How system operating conditions affect CMOS op amp open-loop gain and output impedance

 

Operational amplifiers (op amps) are used in many systems across a broad range of applications. These systems operate over a wide range of conditions, including , supply voltages, common-mode voltages, output loading, and product lifetimes. Op amp (AOL) and (ZO) are the two main small-signal AC parameters that affect the stability and frequency response of an op amp circuit.

In this article we will explore the effects of temperature, power supply voltage, common-mode voltage, output loading, and semiconductor process variation on these two critical parameters. We will provide additional insight beyond what is typically available in manufacturer data sheets regarding how these specifications vary over all of these operating conditions. The variations of these two parameters over the operating conditions can be used to develop a parameter variation circuit analysis model that can be used to create robust designs, using methods as extreme value analysis (EVA), root sum squared (RSS), or Monte Carlo analysis (MCA). The final results are listed in summary tables, which can be used to help identify possible design problems and help to create feasible alternatives.

Simplified small-signal AC model for op amps

A simplified small-signal AC model for an op amp configured for a non-inverting gain is shown in Figure 1. In this simplified model, the differential input voltage (VDIFF) is amplified by the AOL gain block and then passes through ZO to the amplifier’s output pin. Although shown in Figure 1, input impedance (RIN) is not discussed in this article.

While an ideal op amp has infinite open-loop gain and infinite bandwidth capabilities, real op amps have design and manufacturing limitations on both the maximum gain and bandwidth. These limitations are expressed in the magnitude and phase curves of the op amp AOL. Ideal op amps feature zero output impedance, which is not possible in real op amp designs. The actual non-zero output impedance is described in the op amp’s ZO curve.

The magnitude and phase of the op amp AOL and feedback factor (β) will determine the final closed loop gain (ACL) of the circuit as described in the familiar Equation 1.

The product of AOL and β (AOL* β) typically is called loop gain. This critical circuit parameter defines the accuracy and linearity of the closed-loop gain. The loop-gain phase at the intersection of the AOL and β curves is referred to as the circuit phase margin, which is critical in the stability analysis of the circuit. The phase margin determines the circuit’s percent overshoot and settling time in response to an input voltage or output current step.

Interactions between the ZO and the output load affect the final open-loop gain presented at the amplifier’s output pin. These interactions can change the system phase margin, possibly resulting in circuit stability issues.

The modern single-supply, three-stage rail-to-rail output CMOS op amp is significantly evolved from its older high-voltage BJT predecessors, and features complex AOL and ZO behavior, which is detailed in the following sections (when we discuss op amps we’re referring to modern three-stage CMOS amplifiers with complex AOL and ZO behavior).

Figure 1 | Simplified small-signal op amp AC model

CMOS op amp AOL

The frequency behavior of an op amp AOL magnitude and phase is displayed in Figure 2.

Figure 2: Op amp gain (AOL) and phase response (φ) over frequency

AOL_DC is the DC change in output voltage (VOUT) versus the change in the differential input voltage (VDIFF). In Figure 2 the value of AOL_DC is 195 dB, as expressed in the following equation:

AOL_DC=195 dB

The frequency where AOL equals 1 V/V, or 0 dB, is defined as the unity-gain crossover frequency and is marked as fu in Figure 2:

fU=1.8 MHz

The frequency behavior of AOL is largely defined by the low-frequency dominant pole located at frequency ω1 or f1. At the dominant pole frequency, AOL has decreased 3 dB from AOL_DC and the phase has shifted by -45 degrees:

f1=0.37 mHz

The complete frequency behavior of op amp AOL curves also can be shaped by mid-frequency pole-zero pairs, and higher frequency zeros and poles. In Figure 2, fXP1 and fXZ1 describes a mid-frequency pole-zero pair. Additionally, there is a zero at fXZ2 and a high-frequency triple-pole, fXP2. These poles and zeros in the AOL transfer function determine the fU frequency of 1.8 MHz. The following equations list the frequencies of these poles and zeros:

fXP1=142 kHz

fXZ1=274 kHz

fXZ2=1.24 MHz

fXP2=4.88 MHz

To create a robust design, you need to understand how AOL changes as the system operating conditions change. System operating conditions that affect the performance of the AOL curve include temperature, output load, power supply voltage, and process variation.

Temperature effects on AOL

As an example, we’ll focus on how the operating temperature affects the frequency behavior of the op amp AOL curve. Our reasoning is that, out of all system operating conditions, temperature commonly has the largest impact and varies in many applications (the effects of the other operating conditions are summarized in Table 1).

Many op amps are specified over an extended operating temperature range of -40 ºC to 150 ºC. The operating temperature affects both the DC and frequency behavior of the AOL curve (Figure 3).

Figure 4 zooms in on the temperature effects on AOL_DC. Over the operating temperature range, AOL_DC can vary from 214 dB to 149 dB. The 65 dB change in AOL_DC changes the op amp loop gain at low frequencies, impacting the accuracy of the circuit’s closed-loop gain.

The variation of the unity-gain frequency, fu, over the operating temperature range is shown in Figure 5. Over the operating temperature, fu, can vary from 1.26 MHz to 2.75 MHz. This variation affects circuit closed-loop bandwidth and loop gain phase margin, which impacts circuit response and settling times.

Figure 5 | Variations in the unity-gain frequency versus temperature.

CMOS op amp ZO

Figure 6 shows the typical frequency behavior of the op amp ZO magnitude.

As mentioned, this op amp features a three-stage architecture, which results in three distinct ZO regions as seen in the ZO magnitude. At low frequencies the ZO curve is defined by a low-frequency resistance value, RLOW_F. As frequency increases, ZO becomes capacitive. In that region, ZO is defined by a low-frequency capacitance value, CLOW_F. At mid-frequencies, ZO becomes resistive again and is defined by a mid-frequency resistance value, RMID_F. ZO then becomes inductive and is defined by an open-loop inductance value, LO. This inductive region is the most important for stability analysis because capacitive loading on the output can interact with the inductance. This results in resonance and stability issues that are difficult to compensate. The inductive region turns resistive again at higher frequencies, and can be defined by a high-frequency resistance value, RHIGH_F. Finally, at high frequencies near the end of the region of interest, ZO turns capacitive again and can be defined by a capacitance, CHIGH_F.

Component values for the op amp featured in Figure 6 are listed in the equations below:

RLOW_F=4.87 MΩ

CLOW_F=1.57 mF

RMID_F=4.09 MΩ

LO=1.23 mH

RHIGH_F=1.03 kΩ

CHIGH_F=20.14 pF

To create a robust design, you need to understand how ZO changes as the system operating conditions change. System operating conditions that affect the performance of the ZO curve include temperature, power supply voltage, common-mode voltage, output loading, and process variation.

Temperature effects Zo

Similar to the section on AOL, here we examine how the operating temperature affects the frequency behavior of the op amp ZO curve because commonly it also has the largest impact of all other operating conditions (effects of the other operating conditions are summarized in Table 2).

The effects of the operating temperature on frequency behavior of the op amp ZO curve over the full -40 ºC to 150 ºC temperature range are shown in Figure 7.

Worst-case analysis

Table 1 lists the variations of AOL over the system operating conditions that have the greatest effect on the curve. The worst-case results are generated by operating the op amp in the worst conditions simultaneously, which is why the shifts are beyond the levels for any of the individual conditions. Figure 8 shows the results generated from the worst-case analysis.

 

Table 2 lists the variations of ZO over the system operating conditions that have the greatest impact on the curve. Similar to the worst-case AOL results, these results are generated by operating the op amp in the worst scenarios simultaneously. Figure 9 shows the envelope of the results generated from the worst-case analysis.

 

Conclusion

The AOL and ZO of op amps are two key specifications in the understanding of small-signal behavior of op amps, including circuit closed-loop gain, bandwidth, settling time, and stability. The typical magnitude and phase response of the AOL and ZO curves change with variations in the system operating conditions. Some system operating conditions that affect these parameters are temperature, output load, power supply voltage, common-mode voltage, and semiconductor processing variations.

The changes in AOL and ZO over these system operating conditions were presented in this article over the full operating range of an example op amp. The worst-case changes over the same operating range that may occur were also shown. This provides additional insight beyond what is typically available in manufacturer data sheets.

Variations of these two parameters over the operating conditions can be used to develop a parameter variation circuit analysis model that can be used to create robust designs, using methods as EVA, RSS, or MCA. Engineers can use information in this article to create a robust design over the expected application operating conditions.

Miroslav Oljaca is the end equipment lead for building automation applications and system solutions. Miro has nearly 30 years of engineering experience and has been granted at least a dozen patents, several related to high-performance . He has written many articles about the same. Miro received his BSEE and MSEE from the University of Belgrade, Serbia.

Collin Wells is an Applications Engineer in the Precision Linear group at Texas Instruments, where he supports industrial products and applications. Collin received his BSEE from the University of Texas at Dallas. For questions about this article, Collin can be reached at ti_collinwells@list.ti.com.

Texas Instruments

www.ti.com

@TXInstruments

LinkedIn

Facebook

Google+

References:

1. Jerald G. Graeme, Optimizing op amp performance, ISBN: 978-0071590280.

2. Sergio Franco, Design with operational amplifiers and integrated circuits, ISBN: 978-0078028168.

3. Miro Oljaca, Collin Wells, Tim Green, Understanding open loop gain of the PGA900 DAC gain amplifier, TI Application Report (SLDA031), April 2015.

4. Miro Oljaca, Collin Wells, Tim Green, Understanding open loop output impedance of the PGA900 DAC gain amplifier, TI Application Note (SLDA033A), May 2015.

Solving op amp stability issues, TI E2E™ Community Forum, October 14, 2015.