In today's electronic systems, the negative power supply isdisappearing, while the positive power supply voltage is decreasing.This trend is driving the proliferation of rail-to-rail amplifiers.While the supply voltage is changing, the signal level often remainsthe same.
For example, the standard video signal is 2V. When the power supplycomes down to 2V, the amplifier/buffer must be linear or accuratewithin the full 2V supply range.
This article focuses on the development of the rail-to-railamplifier's input stage and discusses in detail the input enhancementcircuit that overcomes the deficiencies associated with rail-to-railamplfiers. For simplicity, this discussion is limited to MOSFETamplifiers.
|Figure1: This topology provides differential gain and rejects common-modesignals, but its limitation is in its operating range.|
The input stage from a basic op amp is shown in Figure 1, above . A transistor pair,called a differential pair, sits atop a current source to accommodatethe differential input.
While this topology provides differential gain and rejectscommon-mode signals, its limitation is in its operating range. Theinput voltage range is 0-1.5V with a 3V single-supply. If the inputvoltage is raised above 1.5V, the current source is forced out ofsaturation. Once the current source leaves the saturation region, thegain is distorted.
For a sample application such as current or voltage sensing by anelectrocardiogram (EKG), the quality of the design is directly relatedto the voltage range of signals that can be processed. A standardtextbook rail-to-rail opamp topology accommodates this challenge with adual-input stage (Figure 2, below ).
|Figure2: In sensitive applications like the EKG, any variation in offsetvoltage jeopardizes the accuracy of the system.|
When the input voltage approaches the lower supply rail, the PMOStransistor pair amplifies the signal. Conversely, the NMOS differentialpair amplifies input signals that approach the upper supply rail. Inthis way, the input can span the entire supply voltage range. The mostobvious trade-off to achieve this improvement in input range is theextra power required to bias the complementary differential pair.
A less obvious trade-off exists in the offset voltage with respectto the input bias voltage. The offset for the NMOS pair does notnecessarily match the offset of the PMOS pair and occurs with opposingpolarity.
Somewhere near mid-supply, there is a hand-off from one pair to theother. During the hand-off, the offset voltage is the average of theoffset from each pair. This creates a stair-step characteristic.
For added insight, the offset is plotted for various temperatures.The PMOS input pair, active for low common-mode input voltages,exhibits a wide range of offset voltage vs. temperature. The variationin the NMOS pair causes the distribution on the right side of the plotfor high common mode input voltages.
In sensitive applications like the EKG, any variation in offsetvoltage jeopardizes the accuracy of the system. The signal must firstbe amplified well beyond the offset voltage level to take advantage ofrail-to-rail amplifiers with an input topology shown in Figure 2.
|Figure3: An input-range enhancement circuit has been included in op amps likethe EL8178 to adjust the bias internally provided to the current source.|
Full range of inputs
In precision and low-power applications, a new type of rail-to-railamplifier is needed. The goal is to achieve a full range of inputvoltages without the crossover distortion in the offset voltage thatoccurs during the hand-off region of the dual differential design.
Let's return to the single differential design. The input range ofthe topology in Figure 1 does not allow for full range input operation.
A portion of the input range is preserved for biasing the currentsource in the saturation region. Can the biasing for the current sourcebe accomplished in a manner that allows the input to span between thesupply rails? An input-range enhancement circuit has been included inop amps like the EL8178 to adjust the bias internally provided to thecurrent source (Figure 3, above ).
Inside the enhancement circuit is a charge pump. While the mentionof a charge pump usually brings up noise issues, the charge pump'soperating frequency is beyond the bandwidth of the amplifier. Thus,there is no measurable change in the noise performance of theamplifier. Moreover, we must revisit the issue of offset voltage.
|Figure4: The input range enhancement circuit allows a single differentialpair to provide rail-to-rail operation without the need for a second,complementary differential pair.|
Figure 4, above , fulfillsour goal of maintaining the offset voltage. The input range enhancementcircuit allows a single differential pair to provide rail-to-railoperation without the need for a second, complementary differentialpair.
The offset voltage is completely dependent on the mismatch of onlyone set of transistors, so there is no crossover region. Careful layoutand trimming can ensure that the input-referred offset voltage is lessthan 100 microvolts.
So far, the discussion has been limited to MOSFET implementations.Bipolar technologies can also benefit from this configuration. Besidesimprovement in offset voltage, a bipolar version would also exhibit asimilar improvement in input bias current. The input bias current wouldfeed only one matched differential pair, not two pairs with a crossoverregion.
|Table1. Table 1: The EL8178 provides the specifications needed by low-power,high-resolution systems like a portable EKG machine.|
An example of the evolution of rail-to-rail amplifiers is shown in Table 1, above. It summarizes theperformance of three op-amp examples. The final version, the EL8178,provides the specifications needed by low power, high-resolutionsystems like a portable EKG machine.
The basic input stage, which is composed of a single-differentialpair, doesn't allow a full range of voltages at the input. A dualdifferential pair extends the input voltage range to the supplies, butsuffers nonlinearity in the offset voltage (and input bias current inbipolar junction transistors) because of the hand-off between the twopairs.
The third solution includes an internal enhancement to adjust thebias on the current source of a single differential pair to allowrail-to-rail operation with no discontinuity in offset voltage
Mike Wong is VP of ApplicationsEngineering and Tamara Schmitz is principal Applications Engineer at Intersil Corp. To read a PDFversion of this story go to Rail-to-railamp improves input offset accuracy.