8.3.1 Capacitive Load

The LMP770x can each be connected as a non-inverting unity gain follower. This configuration is the most sensitive to capacitive loading.

The combination of a capacitive load placed on the output of an amplifier along with the amplifier's output impedance creates a phase lag which in turn reduces the phase margin of the amplifier. If the phase margin is significantly reduced, the response will be either underdamped or it will oscillate.

To drive heavier capacitive loads, an isolation resistor, R_ISO, in Figure 40 should be used. By using this isolation resistor, the capacitive load is isolated from the amplifier's output, and hence, the pole caused by C_L is no longer in the feedback loop. The larger the value of R_ISO, the more stable the output voltage will be. If values of R_ISO are sufficiently large, the feedback loop will be stable, independent of the value of C_L. However, larger values of R_ISO result in reduced output swing and reduced output current drive.

Figure 40. Isolating Capacitive Load

8.3.2 Input Capacitance

CMOS input stages inherently have low input bias current and higher input referred voltage noise. The LMP770x enhance this performance by having the low input bias current of only ±200 fA, as well as, a very low input referred voltage noise of 9 nV/√Hz. To achieve this a larger input stage has been used. This larger input stage increases the input capacitance of the LMP770x. The typical value of this input capacitance, C_IN, for the LMP770x is 25 pF. The input capacitance will interact with other impedances such as gain and feedback resistors, which are seen on the inputs of the amplifier, to form a pole. This pole will have little or no effect on the output of the amplifier at low frequencies and DC conditions, but will play a bigger role as the frequency increases. At higher frequencies, the presence of this pole will decrease phase margin and will also cause gain peaking. To compensate for the input capacitance, care must be taken in choosing the feedback resistors. In addition to being selective in picking values for the feedback resistor, a capacitor can be added to the feedback path to increase stability.

The DC gain of the circuit shown in Figure 41 is simply –R₂/R₁.

Figure 41. Compensating for Input Capacitance

For the time being, ignore C_F. The AC gain of the circuit in Figure 41 can be calculated as follows:

Equation 1.

This equation is rearranged to find the location of the two poles:

Equation 2.

As shown in Equation 2, as values of R₁ and R₂ are increased, the magnitude of the poles is reduced, which in turn decreases the bandwidth of the amplifier. Whenever possible, it is best to choose smaller feedback resistors. Figure 42 shows the effect of the feedback resistor on the bandwidth of the LMP770x.

Figure 42. Closed-Loop Gain vs Frequency

Equation 2 has two poles. In most cases, it is the presence of pairs of poles that causes gain peaking. To eliminate this effect, the poles should be placed in Butterworth position, because poles in Butterworth position do not cause gain peaking. To achieve a Butterworth pair, the quantity under the square root in Equation 2 should be set to equal −1. Using this fact and the relation between R₁ and R₂, R₂ = −A_V R₁, the optimum value for R₁ can be found. This is shown in Equation 3. If R₁ is chosen to be larger than this optimum value, gain peaking will occur.

Equation 3.

In Figure 41, C_F is added to compensate for input capacitance and to increase stability. Additionally, C_F reduces or eliminates the gain peaking that can be caused by having a larger feedback resistor. Figure 43 shows how C_F reduces gain peaking.

Figure 43. Closed-Loop Gain vs Frequency With Compensation

8.3.3 Diodes Between the Inputs

The LMP770x have a set of anti-parallel diodes between the input pins, as shown in Figure 44. These diodes are present to protect the input stage of the amplifier. At the same time, they limit the amount of differential input voltage that is allowed on the input pins. A differential signal larger than one diode voltage drop might damage the diodes. The differential signal between the inputs needs to be limited to ±300 mV or the input current needs to be limited to ±10 mA.

Figure 44. Input of LMP7701

8.4.1 Precision Current Source

The LMP770x can each be used as a precision current source in many different applications. Figure 45 shows a typical precision current source. This circuit implements a precision voltage controlled current source. Amplifier A1 is a differential amplifier that uses the voltage drop across R_S as the feedback signal. Amplifier A2 is a buffer that eliminates the error current from the load side of the R_S resistor that would flow in the feedback resistor if it were connected to the load side of the R_S resistor. In general, the circuit is stable as long as the closed loop bandwidth of amplifier A2 is greater then the closed loop bandwidth of amplifier A1. If A1 and A2 are the same type of amplifiers, then the feedback around A1 will reduce its bandwidth compared to A2.

Figure 45. Precision Current Source

The equation for output current can be derived as shown in Equation 4.

Equation 4.

Solving for the current I results in the Equation 5.

Equation 5.