CN101931323B - Method for enhancing non-uniform variation grid width of light load efficiency of integrated switch DC-DC converter - Google Patents

Abstract

The invention discloses a method for enhancing the non-uniform variation grid width of the light load efficiency of an integrated switch DC-DC converter, a Buck-Boost converter in the invention adopts the design of a CSMC 0.5mu m CMOS process library, whole-circuit integration is realized except for a passive filter, the external filter inductance is 2.2mu H, and the filter capacitance is 1mu F. According to the requirements of input voltage and output voltage, the converter can work in three modes: Buck (voltage reducing), Buck-Boost (voltage reducing and boosting) and Boost (voltage boosting), the range of the input voltage is 2.5V-4.2V, the range of the output voltage is 1.5V-5V, and the working frequency is 5MHz. A non-uniform grid width modulation method is adopted in the whole load current range of 10mA-650mA. When the converter works at high frequency of 5MHz, the efficiencies of medium load and heavy load keep above 90 percent all the time, and the efficiency of light load (10mA) can reach above 80 percent. As the grid width of a switching tube is only changed and a control link of the working efficiency of an extra switching tube is not adopted, negative effects caused by frequency conversion control are eliminated fundamentally.

Description

A kind of method that improves the light load efficiency of integrated switch DC-DC converter non-uniform variation grid width

Technical field:

The invention belongs to DC-DC converter field, relate to a kind of method that improves the light load efficiency of integrated switch DC-DC converter non-uniform variation grid width

Background technology:

Along with the continuous progress of portable type electronic product, the power management chip market development is rapid.Because the requirement of the continuous miniaturization of portable set, the integrated trend of Switching Power Supply product is more and more obvious.Improve the size that switching frequency can effectively reduce inductance, capacity cell, however high-frequency limitations the conversion efficiency of Switching Power Supply, the especially efficient at underload place.The general portable set most of the time is operated under " standby " state, and the lifting of the light-load efficiency of converter is most important for the prolongation of stand-by time.

The switch DC-DC converter theoretical efficiency can reach 100%, but has loss owing to non-ideal factor in the side circuit.Can be write the total losses of switch DC-DC converter as following form:

$P_{\mathrm{loss}}=k_1\frac{I_{L,\mathrm{rms}}^2}{W}+k_{2}W{f}_s---\left(1\right)$

Wherein, W is a switch MOS pipe grid width, I _{L, rms}Be rms current through switching tube, f _sBe switching frequency, k ₁And k ₂Be and grid width W and frequency f _sIrrelevant coefficient.Can find out in the expression formula that total losses roughly can be divided into two: first is conduction loss, and electric current transmits the loss that causes, its value and the grid width W relation of being inversely proportional to through the dead resistance of switch MOS pipe when being the conducting of switch MOS pipe; Second is switching loss, mainly is because the parasitic capacitance that the conducting of switch MOS pipe is turn-offed itself and drive circuit thereof discharges and recharges the loss that causes, its value and grid width W and switching frequency f _sDirect ratio.

Do not reduce because the switching loss of power switch pipe does not reduce with the converter output loading, but be directly proportional with switching frequency.Therefore, grid drive loss and switching loss become the leading factor that high-frequency switch converters light-load efficiency worsens.

For the efficiency at underloading place, the solution of main product is to introduce pulse frequency modulated (PFM) pattern on the market, during underloading converter is transformed under the PFM control model from pulse width modulation (PWM).At this moment, switching frequency reduces and reduces along with load current, thereby has alleviated the deterioration of efficient under the underloading situation.Document for example: " the high efficiency dc-dc boost transducer of double mode PWM/PFM control, " (Chen Dongpo, He Lenian; Tight dawn is unrestrained; The 29th volume in August, 2008 the 8th phase, the semiconductor journal) adopted the double mode control of PWM/PFM, output loading is still remained on more than 55% when 20mA.Yet may cause unnecessary sharp cutting edge of a knife or a sword electric current during two kinds of modulation system transition.In addition; Because PFM control switch frequency changes continuously; And be embodied in the harmonic noise that has just become broad spectrum in the output voltage of switch converters, and can cause electromagnetic interference to system, therefore this converter of frequency conversion that needs is difficult to be applied in the system to noise-sensitive.

The solution of fixed frequency also is suggested; As document " A Constant Frequency Method for ImprovingLight-Load Efficiency in Synchronous Buck Converters; " (M.D.Mulligan.B.Broach, and T.H.Lee, IEEE Power Electronics Letters; VOL.3; NO.1, MARCH 2005.S.) proposed the method for the dynamic grid swing of switching tube, hour adopted the charging/discharging voltage that reduces on the switch MOS pipe to reduce its switching loss at output current.

The solution of another kind of fixed frequency is the method for the dynamic grid width adjustment of switch MOS pipe.Shown in (1), two kinds of losses of converter are directly proportional and inverse ratio with switch MOS pipe grid width respectively, therefore can find an optimum grid width value to make the loss of converter reach minimum.Differentiate can obtain best grid width and makes loss reach minimum formula (1) to grid width W, and the optimum grid width expression formula of this moment is following:

$W_{\mathrm{opt}}=\sqrt{\frac{k_1}{k_2{f}_s}}{I}_{L,\mathrm{rms}}---\left(2\right)$

Because rms current is approximately equal to load current, so can find out from formula (2), optimum grid width reduces along with load current and reduces.In order to realize the high efficiency in the wide loading range, must adopt the switching tube grid width that changes with load current.Show that like Fig. 1 (W1, W2 W3), can make converter in wide load current range, keep high conversion rate to adopt different grid width in different current ranges.Document " Improvement of Light-Load Efficiency UsingWidth-Switching Scheme for CMOS Transistors; " (Musunuri; And P.L.Chapman, IEEEPower Electronics Letters, VOL.3; NO.3, SEPTEMBER 2005) this method of raising the efficiency at first proposed.Document " Dithering Skip Modulation; Width and Dead Time Controllers in Highly EfficientDC-DC Converters for System-On-Chip Applications, " (Hong-Wei, H.; C.Ke-Horng; IEEEJournal of Solid-State Circuits, 42 (11): 2451-2465,2007) in the structure identical design philosophy is arranged also.

Since the grid width of switch MOS pipe can not be continuous variation, therefore can only be employed in corresponding fixing grid width in certain load current range.Its concrete realizing method is that some switch MOS pipes are connected in parallel, and converter work the time can make the paralleling MOS pipe of part turn-off fully according to the difference of load current, and remaining metal-oxide-semiconductor operate as normal, reaches the effect of the total grid width of dynamic adjustments switching tube.What dynamically the grid width scheme all adopted in the above document is that the identical metal-oxide-semiconductor of size is connected in parallel and forms total metal-oxide-semiconductor, thus grid width no matter in heavy duty still under light load condition, the changing value Δ W of each grid width fixes.But research shows, when heavy duty and underload work, is different to the requirement of grid width changing value Δ W, and therefore, when how to confirm the different loads situation, the optimum variation size of switching tube becomes and further reduces power consumption, simplifies the key of circuit design.

Summary of the invention:

The efficient expression formula of DC-DC converter is following:

$\mathrm{\η}=\frac{P_{\mathrm{out}}}{P_{\mathrm{out}}+P_{\mathrm{loss}}}---\left(3\right)$

P _OutFor effectively being transferred to the power in the load, suppose at certain load current I _OutThe place makes transducer effciency reach the optimum switch metal-oxide-semiconductor and is of a size of W _Opt, it satisfies expression formula (2), and corresponding at this moment optimum loss (P _Loss) _OptWith optimum efficiency η _OptExpression formula following:

${\left(P_{\mathrm{loss}}\right)}_{\mathrm{opt}}=2\sqrt{k_1{k}_2\mathrm{fs}}{I}_{L,\mathrm{rms}}---\left(4\right)$

${\mathrm{\η}}_{\mathrm{opt}}=\frac{P_{\mathrm{out}}}{P_{\mathrm{out}}+{\left(P_{\mathrm{loss}}\right)}_{\mathrm{opt}}}---\left(5\right)$

If the grid width W of switching tube is not at its optimal value W _OptThe place, computational transformation device efficient remains on the span of the switching tube grid width W of optimum efficiency more than 95%, thereby confirms the size of one group of switch MOS pipe size of parallel connection.

η＞η _opt×95％ (6)

To calculate simply in order making, can to make

P _out＝n×(P _loss) _opt (7)

Wherein, n is a proportionality coefficient, when output voltage and load current are confirmed, is a determined value.In (5) and (7) formula substitutions (6) formula, obtain following result,

$0.475\&times;\frac{W_{\mathrm{opt}}}{W}+0.475\&times;\frac{W}{W_{\mathrm{opt}}}<0.05\&times;n+1---\left(8\right)$

Rule of thumb, efficient generally greater than 90%, then can be got n=10 in the reasonable scope in the value at heavy duty place, and the span that obtains W is suc as formula (9),

0.357×W _opt＜W＜2.8×W _opt (9)

From formula (9) can find out if the grid width of switching tube one greater than 0.357 times of optimum grid width size, less than changing between 2.4 times of optimum grid width sizes, can both guarantee 95% η _OptConversion efficiency.But, W _OptProportional with load current; During large load current, corresponding changed size range is big, and reducing along with load current; This guarantees also corresponding the dwindling of grid width size range of high conversion efficiency; At this moment, the less change of load current just possibly make the actual grid width of switch MOS pipe exceed this scope, causes the rapid decline of conversion efficiency.

To the problems referred to above, the present invention proposes the non-homogeneous grid width method of adjustment that grid width adjustment changing value Δ W changes with load current, and give the circuit realization.In the DC-DC converter that the present invention can be applied to is dissimilar (Buck, Buck-Boost and Boost converter, but be not limited thereto three types).Non-homogeneous grid width adjustment method is primarily aimed at heavy duty, and work is the different proposition that requires to the grid width variation with underloading; Converter is operated in fixedly under the high frequency; Also can improve light-load efficiency effectively, the problems of avoiding the common common PFM modulation of light-load efficiency solution to bring through the method.

When converter was operated in large load current, research showed that the total grid width of maintained switch metal-oxide-semiconductor is constant, and efficiency curve is in the excursion of a bigger load current, and changing is not clearly.Therefore, between this load region, do not need to change continually the size of switching tube, the grid width adjustment technology of the present invention's design in heavily loaded scope, adopts bigger grid width changing value Δ W.In fact, adopt during heavy duty big Δ W value also can reduce the energy loss that frequent adjustment grid width size causes.And converter is when being operated in little load current, efficiency curve with load current reduce worsen very obviously, then need select littler grid width changing value Δ W.During design, in whole loading range, grid width changing value Δ W and load current variation in direct ratio, thus the method that can utilize grid width to modulate fully makes the transformation efficiency integral body of converter remain on higher level.

The present invention is directed to the low problem of high frequency switch DC-DC conversion device light-load efficiency; A kind of non-uniform variation grid width that utilizes has been proposed; Realize fixed frequency, interior, the high efficiency solution of wide loading range, and be applied in the four switching mode Buck-Boost switch DC-DC converters.

For reaching above purpose; The present invention takes following technical scheme to be achieved: the shortcoming that the objective of the invention is to overcome above-mentioned prior art; A kind of method that improves the switch DC-DC converter non-uniform variation grid width of light load efficiency is provided, and its implementation comprises power MOS pipe grid width rearrangeable switch DC-DC converter circuit structure and non-homogeneous grid width optimization method two parts.

The adjustable DC-DC converter of a kind of power MOS pipe grid width; The efficient DC-DC converter of this wide load current range comprises: power stage main circuit and driver module thereof; The PWM control module, current detection module in the mode selection module, sheet; Delay line ADC module, switching tube grid width control module and Dead Time control module.

Said power stage main circuit and driver module thereof, V _InPort is connected with power input, V _OutPort links to each other with the converter voltage output port, and the GND port links to each other with external ground wire; V _xEnd and V _yEnd is connected V with external inductance L two ends _xEnd links to each other with current detection module simultaneously, for it provides detection signal; Power stage main circuit and driver module thereof and switching tube grid width control module are through grid width numeral option code D ₀～D _M-1, D _ExtraConnect, grid width numeral option code is used to adjust the power switch pipe size; (PWM_P2 PWM_N2) connects for PWM_P1, PWM_N1, and pwm signal is the control signal of 4 adjustable grid width switching tubes in the power stage main circuit through four road pwm signals for power stage main circuit and driver module thereof and Dead Time control module.

Said PWM control module, its input and output feedback signal bV _OutBe connected signal bV _OutPass through resistance R by the converter output voltage _B1, R _B2Dividing potential drop obtains; The PWM_C signal that its output produces is connected with the Dead Time control module.

Said mode selection module, V _InPort is connected with power input, V _OutPort links to each other with the converter voltage output.Mode selection module is through comparing the size of input voltage and output voltage, and acquisition model is selected signal S _Mode, be used for confirming the mode of operation of converter, and pass through S _ModeSignal is connected with switching tube grid width control module with the Dead Time control module respectively.

Said interior current detection module, V _InPort is connected with power input, V _xPort is connected with power stage main circuit and driver module thereof; Output port produces and the directly proportional detectable voltage signals V of inductive current _Sense, and pass through V _SenseSignal is connected with the Dead Time control module with delay line ADC module.

Said delay line ADC module, input port passes through V _SenseSignal is connected with the interior current detection module of sheet.Delay line ADC module is with analog voltage signal V _SenseConvert m megadyne temperature number of degrees character code Q to ₀～Q _M-1, and be connected with switching tube grid width control module through digital code.

Said switching tube grid width control module is through digital code Q ₀～Q _M-1Be connected with switching tube grid width control module, through mode select signal S _ModeBe connected with mode selection module, through grid width numeral option code D ₀～D _M-1, D _ExtraBe connected with power stage main circuit and driver module thereof.

Said Dead Time control module is through mode select signal S _ModeBe connected with mode selection module, be connected with the PWM control module through the PWM_C signal, through V _SenseSignal is connected with the interior current detection module of sheet, and (PWM_P2 PWM_N2) is connected with power stage main circuit and driver module thereof for PWM_P1, PWM_N1 through four road pwm signals.The Dead Time control module obtains four road pwm signal (PWM_P1 with reference to the different working pattern through the PWM_C signal is added certain Dead Time; PWM_N1; PWM_P2, PWM_N2), and reference work pattern and current detecting acquisition voltage adjustment Dead Time size.

In the technique scheme, power stage main circuit and driver module thereof are realized M through the switching tube parallel connection of different size _N1, M _P1, M _N2And M _P2Four non-homogeneous adjustment of switch MOS pipe.For the load current excursion is (I _Min, I _Max) four switch Buck-Boost converters, through non-homogeneous grid width optimization method with four switching tube M _N1, M _P1, M _N2And M _P2Size be divided into not isometric m+1 section.Wherein, form M _P1The source of every section paralleling MOS pipe of pipe is leaked and is connected on V respectively _InAnd V _xPort is formed M _N1The source of every section paralleling MOS pipe of pipe is leaked and is connected on GND and V respectively _xPort is formed M _P2The source of every section paralleling MOS pipe of pipe is leaked and is connected on V respectively _OutAnd V _yPort is formed M _N2The source of every section paralleling MOS pipe of pipe is leaked and is connected on GND and V respectively _yPort.Any one section paralleling MOS Guan Douyou independently is connected driving switch pipe normally and shutoff with the drive circuit that its size adapts with its grid.Form four switching tube M _N1, M _P1, M _N2And M _P2The drive circuit input of paralleling switch pipe is exported signal PWM_N1 with above-mentioned Dead Time control module respectively, PWM_P1, and PWM_N2, PWM_P2 is connected.Simultaneously, the drive circuit of every section paralleling MOS pipe all has the enable port, respectively with grid width numeral option code D ₀～D _M-1, D _ExtraBe connected.With switching tube M _P1Be example, it is by M _P1(0) to M _P1(m) m+1 not isometric metal-oxide-semiconductor is formed in parallel altogether, wherein, and a metal-oxide-semiconductor M _P1(m) but remain operating state, and other metal-oxide-semiconductor M _P1(0)-M _P1(m-1) respectively by pairing digital code D ₀-D _M-1Control, digital code enables to control single paralleling MOS pipe through driving and is in work or off state.Switching tube M _N1, M _N2Control and realization grid width selection mode and M _P1Manage identical.And be the operate as normal that guarantees each pattern downconverter, switching tube M _P2Except above-mentioned m+1 section paralleling MOS pipe, also need add a tube-carrier M who only is used to be operated in the Buck pattern _P2(extra), its connected mode and M _P2Other section parallel transistor is the same, drives Enable Pin and D _ExtraBe connected.

A kind of aforesaid non-homogeneous grid width optimization method obtains to form power switch pipe M based on above-mentioned four switch Buck-Boost changer systems _N1, M _P1, M _N2And M _P2The number m+1 and the size W separately of paralleling switch pipe _M(0)-W _M(m).Because M _N1, M _P1, M _N2And M _P2Four switching tube size design methods are just the same, for the purpose of making things convenient for, adopt the M pipe to replace.If total grid width size of M pipe is W _Total, to form by m+1 not isometric paralleling MOS pipe M (0)-M (m), correspondingly-sized is respectively W _M(0)-W _M(m).The adjustment of control M pipe adjustment grid width be digital code, from M (0) to M (m) along with load current reduces to turn-off successively, can realize W successively _Opt(0) to W _Opt(m) be total to m+1 kind grid width size.If I _Out(0) to I _Out(m) be the corresponding W of difference _Opt(0) to W _Opt(m) different size reaches the optimal load electric current of peak efficiency.

Foregoing realization improves the method for light load efficiency of integrated switch DC-DC converter non-uniform variation grid width, according to following steps:

(a) build four switch Buck-Boost changer systems according to above-mentioned connected mode;

(b) according to required maximum load current I _MaxConfirm power switch pipe M _N1, M _P1, M _N2And M _P2Overall size; Confirm I _Out(0) ≈ 0.6 * I _MaxThe time corresponding optimal size W _Opt(0), the overall size W of switch MOS pipe just _Total(c) according to overall size W _TotalWith required minimum load current I _Min, adopt grid width changing value Δ W=0.5 * W _OptConfirm the size of M (0)-M (m), realize non-homogeneous variable grid width adjustment.Confirm to remove the size W of the outer remaining switching tube of M (0) pipe when confirming the size of switching tube M (0) _Opt(1),

$W_M\left(0\right)=W_{\mathrm{opt}}\left(1\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2}---\left(11\right)$

W _OptLoad current I when (1) reaching its efficient and reach optimal value _Out(1),

$I_{\mathrm{out}}\left(1\right)=\frac{I_{\mathrm{out}}\left(0\right)}{2}---\left(12\right)$

Derive according to above-mentioned method successively and can obtain the size M (i-1) of each pipe of required parallel connection, i=1,2,3 ...,, make I up to satisfying i=m _Out(m+1)＜I _Min, confirm the number m+1 of paralleling switch pipe;

$W_M(i-1)=W_{\mathrm{opt}}\left(i\right)=\frac{W_{\mathrm{opt}}(i-1)}{2}=\frac{W_{\mathrm{opt}}\left(0\right)}{2^i},i=\mathrm{1,2,3},\&CenterDot;,m---\left(13\right)$

And rest parts promptly is the size of last switching tube M (m),

$W_M\left(m\right)=W_{\mathrm{opt}}\left(m\right)=W_{\mathrm{opt}}\left(0\right)-\underset{i=1}{\overset{m-1}{\mathrm{\&Sigma;}}}{W}_M\left(i\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2^m}---\left(14\right)$

W _Opt(0) to W _Opt(m) be the different grid width size that can realize through digital code in the converter, they are when efficient reaches optimum separately, the corresponding load electric current I _Out(0) to I _Out(m) as follows:

$I_{\mathrm{out}}\left(n\right)=\frac{I_{\mathrm{out}}(n-1)}{2}=\frac{I_{\mathrm{out}}\left(0\right)}{2^m}---\left(15\right);$

(d) design the size of each paralleling switch pipe, confirm each grid width value corresponding load current range of non-uniform change so that the conversion efficiency of converter keeps optimum; The function of turn-off delay line ADC directly adds the corresponding different grid width of different digital codes to switching tube grid width control module, obtains the efficiency curve that changes with load current; In all load current range, choose the digital code when reaching most effective, confirm each grid width value corresponding load current range;

(e) control lag line ADC output digital code Q ₀～Q _M-1With above-mentioned detection voltage V _SenseVariation relation, make ADC output digital code Q ₀～Q _M-1Corresponding with the load current range that calculates in the step (d);

(f) realize the self adaptation adjustment of grid width with the load current non-uniform change.

The Buck-Boost converter of the non-homogeneous grid width changing method of the high frequency that the present invention adopts the present invention to design; Adopt the design of CSMC0.5 μ m CMOS technology library; The whole circuit of realization except that the passive filtering device is integrated, and external filter inductance is 2.2 μ H, and filter capacitor is 1 μ F.According to the requirement to input and output voltage, converter can be operated under Buck, Buck-Boost, three kinds of patterns of Boost, input voltage range 2.5V-4.2V, output voltage range 1.5V-5V, operating frequency 5MHz.In whole load current scope 10mA-650mA, adopt grid width modulator approach heterogeneous; When converter equals under the high frequency of 5MHz work at switching frequency; Its medium load and heavy duty efficient remain at more than 90%; And light-load efficiency has obtained significantly improving, and for Buck pattern conversion and Boost pattern conversion, efficient can reach 80% and 84% respectively when load current was 10mA.

This invention draws under the different loads condition through the power consumption analysis to the HF switch converter, the rule and the method for the modulation of switching tube grid width, thus make total solution.Realized significantly improving to light-load efficiency.Owing to just changed the grid width of switching tube, do not take the controlling unit of extra switch plumber working frequency, fundamentally eliminated the adverse consequences that VFC is brought.

Description of drawings:

Fig. 1 is the basic principle and the foundation of the modulation of switching tube grid width.

The physical circuit implementation of non-uniform variation grid width modulation circuit in Buck-Boost DC-DC converter that Fig. 2 designs for the present invention.

Fig. 3 is design of non-uniform variation grid width switching tube and a drive circuit thereof among Fig. 2.Wherein Fig. 3 (a) is power switch pipe M _P1And drive circuit; Fig. 3 (b) is power switch pipe M _N1And drive circuit; Fig. 3 (c) is power switch pipe M _P2And drive circuit; Fig. 3 (d) is power switch pipe M _N2And drive circuit.

The analog result figure that Fig. 4 changes with load current for the DC-DC converter conversion efficiency that adopts the present invention's acquisition.Wherein Fig. 4 (a) be under the Buck pattern conversion efficiency with load current change modeling figure as a result; Fig. 4 (b) be under the Boost pattern conversion efficiency with load current change modeling figure as a result.

Embodiment:

Below in conjunction with accompanying drawing the present invention is done and to describe in further detail:

The concrete non-homogeneous grid width adjustment Buck-Boost switch DC-DC converter integrated circuit structure that realizes of the present invention is as shown in Figure 2; Comprise power stage main circuit and driver module thereof; The PWM control module, current detection module in the mode selection module, sheet; Delay line ADC module, switching tube grid width control module and Dead Time control module.Inductance L among Fig. 2, capacitor C are external discrete device, R _LBe load; Each block diagram is the module of a certain specific function of realization among Fig. 2, and all modules are the IC internal circuit.

As shown in Figure 2, power stage main circuit and driver module thereof are realized Buck, the power delivery function of the different voltage alternative types with Buck-Boost of Boost.Said power stage main circuit and driver module thereof and other module connected mode are following: V _InBe connected V with power input _OutBe connected with power output end, GND links to each other with external ground; V _xAnd V _yBe connected V with external inductance L two ends _xEnd links to each other with current detection module simultaneously; Through grid width numeral option code D ₀～D _M-1, D _ExtraBe connected with switching tube grid width control module; (PWM_P2 PWM_N2) is connected with the Dead Time control module for PWM_P1, PWM_N1 through four road pwm signals.Its inner M _N1, M _N2, M _P1, M _P2Four are adopted the basic connected mode of grid width rearrangeable switch metal-oxide-semiconductor following: M _P1The pipe source is leaked and is connected on V respectively _InAnd V _xPort, M _N1The source is leaked and is connected on GND and V respectively _xPort, M _P2The pipe source is leaked and is connected on V respectively _OutAnd V _yPort, M _N2The pipe source is leaked and is connected on GND and V respectively _yPort.M wherein _N1And M _N2Be NMOS, M _P1And M _P2Be PMOS, its basic control mode is divided into following three kinds: if switch M _N2Keep turn-offing M _P2Keep conducting, by PWM control M _N1And M _P1, converter works in the Buck pattern; If switch M _N1Keep turn-offing M _P1Keep conducting, M _N2And M _P2By PWM control, converter works in the Boost pattern; If 4 switch MOS pipes then work in the Buck-Boost pattern simultaneously by pwm signal control.Because power switch metal-oxide-semiconductor size is very big, the chain of inverters that each switching tube all needs to amplify step by step is as driving.As shown in Figure 2, PWM control module and other module connected mode are following: input and output feedback signal bV _OutBe connected signal bV _OutPass through resistance R by the converter output voltage _B1, R _B2Dividing potential drop obtains; Output is connected with the Dead Time control module through the PWM_C signal.Its major function is according to output feedback signal bV _OutThe pwm control signal that obtains the adjustable duty ratio of fixed frequency is used for power controlling switch MOS pipe.

As shown in Figure 2, mode selection module V _InPort is connected with power input, V _OutPort links to each other with the converter voltage output, through S _ModeSignal is connected with switching tube grid width control module with the Dead Time control module respectively.Mode selection module is according to input voltage V _InAnd output voltage V _OutThe comparison acquisition model select signal S _Mode, realize the selection of system works pattern.

As shown in Figure 2, current detection module is to realize a main modular of the present invention in the sheet.Its V _InPort is connected with power input, V _xPort is connected with power stage main circuit and driver module thereof, through V _SenseSignal is connected with the Dead Time control module with delay line ADC module.Its major function is through detection power switching tube M _P1The voltage signal at two ends, the change information of acquisition inductive current is through current detection module in the sheet, with voltage signal V _SenseForm output.Signal V _SenseWith load current I _OutTherefore linear change, the situation of change of reaction load accurately.V _SenseSignal offers delay line ADC and Dead Time control module simultaneously.

As shown in Figure 2, delay line ADC module is to realize a main modular of the present invention.Delay line ADC module is passed through V _SenseSignal is connected with the interior current detection module of sheet, through digital code Q ₀～Q _M-1Be connected with switching tube grid width control module.Its major function is to realize will be by the small V of variation of current detection module output in the sheet _SenseConversion of signals becomes the temperature digital sign indicating number, the V that promptly number of " 1 " obtains with sampling in the output code _SenseLevel increases and increases continuously.Obtain m digit numeric code Q _M-1... Q ₂Q ₁Q ₀Only possibly be a group (00 in the following data ... 000,10 ... 000,11 ... 000 ..., 11 ... 100,111 ... 10,1111 ... 1), and along with the increase of load current, digital code changes according to above-mentioned order successively.Such temperature digital sign indicating number need not be encoded again and can directly passed through to add in the drive circuit.Because digital code is from Q _M-1To Q ₀The size of the switching tube of being controlled increases gradually, promptly can realize the adjustment of non-homogeneous grid width.

As shown in Figure 2, switching tube grid width control module is through digital code Q ₀～Q _M-1Be connected with switching tube grid width control module, through mode select signal S _ModeBe connected with mode selection module, through grid width numeral option code D ₀～D _M-1, D _ExtraBe connected with power stage main circuit and driver module thereof.Its major function is according to the different working pattern digital code of delay line ADC output to be adjusted, and is loaded into the adjustment that realizes non-homogeneous grid width in power stage main circuit and the drive circuit module thereof.

As shown in Figure 2, said Dead Time control module is through mode select signal S _ModeBe connected with mode selection module, be connected with the PWM control module through the PWM_C signal, through V _SenseSignal is connected with the interior current detection module of sheet, and (PWM_P2 PWM_N2) is connected with power stage main circuit and driver module thereof for PWM_P1, PWM_N1 through four road pwm signals.Dead Time control module major function is to avoid PMOS and the conducting simultaneously of nmos switch pipe to cause break-through with adding Dead Time in the control signal that offers PMOS switching tube and nmos switch pipe switch.This module through the PWM_C signal is added Dead Time obtain four road pwm signals (PWM_P1, PWM_N1, PWM_P2, PWM_N2), simultaneously, the size of Dead Time is by V _SenseSignal Regulation realizes that Dead Time reduces with load current and the adaptive control that increases, to reduce the loss of switching tube.In the technique scheme, power stage main circuit and driver module thereof are specifically realized M among Fig. 2 _N1, M _P1, M _N2And M _P2Four non-homogeneous adjustment of switch MOS pipe are shown in Fig. 3 (a-d).For the load current excursion is (I _Min, I _Max) four switch Buck-Boost converters, with four switching tube M _N1, M _P1, M _N2And M _P2Size be divided into not isometric m+1 section.Any one section paralleling MOS Guan Douyou independently is connected with its grid with the drive circuit that its size adapts, and is used for driving switch pipe operate as normal.Form four switching tube M _N1, M _P1, M _N2And M _P2The drive circuit input of paralleling switch pipe is exported signal PWM_N1 with above-mentioned Dead Time control module respectively, PWM_P1, and PWM_N2, PWM_P2 is connected.Simultaneously, the drive circuit of every section paralleling MOS pipe all has the enable port, respectively with grid width numeral option code D ₀～D _M-1, D _ExtraBe connected.With switching tube M among Fig. 3 (a) _P1Be example, it is by M _P1(0) to M _P1(m) m+1 not isometric metal-oxide-semiconductor is formed in parallel altogether, wherein, and a metal-oxide-semiconductor M _P1(m) but remain operating state, and other metal-oxide-semiconductor M _P1(0)-M _P1(m-1) respectively by pairing digital code D ₀-D _M-1Control its operating state, digital code enables to control single paralleling MOS pipe through driving and is in work or off state.If certain section digital code D that metal-oxide-semiconductor is corresponding _i(i=0,1 ..., m-1)) and be " 1 ", represent that corresponding pwm signal can be loaded on this section power MOS pipe through drive circuit, promptly this section power MOS pipe plays effective switching tube effect; If digital code is " 0 ", represent that then this section metal-oxide-semiconductor keeps off state at this moment, does not belong to the live part of switching tube.Switching tube M _N1, M _N2Control and realization grid width selection mode and M _P1Manage identical.And metal-oxide-semiconductor M _P2Situation slightly different, except above-mentioned m+1 section paralleling MOS pipe, also need add a tube-carrier that only is used to be operated in the Buck pattern, like the M among Fig. 3 (c) _P2(extra) pipe.Its connected mode and M _P2Other section parallel transistor is the same, just drives Enable Pin and D _ExtraBe connected.The adding of tube-carrier guarantees the pattern at Buck, even under the less situation of output voltage, and M _P2Still have lower conducting resistance, and under Boost and Buck-Boost pattern, tube-carrier turn-offs fully, does not participate in M _P2Grid width adjustment.

A kind of aforesaid non-homogeneous grid width optimization method obtains to form power switch pipe M based on above-mentioned four switch Buck-Boost changer systems _N1, M _P1, M _N2And M _P2The number m+1 and the size W separately of paralleling switch pipe _M(0)-W _M(m).Because M _N1, M _P1, M _N2And M _P2Four switching tube designs are just the same, describe for ease, replace with the M pipe here.If total grid width size of M pipe is W _Total, form (as shown in Figure 3) by m+1 not isometric paralleling MOS pipe M (0)-M (m), correspondingly-sized is respectively W _M(0)-W _M(m).As previously mentioned, control M pipe adjustment grid width size be digital code, from M (0) to M (m) along with load current turn-offs successively, can realize W successively _Opt(0) to W _Opt(m) be total to m+1 kind grid width size.If I _Out(0) to I _Out(m) be the corresponding W of difference _Opt(0) to W _Opt(m) different size reaches the optimal load electric current of peak efficiency.

Based on integral body as shown in Figure 2 four switch Buck-Boost changer systems and required load current excursion (I _Min, I _Max) design paralleling switch pipe number m+1 and size W _M(0)-W _M(m), thus the concrete grammar of realizing the adjustment of non-homogeneous grid width is following:

1) it is as shown in Figure 2 to build four switch Buck-Boost changer system frameworks according to above-mentioned connected mode.

2) according to required maximum load current I _MaxConfirm like power switch pipe M among Fig. 2 _N1, M _P1, M _N2And M _P2Overall size.Based on 1) system of being built, the function of temporary transient turn-off delay line ADC and switching tube grid width control module, selection and definite I _Out(0) ≈ 0.6 * I _MaxThe time corresponding optimal size W _Opt(0), the overall size W of switch MOS pipe just _Total

3) according to overall size W _TotalWith required minimum load current I _Min, adopt grid width changing value Δ W=0.5 * W _OptMethod confirm the size of M (0)-M (m), realize non-homogeneous variable grid width adjustment.Can confirm to remove the size W of the outer remaining switching tube of M (0) pipe when confirming the size of switching tube M (0) _Opt(1),

$W_M\left(0\right)=W_{\mathrm{opt}}\left(1\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2}---\left(11\right)$

W _OptLoad current I when (1) reaching its efficient and reach optimal value _Out(1),

$I_{\mathrm{out}}\left(1\right)=\frac{I_{\mathrm{out}}\left(0\right)}{2}---\left(12\right)$

Derive according to above-mentioned method successively and can obtain the size M (i-1) of each pipe of required parallel connection, i=1,2,3 ...,, make I up to i=m _Out(m+1)＜I _Min, confirm the number m+1 of paralleling switch pipe.

$W_M(i-1)=W_{\mathrm{opt}}\left(i\right)=\frac{W_{\mathrm{opt}}(i-1)}{2}=\frac{W_{\mathrm{opt}}\left(0\right)}{2^i},i=\mathrm{1,2,3},\&CenterDot;,m---\left(13\right)$

And rest parts promptly is the size of last switching tube M (m),

$W_M\left(m\right)=W_{\mathrm{opt}}\left(m\right)=W_{\mathrm{opt}}\left(0\right)-\underset{i=1}{\overset{m-1}{\mathrm{\&Sigma;}}}{W}_M\left(i\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2^m}---\left(14\right)$

W _Opt(0) to W _Opt(m) be the different grid width size that can realize through digital code in the converter, they are when efficient reaches optimum separately, the corresponding load electric current I _Out(0) to I _Out(m) can calculate as follows:

$I_{\mathrm{out}}\left(n\right)=\frac{I_{\mathrm{out}}(n-1)}{2}=\frac{I_{\mathrm{out}}\left(0\right)}{2^m}---\left(15\right)$

4), confirm each grid width value corresponding load current range of non-uniform change so that the conversion efficiency of converter keeps optimum according to the size of each paralleling switch pipe that designs in structure as shown in Figure 3 and the step 3).Turn-off the function of the delay line ADC module in the changer system as shown in Figure 2, directly add the corresponding different grid width of different digital codes, obtain the efficiency curve that changes with load current to switching tube grid width control module.In all load current range, choose the digital code when reaching most effective, just confirmed each grid width value corresponding load current range.

5) control lag line ADC output digital code and detection voltage V _SenseVariation relation, make that the load current range that calculates in ADC output digital code and the step 4) is corresponding.

6) accomplish the integrated circuit design.Circuit can be according to realizing the self adaptation adjustment of grid width with the load current non-uniform change as shown in Figure 2.

The foregoing invention method itself does not rely on the technology of concrete realization.The present invention is with an output load current scope (10mA; DC-DC converter 650mA) is an example; Utilize above method to design, obtaining needing the switch MOS pipe number m+1 of parallel connection is 6, and each ratio (normalization equivalence grid width) of managing size and overall size also can correspondingly obtain like table 2.Table 2 has provided the corresponding grid width of different grid width sizes that finally obtains simultaneously and has selected digital code, and corresponding load current range under each size.Wherein, because M _P2Tube-carrier do not participate in M _P2Grid width adjustment, therefore be not counted in its total grid width.

Corresponding grid width is selected digital code and normalization equivalence grid width in the table 2 different loads current range

The Buck-Boost converter of the non-homogeneous grid width changing method of the high frequency that the present invention adopts the present invention to design; Adopt the design of CSMC0.5 μ m CMOS technology library; The whole circuit of realization except that the passive filtering device is integrated, and external filter inductance is 2.2 μ H, and filter capacitor is 1 μ F.According to the requirement to input and output voltage, converter can be operated under Buck, Buck-Boost, three kinds of patterns of Boost, input voltage range 2.5V-4.2V, output voltage range 1.5V-5V, operating frequency 5MHz.In whole load current scope 10mA-650mA, adopt grid width modulator approach heterogeneous, circuit working is simulation curve such as Fig. 4 under Buck and Boost pattern.Fig. 4 (a) changes simulation curve for conversion efficiency under the Buck pattern with load current; Fig. 4 (b) changes simulation curve for conversion efficiency under the Boost pattern with load current.In the time of can seeing converter 5MHz high-frequency work, medium load and heavy duty efficient remain at more than 90%, and light-load efficiency has obtained significantly improving, and especially for the Boost conversion, efficient can reach more than 84% when load current was 10mA.

This invention draws under the different loads condition through the power consumption analysis to the HF switch converter, the rule and the method for the modulation of switching tube grid width, thus make total solution.Realized significantly improving to light-load efficiency.Owing to just changed the grid width of switching tube, do not take the controlling unit of extra switch plumber working frequency, fundamentally eliminated the adverse consequences that VFC is brought.

Above content is to combine concrete preferred implementation to further explain that the present invention did; Can not assert that embodiment of the present invention only limits to this; Those of ordinary skill for technical field under the present invention; Under the prerequisite that does not break away from the present invention's design, can also make some simple deduction or replace, all should be regarded as belonging to the present invention and confirm scope of patent protection by claims of being submitted to.

Claims (3)

1. a method that improves the light load efficiency of integrated switch DC-DC converter non-uniform variation grid width is characterized in that, according to following steps:

(a) build four switch Buck-Boost changer systems;

(b) according to required maximum load current I _MaxConfirm power switch pipe M _N1, M _P1, M _N2And M _P2Overall size; Confirm I _Out(0) ≈ 0.6 * I _MaxThe time corresponding optimal size W _Opt(0), the overall size W of power switch pipe just _Total

(c) according to overall size W _TotalWith required minimum load current I _Min, the method for employing non-uniform variation grid width is confirmed the number m+1 and the size W of each paralleling switch pipe of composition power switch pipe _M(0) to W _M(m);

(d), confirm each grid width value corresponding load current range of non-uniform change so that the conversion efficiency of converter keeps optimum according to the size of each paralleling switch pipe that designs in the step (c); The function of turn-off delay line ADC directly adds the corresponding different grid width of different digital codes to switching tube grid width control module, obtains the efficiency curve that changes with load current; In all load current range, choose the digital code when reaching most effective, just confirmed each grid width value corresponding load current range;

(e) control lag line ADC output digital code and detection voltage V _SenseVariation relation, make that the load current range that calculates in ADC output digital code and the step (d) is corresponding;

(f) realize the self adaptation adjustment of grid width with the load current non-uniform change.

2. a kind of according to claim 1 method that improves the light load efficiency of integrated switch DC-DC converter non-uniform variation grid width is characterized in that, the structure of four switch converters systems is following in the said step (a):

This DC-DC converter is a wide load current range converter efficiently, comprising: power stage main circuit and driver module thereof, PWM control module; Mode selection module; Current detection module in the sheet, delay line ADC module, switching tube grid width control module and Dead Time control module;

Said power stage main circuit and driver module thereof are realized the power delivery function of Buck, the different voltage alternative types with Buck-Boost of Boost; Said power stage main circuit and driver module thereof and other module connected mode are following: V _InBe connected V with power input _OutBe connected with power output end, GND links to each other with external ground; V _xAnd V _yBe connected V with external inductance L two ends _xSimultaneously with sheet in current detection module link to each other; Through grid width numeral option code D ₀～D _M-1, D _ExtraBe connected with switching tube grid width control module; PWM_P1, PWM_N1, PWM_P2 and PWM_N2 four road pwm signals through receiving the output of Dead Time control module are connected its inner M with it _N1, M _N2, M _P1And M _P2Four are adopted the basic connected mode of the adjustable power switch pipe of grid width following: M _P1The source is leaked and is connected on V respectively _InAnd V _x, M _N1The source is leaked and is connected on GND and V respectively _x, M _P2The source is leaked and is connected on V respectively _OutAnd V _y, M _N2The source is leaked and is connected on GND and V respectively _ySaid M _N1And M _N2Be NMOS, M _P1And M _P2Be PMOS, its basic control mode is divided into following three kinds: if M _N2Keep turn-offing M _P2Keep conducting, by PWM control M _N1And M _P1, converter works in the Buck pattern; If M _N1Keep turn-offing M _P1Keep conducting, M _N2And M _P2By PWM control, converter works in the Boost pattern; If 4 power switch pipes then work in the Buck-Boost pattern simultaneously by pwm signal control; Because the power switch pipe size is very big, the chain of inverters that each switching tube all needs to amplify step by step is as driving;

PWM control module input and output feedback signal bV _OutBe connected signal bV _OutPass through resistance R by the converter output voltage _B1And R _B2Dividing potential drop obtains; Output is connected with the Dead Time control module through the PWM_C signal;

Mode selection module V _InPort is connected with power input, V _OutPort links to each other with the converter voltage output, through S _ModeSignal is connected with switching tube grid width control module with the Dead Time control module respectively; Mode selection module is according to input voltage V _InAnd output voltage V _OutThe comparison acquisition model select signal S _Mode, realize the selection of system works pattern;

Current detection module V in the sheet _InPort is connected with power input, V _xPort is connected with power stage main circuit and driver module thereof, through V _SenseSignal is connected with the Dead Time control module with delay line ADC module; Signal V _SenseWith load current I _OutTherefore linear change, the situation of change of reaction load accurately; V _SenseSignal offers delay line ADC module and Dead Time control module simultaneously;

Said delay line ADC module, input port passes through V _SenseSignal is connected with the interior current detection module of sheet; Delay line ADC module is with signal V _SenseConvert m megadyne temperature number of degrees character code Q to ₀～Q _M-1, and be connected with switching tube grid width control module through digital code;

Said switching tube grid width control module is through digital code Q ₀～Q _M-1Be connected with delay line ADC module, through mode select signal S _ModeBe connected with mode selection module, through grid width numeral option code D ₀～D _M-1, D _ExtraBe connected with power stage main circuit and driver module thereof;

Said Dead Time control module is through mode select signal S _ModeBe connected with mode selection module, be connected with the PWM control module through the PWM_C signal, through V _SenseSignal is connected with the interior current detection module of sheet, is connected with power stage main circuit and driver module thereof through PWM_P1, PWM_N1, PWM_P2, PWM_N2 four road pwm signals;

Power stage main circuit and driver module thereof are realized M through the switching tube parallel connection of different size _N1, M _P1, M _N2And M _P2Four non-homogeneous adjustment of power switch pipe; For the load current excursion is (I _Min, I _Max) four switch Buck-Boost changer systems, through non-homogeneous grid width optimization method with four switching tube M _N1, M _P1, M _N2And M _P2Size be divided into not isometric m+1 section; Any one section paralleling MOS Guan Douyou independently is connected driving switch pipe normally and shutoff with the drive circuit that its size adapts with its grid; Form four switching tube M _N1, M _P1, M _N2And M _P2The drive circuit input of paralleling switch pipe is exported signal PWM_N1 with above-mentioned Dead Time control module respectively, PWM_P1, and PWM_N2, PWM_P2 is connected; Simultaneously, the drive circuit of every section paralleling MOS pipe all has the enable port, respectively with grid width numeral option code D ₀～D _M-1, D _ExtraBe connected.

3. a kind of according to claim 1 method that improves the light load efficiency of integrated switch DC-DC converter non-uniform variation grid width is characterized in that, said step (c) is meant:

According to overall size W _TotalWith required minimum load current I _Min, adopt grid width changing value Δ W=0.5 * W _OptMethod confirm the size of each paralleling switch pipe M (0)-M (m), realize non-homogeneous variable grid width adjustment;

Confirm the size W of switching tube M (0) _M(0) confirms to remove the size W of the outer remaining switching tube of M (0) pipe the time _Opt(1),

$W_M\left(0\right)=W_{\mathrm{opt}}\left(1\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2}---\left(11\right);$

W _OptLoad current I when (1) reaching its efficient and reach optimal value _Out(1),

$I_{\mathrm{out}}\left(1\right)=\frac{I_{\mathrm{out}}\left(0\right)}{2}---\left(12\right);$

Derive according to above-mentioned method successively and can obtain the size M (i-1) of each pipe of required parallel connection, i=1,2,3 ...,, make I up to i=m _Out(m+1)<i _Min, confirm the number m+1 of paralleling switch pipe;

$W_M(i-1)=W_{\mathrm{opt}}\left(i\right)=\frac{W_{\mathrm{opt}}(i-1)}{2}=\frac{W_{\mathrm{opt}}\left(0\right)}{2^i},i=\mathrm{1,2,3},\&CenterDot;\&CenterDot;\&CenterDot;,m---\left(13\right);$

And rest parts promptly is the size of last switching tube M (m),

$W_M\left(m\right)=W_{\mathrm{opt}}\left(m\right)=W_{\mathrm{opt}}\left(0\right)-\underset{i=1}{\overset{m-1}{\mathrm{\&Sigma;}}}{W}_M\left(i\right)=\frac{W_{\mathrm{opt}}\left(0\right)}{2^m}---\left(14\right);$

W _Opt(0) to W _Opt(m) be the different grid width size that can realize through digital code in the converter, they are when efficient reaches optimum separately, the corresponding load electric current I _Out(0) to I _Out(m) can calculate as follows:

$I_{\mathrm{out}}\left(n\right)=\frac{I_{\mathrm{out}}(n-1)}{2}=\frac{I_{\mathrm{out}}\left(0\right)}{2^m}---\left(15\right).$

Images