An amplifier’s PSRR (power-supply-rejection ratio) is among the most commonly characterized parameters when analyzing the performance of an op amp. Examples of some noise sources on an amplifier’s power-supply pins include parasitic supply-line traces, their interaction with currents that the amplifier draws, and the noise that switching circuits sharing the same supply create. Both sources produce voltage-amplitude variations reproduced as noise signals at the amplifier’s input pins.

Characterizing PSRR over frequency commonly involves the use of analyzers equipped with a dc-bias port, such as Agilent’s 8753. To measure negative PSRR, for example, the amplifier’s –V_S pin comes through Port 1, with the negative dc voltage through the bias port, of the 8753 with a superimposed sinusoid. To complete the measurement, you measure the amplifier’s output on Port 2. Unfortunately, the 8753 doesn’t measure frequencies below 30 kHz because of the limitations of the analyzer’s internal bias, T. Additionally, most PSRR-versus-frequency plots begin at frequencies far below 30 kHz.

An alternative technique would involve the use of an analyzer that has no dc-bias port but that can characterize frequency response as low as 10 or even 1 Hz. One such analyzer is the Stanford Research Systems SR785, which can make measurements better than –120 dB. One way of approaching this problem is to connect the output port of the SR785 to a buffer/inverting-summer circuit constructed with an Analog Devices AD8034.

Figure 1 illustrates a negative-PSRR test-circuit configuration. Pin 3 connects to the SR785 source-output port. Pin 1, which is V_OUT of the buffer amplifier, connects to the reference port of the SR785. Here, the first amp isolates the output port of the SR785 from the dc bias and provides the sinusoidal output. The second amplifier within the AD8034 sums the dc bias and sinusoid, which it uses to feed the DUT’s (device under test’s) negative-supply pin. The figure omits all bypass capacitors at the DUT’s negative-supply pin. A 1-kΩ resistor from Pin 3 to ground prevents the noninverting input from floating. The positive terminal of the external dc-power supply feeds Pin 6 through a 1-kΩ resistor. Connecting the DUT’s output to Channel 2A of the SR785 completes the test-circuit configuration.

Building the buffer/inverting summer with an AD8034 dual amplifier is a good choice because it has a supply range of 5 to 24V; a signal-frequency response well beyond 1 MHz; and a large capacitive-load-drive capability, allowing you to neglect the capacitance of test cables. Further, the AD8034 can deliver as much as 40 mA of load current.

To instill confidence that this buffer/inverting-summer configuration works, Figure 2 proves that you can neglect any loss that the AD8034 incurs. The figure demonstrates that the response of the AD8034 buffer/inverting summer of 10 Hz to 10 kHz is approximately flat with a loss of only 0.0025 dB, and the loss from 10 to 100 kHz is approximately 0.024 dB. Figure 3 shows negative-PSRR test results. The Hewlett-Packard HP8753 provides the PSRR-versus-frequency responses beyond 100 kHz. You can measure positive PSRR (figure 4 and figure 5) by connecting Pin 3 to the SR785’s output port. Pin 1, V_OUT of the buffer amplifier, connects to the reference port of the SR785. Here, you use the first amp to isolate the output port of the SR785 from the dc bias and provide the sinusoidal output. The second amplifier within the AD8034 sums the dc bias and sinusoid, which you use to feed the DUT’s positive-supply pin. You must remove all bypass capacitors at the DUT’s positive-supply pin. A 1-kΩ resistor from Pin 3 to ground prevents the noninverting input from floating. Feed the negative terminal of the external dc-power supply to Pin 6 through a 1-kΩ resistor. Connecting the DUT’s output to Channel 2A of the SR785 completes the test-circuit configuration.

For the AD8034, assume that the DUT has a maximum supply voltage of ±15V, that you need to test negative PSRR, and that the DUT supplies ±10V. If you want to accommodate the maximum SR785 output of 5V peak, the first amplifier of the AD8034 needs enough head room to avoid clipping the ±5V signal from the output port of the SR785. In this case, a supply setting for the AD8034 of 6 and –16V is sufficient to prevent any problems. This amount provides enough head room to accommodate the first amp of the AD8034, which handles a ±5V signal centered at ground. The –16V accommodates the dc bias of –10V and the ±5V signal centered at –10V at the output of the second amplifier of the AD8034. Positive PSRR is similar: Just set the AD8034 supplies to 16 and –6V for this example.

You might consider using separate power supplies for the DUT and the AD8034 to simplify matters. However, you can use the same dc-power supply for the DUT to provide the dc-bias voltage at Pin 6 of the AD8034 buffer/inverting summer. Choose the output voltage of the SR785 or whichever analyzer you use so that the DUT operates within its linear region of operation. You can apply this technique to other applications.