The open-loop output impedance curve given in most op amp datasheets is a typical value. The datasheet typical value is also used in the op amp model. The Z_O typical measurement was made at room temperature, with typical power supplies and load conditions. Factors such as output current, temperature, supply voltage, and process variations can cause Z_O to change. Most datasheets do not provide estimates for worst case Z_O, so targeting a conservative phase margin that remains stable even with some variation in Z_O is advisable. This is why TI generally recommends a minimum phase margin of 45° even though 35° can potentially provide an acceptable transient response. Thus, if you design the circuit to a phase margin of 45°, across process and temperature the phase margin can potentially vary from 40° to 50°, which is generally acceptable.

Figure 6-22 illustrates how load current (amplifier output current) shifts Z_O for OPA376. In general, Z_O decreases for higher load current. Generally, the Z_O curve in the datasheet is measured with minimal load current (large R_L). For heavy loads (small R_L), Z_O decreases, which generally improves stability. Some amplifier datasheets, such as OPA376, show Z_O for multiple different loads to show how Z_O changes with load current.

Figure 6-23 illustrates an example of how Z_O changes across temperature for PGA900. The PGA900 is used in the next few examples because the open-loop output impedance of PGA900 is characterized over a wide range of conditions that most devices are not tested under. While each amplifier has different characteristics and sensitivities, the PGA900 data provides a rough idea of what to expect. Figure 6-23 shows that temperature has a more significant effect at lower frequencies. In most circuits with stability concerns, the high frequency Z_O determines the amplifier stability and low frequency Z_O is not important. In this example, the curves are relatively tightly grouped past 10kHz (ΔZ_O(1MHz) ≅ ±20%).

The variation in Z_O versus power supply voltage for PGA900 is really negligible (see Figure 6-24). This makes sense as op amps use internal biasing schemes that keep the bias current for the different internal blocks constant versus power supply voltage.

Figure 6-25 shows the process variation of Z_O for PGA900. The data was collected for multiple different devices across multiple lots and the curves represent the worst-case deviations. As with other curves in this section, the variations at high frequency are not that significant, so the overall impact on stability is not that significant.

Figure 6-26 and Figure 6-27 illustrate the importance of decoupling for stability of op amps. Figure 6-26 shows the test circuit for measuring Z_O. The circuit has a source impedance in the amplifier power supplies labeled R_PS. This can be an inductive reactance, but for simplicity a resistance was used. With decoupling capacitors C₂ and C₃ installed, the power supply resistance is effectively shorted out from an AC perspective. If the decoupling is omitted, the resistance R_PS effectively adds to the open-loop output impedance Z_O. This is why improper decoupling or lack of decoupling often leads to op amp instability. Figure 6-27 illustrates measured Z_O for OPA827 with the decoupling removed and resistor R_PS added to the supply. Inspecting the graph shows that the source resistance increases the open-loop output impedance. The key point here is that this measurement was made without decoupling capacitors. Adding the decoupling effectively shorts out RPS so that all the Z_O values match the original Z_O regardless of R_PS. Thus, decoupling capacitors are very important from a stability perspective. Use a minimum of 0.1µF decoupling with short connections between the amplifier supply pins and ground for good stability.

The intention of this section is to emphasize that process, load, temperature, and decoupling all impact Z_O and stability. Since most amplifier data sheets and associated SPICE models use the typical value for Z_O, having a design target for phase margin that is conservative is important so that the circuit remains stable across process and temperature. This is really the motivation behind the 45° general guidance.