Background technology
Known class ab ammplifier mainly comprises an input utmost point that is made of differential amplifier 41 as shown in Figure 4, and an output stage that is made of class ab ammplifier 42, with input signal through differential amplification and drive the back and export.
Aforementioned class ab ammplifier 42 is can small-signal analysis as follows:
With C
LThe open loop voltage gain of being ordered to O by the A point for this class ab ammplifier 42 of load capacitance is:
Wherein, r
B=r
DS4//r
DS5Be the D.C. resistance (dcresistance) on the B point, r
O=r
DS2//r
DS6Be the direct current output resistance (dcoutput resistance) of output stage, gm
jRepresent MOS transistor (Mj, j=1,2,3 ...) transfer electric field (transferconductance), C
BRepresent the parasitic capacitance (parasitic capacitance) of the current mirror amplifier that crystal M4, M5 constituted.
When providing a building-out capacitor C to aforementioned class ab ammplifier 42
FWhen applying in the operation amplifier circuit, it has building-out capacitor C
FOutput impedance be:
Wherein, r
ABe the D.C. resistance that A is ordered, also be the output resistance of differential amplifier, if
ω>>1/ (r
Ac
F), and
Then can derive:
At this,
Represent an inductance that circles round and export, wherein,
L
OEQ≈C
B/(gm
1gm
6), (4)
And it has the series resistance of an equivalence:
R
SEQ≈1/(gm
1gm
6r
B), (5)
Therefore can get:
And because gm
1Gm
6r
B" gm
2, z
O(S) will resonate in following condition:
And have a following resonant resistance:
And if
Then:
And if gm
2r
BC
B>>C
L, then:
By above analysis as can be known, because the building-out capacitor C of operational amplifier
FCan cause (gyration) effect of circling round, and with parasitic capacitance C
B=C
DB4+ C
DB5+ C
GD4+ C
GD5+ gm
6r
OC
GD6Circle round into the inductance in the output impedance, cause this resonance output impedance | z
OH(j ω
R) | MHz will induce a peak gain to the frequency place of tens of MHz at number, make that the gain margin (gainmargin) of operational amplifier is not enough even it is negative to become, and then cause serious problem such as vibration.General being designed to avoids this problem to take place, must be with the gain margin design of the differential amplifier of the input stage of operational amplifier very low, but so can cause the gain frequency range to become very low, and can cause dc offset (dcoffset) and become problem such as big.
In the document, publication number is that the Japan Patent of 09-018253 proposes a kind of operation amplifier circuit, though it can reduce r in known patent
BEquivalence value alleviate foregoing problems, but can cause the problem of output driving force deficiency, and make that the effect of improving is extremely limited.Therefore, aforementioned known operation amplifier circuit gives improved necessity in fact.
Summary of the invention
Purpose of the present invention is providing a kind of CMOS (Complementary Metal Oxide Semiconductor) class ab ammplifier, with under the situation of unnecessary sacrifice gain frequency range, reaches the performance of power saving and big driving force.
According to one of characteristic of the present invention, its described CMOS (Complementary Metal Oxide Semiconductor) class ab ammplifier, it is right to have an adaptive level shift circuit, a building-out capacitor and an output transistor, to receive the output of a differential amplifier circuit, and provide amplification required Dc bias and direct current amplification, and after the phase compensation of building-out capacitor, export driving by this output transistor by this adaptive level shift circuit, wherein, this adaptive level shift circuit comprises:
One current mirroring circuit comprises first nmos pass transistor and second nmos pass transistor that grid links to each other;
One NMOS crystal diode, its grid links to each other with drain electrode;
One nmos switch transistor, its source electrode is connected to the drain electrode of second nmos pass transistor of this current mirroring circuit, the drain electrode of first nmos pass transistor of its this current mirroring circuit of drain electrode connection and the source electrode of this NMOS crystal diode, its grid then connects the grid leak utmost point connecting place of this NMOS crystal diode;
One PMOS input amplifier transistor, its drain electrode connects the drain electrode of this NMOS crystal diode, and its grid connects the output of a differential amplifier circuit, and its source electrode connects voltage source; And
One PMOS current source transistor, its drain electrode connects the transistorized source electrode of this nmos switch, and its grid connects a bias terminal, and its source electrode connects voltage source.
Wherein this output transistor is constituted PMOS transistor and the nmos pass transistor that is linked to each other by drain electrode, the transistorized grid of this PMOS connects the transistorized grid of this PMOS input amplifier, its source electrode is connected in voltage source, the grid of this nmos pass transistor connects the transistorized source electrode of this nmos switch, and its source electrode is connected in system's electronegative potential.
Wherein this building-out capacitor is connected between this PMOS transistor drain and the transistorized grid of this PMOS input amplifier.
According to another characteristic of the present invention, its described CMOS (Complementary Metal Oxide Semiconductor) class ab ammplifier, it is right that it has an adaptive level shift circuit, a building-out capacitor and an output transistor, to receive the output of a differential amplifier, and provide amplification required Dc bias and direct current amplification, and after the phase compensation of building-out capacitor, export driving by this output transistor by this adaptive level shift circuit, wherein, this adaptive level shift circuit comprises:
One current mirroring circuit comprises a PMOS transistor and the 2nd PMOS transistor that grid links to each other;
One PMOS crystal diode, its grid links to each other with drain electrode;
One PMOS switching transistor, its source electrode is connected to the 2nd PMOS transistor drain of this current mirroring circuit, its drain electrode connects a PMOS transistor drain of this current mirroring circuit and the source electrode of this PMOS crystal diode, and its grid then connects the grid leak utmost point connecting place of this PMOS crystal diode;
One NMOS input amplifier transistor, its drain electrode connects the drain electrode of this PMOS crystal diode, and its grid connects the output of a differential amplifier circuit, its source electrode connected system electronegative potential; And
One NMOS current source transistor, its drain electrode connects the source electrode of this PMOS switching transistor, and its grid connects a bias terminal, its source electrode connected system electronegative potential.
Wherein this output transistor is constituted nmos pass transistor and the PMOS transistor that is linked to each other by drain electrode, the grid of this nmos pass transistor connects the transistorized grid of this NMOS input amplifier, its source electrode is connected in system's electronegative potential, the transistorized grid of this PMOS connects the source electrode of this PMOS switching transistor, and its source electrode is connected in voltage source.
Wherein this building-out capacitor is connected between the drain electrode and the transistorized grid of this NMOS input amplifier of this nmos pass transistor.
Embodiment
For trying to achieve CMOS (Complementary Metal Oxide Semiconductor) class ab ammplifier of the present invention, at first, the problem of aforementioned known techniques by analysis, as make CB go to zero to constrain covibration, then can get a desirable output impedance:
y
OH_Ieal=1/z
OH_Ieal=gm
2+gm
1gm
6r
B+sC
L, (12)
Therefore, can derive impedance makes a variation as follows:
Then can derive:
And if gm
2r
BC
B>>C
L, the maximum loss that then can obtain the nargin gain is:
Reach (15) as can be known with reference to formula (14), the method for idealization of compacting resonance gain is with C
BValue be kept to zero, but under actual conditions, C
BValue can not be zero, and therefore, the practicable best approach is to reduce r
BValue.For reaching this purpose, please refer to CMOS (Complementary Metal Oxide Semiconductor) class ab ammplifier circuit diagram of the present invention shown in Figure 1, it moves (Level shift) circuit 11, a building-out capacitor C by an adaptive level
FAnd an output transistor is to 12 formations, reception is from the output of a differential amplifier circuit 13, to provide the required Dc bias of amplification to reach and the coaxial direct current amplification of differential input I/P by this adaptive level shift circuit 11, after the phase compensation of building-out capacitor CF, drive output to 12 by this output transistor.
As shown in Figure 1, this adaptive level shift circuit 11 comprises a PMOS transistor M1 and the PMOS transistor M5 as current source as input amplifier, two nmos pass transistor M3 and M4 as current mirror, one a nmos pass transistor M8 and a nmos pass transistor M7 as diode as diverter switch, wherein, the output (A point) that connects differential amplifier circuit 13 as the grid of the transistor M1 of input amplifier, grid as the transistor M5 of current source is connected in a bias terminal VB1, the source electrode of two transistor M1 and M5 then connects voltage source V DD, and its drain electrode then connects the drain electrode of transistor M7 and the source electrode of switching transistor M8 respectively.Grid as the transistor M3 of current mirror and M4 links to each other, and links to each other with the drain electrode of M3, and these transistors M1, M3, M4 and M5 are promptly in order to the Dc bias and the direct current amplification of the level shift circuit known to providing generally.
This output transistor is constituted 12 PMOS transistor M2 and the nmos pass transistor M6 that linked to each other by drain electrode, and wherein, the grid of transistor M2 connects the grid of transistor M1, and the grid of transistor M6 then connects the source electrode of transistor M8.This building-out capacitor CF is connected between the grid of the drain electrode of transistor M2 and transistor M1.
Between the transistor M3 of the present invention by output that transistor M7 is arranged at transistor M1 and current mirror, so that being provided, Dc bias comes oxide-semiconductor control transistors M8, as shown in the figure, the grid of transistor M7 links to each other with drain electrode and forms a diode, the source electrode of transistor M8 is connected to the drain electrode (B point) of transistor M4, its drain electrode is connected to the drain electrode of transistor M3 and the source electrode of transistor M7 (C point), and its grid then connects the grid leak utmost point connecting place (D point) of transistor M7.
When class ab ammplifier of the present invention worked in small-signal, the drain-source voltage Vds of transistor M8 was quite little, that is transistor M8 works in linear zone (linear region, or be called trioderegion), so have quite little drain-source electrode resistance r
DS7, the output impedance that therefore can try to achieve class ab ammplifier is:
Wherein, C
B=C
GD8+ C
DB8+ C
DB4+ C
DB5+ C
GD4+ C
GD5+ gm
6r
OC
GD6, its outputting inductance that will circle round out:
L
OEQ≈2C
B/(gm
1gm
6), (17)
This inductance has an equivalent series resistance:
R
SEQ=2/(gm
1gm
6r
DS8), (18)
And if ω>>1/ (r
AC
F) and
Then can be reduced to:
And if gm
1Gm
6r
DS8/ 2>>gm
2, then output impedance has a resonance frequency:
Reach a resonance impedance:
Compared to ideal state (C
BLevel off to 0 to constrain covibration fully) time output impedance:
y
OH_Ieal=1/z
OH_Ieal=gm
2+gm
1gm
6r
B+sC
L, (23)
Can derive following impedance variation:
Therefore,
And if gm
2r
DS8>>C
L, can obtain the maximum loss of gain margin (gain margin):
With reference to aforesaid formula (14), (15), (25) and (26), the r of class ab ammplifier of the present invention as can be known
DS8For operating in the resistance of the range of linearity, the about 110K Ω of its resistance value, and the resistance r of class ab ammplifier in the known technology
BFor operating in the saturation region, its resistance is 1.2M Ω, because r
DSMuch smaller than r
B, therefore, the gain peak in class ab ammplifier of the present invention can be suppressed effectively.
When class ab ammplifier of the present invention works under the large-signal, and when the big source of desire output goes out electric current (source current), the A point of input is dragged down (pulled low), so can exporting big source, transistor M2 goes out electric current, and transistor M1 can pass through transistor M7 and M3 by the big transient current of conduction ratio quiescent current, and because transistor M4 and M3 are current mirrors, so transistor M4 also can conducting big transient current like this, and transistor M4 can conducting the transient current stable bias current that can be provided greater than transistor M5, so the B point can be dragged down, transistor M6 can be near closing (turned off), this moment, amplifier went out electric current with the big source of category-B pattern output, in this transient state, though transistor M8 and M7 also are the connections of electric current mirror, but,, so can not limit the source output capacity of class ab ammplifier so transistor M7 only can the conducting micro-current and do not influence drop-down that B order because its gain is much smaller than the gain of transistor M4 than M3.
Working as the present invention in addition is operated under the large-signal, and desire conducting big sink electric current (sink current) time, the A point is drawn high (pulled high), so the conducting electrorheological of transistor M1 and M2 gets very little, therefore the conducting electric current of transistor M7, M8, M3 and M4 also becomes very little, and the stable bias current that is provided less than transistor M5, so the B point can be drawn high near voltage source V DD, transistor M6 can the big electric current of conducting, therefore, amplifier will import the big electric current that sinks from load end with the category-B pattern.
With reference to shown in Figure 2, it is presented at load capacitance C
LThe phase curve (B) of phase curve of the present invention (A) and known amplifier during for 100pF, can obvious class ab ammplifier of the present invention very effective to the inhibition of resonance peak.
Fig. 3 is another preferred embodiment of the present invention, and it is by an adaptive level shift circuit 31, a building-out capacitor C
LAnd an output transistor is to 32 formations, as shown in the figure, this adaptive level shift circuit 31 comprises respectively a PMOS transistor M8 and a PMOS transistor M7 as diode who switches switch as two nmos pass transistor M1 of input amplifier and current source and M5, as the two PMOS transistor M3 and the M4, of current mirror, wherein, an output (A point) and a bias terminal VB1 as input amplifying circuit 33, its source electrode is connected system electronegative potential VSS then, and its drain electrode then connects the drain electrode of transistor M7 and the source electrode of transistor M8 respectively.Link to each other as the transistor M3 of current mirror and the grid of M4, and link to each other with the drain electrode of M3.
This output transistor is constituted 12 nmos pass transistor M2 and the PMOS transistor M6 that linked to each other by drain electrode, and wherein, the grid of transistor M2 connects the grid of transistor M1, and the grid of transistor M6 then connects the source electrode of transistor M8.This building-out capacitor C
FBe connected between the grid of the drain electrode of transistor M2 and transistor M1.Transistor M7 is arranged between the transistor M3 of the output of transistor M1 and current mirror, so that being provided, Dc bias comes oxide-semiconductor control transistors M8, as shown in the figure, the source electrode of transistor M8 is connected to the drain electrode (B point) of transistor M4, its drain electrode is connected to the drain electrode of transistor M3 and the source electrode of transistor M7 (C point), and its grid then connects the grid leak utmost point connecting place (D point) of transistor M7.
It is replaced the circuit of previous embodiment PMOS, nmos pass transistor respectively with different only being of last embodiment with NMOS, PMOS transistor, the working method of its circuit is then identical with last embodiment, therefore, no longer repeat to describe in detail the workflow of this embodiment.
By above explanation as can be known, the present invention by seal in transistor M7 as bias voltage with driving transistors M8, and provided enough low linear zone resistance r
DS8As feedback, therefore reduce the Q value of the inductance that circles round effectively, and suppress the generation of peak gain effectively, gain and keep required nargin, and the adding of this transistor M7 and M8 and mode of connection still can make the driving voltage of output transistor M6 not be subjected to strangulation when large-signal, still can make amplifier with the work of AB quasi-mode, reach the performance of power saving and big driving force, and needn't sacrifice the gain frequency range.
It should be noted that above-mentioned many embodiment are the present invention and giving an example for convenience of explanation only, and unrestricted the present invention.