. . . The operational amplifier is the work horse of the analog world. Essential in both digital and analog systems, it is found in applications ranging from cellular phones to laptop computers to smoke detectors. The design constraint common to op. amps. in most of these applications is board space -- especially when the device is used as a "glue element" or a quick fix for problems not anticipated in the original design. Maxim's new Gain-Amps offer the smallest fix available.

The need for more and more functions in ever smaller and more portable products is constantly forcing circuit designers to conserve space and reduce complexity. Yet, the last-minute addition of an op. amp. for signal buffering or gain can always aggravate the allocation of board space. To address these situations, Maxim has introduced a new family of monolithic ICs called Gain-Amps. They combine a high-performance rail-to-rail op. amp. with precision laser-trimmed resistors in a SOT23 package.

Regardless of background or experience, any engineer can apply Gain-Amps with almost no design effort. Their low-power operation and rail-to-rail output capability offers the best combination of convenience and small size available in an op. amp. (see Fig.1)

The Gain-Amp's factory-trimmed internal resistors provide 0.1% gain and an internal V_CC/2 bias.

Integrating an op. amp. and factory-trimmed resistors on the same chip offers multiple benefits. Chief among these is the reduction in board space and component count, but additional secondary benefits are inherent in the device. The combination of internal resistors and Maxim's proprietary trimming technology, for example, enables gain trims to within 0.1%. This level of accuracy eliminates the headaches of cost and availability associated with high-accuracy, external surface-mount resistors.

Factory-trimmed gain also preserves full bandwidths: knowing the gain value of each amplifier at production prevents over-compensating its frequency response. At gains of 100 V/V, Gain-Amps offer a 22 MHz gain-bandwidth product while drawing only 325 ýA of supply current.

Other Gain-Amp features include single-supply operation in the range 2.5 V to 5.5 V, rail-to-rail outputs, a non-inverting input range that includes ground, ý17 V input fault protection, and the capability for direct drive of capacitive loads beyond 470 ýF. To implement inverting and non-inverting gains in the range 0.25 V/V to 100 V/V, each of the MAX4174/ MAX4175 and MAX4274/MAX4275 Gain-Amps come in 27 versions according to their internal R_F/R_G ratios (see Fig. 1, again.) For further savings in external components, the MAX4175 and MAX4275 include V_CC/2 bias resistors at the non-inverting inputs.

Also included in the Gain-Amp family are single/dual/quad, open-loop, unity-gain-stable amplifiers with no internal resistors (MAX4281/MAX4282/MAX4284.) The designer can use these devices in prototype development, and then replace them with factory-trimmed Gain-Amps.

On-chip Resistors Improve Cost, Space, and Accuracy

In the past, costly laser trimming was reserved primarily for data converters and other precision devices. Today, trimming the offset voltage and supply current in op. amps. is a standard practice in the industry, even though it increases the IC's cost, size, and complexity. Until the advent of modern trimming technology, the drawbacks introduced by internal gain resistors outweighed any benefits. Nor did market demands warrant the integration of on-chip resistors.

Today's trend toward miniaturization has tipped the scales. Using the smallest available components to achieve tightly-controlled accuracy is expensive, and the necessary components are not always readily available. Two such 0.1% surface-mount resistors can add more than $0.20 to the cost of a simple amplifier and contribute 0.2% to its gain-error, exclusive of any gain-error inherent in the op. amp. The alternative -- internally trimmed gain for space-sensitive applications -- has become more attractive, thanks to the availability of efficient trimming techniques, more compact amplifier cores, and lower costs.

A comparison of printed circuit board areas shows that the Gain-Amp offers dramatic improvement over previous designs based on an op. amp. (5-pin SOT23 or SC70 package) and the four external surface-mount resistors necessary for gain and biasing. Even when these designs incorporate the smallest surface-mount resistors available today, the Gain-Amp offers better accuracy and performance while occupying only 30% to 50% of the board area. Space-critical designs have no better alternative.

The capability for trimming overall gain-error (rather than just matching the resistors) is another benefit inherent in on-chip gain resistors. For example, specifying a typical overall gain error of 0.1% (for V_OUT within 25mV of either rail) requires an R_F/R_G ratio with accuracy better than 0.1%. Because on-chip resistors offer excellent matching -- not only at room temperature but also as the temperature changes -- the Gain-Amp's monolithic design provides an R_F/R_G ratio with inherently low drift. Low drift guarantees overall gain errors no higher than 0.5% over the extended temperature range -- a claim difficult to make for discrete components.

Optimizing Frequency Compensation To Maximize Bandwidth

Gain-Amps include optimized frequency compensation for each range of gains. Like any system with feedback, an op. amp. has the potential to ring or oscillate. Indeed, the parameters that contribute to better performance (high dc open-loop gain and wide bandwidth) can also contribute to unstable behavior. To prevent oscillation, most op. amps. include a dominant pole that ensures good low-frequency performance while reducing the open-loop gain at high frequencies. The low-frequency pole reduces bandwidth, but most amplifiers require it for stable operation. A quick analysis shows the tradeoff between stability and bandwidth.

A basic algebraic derivation yields block diagrams for the common inverting and non-inverting configurations (see Fig.2) and Black's formula can determine the overall transfer function for each system. The factor R_GA(S)/(R_G+R_F) (called the loop gain of the system) represents the amount of output signal fed back and subtracted from the input. If the open-loop gain A(S) is much greater than unity, the transfer function reduces to the reciprocal of the feedback factor: (R_G+R_F)/R_G. As long as A(S) remains sufficiently high, the closed-loop gain depends entirely on the values of R_F and R_G. Note that decreasing the feedback factor increases the system's closed-loop gain.

The feedback dynamics for inverting and non-inverting amplifiers are identical.

The closed-loop system remains stable as long as its loop gain subtracts from the input signal. (The action known as negative feedback.) Instability occurs when frequency-dependent phase shifts reverse the feedback polarity. As stated in control theory for monotonically decreasing gain and phase, a closed loop remains stable as long as the phase of the loop gain does not shift by 180ý at frequencies for which the gain is greater than unity. When the loop gain inverts a sinusoidal signal, the feedback reverses polarity and becomes positive. Under that condition, oscillation at the sinusoidal frequency will be sustained even if the input signal is completely removed.

For the criteria just stated, phase shift will not exceed 90ý and the system will be unconditionally stable if the op. amp's. open-loop gain has only one pole. Because an op. amp. is a very complicated, high-gain, multi-pole system, however, its first and second natural poles usually occur at frequencies for which the gain is greater than unity. Op. amp. designers therefore introduce an additional, low-frequency pole to counteract the effect of these inherent poles. By rolling off the open-loop gain, the low-frequency pole assures that the gain is much less than unity when the feedback signal undergoes a 180ý phase shift.

In the loop gain and phase of a Gain-Amp (see Fig. 3a) compensated and configured for unity gain, point A indicates the loop gain's crossover frequency, and point B indicates where the phase shift reaches 180ý. The lower the gain at point B, the more stable the system becomes. The phase difference between points A and B is called the phase margin, and higher phase margin denotes greater stability. Note that higher closed-loop gain increases the phase margin by decreasing the magnitude of R_GA(S)/(R_G+R_F).

Curve A (see Fig. 3b) shows the additional phase margin obtained by increasing the closed-loop gain from unity to 100 V/V. High closed-loop gains are inherently more stable. Using a unity-gain-stable op. amp. in a high-gain configuration wastes bandwidth, but optimizing the compensation can preserve useful bandwidth. For this case, curve B shows the loop-gain bandwidth regained by de-compensating the amplifier. A decade of bandwidth is added to the closed-loop response as well. Because the feedback dynamics for inverting and non-inverting configurations are identical, this analysis holds for both cases.

For closed-loop systems, the phase margin is a measure of stability.

Operating a unity-gain-compensated op. amp. in a high closed-loop gain configuration yields a high phase margin but wastes valuable bandwidth.

Adjusting the compensation network for each gain is expensive and inefficient in practice. Instead, Maxim designates five different compensation networks, which are distributed over the 27 gain versions available for each device. The five compensation break points are optimized for gains of 1.25 V/V, 3 V/V, 5 V/V, 10 V/V, and 25 V/V. Unlike the constant gain-bandwidth product (GBW) of conventional unity-gain amplifiers, that of a Gain-Amp is re-optimized (maximized) at each gain breakpoint (see Fig. 4.) The Gain-Amp's GBW near 100 V/V is 22MHz. The plot shows that most benefits of the optimized frequency response are realized at low gains. Above 25 V/V, de-compensating the Gain-Amp is useful but offers diminishing returns. Regardless of experience, every designer can take full advantage of the Gain-Amp's bandwidth without manipulating complicated external compensation networks.

Unlike most conventional op. amps., frequency-optimized Gain-Amps actually recover much of their useful bandwidth when configured for higher gains.

Innovative Design Delivers Rail-to-Rail Output

The trend toward lower power consumption and less complex systems has diminished the popularity of dual and high-voltage supplies. As a consequence, today's market demands lower supply voltages than ever before. To sustain a good dynamic range while operating from low supply voltages, Gain-Amps employ an output stage that swings rail-to-rail.

Conventional output stages comprise emitter-followers or source-followers. These stages form excellent buffers, but they have difficulty in swinging to within one volt of the rails, especially when loaded. An emitter-follower stage in a 2.5V system offers a total swing of only 500 mV. The rail-to-rail output of a Gain-Amp, on the other hand, operating with a 100 kW load over the extended temperature range, is guaranteed to swing within 5 mV of each rail. With a heavier load of 1 kW, the Gain-Amp output is guaranteed to swing within 250 mV of the positive rail and within 100 mV of the negative rail, over temperature.

Achieving rail-to-rail performance precludes the use of source followers or emitter followers in the output buffer. Only a common-emitter or common-source stage can swing so close to the rails (see Fig. 5.) The output stage must actually provide load-driving capability as well as gain. The gain of a common-source stage (unlike that of a source-follower) drops with increasing drain current, producing gain that varies with the output current. Load-dependent gain under heavy load introduces error, and it also increases distortion due to fluctuation in the output-stage transconductance.

The common-source output stage makes rail-to-rail swings possible.

To increase open-loop gain and reduce the effects of output loading, most low-voltage rail-to-rail op. amps. are obliged to incorporate a 3-stage design. Because the intermediate stage provides additional gain, fluctuations in the output transistors' transconductance have less effect on the overall loop gain. But, three-stage designs have more complex circuitry, require a larger die size, require more supply current, and introduce poles in the frequency response that make compensation more difficult. Three-stage compensation often kills much of the useful bandwidth in an amplifier. The Gain-Amp's required accuracy and low-voltage operation called for a high-gain rail-to-rail architecture, but space and bandwidth constraints eliminated consideration of the 3-stage design.

The key advantage of a Gain-Amp lies in the proprietary 2-stage design (patent pending) that overcomes the open-loop-gain fluctuations characteristic of rail-to-rail amplifiers. Like most op. amps., the Gain-Amp incorporates a MOS-input, folded-cascode differential pair with a common-source second stage. Unlike conventional amplifiers, it has internal circuitry that compensates for the effect of output loading. As the output transconductance falls, feedback boosts the open-loop gain. The result is a compact, low-power op. amp. that maintains high open-loop gain and is simple to compensate.

High-Voltage (ý17 V) Fault Protection

Providing input protection to ý17V internally makes the Gain-Amp flexible, and for applications in which high voltage may be present, also lowers the external component count. The internal op. amp. exhibits a common-mode rejection ratio of 60 dB for inputs ranging from 150 mV below V_EE to 1.2 V below V_CC. Although it cannot meet this CMRR at higher voltages, the output is guaranteed not to undergo a phase reversal for inputs within ý17 V of ground.

In the input structures (see Fig. 6) that provide protection against high-voltage faults, although the IC's absolute maximum rating for supply voltage is 6 V, the input silicon-controlled rectifiers (SCRs) can sustain a minimum of 17 V above or below V_EE before clamping. Internal diodes clamp the inverting and non-inverting inputs to V_CC and V_EE. Applying 0.3 V (or more) above the positive or below the negative rail causes current flow by forward-biasing the protection diode. Over-voltage would normally damage these structures, but current through each diode is limited by a series resistance (RG at the inverting input; 5 kW at the non inverting input.) The path from either input to its SCR contains no such resistance, so exposure to inputs in excess of ý17V will cause conduction and the possibility of being electrically overstressed.

Input voltages that exceed the supply rails are current-limited by a 5 kW resistor at the non-inverting input and by R_G at the inverting input.

Input voltage levels don't normally exceed the supply rails, but two of the Gain-Amp options act as attenuators (A_V = -0.5 V/V and A_V = -0.25 V/V), which allow input voltages to exceed the supplies without clipping the output or violating the internal amplifier's common-mode requirements. For a 5 V supply, this capability allows input swings from +7.5 V to -2.5 V (A_V = -0.5 V/V) and input swings from +9.75 to -5.5 V (A_V = -0.25 V/V, when biased near 2.5 V.) These limits hold for inverting gains only, because the op-amp inputs are held at a constant bias voltage and cannot vary as the signal varies.

Power-Supply Rejection and Bypassing

Gain-Amps exhibit excellent immunity to low-frequency variations in the power-supply voltage. Their high dc power-supply rejection (102 dB) is guaranteed from 2.5 V to 5.5 V over the operating temperature range. For the maximum-allowed power-supply excursion of 2.5 V to 5.5 V and a full-scale output voltage of 2.5 V, this PSR yields an output change of DV_OUT = [10^(102dB/20)]/(5.5 V-2.5 V) = 24 ýV: an error of less than 0.001%. In conjunction with low voltage, low supply current, and rail-to-rail operation, this level of power-supply immunity makes the Gain-Amp well suited for use in portable and battery-powered instruments and equipment. It enables these systems to maintain accuracy and initial calibrations while operating directly from a single lithium-ion cell or a stack of three NiCd, NiMH, or alkaline cells.

Gain-Amps also provide excellent ac power-supply rejection up to 20kHz (see Fig. 7.) The rejection of 50 Hz, 60 Hz, 100 Hz, and 120 Hz ripple is critical in wall-powered applications for which the pre-regulation is not adequate. At these low frequencies, ripple immunity inherent in the IC eliminates the need for bulky, low-frequency filter capacitors. Like most amplifiers, the Gain-Amp requires high-frequency power-supply decoupling, in this case using a compact 0.1 ýF ceramic capacitor. To improve the circuit's high-frequency performance when operating with a noisy power supply, add an external lowpass RC network at V_CC to reduce noise coupling through the supply.

Gain-Amps exhibit excellent power-supply rejection to 20 kHz.

MAX4175/MAX4275 Gain-Amps save additional components in single-supply applications by self-biasing the non-inverting input halfway between V_CC and ground, using an internal resistor divider (see Fig. 1, again.) The amplifier itself rejects power-supply changes, but the divider at the amplifier's non-inverting input attenuates V_CC fluctuations by only a factor of two. A supply-voltage change from 2.5V to 5.5V, for instance, shifts the output as much as 1.5 V.

Low-frequency V_CC fluctuations cannot be filtered, because the divider feeds the non-inverting input directly. Improving the power supply's dc performance is the only viable solution. To minimize the coupling of ac signals through the bias resistors to the non-inverting input, bypass the input to ground with a 1 mF ceramic capacitor. The remaining, residual power-supply feedthrough (with the IN+ pin floating and bypassed to ground) decreases with frequency.

Capacitive Drive

As in many other applications, the input buffer for an analog-to-digital converter must drive a highly capacitive load. This load capacitance can create ringing and oscillation by reducing phase margin in the buffer's frequency response. One antidote to the problem is to choose an amplifier whose output stage is designed to drive capacitive loads. For example, a MAX4174 Gain-Amp set for 3 V/V gain and driving a 1 kW load exhibits little more ringing with a 470pF load than without (see Fig. 8a and 8b.) Adding an isolation resistor lets the device drive more capacitance. To maintain high dc accuracy, this resistor should be 1000 times smaller than the load resistance. Combining a 30W isolation resistor with a 100 kW load, for instance, introduces a gain error of only 0.03%.

The output of a Gain-Amp with no capacitive load (a) and loaded with 470 pF (b) shows minimal additional ringing.

Thus, the Gain-Amp's convenience and simplicity targets applications that require compact design. By maximizing accuracy, bandwidth and space without the manufacturing headaches that plague conventional op. amps., it makes an ideal choice for use in tight spaces. Gain-Amps are the first line of defense against amplifier-design obstacles and are perfect for last-minute fixes.

Editor's Note: For details of the full range of Maxim's Gain-Amp products, and for a selection guide, please contact the company at (408) 737-7600 or http://www.maxim-ic.com.