Auto-zero amplifiers ease the design of high …

1 19 Analog Applications JournalTexas instruments IncorporatedAmplifiers: Op Amps2Q and Mixed-Signal ProductsAuto- zero amplifiers ease the design ofhigh-precision circuitsA wide variety of electronic applications deal with the conditioning of small input signals. These systems requiresignal paths with very low offset voltage and low offsetvoltage drift over time and temperature. With standard linear components, the only way to achieve this is to usesystem-level auto -calibration. However, adding auto -calibration requires more complicated hardware and soft-ware and can slow down time to market for new alternative is to use components with low offset andlow drift. The amplifiers with by far the lowest offset anddrift available are the Auto-zero amplifiers (AZAs). Theseamplifiers achieve high dc precision through a continuouslyrunning calibration mechanism that is implemented a typical input offset of 1 V, a temperature-relateddrift of 20 nV/ C, and a long-term drift of 20 nV/month,these amplifiers satisfy even the highest requirements ofdc s AZAs differ neither in form nor in the applicationfrom standard operational amplifiers .

2 There is, however,some hesitation when it comes to using AZAs, as mostengineers associate them with the older chopper ampli-fiers and chopper-stabilized amplifier designs. This stigmahas been perpetuated either by engineers who workedwith the older chopper amplifiers and remember the diffi-culties they had with them, or younger engineers wholearned about chopper amplifiers in school but probablydid not understand them very original chopper amplifier heralded the beginning ofthe new era of self-calibrating amplifiers more than 50 yearsago. This amplifier provided extreme low values for offsetand drift, but its design was complicated and expensive. Inaddition, ac performance was limited to a few hertz of inputbandwidth accompanied by a high level of output the years, unfortunately, the term chopper amplifier became a synonym for any amplifier with internal calibra-tion capability. Therefore, AZAs, often wrongly designatedas chopper or chopper-stabilized amplifiers , are associatedwith the stigma of the older chopper article shows that the Auto-zero calibration tech-nique is very different from the chopper technique and isone that, when implemented through modern processtechnology, allows the economical manufacturing of wide-band, high-precision amplifiers with low output noise.

3 Thefollowing discussion presents the functional principles ofthe chopper amplifier , thechopper-stabilized amplifier ,and the AZA. It then com-pares the efficiencies oflow-frequency filteringwhen applied to AZAs and standard operationalamplifiers. Finally, threeapplication examplesdemonstrate the use of anAZA as a signal amplifierand as a calibrating ampli-fier in dc and widebandac chopper amplifierFigure 1 shows a simpli-fied block diagram of achopper dc input signal ischopped into an ac voltageand amplified by an ac-coupled amplifier . A phase-sensitive demodulator converts the output of A1By Thomas Kugelstadt (Email: Systems Engineer, Industrial SystemsVINVOUTS1S2R1 VIN,VOUT V1V2V3with OffsetV4V5 VOUT0000ttttV1V2V3V4V5 WidebandAmplifierLow-Pass FilterIntegratorR20tVINVOUT 00ttA1 OscillatorA2+ Figure 1. Chopping principle in the time domainNote: Labels for amplifiers such as AIN, GB, and AMare used to identify amplifiers in figures and to representamplifier gain in instruments IncorporatedAmplifiers: Op Amps20 Analog Applications JournalAnalog and Mixed-Signal 2005back to dc.)

4 The demodulator consists of a switch S2that issynchronously driven to S1. An integrator then smoothsthe switch output and presents the final dc circuit benefits from high overall dc gain and lowbaseband noise. The dc gain, being the product of the acstage and the dc gain of the integrator, easily reaches anopen-loop gain of 160 dB and reduces the gain error,to almost baseband is defined as the maximum usable inputbandwidth. Baseband noise consists of the input offsetvoltage (also known as dc noise), the 1/f noise, and low-frequency white noise. The reduction of baseband noisehappens in several steps: Offset and drift in the output integrator stage are nulledby the dc gain of the preceding ac stage. dc drifts in the ac stage are also irrelevant because they are isolated from the rest of the amplifier by thecoupling capacitors. The 1/f noise of the ac amplifier is modulated to higherfrequencies via the 2 clarifies the process of noise reduction by demon-stratingthe effects of chopping in the frequency chopping of the input signal constitutes an ampli-tude modulation (AM), with the chopping frequency, fCH,being the carrier, and the input voltage representing themodulating signal.

5 Both switches, S1and S2, are replacedby the modulators, M1and (f) in Figure 2 shows that the modulation of a squarewave causes sidebands of the input signal to appear onVAAOUT12,both sides of the odd harmonics of the chopper amplitudes of the harmonics and their sidebandsdecrease following a 1/n function, with n indicating theorder of the 1/f noise of A1present in the baseband adds to themodulated input signal after the first modulation stage,M1. The combined signal is amplified by A1and fed intothe demodulator, M2. The 1/f noise, experiencing its firstmodulation through M2, introduces sidebands on bothsides of the odd harmonics of fCH. For the modulatedinput signal, however, M2represents the second modulat-ing stage. VM1is now demodulated, causing sidebands ofthe input signal to occur around the even harmonics offCH. The input signal reappears in the baseband, and theroll-off of the subsequent low-pass filter limits the base-band to frequencies far below the chopper that the AM does not change the spectral densityof the white noise.

6 The residual baseband noise is there-fore limited, low-frequency white the small values for offset, drift and basebandnoise, this approach has some drawbacks. First, the ampli-fier has a single-ended, noninverting input and cannotaccept differential signals without additional circuitry at thefront end. Second, the carrier-based approach constitutesa sampled data system, and overall amplifier bandwidth islimited to a small fraction of the chopper frequency. Thechopper frequency, in turn, is restricted by ac amplifiergain-phase limitations and errors induced by switchresponse time. Maintaining good dc performance involveskeeping the effects of these considerations small. Chopperfrequencies are therefore in the low-kilohertz range, dictat-ing low overall (f)fCHfCHVM1VM2 VINVOUT(f)VM2(f)VIN(f)VM1(f)VOUT00001123 34556f/fCHf/fCHf/fCHf/fCHM1M2 VNoiseA1 Figure 2. Chopping principle in the frequency domainTexas instruments IncorporatedAmplifiers: Op Amps21 Analog Applications Journal2Q and Mixed-Signal ProductsThe chopper-stabilized amplifierThe classic chopper-stabilized amplifiersolves the chopper amplifier s low-bandwidthproblem.

7 It uses a parallel path approach(Figure 3) to provide wider bandwidth whilemaintaining good dc characteristics. The stabilizing amplifier , a chopper type, biasesthe fast amplifier s positive terminal to forcethe summing point to signals directly drive the ac amplifier ,while slow ones are handled by the stabilizingchopper amplifier . The low-frequency cutoffof the fast amplifier must coincide with thehigh-frequency roll-off of the stabilizingamplifier to achieve smooth overall gain-frequency characteristics. With proper design ,the chopper-stabilized approach yields band-widths of several megahertz with the low-driftcharacteristic of the chopper , because the stabilizing amplifiercontrols the fast amplifier s positive terminal,the classic chopper-stabilized approach isrestricted to inverting operation addition, the high residual output noiseof the chopper amplifier is amplified by thefast amplifier s noise gain. Keeping outputnoise small dictates additional filter effort,thus increasing complexity and cost of thechopper-stabilized Auto-zero amplifier (AZA)Similar to the chopper-stabilized approach,the AZA uses a main amplifier for widebandsignal amplification and a nulling amplifier foroffset correction.

8 Figure 4 shows a block diagram of the TLC2654, an AZA developedby texas instruments in the the calibration path lying in parallelwith the signal path, both inputs of the mainamplifier are available for differential main amplifier , AM, and the nulling amplifier , AN,each have an associated input offset voltage (VOSMandVOSN, respectively) modeled as a dc offset voltage in serieswith the noninverting input. The open-loop gain of the signal inputs is given as AMand AN. Both amplifiers alsohave additional voltage inputs with the associated open-loop gains of +BMand correction of the overall amplifier occurs withinone cycle, fAZ, of the Auto-zero clock and is split into twomodes of operation: an Auto-zero phase and an amplificationphase. The oscillator, generating fAZ, initiates the auto -zerophase by driving both switches into position 1. The inputsof the nulling amplifier are shorted together, while its out-put is connected to capacitor C1.

9 In this configuration ANmeasures its input offset voltage and stores it via we can express the voltage at C1 aswhich, by simple rearrangement, is(1)This shows that the offset voltage of the nulling amplifiertimes a gain factor appears at the output of ANand thuson the C1 the amplification phase, when both switches are inposition 2, this offset voltage remains on C1 and essentiallycorrects any error from the nulling amplifier . ANamplifiesVVABCOSNNN11=+ .VAV BVCNOSNNC11= ,VINVOUTR2R1*Similar to the chopper amplifier in Figure 1 SummingPointStabilizingAmplifier*Wideban dAmplifier+ Figure 3. Chopper-stabilized amplifierVINO scillatorS1 VOSMS2 VOUT1221C1C2 BN+BMVC1VC2 AMVOSNAN+ + Figure 4. Simplified TLC2654 block diagramNote: Labels for amplifiers such as AIN, GB, and AMare used to identifyamplifiers in figuresand to represent amplifier gain in instruments IncorporatedAmplifiers: Op Amps22 Analog Applications JournalAnalog and Mixed-Signal 2005VC1by the factor BNand subtracts it from the amplifiedinput signal,At the same time, the output of ANcharges capacitor C2 toReplacing VC1with Equation 1 results in(2)Equation 2 shows that VOSNhas been reduced by a factor1 + BN, indicating how the nulling amplifier reduces itsown offset voltage error even before correcting the mainamplifier.

10 The potential, VC2, now serves the main amplifieras an offset correcting voltage, forcing its output, and thusthe output of the complete AZA, toReplacing VC2with Equation 2 and combining terms gives us(3)The Auto-zero architecture is optimized in such a waythat AM= AN, BM= BN, and BN >> 1. This allows Equation 3to be simplified to(4)Most obvious is the gain product of both the main andnulling amplifiers . The ANBN term in Equation 3 explainswhy AZAs have extremely high open-loop gain. To under-stand how VOSMand VOSN relate to the overall effectiveVVABAVVOUTIN N NNOSMOSN=++().V VAAB VAVABBOUTINMN NOSM MOSNNMN=++ ++ ().1 VAVV BVOUTMINOSMM C=++().2 VAVVBCNINOSNN21=++ .VVAVV BVONCNINOSNN C==+ 21().AV VNIN OSN().+input offset voltage of the complete amplifier , we shouldset up the equation for the generic amplifier in Figure 5:(5)where k is the open-loop gain of the amplifier and VOS_Effis its effective offset Equation 4 into the form of Equation 5 gives usFrom here it is easy to see that k = ANBNandThus, the offset voltages of both the main and thenulling amplifiers are reduced by the gain factor BN.