in magnitude to the input signal.
Figure 7. Using the cross-connect technique to generate a differential in-amp output signal, measured at gain=-1.
As mentioned earlier, one limitation of other techniques is that attenuation cannot be achieved. According to the gain equation, using R2 = 98.8kΩ, the circuit attenuates the input signal by a factor of two.
Figure 8. Using the cross-connect technique to generate a differential in-amp output signal, measured at gain = 1/2.
Finally, to demonstrate high gain, R2=494Ω was chosen to achieve G=100.
Figure 9. Using the cross-connect technique to generate a differential in-amp output signal in-amp, measured at gain = 100.
The performance of this circuit is described by the gain equation. For best performance, some precautions should be taken when using this circuit. The accuracy and drift of the gain resistors will add to the gain error of the instrumentation amplifier, so choose the appropriate tolerance based on the error requirements. Since the capacitance on the Rg pin of the in-amp can cause poor frequency performance, these nodes should be noted if high frequency performance is required. Also, the temperature mismatch between the two instrumentation amplifiers can cause system offsets due to offset drift, so attention should be paid to layout and loading here. Using a dual-channel instrumentation amplifier, such as the AD8222, can help overcome these potential problems.
The cross-connect technique maintains the desired characteristics of the instrumentation amplifier while providing additional functionality. Although all of the examples discussed in this article implement differential outputs, in a cross-connect circuit, the common mode of the outputs is not affected by resistor pair mismatch, unlike other architectures. Therefore, a true differential output is always achieved. Also, as shown in the gain equation, differential signal attenuation is possible, which eliminates the need for a funnel amplifier, which was previously necessary. Finally, the polarity of the output is determined by the position of the gain resistor (using R2 or R3), which adds more flexibility to the user.
about the author:
Matthew “Rusty” Juszkiewicz [[email protected]