Application Hints

AMPLIFIER TOPOLOGY

The topology chosen for the LMC660, shown in Figure 1,is

unconventional (compared to general-purpose op amps) in

that the traditional unity-gain buffer output stage is not used;

instead, the output is taken directly from the output of the

integrator, to allow rail-to-rail output swing. Since the buffer

traditionally delivers the power to the load, while maintaining

high op amp gain and stability, and must withstand shorts to

either rail, these tasks now fall to the integrator.

As a result of these demands, the integrator is a compound

affair with an embedded gain stage that is doubly fed forward

(via C

driver. In addition, the output portion of the integrator is a

push-pull configuration for delivering heavy loads. While

sinking current the whole amplifier path consists of three

gain stages with one stage fed forward, whereas while

sourcing the path contains four gain stages with two fed

forward.

The large signal voltage gain while sourcing is comparable

to traditional bipolar op amps, even with a 600

Ω load. The

gain while sinking is higher than most CMOS op amps, due

to the additional gain stage; however, under heavy load

(600

Ω) the gain will be reduced as indicated in the Electrical

Characteristics. Avoid resistive loads of less than 500

Ω,as

they may cause instability.

COMPENSATING INPUT CAPACITANCE

The high input resistance of the LMC660 op amps allows the

use of large feedback and source resistor values without

losing gain accuracy due to loading. However, the circuit will

be especially sensitive to its layout when these large-value

resistors are used.

Every amplifier has some capacitance between each input

and AC ground, and also some differential capacitance be-

tween the inputs. When the feedback network around an

amplifier is resistive, this input capacitance (along with any

additional capacitance due to circuit board traces, the

socket, etc.) and the feedback resistors create a pole in the

feedback path. In the following General Operational Amplifier

circuit,Figure 2 the frequency of this pole is

where C

including amplifier input capacitance and any stray capaci-

tance from the IC socket (if one is used), circuit board traces,

etc., and R

formula, as well as all formulae derived below, apply to

inverting and non-inverting op amp configurations.

When the feedback resistors are smaller than a few k

Ω, the

frequency of the feedback pole will be quite high, since C

generally less than 10 pF. If the frequency of the feedback

pole is much higher than the “ideal” closed-loop bandwidth

(the nominal closed-loop bandwidth in the absence of C

the pole will have a negligible effect on stability, as it will add

only a small amount of phase shift.

However, if the feedback pole is less than approximately 6 to

10 times the “ideal” −3 dB frequency, a feedback capacitor,

C

ing input of the op amp. This condition can also be stated in

terms of the amplifier’s low-frequency noise gain: To main-

tain stability a feedback capacitor will probably be needed if

where

is the amplifier’s low-frequency noise gain and GBW is the

amplifier’s gain bandwidth product. An amplifier’s low-

frequency noise gain is represented by the formula

regardless of whether the amplifier is being used in inverting

or non-inverting mode. Note that a feedback capacitor is

more likely to be needed when the noise gain is low and/or

the feedback resistor is large.

If the above condition is met (indicating a feedback capacitor

will probably be needed), and the noise gain is large enough

that:

the following value of feedback capacitor is recommended:

If

the feedback capacitor should be:

00876704

FIGURE 1. LMC660 Circuit Topology (Each Amplifier)

www.national.com

7