To reduce product size, reduce cost, extend battery life, and improve the performance of battery-powered systems, thermal engineers have accelerated the development and application trend of low-voltage, single-supply systems. This trend is beneficial to consumers, but it complicates the selection of the right operational amplifier for a specific application.
Single-supply operation is often the same as low-voltage operation. Moving from ±15V or ±5V to a single 5V or 3V supply reduces the usable signal range. Therefore, common-mode input range, output voltage swing, CMRR, noise, and other op amp limitations become important. In all engineering designs, it is often necessary to sacrifice one aspect of system performance to improve another. The following discussion of trade-offs in single-supply op amp specifications also illustrates how these low-voltage amplifiers differ from traditional high-voltage products. Input Stage Considerations
Input common-mode voltage range is the first consideration that designers should consider when specifying a single-supply op amp. It is important to emphasize that rail-to-rail input capability can solve this problem. However, true rail-to-rail operation comes at a cost.
Most of Maxim's low-voltage op amps allow for common-mode input voltages that include the negative supply voltage (Table 1), but only a few extend to the positive supply voltage. Typically, the input voltage is allowed to within 1V or 2V of the positive supply voltage. An operational amplifier that allows the signal to reach the negative power supply voltage is called a ground-sensing amplifier, and an operational amplifier that allows the signal to reach both the positive and negative power supply voltages is called a full-rail input amplifier.
Table 1. Maxim's low-voltage op amps
VOS and IB Considerations
In many applications, an amplifier can provide a gain of +2V/V or more to a ground-referenced signal. In these cases, a ground-sensing amplifier can adequately handle the common-mode range of the signal and, for this application, can achieve better performance than a rail-to-rail input op amp. A typical rail-to-rail input stage uses two differential input pairs instead of one (Figure 1).
As the input signal moves from one supply rail to the other, the amplifier moves from one input differential pair to the other. At the crossover point, this movement causes changes in the input bias current and offset voltage, affecting the magnitude and polarity of these parameters. The change in offset voltage generally degrades the distortion performance and accuracy specifications of a rail-to-rail amplifier (compared to a ground-sensing amplifier). To minimize the change in offset voltage and achieve a smooth transition from one input differential pair to the other, Maxim trims the offset at both the high and low ends of the common-mode input range of its rail-to-rail amplifiers.
To reduce the offset voltage caused by input bias current, designers should keep the impedance of the op amp's non-inverting and inverting terminals matched. Because input bias current is usually larger than input offset current, impedance matching is a good solution not only for full-rail input amplifiers, but also for all other amplifiers. To reduce the offset voltage caused by input bias current, designers should keep the impedance of the op amp's non-inverting and inverting terminals matched. Because input bias current is usually larger than input offset current, impedance matching is a good solution not only for full-rail input amplifiers, but also for all other amplifiers.
To illustrate this point, Figure 2 shows the input bias current vs. common-mode voltage curve for the MAX4122-MAX4129 series of op amps (with full rail-to-rail inputs and outputs). As the common-mode input voltage slowly rises from 0V to 5V, the absolute change in input bias current is 85nA (from -45nA to +40nA). The input offset current in the technical specification is only ±1nA. Therefore, despite the large changes in the magnitude and polarity of the bias current, the curves for the inverting and non-inverting inputs are very close to each other (input offset current). By keeping the impedance of the non-inverting and inverting terminals matched, the offset voltage caused by the change in input bias current can be minimized.
Figure 3 shows how to keep the impedances of the inverting and noninverting structures matched in a typical op amp. The inverting structure (Figure 4) cancels out input bias current changes by keeping the common-mode input voltage of the amplifier at the reference voltage (VREF). The output is VOUT = (-VIN x R2/R1) + VREF (1 + R2/R1). If R2 = R1, the equation becomes VOUT = -VIN + 2VREF. If VREF = 2V and VIN is between 0V and 3V, VOUT ranges from 4V to 1V. Since the common-mode range is fixed, CMR errors can also be canceled. Table 2 lists reference values for low voltage systems.
Slew rate
is also affected when replacing a ground-sensing amplifier with a rail-to-rail input amplifier. The simple input stage of the ground-sensing amplifier has a variety of processes that increase the slew rate, which cannot be used for a rail-to-rail input amplifier with two differential pairs. For example, the MAX4212 series of op amps (Table 1) has ground-sensing inputs and can achieve a slew rate of 600V/μs and a bandwidth of 300MHz at a maximum supply current of 7mA. If it is allowed to provide rail-to-rail inputs and all other parameters remain unchanged, the slew rate will be reduced by several times.
Output Stage Considerations
Low-voltage designs may not require rail-to-rail input characteristics, but rail-to-rail outputs are required to maximize dynamic range. Because op amps provide amplification in most applications, the output voltage is usually greater than the input voltage. Therefore, rail-to-rail inputs are not always required, but rail-to-rail output stages are often required, which are different from the output stages in dual-supply op amps.
Rail-to-rail output stages generally consist of a common-emitter amplifier, and the standard output stage is usually an emitter follower (see Figure 5). Common-emitter output amplifiers have relatively low input and output voltage drops (collector-emitter saturation voltage, or VCE(SAT)), but the typical emitter follower output is more than the sum of VCE(SAT) (generated by the current source) and VBE (generated by the output transistor).
Because the VCE(SAT) of a bipolar transistor depends on the current flowing through the transistor, the output swing of a bipolar op amp is related to the load current. It can be seen that although the amplifier is nominally full-rail output, its output stage cannot actually reach the full power supply amplitude. For example, the load resistance of the MAX4122 is 100k, and the maximum swing is 12mV different from the positive power supply voltage and 20mV different from the negative power supply voltage. However, when the load is 250, the swing can only reach within 240mV of the positive power supply voltage and within 125mV of the negative power supply voltage.
For a CMOS output stage, the collector-emitter voltage of the bipolar transistor corresponds to the drain-source voltage of the MOSFET, which is the product of the MOSFET on-resistance and the channel current. Therefore, the voltage swing of the MOSFET output stage is also a function of the load.
Gain vs. Load
In addition to the low input-output voltage difference, the common-emitter circuit of a rail-to-rail amplifier differs from the emitter-follower circuit in other parameters. The common-emitter circuit provides voltage gain with relatively high output impedance; the emitter-follower circuit provides unity gain with low output impedance. Therefore, rail-to-rail op amps usually have an output node for compensation, whereas standard op amps usually have compensation located in the previous stage. For rail-to-rail op amps, the resulting gain is affected by the load current, making them unstable when driving capacitive loads.
The performance of these rail-to-rail output amplifiers can be improved by careful op amp design; the trade-off is to increase the supply current and consume more current than the op amp with an emitter-follower output stage. The MAX4122-MAX4129 family of op amps excels in driving capacitive loads (see Table 1). These op amps are stable with rail-to-rail inputs and outputs up to 500pF, making them ideal for driving poorly terminated cables and capacitive input stages of analog-to-digital converters. Because they can drive large capacitive loads, they have good large-signal voltage gain, even under heavy loads.
Open-Loop Gain and Output Swing
As with all op amps, the open-loop gain of a rail-to-rail output amplifier is a function of the output voltage swing. Therefore, to evaluate rail-to-rail output amplifiers, it is necessary to specify the gain at a specified voltage and load. This is how Maxim specifies gain, which is not available in some manufacturers' data sheets. For example, some op amps can have an open-loop gain of 106dB and can swing to within 125mV of the supply voltage when driving a 250Ω load, but they cannot guarantee both. For example, the MAX4122-MAX4129 data sheet clearly specifies large-signal voltage gain and output voltage swing in its "Electrical Characteristics Table" (Figure 6). The large-signal voltage gain of these devices as a function of output voltage and load is shown in Figure 7.
The MAX4162 series of charge-pump
op amps solve the problem of providing rail-to-rail outputs with standard output stages in an innovative way. The op amps use a typical emitter-follower output stage, but the internal charge pump provides the bias voltage for the output stage, thereby achieving rail-to-rail outputs. The charge pump also powers the rest of the amplifier circuitry, so that the input can vary between ground and VCC when the input stage is a standard ground-sensing structure. The specifications for this series of op amps are shown in Table 1, and each device draws only 35μA (including the charge pump) to provide a 200kHz bandwidth. The amplifiers can drive relatively large loads of 20 and 500pF while maintaining low supply current.
The introduction of the charge pump allows the amplifier to use a standard input and output structure, so the performance of these amplifiers is better than that of rail-to-rail op amps. The common-mode rejection ratio of the charge-pump op amp is very high, and the single input transistor pair does not have the offset voltage change caused by switching between the dual differential pairs. In addition, the typical emitter-follower output still ensures high open-loop gain even under relatively large loads, and the amplifiers can remain stable even when driving large capacitive loads.
Common Problems
Single-supply operation also makes noise, bias, and distortion issues more severe.
Single-supply applications typically have
very low voltages, which forces designers to reduce noise to maintain the system's signal-to-noise ratio. Unfortunately, low voltages usually require low power consumption, and as the supply current decreases, the amplifier noise increases. All else being equal, low-noise amplifiers consume more power.
To estimate the noise of an op amp, all noise sources must be considered: input voltage noise, input current noise, and thermal noise due to gain-setting resistors. Figure 8 shows the noise sources for a voltage feedback op amp. C1 is the parasitic capacitance at the op amp's inverting input, C2 limits the noise gain and signal bandwidth at high frequencies, R1/R2 are standard gain-setting resistors, and R3 is used to balance the resistance at the inverting and non-inverting inputs.
At low frequencies, the noise gain is 1+R2/R1 (Figure 9). The first zero of the noise gain is at 1/2R1C1, and it increases at a slope of 6dB per decade before reaching the pole created by C2; at the pole 1/2R2C2, the noise gain becomes flat and equal to 1+C1/C2. The noise gain curve then intersects the amplifier open-loop gain curve and begins to roll off at a slope of 6dB per decade (the standard single-pole roll-off of the amplifier open-loop gain).
Because the input voltage noise, the non-inverting current noise, and the noise due to R3 are integrated over the entire closed-loop bandwidth and multiplied by the current noise gain, it can be seen (from the noise gain and open-loop gain plots) that the circuit noise is minimized by choosing an op amp with a low unity-gain crossover frequency. For the inverting input, the current noise and thermal noise due to R1 and R2 are integrated only over the signal bandwidth (1/2R2C2). Because there is no capacitor C2 in a current feedback op amp, the noise of this type of op amp is integrated only over the entire closed-loop signal bandwidth.
Distortion
Proper amplifier loop gain minimizes distortion, which would otherwise create nonlinearities in its input-output transfer function. Because the amplifier gain decreases at high frequencies, its harmonic distortion increases.
Good harmonic performance is achieved if the op amp operates in its linear region and the loop gain is maximized at a given frequency. This requires biasing the output away from the supply voltage, as shown in Figure 4 (signal inversion and a bias voltage added) or Figure 10 (bias voltage applied, but signal not inverted).
The inverting configuration shown in Figure 4 eliminates common-mode nonlinearity by keeping the common-mode input voltage constant. This is particularly useful in rail-to-rail input amplifiers where nonlinearity is caused by changes in the common-mode input as the input stage transitions from one input differential pair to the other.
Let's focus on the output stage. Since gain is a function of load current, light loads help improve the harmonic performance of rail-to-rail amplifiers. The amplifier voltage offset also affects distortion. All op amps benefit from minimal voltage drift (internal operating points do not need to be shifted too much, they remain in the linear region). The amplifier slew rate is related to the full power bandwidth and also affects harmonic distortion. When the amplifier operates outside the full power bandwidth, the associated slew rate limitation will produce severe nonlinearity.
Generating Another Power
SupplyHigh-performance, single-supply op amps are becoming more and more popular, but to maximize performance, it is sometimes necessary to select dual-supply amplifiers. Because dual-supply op amps are designed without the limitations of single-supply designs, there are more dual-supply products to choose from.
There are many ways to derive negative power from the positive supply. Switching regulators are the most flexible, while charge pump converters are the simplest, smallest, and cheapest. Because charge pumps use external capacitors (rather than inductors) to provide voltage conversion, they work best when providing integer multiples of the input voltage (-VIN, +2VIN, etc.). The output voltage is generally not regulated, but if the load current is relatively small, the output voltage can be very close to an integer multiple of the input voltage.
Because the quiescent current of the charge pump can be very small, the efficiency is very high at light loads. As shown in Figure 11, the charge pump is configured to generate a negative voltage that is equal to the input voltage but opposite in polarity. The internal oscillator frequency can be set to 13kHz, 100kHz, or 250kHz through pin configuration, allowing the designer to trade off parameters such as quiescent current, charge pump capacitor size, or output voltage ripple.
提升卡
变色卡
千斤顶
京公网安备 11010802033920号
