10.2.2 Detailed Design Procedure

The OPA858 is compensated to have less than 1 dB of peaking in a gain of 7 V/V. Using the device in lower gains results in increased peaking and potential instability. Figure 59 shows the OPA858 configured in a signal gain of –2 V/V. The DC noise gain (1/β) of the amplifier is affected by the 62-Ω termination resistor and the 50-Ω source resistor and is given by Equation 1. At higher frequencies the noise gain is affected by reactive elements such as inductors and capacitors. These include both discrete board components as well as printed circuit board (PCB) parasitics.

Equation 1.

The stability and phase margin of the amplifier depend on the loop gain of the amplifier, which is the product of the A_OL and the feedback factor (β) of the amplifier. The β of a negative-feedback loop system is the portion of the output signal that is fed back to the input, and in the case of an amplifier is the inverse of the noise gain. The noise gain of the amplifier at high frequencies can be increased by adding an input capacitor and a feedback capacitor as Figure 60 shows. If done carefully, increasing 1/β improves the phase margin just as any amplifier is more stable in a high gain configuration versus a unity-gain buffer configuration. The modified network with the added capacitors alters the high-frequency noise gain, but does not alter the signal gain. The AN-1604 Decompensated Operational Amplifiers application report provides a detailed analysis of noise gain-shaping techniques for decompensated amplifiers and shows how to choose external resistors and capacitor values.

Figure 61 shows the uncompensated frequency response of the OPA858 configured as shown in Figure 59. Without any added noise gain shaping components, the OPA858 shows approximately 13 dB of peaking.

Figure 62 shows the noise gain compensated frequency response of the OPA858 configured as shown in Figure 60. The noise gain shaping elements reduce the peaking to less than 1.5 dB. The 2.7-pF input capacitor, the input capacitance of the amplifier, the gain resistor, and the feedback resistor create a zero in the noise gain at a frequency f, as Equation 2 shows.

Equation 2.

where

• R_F is the feedback resistor
• R_G is the input or gain resistor (includes the effect of the source and termination resistor)
• C_IN is the total input capacitance, which includes the external 2.7-pF capacitor, the amplifier input capacitance, and any parasitic PCB capacitance.

The zero in Equation 2 increases the noise gain at higher frequencies, which is important when compensating a decompensated amplifier. However, the noise gain zero reduces the loop gain phase which results in a lower phase margin. To counteract the phase reduction due to the noise gain zero, add a pole to the noise gain curve by inserting the 0.5-pF feedback capacitor. The pole occurs at a frequency shown in Equation 3. The noise gain pole and zero locations must be selected so that the rate-of-closure between the magnitude curves of A_OL and 1/β is approximately 20 dB. To ensure this, the noise gain pole must occur before the 1/β magnitude curve intersects the A_OL magnitude curve. In other words, the noise gain pole must occur before |A_OL| = |1/β|. The point at which the two curves intersect is known as the loop gain crossover frequency.

Equation 3.

where

• C_F is the feedback capacitor (includes any added PCB parasitic)

For more information on op amp stability, watch the TI Precision Lab series on stability video.