“Differential signaling is required in applications including driving modern analog-to-digital converters (ADCs), transmitting signals over twisted pair cables, and conditioning high-fidelity audio signals. Higher signal-to-noise ratios are achieved because differential signaling uses larger signals at a specific set of supply voltages, improving common-mode noise rejection and reducing second-harmonic distortion. Because of this need, we need circuit blocks that can convert single-ended signals in most signal chains to differential signals.
Differential signaling is required in applications including driving modern analog-to-digital converters (ADCs), transmitting signals over twisted pair cables, and conditioning high-fidelity audio signals. Higher signal-to-noise ratios are achieved because differential signaling uses larger signals at a specific set of supply voltages, improving common-mode noise rejection and reducing second-harmonic distortion. Because of this need, we need circuit blocks that can convert single-ended signals in most signal chains to differential signals.
Figure 1 shows a simple single-ended-to-differential converter using the AD8476 precision low-power fully differential amplifier (diff-amp) with integrated precision resistors. The differential gain configured inside the differential amplifier is 1, so the transfer function of the circuit is:
The output common mode voltage (VOP + VON)/2 is set by the voltage on the VOCM pin. If the VOCM pin is allowed to float, the output common-mode voltage will float to mid-supply voltage due to the internal 1 MΩ resistors forming the resistive divider of the supply. Capacitor C1 filters out the noise of the 1 MΩ resistor to reduce output common-mode noise. Because of the AD8476’s internal laser-trimmed gain-setting resistors, the gain error of the circuit is only 0.04% maximum.
Figure 1. Simple single-ended-to-differential converter.
For many applications, the circuit in Figure 1 is sufficient to perform a single-ended-to-differential conversion. For applications requiring higher performance, the single-ended-to-differential converter shown in Figure 2 has a high input impedance, a maximum input bias current of 2 nA, a maximum offset (RTI) of 60 µV, and a maximum offset drift of 0.7 µV /°C. This circuit achieves this level of performance by cascading an OP1177 precision operational amplifier (op amp) with the AD8476 and feeding back the positive output voltage of the AD8476 to the inverting input of the op amp. This feedback allows the op amp to determine the accuracy and noise performance of the configuration because it connects the differential amplifier within the feedback loop to the large open-loop gain of the preceding op amp. Therefore, when referenced to the input, this large gain reduces the AD8476’s errors, including noise, distortion, offset, and offset offset.
Figure 2. Improved single-ended-to-differential converter.
The circuit in Figure 2 can be represented by the following formula:
Simultaneous (1) and (3)
Equation 3 shows two important properties about the circuit: First, the circuit has a single-ended-to-differential gain of 2. Second, the VREF node acts as a reference to the input signal, so it can be used to remove offsets in the input signal. For example, if the input signal has a 1 V bias, applying 1 V to the REF node can remove the bias.
If the target application requires a gain greater than 2, the circuit in Figure 2 can be modified as shown in Figure 3. In this case, the single-ended-to-differential gain of the circuit depends on the external resistors RF and RG as follows:
Figure 3. Improved single-ended-to-differential converter with resistor-programmable gain.
Similar to the circuit in Figure 2, this improved single-ended-to-differential converter suppresses the error of the differential amplifier by placing the differential amplifier inside the op amp’s feedback loop. As with any feedback connection, we must be careful to ensure that the system is stable. Referring to Figure 2, the cascade of the OP1177 and AD8476 forms a composite differential output op amp. The open-loop gain over frequency is the product of the op-amp’s open-loop gain and the differential amplifier’s closed-loop gain. Therefore, the closed-loop bandwidth of the AD8476 adds a pole to the open-loop gain of the OP1177. To ensure stability, the bandwidth of the differential amplifier should be higher than the unity-gain frequency of the op amp. In the circuit shown in Figure 3, this requirement is relaxed because the resistive feedback network effectively reduces the unity-gain frequency of the OP1177 by a factor of RG/(RG + RF). Since the D8476 has a bandwidth of 5 MHz and the OP1177 has a unity gain frequency of 1 MHz, the circuit shown does not suffer from stability issues. Figure 4 shows an oscilloscope of the input and output signals of the circuit in Figure 2, driven by a ground-referenced 10 Hz, 1 V pp sine wave. For simplicity, the VREF node is grounded.
Figure 4. Input and output signals of the circuit in Figure 2 when driven by a ground-referenced 10 Hz, 1 V pp sine wave.
If the unity-gain frequency of the op amp used is much larger than the bandwidth of the differential amplifier, a bandwidth-limiting capacitor, CF, can be inserted, as shown in Figure 3. Capacitor CF and feedback resistor RF form an integrator, so the bandwidth of the entire circuit is calculated as:
The ? in the bandwidth formula is because the feedback is single-ended, not differential, which cuts the feedback and bandwidth in half. If the reduced bandwidth is lower than the closed loop bandwidth of the differential amplifier, the circuit will be very stable. This bandwidth limiting technique can also be used with a gain of 2, leaving RG open.