Precision (Accuracy) related to numerical value: Systematic character and sheer true real number are worth the difference between. Nicety (Precision) be with digital form denotive numeric deepness. In the article, we will use precision one word, it includes sum of errors of noise, deflection, gain nonlinear all degree of limitation that wait for a system to measure. The certain error of a lot of operation amplifier measures level in Ppm, but all errors that do not have an operation amplifier achieved Ppm to measure level. For example, behead wave amplifier can provide degree of linearity of the maladjusted voltage of Ppm class, dc and low frequency noise, but their input slants buy electric current and problem of existence of frequency degree of linearity. Bipolar amplifier has low broadband noise and good degree of linearity, but its input electric current to still may cause in-house circuit error (to in-house circuit, we use use “ one word of ” ) . MOS amplifier is had slant outstandingly buy electric current, but exist in low frequency noise and domain of degree of linearity normally blemish.

 

In the article, we will be used in changing function what be equivalent to 1 Ppm roughly is nonlinear the –120 DBc with degree of expression lack fidelity harmonic is lack fidelity.

 

Be not PPm amplifier type

 

Let us see the amplifier kind that is not tall degree of linearity. The type of lowest of degree of linearity namely so called video or line drive amplifier. These are the broadband amplifier with dc not quite good precision: Deflection amounts to a few millivolt, slant buy electric current comes in 1µA inside 50µA limits, and 1/f noise function normally poorer. Ideal dc precision is in 0.3% to 0.1% between, but hand in prediction of a person’s luck in a given year to be able to upright to come at –55dBc really –90dBc (degree of linearity: 2000ppm comes 30ppm) between.

 

A classification is traditional and classical carry puts a design, for example OP-07, the likelihood has tall gain, CMRR, PSRR and good maladjusted voltage and noise property, but its are lack fidelity however cannot excel –100dBc, it is especially under the condition that obtains 1kΩ or higher load.

 

Next, still have a few or new or old cheap amplifier, its are lack fidelity below the case that surpasses 10kΩ in load cannot excel –100dBc.

 

In addition, still have frequency amplifier kind operation amplifier. They are quite substantial, and lack fidelity expression may be first-rate. But, their design is improper and cannot provide good maladjusted voltage and 1/f noise performance. In addition, their lack fidelity be in probably after be more than 10kHz also immutable is better.

 

Some operation amplifier aims to support the degree of linearity of MHz signal. They are normally bipolar, have bigger input to slant buy electric current and 1/f noise. The field uses in this, what operation amplifier more goes after is –80dBc the function to –100dBc degree, implementation Ppm property is not quite real.

 

No matter the broadband is reached,press place rate is how old, voltaic feedback amplifier also cannot support deep degree of linearity, it is measurable precision even. Their input class has cause of a lot of errors, and gain, input is mixed power source controls function not tall. Voltaic feedback amplifier still has hot drift effect, can extend normal set-up time considerably.

 

Next, we are had contemporary general-purpose amplifier. They have the deflection of 1mV and noise of 1/f of class of small bend over commonly. Supportive –100dBc is lack fidelity, but cannot come true normally when high load.

 

The error source of operation amplifier

 

Graph 1 those who show is simplifying operation amplifier block diagram, added communication and dc error source. Develop attacks to guide to contain an input to cross the amplifier of only the limit of Gm, drive outputs the gain node of amortize unit. Although have a lot of operation amplifier,develop attacks, but shows error cause is all to them applicable.

 

Graph 1. Simplifying operation amplifier and error source.

 

Input noise

 

VNOISE E of some input noise voltage include broadband and 1/f spectrum part. If the range of noise is similar or exceed systematic LSB, cannot measure signal well and truly. For example, if broadband noise is 6nV/√Hz, systematic bandwidth is 100kHz, so the virtual value noise that inputs end can achieve 1.9µV. We can use filter to reduce noise: For example, fall bandwidth to be able to make to 1kHz noise falls to 0.19µV Rms or 1µVp-p (summit summit) the left and right sides. The low wave connecting filter of frequency domain can reduce noise scope, like follow time elapse like ADC output and be being changed on average.

 

Nevertheless, because rate is too slow, 1/f noise cannot filter actually or leveling. 1/f noise uses 0.1Hz to come normally the noise of voltage of peak peak value that makes inside 10Hz spectrum limits is reflected. The low frequency noise of most operation amplifier uprights to come at 1µVp-p between 6µVp-p, suit high to dc precision requirement Ppm level not quite consequently, below the case that providing gain especially.

 

Graph the 2 electric current that those who show is LT1468 of admirable high accuracy amplifier and voltage noise.

 

Graph 2. LT1468 inputs voltage and voltaic noise.

 

In the graph the input of 1 is carried, still have slant INOISE+ of source of noise of buy electric current and INOISE-– . They include broadband and 1/f spectrum part. INOISE is multiplied can produce more to input voltage noise with equivalent resistor. Generally speaking, each other is not relevant between inphase end and the noise of two electric current that carry instead, won’t input resistance as two end equal and quits, increase with Rms means however. INOISE multiplies the noise tension that produces with inputting equivalent resistor to often can exceed the VNOISE of 1/f area.

 

Input the model is restrained and slant in all buy error

 

Source of below one kind of error is VCMRR. This body now in all the model is restrained go up than index parameter, among them maladjusted voltage is met as relative to the input n at course of two power supply change (so called in all modular voltage, VCM) . Use symbol indicates the power source of arrowhead direction to be affected each other, through its cent secant states its are alterable, but the likelihood is nonlinear change. CMRR depends on making linear part and gain error cannot distinguish to the main effect of signal. Nonlinear part will lack fidelity. Graph the 3 CMRR that showed LT6018. Addition line and CMRR curve split up in this curve the utmost before overload intersects. The CMRR = 133 DB that the slope of this line offers. Limits every differ 30 V, the deviation between CMRR curve and good line is 0.5 µV about only, the input that shows Ppm the following level is very successful. The curvature of other amplifier may be bigger.

 

Graph 3. LT6018 inputs maladjusted voltage and VCM.

 

Maladjusted voltage (VOS) ) the CMRR classify here. The input of behead wave amplifier is maladjusted voltage under 10µV, input signal to the model of 10Vp-p at 2Vp-p relatively, adjacent at odd Ppm error. Even, the maladjusted voltage of optimal ADC can amount to 100µV normally. So, the maladjusted voltage of 10uV class won’t create too great burden to operation amplifier oneself; Anyhow, systematic itself will be automatic zero. With input signal in all related modular n is ICMRR, input namely slant buy electric current and the metabolic case that its follow power source. Break a line to make clear slant buy electric current can change along with voltage, and also may not be linear change. Share 4 ICMRR, because two inputs carry those who have independence to slant,buy electric current is mixed n dependency, and every input carries the change along with two kinds of power source differ. The whole that ICMRR multiplies the block value with applying resistor to be able to add circuitry is maladjusted voltage. Graph 4 showed LT1468 slant buy electric current and) of VCM(ICMR norms. The slope that additive line place shows is ~8nA/V, using 1kμΩApply resistor or 8µV/V will is below the circumstance of low Ppm error. It and linear deviation are 15nA about, leave the error that can produce 15µV inside 26V limits in 1kμΩ application environment from this, or nonlinear degree amount to 0.6ppm.

 

Graph 4. LT1468 input slants buy electric current and VCM.

 

Input class is lack fidelity

 

Graph 1 showed input level, they are normally by a pair the transistor that need branch is designed cross electric path. Graph the collector that 5 coping showed to all sorts of difference distribute amplifier kind or leakage current and voltage of the input that need branch. Our imitate simple bipolar right, cross linear circuit (we say for “ intelligence ambipolar ” ) , a low threshold value (namely very big) MOS difference divides a right, belt to blast off extremely of resistor bipolar right (pursue 5. The imitate of end electric current of use 100 μA is all the amplifier that need branch.

 

Graph 5. The output electric current of all sorts of amplifier that need branch and cross guide and input voltage.

 

What in indication graph 5 bottom show cross guide with VIN before, clear message is not much. Cross guide (Gm) is photograph of output electric current the derivative to inputting voltage, use LTspice® simulator is generated. D() is included among grammar, its are equal on maths at D()/d(VINP) . The blame flatness of Gm namely the basic and lack fidelity mechanism that operation amplifier issues in frequency.

 

To dc, the gain of open loop voltage of operation amplifier is ~gm(R1 about | | R2) , but premise is output buffer gain about 1. R1 and R2 express the output impedance of all sorts of transistor in signal method, every resistor all receives course of a power supply or other unit repeatedly. This is the gain in operation amplifier accepts the base that be restricted. R1 and R2 are unwarrantable for linear; They may be brought about carry for nothing lack fidelity or nonlinear degree. Except degree of linearity, we need to gain is achieved or exceed million, ability realizes the gain precision of Ppm class.

 

Observation standard is bipolar transistor curve, we can see it be crossed mediumly in this groups guide highest, but should cross guide can begin change from 0 bend over as the input and fast subsidise. This make a person anxious, because the main demand of degree of linearity is gain or Gm constant. On the other hand, the voltage gain that who can care about amplifier is such tall, does as a result follow the volt class that outputs voltage to increase to you can realize class of small bend over to increase only at differring minute of input? CCOMP is below.

 

The parallel of CCOMP(CCOMPP and CCOMPM)M) can absorb Gm to export electricity in the great majority inside frequency limits. It but the gain bandwidth product of set amplifier (GBW) . GBW but set: Below frequency F, the open loop gain of amplifier is GBW/f. If this amplifier is 1Vp-p in the output when F=GBW/10, closed circuit gain is 10, there will be 100mVp-p between the input so. Namely, balance ±50mV. Ask an attention, graph 5 in indication level is bipolar the curve is when ±50mV loss the gain that arranges an in part, assured thereby large-scale lack fidelity. Nevertheless, intelligence ambipolar only loss the gain of 13% , loss of the MOS below threshold 26% , degrade ambipolar loss 12% , loss of square law MOS 15% .

 

Graph 6 those who showed input level is lack fidelity with amplitude. These information will show when applied circuit is outputted (by with noise gain) . Output is lack fidelity and OK continue to increase, but cannot decrease.

 

Graph 6 those who showed input level is lack fidelity with amplitude. These information will show when applied circuit is outputted (by with noise gain) . Output is lack fidelity and OK continue to increase, but cannot decrease.

 

Except class of intelligent bipolar input, the amplifier needing branch that inputs class shows lack fidelity become direct ratio with the square of the input. In the application that agrees in gain, output is lack fidelity as lack fidelity as the input effect is same. This is the main and lack fidelity origin of most operation amplifier.

 

Consider the buffer with a consistent gain that uses ambipolar input please. If output voltage of VOUT peak peak value, the input differs minute of signal to will be

 

 

Our estimation

 

 

And

 

 

Among them, GNOISE is noise gain of application.

 

1ppm is nonlinear degree be equivalent to –120dBc harmonic lack fidelity, scale is 0.0001% . Assume an amplifier is used bipolar input class, GBW is 15MHz, the output that serves as buffer is 5Vp-p, through equation the 2 the biggest frequency that can be informed this degree of linearity are 548Hz only. Afore-mentioned hypothesis premise is the minimum of degree of linearity that amplifier leaves in inferior frequency. Of course, when amplifier provides gain, noise gain increases, and the frequency of –120dBc can drop.

 

The –120dBc frequency that thresholds next MOS input class to support is highest for 866Hz, 1342Hz of highest support of square law MOS, degrade 1500Hz of ambipolar and highest support. Intelligent bipolar is lack fidelity do not accord with forecast mode, people must undertake estimation according to data manual.

 

We can use simpler formula

 

 

Among them, k can be found from inside the lack fidelity curve of manual of operation amplifier data.

 

A bit adder, a lot of operation amplifier are to use course to input class to course. The input class that most amplifier becomes independent through two can realize this function, inputting namely in all inside modular limits, can change between different input class. This kind of changeover can cause maladjusted voltage change, bring about possibly still slant buy electric current, noise and even bandwidth change. In addition, switch transient phenomena appears when still basically can bringing about output. If signal always is crossed,hand in jump over an area, cannot use these amplifier to low lack fidelity applying so. Nevertheless, to contrary application the circumstance can use them.

 

We had not discussed pressure place increase model amplifier. These designs input bigger case to issue won’t extinct electric current in poor cent. Regretful is, the input extent that poor cent inputs lesser circumstance to still can bring about Gm to appear to discuss with place changes similarly, and low lack fidelity still need to have gain of bigger frequency annulus road.

 

Because we want the distortion factor that those who search is Ppm class, so we won’t be pressed with be close to place rate any kind that are restricted to be worth move amplifier, so very unusual pressure place the important parameter that leading is not degree of linearity of Ppm class frequency, consider GBW only can.

 

In front, we discussed sheet pole to compensate the open loop gain that designs pattern. Not be all operation amplifier provides compensation with this means. Normally, open loop gain can be found from inside the curve of data manual, and the GBW/(GNOISE in equation×FSIGNAL)L) is the open loop gain of frequency.

 

Gain node error

 

Next, pursue in light of us 1 medium R1 and R2. These resistor along with input Gm to provide the gain of open loop dc of amplifier: Gm×(R1 | | R2) . These resistor of the scale in principle graph are contained alterable nonlinear delete a line. Of these resistor nonlinear spend the sky that reflected amplifier to carry distortion factor. And, r1 can bring to bear on from positive electricity source influence, as a result is restrained at voltage of source of dc positive electricity than (PSRR+) approximates Gm×R1. Manage together, r2 is in charge of PSRR– . Ask an attention, why is the extent of PSRR equal to open loop gain almost? CCOMPP and CCOMPMM are mixed to R1 the power source signal with R2 similar infuse; They install PSRR+ and PSRR– inside frequency limits.

 

Gain is modest (the degree of linearity of the amplifier of <<106)) may be very good, but modest gain can restrict gain precision.

 

Power source port may be brought about lack fidelity. If the load of output level drive is bigger, among them a certain power source can offer load current. Below certain frequency, far end the long-range modulation capacity of power source may be very small, as a result becomes real power source at the bypass capacitance of operation amplifier. After passing bypass capacitance, source current drops. Drop extent depends on ESR, ESL and reactance, and they can cause electrical interference. Because output is AB kind, the weaveform of output electric current that has an in part only so is met modulation power source, form smooth harmonic lack fidelity. The PSRR inside frequency limits can reduce electrical interference. For example, if we observe electrical interference is 50mVp-p, hope PSRR restrains power source to input interference to make its fall in output end to under 5µVp-p, criterion PSRR needs to achieve 80dB below signal frequency. Estimation PSRR(f)~Avol(f) , the amplifier that GBW is 15 MHz is falling to be able to have enough PSRR under the frequency of 1500Hz.

 

Output level is lack fidelity

 

Graph 1 it is output level last medium, output level is regarded as buffer in the article. Graph the 7 output level that showed a pattern change function.

 

Graph 7. The outputs buffer changeover function of different load.

 

To different load, we can see 4 kinds of errors. It is to cut above all wave: What although assume,should output class is nominal gain is 1, but it is not course outputs class to course completely. Below this kind of circumstance, when holding output for nothing even, course of every power source also can cut wave 100mV. Increase as load (reduce laden resistance) , output voltage can be cut down stage by stage. Apparent, cut a can serious effect lack fidelity, and must reduce output amplitude ability to avoid to cut wave.

 

Below one kind of error is gain is compressed, when the curvature that changes function obtains signal limit condition, we can see this kind of phenomenon. Increase as load, in voltage inchoate phase can appear compress. Cut together wave is same, below this kind of mechanism, cannot achieve Ppm level normally lack fidelity. This kind is compressed is normally by output class is lesser and the voltaic be caused by that satisfies output need hard. Best solution is, the linear that makes amplifier is offerred, without 35% what compress electric current of the biggest output to be output short-circuit current about only.

 

Another kind of remarkable lack fidelity origin depends on handing in jump over an area to be VIN=0 about. When carrying for nothing, hand in more twist together may be so not apparent. But increase as load, we can see the twist together of green curve increases. Estimation is handed in more lack fidelity the source current with normally powerful need.

 

Last kinds lack fidelity understand hard quite. Because some of amplifier circuit exports positive electricity pressing and electricity, still a few output lose signal, cannot make sure they have same gain so, when taking load especially. Graph the 7 gain that signal loses when showing load reduce a circumstance.

 

Can reduce through annulus road gain all these are lack fidelity. If output class lack fidelity for 3% , so annulus road gain needs to be 30, 000 ability achieve –120dBc n. Of course, this kind of circumstance happens in GBW/(30, 000×GNOISE) frequency is the following, it is 1kHz mechanism normally to the amplifier of 15MHz.

 

What some output class is lack fidelity concern with frequency, but also a lot of output level and frequency have nothing to do. Open loop gain can restrain output class is lack fidelity, but this gain can follow frequency and drop. If output is lack fidelity,change, criterion gain loss can generate output lack fidelity, follow frequency and linear increases. In the meantime, input lack fidelity meeting to bring about total output lack fidelity follow frequency and increase. Below this kind of circumstance, overall closed circuit outputs lack fidelity likelihood to basically be an input lack fidelity, mask the effect with output lack fidelity level thereby.

 

On the other hand, if output class to break,linear changes, so annulus road gain drops except bring about an input lack fidelity besides, still can bring about another kind of output lack fidelity, this are lack fidelity follow the square of frequency and change, and cannot as lack fidelity as the input area comes apart.

 

The output form that amplifier of operation of low power comsumption includes normally less, static electric current is small. Outputting lack fidelity likelihood basically is by the output of these amplifier class is brought about, is not input class. So, need ability of 2mA source current to obtain amplifier of small lack fidelity operation at least, this kind of view is correct on certain level.

 

The norms of Ppm class precision asks

 

In actual n changeover, attenuation / in gain and active filter circuit, operation amplifier needs to satisfy ability of a few main demands to support ±5V signal, apply to 1kΩ environment and the 1ppm degree of linearity that implementation watch shows 1 times.

 

Express 1. The list of extent of operation amplifier sum of errors that Ppm precision place wants

 

Now, we understood operation amplifier to be in the limitation of Ppm precision domain, so how should we improve them?

 

Noise: Apparent, should choose voltage of an input noise above all not the operation amplifier of noise of combination of resistor of prep above application. Can reduce the total impedance that uses circuit so, reduce noise thereby. Of course, drop as applied impedance, the signal electricity that carries them can increase, what make load is caused possibly is lack fidelity increase. For love or money, need not make the output noise of operation amplifier level far the input noise under its drive level.

 

Voltaic noise can be multiplied in order to apply impedance, form more voltage noise then. It is in the application with voltaic very low noise, MOS input is very absorbing, but their 1/f voltage noise is compared normally bipolar the input is big. Bipolar the voltaic noise of the input is PA/√Hz level, may produce bigger applied noise, but the 1/f voltage noise that the applied voltage noise that content of 1/f electric current makes may be more than amplifier. Generally speaking, applied impedance should be less than the applied noise that the VNOISE/INOISE of amplifier gives priority to with avoiding IBIAS. Bipolar the VNOISE of amplifier is lower, INOISE is taller.

 

Help operation amplifier realizes optimum behavior

 

◆ reduces input error

 

Except the operation amplifier with choice admirable CMRR, design personnel still can choose to be put with carry turn over photograph enlarge circuit and building is not inphase enlarge circuit. In turning over photograph circuit, the input can be linked together with the ground or source of a few fiducial voltage, won’t cause CMRR error completely. Nevertheless, not be all application circuit can be turned over, and normally negaive electricity source cannot be used at losing signal deflection. Graph 8 showed blame oppose photograph circuit and the double the limit that in turning over photograph circuit, apply Sallen-Key filter.

 

Graph 8. Blame turn over (left) and instead (right) Sallen-Key active filter.

 

If two input end to all include applied resistance, criterion every input carries slant buy electric current multiplies the voltage error that arises with corresponding resistor to be able to be in output end is quits, because of error of this OK also and quits ICMR. For example, if the amplifier gain of the setting is 10, accessary 900Ω feedback and 100Ω receive terrestrial electricity block, inputting end to find a place for the 90Ω of series connection (900Ω | | 100Ω ) resistor can offset completely equal output slants the voltage error that buy electric current produces. Great majority is bipolar of operation amplifier slant collocation of buy electric current is very appropriate, make choice 0.1% (is not common 1% ) resistor can realize optimal ICMR to restrain. In the graph 4 in, compensation resistor and turn over photograph input to carry –input series connection to place. They should be passed by bypass quite. Because extra input resistance can cause noise,increase (voltaic noise is multiplied with connective equivalent resistor) .

 

Turn over photograph gain to let us can use the operation amplifier that includes course to be inputted to course, and need not let signal cross switch to nod (assume we already slanted buy power source and in all the model inputs n, in order to avoid switch voltage) .

 

Note of ◆ power source

 

Output electric current will move the power supply source of abridged edition ground. Power source signal will transmit input end through PSRR. The input that is affected can generate output signal, around its annulus road moves. Below 1kHz frequency, the impedance of capacitance of bypass of 1μF this locality is 159Ω , far under the impedance that circuit adds electrify source itself between power source. Accordingly, capacitance of this locality bypass is falling under the frequency of 100kHz actually without the effect. Below 1kHz frequency, adjusting control circumstance is controlled by long-range power source. Below 1kHz frequency, amplifier may achieve 90dB power source to restrain than. Ask an attention, the major electric current of port of operation amplifier power source included many signal harmonic, so we hope to arrive from output the gain of supply power supply under 30dB, in order to achieve the goal of 120dBc. Want to realize the gain of 30dB, need laden impedance of <30× of power source impedance. Accordingly, 500Ω load needs the impedance of power source to be less than 17Ω . This kind of circumstance is feasible, but cannot establish ties between power source and operation amplifier so resistor and inductance. Below 10kHz frequency, the requirement is more strict; PSRR will fall from 90dB to 70dB, and power source impedance must fall to 1.7Ω . Feasible, but requirement severe exacting. Use bypass of large this locality to be able to provide a help.

 

From the point of layout angle, the way that knows way of annulus of output electric current is very important, if pursue,9 are shown.

 

Graph 9. Load and road of source current annulus.

 

Graph 9 left chart showed drive comes electric current of laden positive electricity source, return to load through the ground again next. In the likelihood in method of whole ground connection existence pressing falls, as a result at Ou Xie wave the voltage of source current falls from signal source to output, from feedback scale down implement fall to output or input ground. Nevertheless, here is not those ground. Graph 9 on the right side of the better pattern that showed a kind of transmission source current. Source current comes out from input and feedback node.

 

Below the taller frequency of prep above 100kHz, the magnetism radiate of road of power supply cord may become lack fidelity source. The electric current of Ou Xie wave of power source can pass the input of magnetic means coupling to feedback network, make thereby lack fidelity increase considerably along with frequency. It is under these frequency, careful position is crucial. The dispute level that some amplifier use cites a base; Their power source brings a foot to be far from an input, some can offer additional output port in input side even, in order to avert magnetism interference.

 

What ◆ reduces load to give priority to is lack fidelity

 

Below tall laden environment, the output level of a lot of operation amplifier can become main lack fidelity source. You can improve load through a few skill lack fidelity. One of, use compound amplifier, namely output of an amplifier drive, another amplifier undertakes controlling, if pursue,10 are shown.

 

Graph 10. Compound amplifier and single amplifier are lack fidelity test.

 

This circuit emulates design implementation through LTspice. Include in the Spice model file of LTC6240 and LT1395 lack fidelity a grand model that puts a function. Grand model won’t try great majority show lack fidelity case, although show, emulation outcome forbids possibly also truly. This tool (LTspice) the text document that can examine grand model, really such, the result of lack fidelity imitate of these grand models is very right.

 

Graph 10 on the right side of it is LTC6240, offerred gain is 2, drive resistor is 100Ω , to this amplifier character load is bigger. Graph 10 left it is a compound amplifier, input end sets a LTC6240 additionally, a favorable broadband report sheds feedback amplifier (CFA) comes as independent amplifier the load with same drive. The concept of compound amplifier is, output operation amplifier already was had measurable low lack fidelity, and through the input amplifier is in the annulus road gain inside frequency limits can reduce this further lack fidelity. To independent amplifier and compound amplifier, our closed circuit gain is 2, but in compound amplifier, the gain that can install its oneself alone to LT1395 (it is through Rf1 and Rg1 setting 4) , in order to reduce the output amplitude of control amplifier. What cause as a result of the input is lack fidelity increase along with the square that outputs amplitude, what can reduce control operation amplifier further from this is lack fidelity.

 

Graph the 11 spectrum that showed 10 KHz, 4 V P-p is outputted.

 

Graph 11. The lack fidelity spectrum of compound amplifier and groovy amplifier.

 

The computational means with lack fidelity harmonic is: Every harmonic n (DB) is subtractive base wave n (fall in 10kHz frequency) . If what graph bottom shows, those who input signal is lack fidelity it is –163dBc about, first-rate, enough lets a person believe imitate result. V(out2) comes from the LTC6240 at independence, lack fidelity for –78dBc. Pretty good also, but did not reach Ppm level of course.

 

Graph what 11 coping showed compound amplifier is lack fidelity, –135dBc, quite outstanding. So good outcome, we whether believe? To try test and verify, what the node on principle graph showed partly among is lack fidelity. What if compound amplifier is outputted,carry is lack fidelity and adjacent Yu Ling, but those who output amplifier itself is lack fidelity really finite, so feedback process can input end in its (among) to output amplifier lack fidelity setting loses a value. Of intermediate part lack fidelity for –92 DBc, this matchs with the curve of LT1395 data manual actually! I still can think, if the body in grand model reveals physical LTC6240 to input CMRR or ICMR curvature, they still can add actual circuitry possibly lack fidelity.

 

Regretful is, rarely grand model is included lack fidelity. You must read the caption of file of grand model.cir to examine its to whether get support. Wanting understanding to break is really deny the curve with data manual to match, need undertakes a few imitate.

 

The compensation of compound amplifier may be a bit intractable, but in the give typical examples in us, the bandwidth of the 2nd amplifier compares input amplifier tower above 10 times above, need a few Cf to be able to offer circuit to compensate only. In compensating a framework here, if control the bandwidth that BW includes in the overall gain of amplifier, so output the bandwidth of amplifier to answer > 3×BW, and overall bandwidth should guard a setting to approximate BW/3.

 

To avoid bandwidth loss, we can use the method that enhances amplifier. The improvement with such truer to breaking than complex plan photographs is lesser, but bandwidth and set-up time can a hair nondestructive. Graph 12 showed test principle pursues.

 

Graph 12. Increase model the imitate setting of amplifier and independent amplifier.

 

Graph 12 on the right side of showed U2, namely our independent LTC6240; Left showed two LTC6240 amplifier. U1 becomes independent together amplifier is similar, control output, gain is 2; The gain of U2 is 3. U2 is more than the corresponding voltage of U1 in the output tension that heightens node, so U2 can carry drive to carry electricity to output. The gain of RBOOST and U2 can be configured, in order to make U2 carries the load current of 96% to Rl drive, make U1 keeps light carry, improve thereby lack fidelity. We need to ensure U2 contains sufficient abundant amount, in order to bear the weight of additional amplitude.

 

LTC6240 breaks Allah to want to be an input in the load inside KΩ limits lack fidelity, but basically be output to 100Ω load class is lack fidelity.

 

Graph 13 showed spectrum result.

 

Graph 13. Increase model the lack fidelity spectrum of amplifier and groovy amplifier

 

Same, what independent amplifier falls in 10kHz frequency is lack fidelity for –78dBc. Increase model what amplifier offers is lack fidelity for –106dBc; Unlike is compound amplifier is so good, but than independent amplifier almost tower above 30dBc. Nevertheless, increase model the bandwidth of amplifier can reduce a few only.

 

Ask an attention, RBOOST fine tuning one; If its instead 52±2Ω , increase model lack fidelity drop 10dBc, but the change that produces subsequently is minor, highest for ±10Ω . It seems that the measurable load that U1 has a few to anticipate polarity. Ideal (without load) or additional buildup electric current can be brought about lack fidelity increase.

 

Had better be, u2 and U1 have group of identical group delay, appear at the same time with making enhance signal and output. The gain of U2 is taller than U1 50% , closed circuit bandwidth is consequently less, this is meant enhance output to be able to make the main output defer inside frequency limits. The resistor of end is inputted in U1 through bond, can fall the bandwidth of U1 to as same as U2 standard. Can make the noise gain of U1 is equal to U2 so, realize group of identical group defer thereby. This simulator does not have improvement below 10kHz frequency; U1 is offerred optimal and lack fidelity, without defer balanced. You need to try, whether ability understanding also is this kind of circumstance below taller frequency. If amplifier is voltaic feedback type, can make through reducing Rf1 and Rg1 the bandwidth of U2 raises the level to U1 so.

 

Amplifier of Ppm level quality is recommended

 

Express the 2 immediateness that showed a few proposals the important norms of the amplifier of Ppm degree of linearity. Gules entry aims to hint reader: This parameter may not accord with Ppm class lack fidelity. This groups in the center most the better product that uses easily is AD8597, ADA4807, ADA4898, LT1468, LT1678, with LT6018.

 

Express 2. The immediateness that suggests partly the important norms of the amplifier of Ppm degree of linearity

 

Express 3. Operation amplifier is comparative (add)

 

Some amplifier need to solve its to input a problem (inphase enlarge applies a likelihood to be put in the problem) , but still can offer good lack fidelity: AD797, ADA4075, ADA4610, ADA4805, ADA4899, with LTC6228.

 

Conclusion

 

Regretful is, business is used model Ppm precision amplifier is found hard (if can be found) . Ppm linear amplifier exists on the market, but the input electric current that must notice these amplifier, they may apply impedance mediumly through circuit generation is lack fidelity. These impedance can be reduced, but in the drive in feedback they can bring about operation amplifier to input end to produce lack fidelity risk. Operation amplifier is used below particularly small input electric current and fluctuant environment, can apply impedance mediumly through adjusting circuit in order to make operation amplifier is obtained optimal and lack fidelity, but such meetings increase systematic noise. Want to achieve Ppm class degree of linearity and noise, need chooses operation amplifier seriously and optimize applied circuit.