WO2011150732A1 - Modulator and design method thereof - Google Patents

Abstract

A modulator and a design method thereof are provided. The modulator includes an analog filter (11) for pre-modulating input analog signals; an analog adder (15) for summing the analog signals; a multi-bit quantizer (13) for converting the sum of the analog signals into multi-bit digital signals output; a digital loop filter (12) for filtering the multi-bit digital signals output from the multi-bit quantizer (13); and a multi-bit feedback digital-to-analog converter (14) for converting the filtered multi-bit digital signals to feedback analog signals. Steady modulators may be constructed rapidly according to this solution. The problem of non-linearity of the multi-bit feedback digital-to-analog converter can be solved by adopting the digital loop filter to shape the quantization noise and the mismatch noise simultaneously. The modulator has the stability and the ability of noise shaping, and can be used in high speed and high precision digital-to-analog conversion.

Description

 Modulator and design method thereof

Technical field

 This invention relates to analog to digital conversion, and more particularly to a modulator and method of designing same. Background technique

 Since the Delta-Sigma (Δ_∑) modulator requires oversampling during operation, the Delta-Sigma modulator-based ADC (modular-to-digital converter) is mainly used in low-speed, high-precision fields such as audio systems and instruments. Equipment, etc. As communication systems move toward broadband, the demand for high-speed, high-precision ADCs is increasing. To achieve high-speed, high-precision performance, the Delt book a-Sigma modulator must reduce the oversampling rate. At the same time, it is necessary to increase the order of the loop filter or increase the number of bits of the quantizer to compensate for the loss of signal-to-noise ratio.

 Although increasing the order of the loop filter or increasing the number of bits of the quantizer is beneficial to improve the performance of the modulator, under the condition that the quantizer has a bit number of 1, the loop filter of more than 2 orders is very It is difficult to ensure the stability of the system; increasing the number of quantizers not only reduces the quantization noise, but also enhances the stability of the system. So multi-bit quantization is the development trend of high performance Delta-Sigma modulators.

 Multi-bit Delta-Sigma modulators have a very demanding requirement: The accuracy of a multi-bit feedback DAC (digital-to-analog converter) needs to meet or exceed the accuracy of the entire modulator, but existing process manufacturing techniques are difficult to make multi-bit feedback DACs Achieve high-precision indicators (such as 16-bit or higher). This is due to the fact that the matching error (or mismatch) between the basic unit components causes the nonlinearity of the multi-bit feedback DAC, which causes the modulator to produce harmonic distortion, which appears as a sharp harmonic in the spectrum of the modulator output signal. Glitch, in-band harmonic glitch can cause the modulator signal-to-noise ratio to be greatly reduced, so the main problem with multi-bit Delta-Sigma modulators is the nonlinearity of the feedback DAC.

 For the nonlinear problem of multi-bit feedback DAC, the existing solutions include: laser correction or analog correction, digital correction, dual quantizer structure, DEM (Dynamic Element Matching) technology, PWM (Pulse WidthModulation, pulse width) Modulation) technology and so on. The most mainstream solutions to date have three:

 First, the background number is corrected. Background digital correction can eliminate the nonlinear effects of DAC, but the circuit structure is very complicated.

Second, the cascaded dual quantizer structure (ie, the dual quantizer MASH structure). Cascaded dual quantizer The structure generally uses a 1-bit feedback DAC for the first few stages and a multi-bit feedback DAC for the final stage. This structure can shape the mismatch noise of the multi-bit feedback DAC, but the shaping order of the mismatch noise is at least lower than the shaping order of the quantization noise. The first order, and the interstage coupling coefficient amplifies the quantization noise and mismatch noise, which all lead to a decrease in modulator performance. In addition, this structure places high demands on analog circuits in order to avoid excessive noise leakage.

 Third, DEM technology. According to the different order of the feedback DAC mismatch noise shaping, DEM technology can be divided into two categories, first, low-order mismatch noise shaping DEM technology (including 0th order and 1st order); second, high-order mismatch noise shaping DEM technology (including 2nd order and higher order). Low-order mismatched noise shaping DEM technology can weaken the nonlinear effects of multi-bit feedback DACs to varying degrees, but they have high noise floor, large in-band harmonic spurs, and DAC noise and harmonic distortion with input signal amplitude fluctuations. Problem; High-order mismatched noise shaping DEM technology can better attenuate the nonlinear effects of multi-bit feedback DACs, but they are not only complicated in circuit structure, but also unstable.

 In short, there is still room for improvement in how to better solve the nonlinear problem of multi-bit feedback DACs. Summary of the invention

 The technical problem to be solved by the present invention is to provide a modulator and a design method thereof, which can better solve the nonlinear problem of the multi-bit feedback DAC.

 In order to solve the above technical problems, the present invention adopts the following technical solutions:

 A modulator for analog to digital conversion, comprising an analog filter, an analog adder, a multi-bit quantizer, a digital loop filter, a multi-bit feedback digital-to-analog converter, the analog filter is used to input an analog signal Premodulating and outputting to an input of the analog adder, the analog adder is used for summing analog signals and outputting to the multi-bit quantizer, the multi-bit quantizer is used to convert the analog signal into a multi-bit digital signal output; the digital loop filter is configured to filter a multi-bit digital signal output by the multi-bit quantizer and output to the multi-bit feedback digital-to-analog converter, wherein the multi-bit feedback digital-to-analog converter is used The filtered multi-bit digital signal is converted to a feedback analog signal and output to the other input of the analog adder.

 In an embodiment of the present invention, the modulator further includes at least one shunt feedback branch, and each of the shunt feedback branch inputs is connected to one of the digital loop filters, and outputs The terminal is connected to a corresponding shunt signal feeding node of the analog filter, and each of the shunt feedback branches is provided with a signal shunt, a shunt feedback digital-to-analog converter and an adder from the input end to the output end.

f too *昍—^? s: ^Αΐ, . 3⁄4 W 々軎古 S夂frfr into the right county 1-bit digital signal or multi-bit digital signal.

 In one embodiment of the invention, the modulator is a Delta-Sigma modulator.

 The present invention also provides an analog to digital converter comprising the modulator described above.

 The present invention also provides a method of constructing a modulator for constructing a second modulator from a first modulator, the first modulator comprising an analog loop filter, a multi-bit quantizer, and a multi-bit feedback digital-to-analog conversion An input end of the analog loop filter receives an analog signal input, and an output end is connected to an input end of the multi-bit quantizer, and an output end of the multi-bit quantizer outputs a multi-bit digital signal, and is connected to the An input end of the multi-bit feedback digital-to-analog converter, the output end of the multi-bit feedback digital-to-analog converter is connected to another input end of the analog loop filter, and the constructing method comprises the following steps: The path filter is decomposed into a first analog filter and a second analog filter, and an analog adder is added, and the device connection is adjusted as follows: two input ends of the analog adder are respectively connected to the first analog filter and the first An output end of the second analog filter, the output end is connected to the input end of the multi-bit quantizer, the first analog filter accepts an analog signal input, and the second analog filter Into multi-bit output terminal connected to the terminal of the feedback DAC;

 Adjusting the second analog filter to a digital loop filter, adjusting the device connection as follows to construct a second modulator: the input of the digital loop filter is connected to the output of the multi-bit quantizer, and the output The terminal is connected to an input end of the multi-bit feedback digital-to-analog converter, and an output end of the multi-bit feedback digital-to-analog converter is connected to an input end of the adder.

 The invention also provides a signal splitting method for a modulator, the modulator being used for analog-to-digital conversion, including an analog filter, an analog adder, a multi-bit quantizer, a digital loop filter, a multi-bit feedback digital-to-analog converter, The analog filter is configured to pre-modulate an input analog signal and output to an input of the analog adder, the analog adder is used for summing analog signals, and outputting to the multi-bit quantizer, a multi-bit quantizer for converting an analog signal to a multi-bit digital signal output; the digital loop filter for filtering a multi-bit digital signal output by the multi-bit quantizer and outputting to the multi-bit feedback digital-to-analog converter The multi-bit feedback digital-to-analog converter is configured to convert the filtered multi-bit digital signal into a feedback analog signal and output to another input end of the analog adder; the signal shunting method includes the following steps:

 Determining at least one signal shunt node of the digital loop filter;

 And providing at least one shunt feedback branch, wherein the input end of each of the shunt feedback branches is connected to one of the signal shunt nodes, and a signal shunt, a shunt feedback digital-to-analog converter, and an adder are sequentially disposed from the input end;

Inserting an adder of each of the shunt feedback branches into a pair of the analog filters Through the modulator structure provided by the invention, simultaneous shaping of quantization noise and mismatch noise can be realized, and the nonlinear problem of the multi-bit feedback DAC in the modulator can be effectively solved. DRAWINGS

 1 is a diagram showing the original structure of a digital noise shaping multi-bit Delta-Sigma modulator according to an embodiment of the present invention;

 Figure 2 is a block diagram showing the structure of a general-purpose conventional Delta-Sigma modulator;

 3 is a block diagram showing the structure of a conventional conventional Delta-Sigma modulator equivalent to FIG. 2; FIG. 4 is a digital noise shaping multi-bit including an accumulation unit according to an embodiment of the present invention;

Original structure diagram of Delta-Si gma modulator;

 5 to 7 are digital noise shaping multi-bits including accumulation units according to an embodiment of the present invention

Schematic diagram of the internal signal shunting process of the De lta-Si gma modulator;

 8 is a structural diagram of a digital noise shaping multi-bit De 11 a-S i gma modulator with accumulation unit optimized by signal shunting according to an embodiment of the present invention;

 Figure 9 is a block diagram of a conventional fourth-order multi-bit Delta-Sigma modulator;

 Figure 10 is a structural diagram of a conventional fourth-order low-pass multi-bit Delta-Sigma modulator equivalent to Figure 9;

 11 is a diagram showing an original structure of a fourth-order low-pass digital noise shaping multi-bit De 11 a-S i gma modulator according to an embodiment of the present invention;

 12 to 14 are schematic diagrams showing the internal signal shunting process of a fourth-order low-pass digital noise shaping multi-bit De lta-Si gma modulator according to an embodiment of the present invention;

 15 is a structural diagram of a fourth-order low-pass digital noise shaping multi-bit De 1 ta-Sigma modulator optimized by signal shunting according to an embodiment of the present invention;

 Figure 16 is the spectrum when the three modulator output signals SNDR reach the maximum;

 Figure 17 is a plot of the three modulator output signals, SNDR, versus the input signal amplitude. detailed description

 The present invention will be further described in detail below with reference to the accompanying drawings.

 The present invention relates generally to analog-to-digital conversion, and more particularly to a Delta-Sigma analog-to-digital converter, and in particular to a core component thereof, a multi-bit Delta-Sigma modulator, for the purpose of solving multiple bits.

Nonlinear problems with multi-bit feedback DACs in De lta-Si gma modulators.

Delta-Sigma analog-to-digital converters are used in a wide range of applications, such as: security, audio, Such as audio systems, equipment and so on. The Delta-Sigma modulator is a core component of the Delta-Sigma analog-to-digital converter. The De 1 ta-S i gma analog-to-digital converter based on the multi-bit De 11 a- S i gma modulator of the present invention can be applied to high speed. , high precision analog to digital conversion.

 The specific embodiments of the present invention are described below. Since the basic inventive concept of the present invention is to implement noise shaping (including quantization noise, feedback DAC mismatch noise) by using a digital loop filter, the modulator is called digital noise shaping. Bit Delta-Sigma modulator, and will introduce the modulator mainly from three aspects, including: digital noise shaping multi-bit Delta-Sigma modulator, constructing digital noise shaping multi-bit Del from traditional Delta-Si gma modulator The ta-S i gma modulator method and the digital noise shaping method of the multi-bit Delta-Sigma modulator internal signal shunt.

In a first aspect, the present invention provides a digital noise shaping multi-bit Delta-Sigma modulator, which is a universally-used serialized multi-bit De 11 aS i gma modulator, which can be implemented by a user by designing a filter as needed. A series of specific performance signal transfer functions (STF) and noise transfer functions (NTF). The original structure of the digital noise shaping multi-bit Delta-Sigma modulator is shown in FIG. 1. It mainly includes an analog filter 11, a digital loop filter 12, a multi-bit quantizer 13, a multi-bit feedback DAC 14, and an analog adder 15. Where U(z) represents the input analog signal of the modulator, V(z) represents the output digital signal of the modulator, Y(z) represents the input analog signal of the multi-bit quantizer 13, and E _Q (z) represents multi-bit quantization The quantization noise of the device 13, E _D (z), represents the mismatch noise of the multi-bit feedback DAC 14. The analog filter 11 is used to pre-modulate the input analog signal U(z) to L. *U(z) signal, digital loop filter 12 is used to shape quantization noise E _Q (z), mismatch noise E _D (z), and pre-modulation signal L. * U (z), the multi-bit quantizer 13 is used to convert the analog signal Y (z) into a corresponding multi-bit digital signal, and the multi-bit feedback DAC 14 is used to convert the multi-bit digital signal output from the digital loop filter 12 A corresponding analog signal (to distinguish the aforementioned input analog signal, referred to as a feedback analog signal), the analog adder 15 is used for summing the analog filter 11 and the multi-bit feedback DAC 14 output signal (feedback analog signal) (analog signal sum That is, the input analog signal Y(z) of the multi-bit quantizer 13.

When the modulator is operating normally, first, the input analog signal U(z) is pre-modulated to L by the analog filter 11. *U(z) signal, then, L. The *U(z) signal, the quantization noise E _Q (z), and the mismatch noise E _D (z) are injected together into a feedback loop including the digital loop filter 12, and finally, these signals are all subjected to the digital loop filter 12 After shaping, it is output from the modulator. According to Figure 1, after derivation:

V(z)=-^-U(z) + -^—E _Q (z)--^—E _D (z) ( 1 )

 l + ^ l + Z^ l + ^

(where the signal transfer function is: STF i

1 + L, The quantization noise transfer function is: NTF ₀

l + Z^ The mismatch noise transfer function is: NTF _D = - one -)

 1 +

As can be seen from equation (1), a series of specific performances of STF and NTF (such as LP (low pass), BP (band pass), HP (high pass), etc.) can be realized by designing appropriate filters 11 and 12, and Equation 1) also shows that the quantization noise E _Q (z) and the mismatch noise E _D (z) can be shaped at the same time, for example: Take = ₍ -)^ _{= (} - _T .i , substituting them into (1) :

 ΰ Z-1 Z-l

V(z) = U(z) + (lZ- ¹ ) "E _e (z)-(lZ- ¹ )"E _D (z) ( 2 )

Equation (2) shows that the quantization noise E _Q (z) of the multi-bit quantizer 13 and the mismatch noise E _D (z) of the multi-bit feedback DAC 14 are simultaneously subjected to n-order noise shaping, and therefore, digital noise shaping multi-bit Delta-Sigma The modulator can effectively solve the nonlinear problem of multi-bit Delta-Sigma modulator feedback DAC.

 In summary, the modulator structure and signal flow direction of the present invention are: The analog filter 11 pre-modulates the input analog signal U(z) and outputs it to an input terminal of the analog adder 15, and the analog adder 15 performs an analog signal request. And, the analog signal and Y (z) are obtained and output to the multi-bit quantizer 13, and the multi-bit quantizer 13 converts the analog signal and Y (z) into a multi-bit digital signal V (z) output; the digital loop filter 12 The multi-bit digital signal V(z) outputted by the multi-bit quantizer 13 is filtered and output to the multi-bit feedback DAC 14, and the multi-bit feedback DAC 14 converts the filtered multi-bit digital signal into a feedback analog signal and outputs it to the analog adder. The other input of 15.

 In a second aspect, in order to construct a digital noise shaping multi-bit Delta-Sigma modulator as shown in FIG. 1 more quickly, the present invention proposes a digital noise shaping multi-bit Delta-Sigma modulator constructed from a conventional Delta-Sigma modulator. method. The basic idea of the method is: From the traditional Delta-Sigma modulator, the transfer function of the feedback loop is separated from the analog loop filter, and the separated transfer function is transferred to the digital circuit side, and This transfer function is implemented using digital circuitry. The basic principle of the method is: In the feedback loop of the modulator, the injection node of the multi-bit feedback DAC mismatch noise is transferred to the equivalent injection node of the quantization noise. The specific implementation steps of the construction method are:

 i) the first step, separate transfer function

The general structure of the conventional multi-bit Delta-Sigma modulator is shown in Figure 2. It consists of an analog loop filter 21, a multi-bit quantizer 22, and a multi-bit feedback DAC23, where U(z) represents the input analog of the modulator. Signal, W(z) represents the output analog signal of the multi-bit feedback DAC23, V(z) table The output digital signal of the modulator is shown, Y(z) represents the input analog signal of the multi-bit quantizer 22, E _Q (z) represents the quantization noise of the multi-bit quantizer 22, and E _D (z) represents the loss of the multi-bit feedback DAC 23 With noise, and Y (z) and U (z), W (z) have the following relationship:

F(z) = L ₀ [/(z)-Z ₁ W(z) ( 3)

 According to Figure 2, after derivation:

V(z)=^U(z) + -^—E _Q (z)--^-E _D (z) (4)

l + ^ l + Z^ l + ^ From (4), it is known that this modulator cannot normally shape the mismatch noise E _D (z) of the multi-bit feedback DAC23. According to the formula (3), two transfer functions L in the analog loop filter 21 shown in Fig. 2 are used. Separation, resulting in a modulator structure as shown in FIG.

 Ii) the second step, transfer transfer function

 The filter 32 shown in FIG. 3 is transferred to the side of the digital circuit, and a digital noise shaping multi-bit Delta-Sigma modulator as shown in FIG. 1 is obtained (Note: in the process of implementing the modulator design, in order to save power Consumption and area, some types of digital loop filter 12 may require equivalent conversion).

It can be seen that the method can quickly construct a digital noise shaping multi-bit Delta-Sigma modulator. It can be observed from observation that the only change in this construction process is that the position of the filter 32 and the multi-bit feedback DAC 34 are mutually adjusted, which inevitably causes the mismatch noise transfer function NTF _D to change. According to the above, the digital noise shaping multi-bit Delta-Sigma modulator input and output relationship is:

V(z)=-^U(z)+-^—E _Q (z)--^—E _D (z) ( 5 )

 l + ^ l + Z^ l + ^

Comparison (4) (5) Two types show that the digital noise shaping multi-bit Delta-Sigma modulator signal transfer function (STF), quantization noise transfer function (NTF ₀ = ) and traditional tuning

l + Z^ ^Q l + Z^ The controller remains consistent, but the mismatch noise transfer function is from NrF _D = (Note)

The noise is random, the sign can be ignored, and is equal to the quantization noise transfer function. It can be seen that the digital noise shaping multi-bit Delta-Sigma modulator not only inherits the shaping function of the quantization noise of the traditional modulator, but also increases the shaping function of the mismatch noise of the multi-bit feedback DAC. Therefore, the construction method can be traditionally The modulator effectively constructs a digital noise shaping multi-bit De 11 aS i gma modulator. In summary, the above construction method is to construct the modulator (second modulator) of the present invention from a conventional modulator (first modulator), the first modulator including the analog loop filter 21, multi-bit quantization The multi-bit feedback DAC 23, an input end of the analog loop filter 21 receives an analog signal input, the output end is connected to the input end of the multi-bit quantizer 22, and the output end of the multi-bit quantizer 22 outputs a multi-bit digital signal, and An input end of the multi-bit feedback DAC 23 is connected, and an output end of the multi-bit feedback DAC 23 is connected to the other input end of the analog loop filter 21, and the construction method comprises the following steps:

 The analog loop filter 21 is decomposed into a first analog filter 31 and a second analog filter 32, and an analog adder 35 is added to adjust the device connection as follows: The two inputs of the analog adder 35 are respectively connected to the first analog The output ends of the filter 31 and the second analog filter 32 are connected to the input end of the multi-bit quantizer 33, the first analog filter 31 receives the analog signal input, and the input of the second analog filter 32 is connected to the multi-bit feedback. An output of the digital to analog converter 34;

 The second analog filter 32 is adjusted to a digital loop filter, and the device connection is adjusted as follows to construct a second modulator: the input of the digital loop filter is connected to the output of the multi-bit quantizer 33, and the output is connected. An input of the multi-bit feedback digital-to-analog converter 34 is coupled to an input of the adder. That is, the second analog filter 32 and the multi-bit feedback digital-to-analog converter 34 are mutually aligned, and then the second analog filter is transferred from the analog domain to the digital domain, thereby being converted into a digital loop filter.

 In a third aspect, the filters 11, 12 in the modulator shown in Figure 1 may contain units with cumulative effects, such as: integrators, resonators, accumulators, etc., when the modulator is operating, the output of these units It will increase greatly. However, the modulator has limited operating voltage and limited register storage value when the circuit is implemented. If no measures are taken, this contradiction will inevitably lead to unstable operation of the modulator and collapse. In order to ensure that the digital noise shaping multi-bit De ta-S i gma modulator shown in FIG. 1 works stably, the present invention proposes a digital noise shaping multi-bit De ta-S i gma modulator internal signal shunting method. The basic idea of the method is: From the digital filter, the shunt digital signal is derived from the input of the unit whose output is greatly increased, and then converted into a corresponding analog signal by the DAC, and finally the analog signal is fed back to the analog filter. Equivalent node of the device. The basic principle of the method is to: Make the transfer function of the shunt signal consistent before and after being shunted. The specific implementation steps of this method are:

 i) First step, determine the signal shunt node

For convenience of description, the modulator shown in FIG. 4 is taken as an example (note that the signal shunting method is not limited to this example, and the internal signal shunting method of other modulators can be analogized according to this example) f\ - ^ '^ -7T ^ . ii^i *il 3⁄4 ^iil v/ . 3⁄4 41 ^ if ^ ^S夂4 ? i 3⁄4 The bit quantizer 44, the multi-bit feedback DAC 45 and the analog adder 46 are constructed, wherein inside the digital loop filter 42, a certain unit 43 having an accumulation function is located between the node 47 and the node 48, U(z), V( z), E _Q (z), E _D (z) represent the input analog signal of the modulator, the output digital signal of the modulator, the quantization noise of the multi-bit quantizer 44, and the mismatch noise of the multi-bit feedback DAC 45, respectively. It can be observed by observation that when the modulator operates, the output of the unit 43 having the cumulative action will greatly increase, so the node 47 at the input of the determining unit 43 is the signal shunting node.

 Ii) The second step, exporting the shunt signal

In order to derive the shunt signal, a signal shunt is required to be inserted at node 47 of Fig. 4 to obtain a modulator structure as shown in Fig. 5. The signal shunt 54 decomposes ^( into V _nl (z) and V _n2 (z), which There are the following relationships:

V _n (Z)=V z) + V _n2 (z) (6)

Where V _nl (z) is the derived shunt signal that is fed back to the signal feed node of the analog filter 51. According to the number of digital signal bits, the shunt signal V _nl (z) can be divided into two categories. In the first category, V _nl (z) is a 1-bit digital signal, and the modulator containing only such shunt signal is classified as Class A digital noise. Plastic multi-bit Delta-Sigma modulator, this type of modulator does not need to introduce additional DEM technology; In the second category, V _nl (z) is a multi-bit digital signal, and the modulator containing this shunt signal is classified as Class B digital noise. Shaped multi-bit Delta-Sigma modulators, such modulators may require the introduction of additional DEM technology.

 Iii) Step 3, signal type conversion

The shunt signal V _nl (z) shown in FIG. 5 belongs to the digital signal, and before the signal fed back to the analog filter 51 is fed to the node, it needs to be converted into a corresponding analog signal, and the feedback is inserted after V _nl (z). The DAC gets the modulator structure shown in Figure 6.

 Iv) the fourth step, determine the signal feed node

In FIG. 6, it is assumed that the transfer function of the transmission path 611 before the V _nl (z) signal is derived is TF ₂ (z), and the transfer function of the transmission path 612 after the V _nl (z) signal is derived is TF z) , according to Under the premise that the TF z -TF z) is established, the node 610 can be found in the analog filter 601, which is the feed node of the shunt signal, and the adder is inserted at the node 610 of FIG. A modulator structure as shown in Fig. 7 is obtained.

 V) Step 5, closing the shunt signal feedback loop

In FIG. 7, the output of the feedback DAC 709 is connected to the negative input of the adder 703 to obtain a modulator structure as shown in FIG. 8, which is a digital noise shaping multi-bit Delta-Sigma modulator optimized by signal shunting. The interior contains a shunt feedback branch. There are multiple units with greatly increased output, which requires multiple shunt feedback branches. The operation of the modulator is made more stable by the setting of the shunt feedback branch.

 As described above, for the feedback DAC nonlinearity problem of a multi-bit Delta-Sigma modulator, the Class A digital noise shaping multi-bit Delta-Sigma modulator of the present invention does not need to introduce DEM technology, and it is cascaded with Compared with the modulator of the quantizer structure, the quantization noise and the mismatch noise of the multi-bit feedback DAC can be shaped in the same order, and there is no problem that the quantization noise and the mismatch noise are amplified by the interstage coupling coefficient. The Class B digital noise shaping multi-bit De 11 aS i gma modulator of the present invention may need to introduce DEM technology, but it can reduce the number of DACs fed back to the input of the modulator compared with the existing DEM technology. The circuit structure of the feedback DAC is improved; on the other hand, the matching precision between the unit components of the feedback DAC is improved, and conditions are created for DEM technology (such as DWA technology, etc.) using 1st order mismatch noise shaping, which can Further reducing circuit complexity, both reducing the noise generated by digital circuits, but also saving power and area.

 The main features of the modulator of the present invention are:

 Unlike conventional multi-bit modulators, a digital loop filter is inserted between the multi-bit quantizer and the multi-bit feedback DAC in the modulator. The feedback node of the multi-bit feedback DAC output signal is transferred from the input of the modulator to The input of the multi-bit quantizer forms a feedback loop composed of a multi-bit quantizer, a digital loop filter and a multi-bit feedback DAC, so that the injection nodes of the quantization noise EQ (z) and the mismatch noise ED (z) are at Equivalent position, under the action of the digital loop filter, these noises are simultaneously shaped (according to this modulation as "digital noise shaping multi-bit Delta-Sigma modulator"), solving multi-bit Delta The nonlinear problem of the multi-bit feedback DAC in the -Sigma modulator, as a core component of the Delta-Sigma analog-to-digital converter, can be applied to high-speed, high-precision analog-to-digital conversion.

 In addition, in order to design and implement the above-mentioned digital noise shaping multi-bit Delta-S i gma modulator, the present invention also proposes a design method thereof, that is, constructing digital noise shaping multi-bit De l ta-S from a conventional Delta-Sigma modulator. The construction of the igma modulator and the signal shunting method taken to make the modulator work more stable.

 The present invention will be further described below in conjunction with an application example. For illustrative purposes, this embodiment first constructs a fourth-order low-pass digital noise shaping multi-bit De 11 aS i gma modulator from a conventional modulator; secondly, internal signal shunting is performed on the modulator; finally, the modulator is illustrated The working principle, and the performance simulation results of the modulator are given.

The conventional fourth-order pass-through multi-bit Delta-Sigma modulator is shown in FIG. 9, and the method for constructing a digital noise shaping multi-bit De 11 aS i gma modulator described above can be implemented by structural transformation. The path filter 912 contains two transfer functions L. Separating, the modulator structure shown in FIG. 10 is obtained; in the second step, the filter 1018 shown in FIG. 10 is transferred to the side of the digital circuit to obtain a modulator structure as shown in FIG. 11, which is the implementation. The fourth-order low-pass digital noise shaping multi-bit De 11 a-S i gma modulator original structure to be explained.

 As can be seen from the modulator shown in Fig. 11, both the analog filter 1109 and the digital loop filter 1118 are composed of integrators, and the integrators are also connected in series, since the integrator has a cumulative effect, when modulation When the device starts to work, the amplitude of the output voltage of these analog integrators will increase greatly, and the absolute value of the output value of the digital integrator will also expand greatly, especially the output of the post-integrator will explode, according to The modulator internal signal shunting method described above can avoid this by signal shunting. In the first step, it is observed that the nodes 1122, 1124, 1126, and 1128 before the digital integrators 1110, 1111, 1112, and 1113 are signal shunt nodes; and in the second step, the signal shunts are inserted at the nodes 1122, 1124, 1126, and 1128, respectively. The modulator structure shown in Fig. 12 is obtained, in which the signal splitters 1218, 1219, 1220 and 1221 decompose the respective input signals V ( , v\z) ^ V "( and V to obtain the derived shunt signal They are 1⁄4( , 1⁄4'( , , and ( , these signals satisfy the following relationship:

V(z)=V ₁ (z) + V ₂ (z)

V z) = V; (z) + V ₂ (z) _{( ? )}

 V z) =V;(z) + V;(z)

 V" z) = V; (z) + V; {z) The third step, as shown in Figure 12, the derived signals ( , v; (z) , and ( are all digital signals, fed back to the analog filter Before the signals of 1209 are fed into the node, they need to be converted into corresponding analog signals, which are shown in Figure 1⁄4ω, ν;(ζ), νω and (after inserting the feedback DAC)

The modulator structure shown in Fig. 13; the fourth step, in Fig. 13, shows that the shunt signals 1⁄4ω, ( , ( and V' (the transfer functions before the paths 1334, 1335, 1336, and 1337 flow to the adder 1329) are observed. Consistent with the "negative transfer function" before being shunted, therefore, nodes 1330, 1331, 1332, and 1333 as shown in Figure 13 are the corresponding shunt signal feed nodes, and the adders are inserted at these nodes to obtain Figure 14 The modulator structure shown; in the fifth step, in Figure 14, the outputs of the feedback DACs 1425, 1426, 1427, and 1428 are connected to the negative inputs of the adders 1429, 1430, 1431, and 1432, respectively, as shown in FIG. The illustrated modulator structure, which is a fourth-order low-pass digital noise shaping multi-bit Delta-Sigma modulator optimized by signal shunting, ends the internal signal shunting process of the entire modulator. As shown in Figure 15, the fourth-order low-pass digital noise shaping multi-bit Delta-Sigma modulator consists of an analog filter 1509, a digital loop filter 1522, a multi-bit quantizer 1523, a multi-bit feedback DAC 1524, and a shunt signal feedback DAC ( 1525, 1526, 1527, 1528), and analog adder 1533. Among them, the analog filter 1509 adopts a distributed feedforward structure, consisting of four cascaded analog integrators (1501, 1502, 1503, 1504) and four with certain gain coefficients ("1", 1505, 1506, 1507). The feedforward branch, feedforward analog adder 1508 and four shunt signals are fed into the adders (1529, 1530, 1531, 1532); the digital loop filter 1522 also uses a distributed feedforward structure, consisting of four cascades Digital integrator (1510, 1511, 1512, 1513), three feedforward branches with a certain gain factor (1514, 1515, 1516), feedforward digital adder 1517 and four signal splitters (1518, 1519, 1520, 1521), U (z) represents the input analog signal of the modulator, V (z) represents the output digital signal of the modulator, and E _Q (z) represents the quantization noise of the multi-bit quantizer 1523, E _D (z) Indicates the mismatch noise of the multi-bit feedback DAC1524. The analog filter 1509 is used to pre-modulate the input analog signal U(z) into a (^)4[/( _z ) signal, and the digital loop filter 1522 is used to shape the quantization noise E _Q (z), mismatch noise E _D (z) and pre-modulation signal (^) ⁴ [/(, multi-bit quantizer 1523 is used to convert the analog signal output from the analog adder 1533 into a corresponding multi-bit digital signal, and the multi-bit feedback DAC 1524 is used to digitally loop The multi-bit digital signal outputted by the filter 1522 is converted into a corresponding analog signal, and the analog adder 1533 is used to sum the output signals of the analog filter 1509 and the multi-bit feedback DAC 1524, the signal splitter 1518, 1519, 1520, 1521, the shunt signal Feedback DACs 1525, 1526, 1527, 1528 and shunt signals are fed to adders 1529, 1530, 1531, 1532 for signal shunting and feedback. When the modulator is operating normally, first, the input analog signal u(z) is analog filter 1509. Pre-modulation into (^) ⁴ [/(signal, then, (^) ⁴ ω signal, quantization noise E _Q (z), and mismatch noise E _D (z) - injection of feedback containing digital loop filter 1522 Loop, finally, these signals are all digital loop filter 1522 After the shape is output from the modulator. According to Figure 15, after derivation, it can be obtained:

V(z) = U(z) + (lZ- ¹ ) ⁴ E _Q (z)-(lZ- ¹ ) ⁴ E _D (z) ( 8 )

It can be seen from equation (8) that the modulator has a fourth-order noise shaping low-pass performance, and the input signal U(z) can pass directly through the modulator, while the quantization noise E _Q (z) and the mismatch noise E _D (z) At the same time, it is subject to 4th order shaping. In order to understand the noise shaping function of the modulator more intuitively, the SNDR (Siena 1 o Noi s + Di st.ort. i on Ra io) ^ ife can be modeled. According to Figure 15, the corresponding modulator model is established by using Ma tLab S imuL ink. The key parameters of the model are as follows:

 a) Multi-bit quantizer 1523 uses a 5-bit ideal model;

 b) Multi-bit feedback The DAC1524 uses a 5-bit non-ideal model with a 1% mismatch between the DAC unit components;

 c) shunt signal feedback DAC1525, 1526, 1527, 1528 use a one-bit model; d) the input signal frequency is 129394. 53125HZ;

 e) the signal bandwidth is 1MHZ;

 0 oversampling rate is 20.

且具有良好的噪声整形能力， 这种调制器可以有效地解决多位After simulation, a spectrum diagram of the modulator output signal SNDR as shown in FIG. 16 and a relationship between the modulator output signal SNDR and the input signal amplitude as shown in FIG. 17 can be obtained. For the convenience of the following description, three cartridges are pre-defined, and the ideal modulator cylinder with an ideal multi-bit feedback DAC is called an "ideal modulator"; the conventional modulator cylinder with a non-ideal multi-bit feedback DAC is called "non-ideal". Modulator"; A digital noise shaping multi-bit De ta-S i gma modulator cylinder with a non-ideal multi-bit feedback DAC is referred to as a "non-ideal digital noise shaping modulator." In Figure 16, the dashed line, the solid thin line, and the solid thick line represent the output signal spectrum of the "ideal modulator", "non-ideal modulator", and "non-ideal digital noise shaping modulator", respectively. In the "non-ideal modulator" spectrum, the noise floor is raised, and there are significant 3rd and 5th harmonics in the band, causing the maximum SNDR to drop sharply to 72.8dB, and the low-frequency part of the noise exhibits a horizontal shape, which indicates the traditional modulation. The mismatch noise E _D (z) of the multi-bit feedback DAC has no shaping function. However, in the spectrum of the "non-ideal digital noise shaping modulator", the noise floor is almost overlapped with the "ideal modulator", and the maximum SNDR is maintained at 117. 8dB (the maximum SNDR of the "ideal modulator" is 116. 3dB). And the noise rises at a slope of 80dB/decade, which indicates that the digital noise shaping multi-bit De ta-S i gma modulator of the present invention can perform quantization noise E _Q (z) and multi-bit feedback DAC mismatch noise E _D (z) performs good shaping at the same time (this example is fourth-order noise shaping). In FIG. 17, the dotted line, the solid thin line, and the solid thick line respectively represent the relationship between the output signal SNDR of the "ideal modulator", "non-ideal modulator", and "non-ideal digital noise shaping modulator" and the input signal amplitude, Comparing the three, the SNDR of the "non-ideal digital noise shaping modulator" output signal is about 40 dB higher than the SNDR of the "non-ideal modulator" output signal, and the "non-ideal digital noise" is almost the entire range of the input signal amplitude. The shaped modulator's SNDR of the output signal is almost equal to the SNDR of the "ideal modulator" output signal, which indicates that the digital noise shaping multi-bit De ta-S i gma modulator of the present invention has robust stability. In short, in this embodiment, through theoretical analysis and system simulation,

And has good noise shaping ability, this modulator can effectively solve multiple bits

Feedback from the Del ta-S igma modulator DAC nonlinearity problem.

 The above is a further detailed description of the present invention in connection with the specific embodiments, and the specific embodiments of the present invention are not limited to the description. It will be apparent to those skilled in the art that the present invention can be made without departing from the spirit and scope of the invention.

Claims

Claim A modulator for analog-to-digital conversion, comprising: an analog filter, an analog adder, a multi-bit quantizer, a digital loop filter, a multi-bit feedback digital-to-analog converter, and the analog filter The input analog signal is pre-modulated and output to an input of the analog adder, the analog adder is used for analog signal summation, and output to the multi-bit quantizer, the multi-bit quantizer is used Converting the analog signal to a multi-bit digital signal output; the digital loop filter is configured to filter the multi-bit digital signal output by the multi-bit quantizer and output to the multi-bit feedback digital-to-analog converter, the multi-bit The feedback digital-to-analog converter is configured to convert the filtered multi-bit digital signal into a feedback analog signal and output to another input of the analog adder.  2. The modulator of claim 1, wherein the modulator further comprises at least one shunt feedback branch, each of the shunt feedback branch inputs being coupled to the digital loop filter a signal shunt node, the output end is connected to a corresponding shunt signal feeding node of the analog filter, and each of the shunt feedback branches is provided with a signal shunt and a shunt feedback digital mode from the input end to the output end. Converter, adder.  3. The modulator according to claim 2, wherein the input shunt signal of each of the shunt feedback branches is a 1-bit digital signal or a multi-bit digital signal.  The modulator according to any one of claims 1 to 3, wherein the modulator is a multi-bit Del ta- sig modulator.  An analog to digital converter comprising the modulator according to any one of claims 1-4. 6. A method of constructing a modulator for constructing a second modulator from a first modulator, wherein the first modulator comprises an analog loop filter, a multi-bit quantizer, and a multi-bit feedback digital mode a converter, an input end of the analog loop filter receives an analog signal input, an output end is connected to an input end of the multi-bit quantizer, and an output end of the multi-bit quantizer outputs a multi-bit digital signal, and is connected An input end of the multi-bit feedback digital-to-analog converter, wherein an output end of the multi-bit feedback digital-to-analog converter is connected to another input end of the analog loop filter, and the constructing method includes Next steps:  Decomposing the analog loop filter into a first analog filter and a second analog filter, adding an analog adder, adjusting device connections as follows: two inputs of the analog adder are respectively connected to the first An output end of the analog filter and the second analog filter, the output end is connected to the input end of the multi-bit quantizer, the first analog filter accepts an analog signal input, and the input end of the second analog filter is connected The output of the multi-bit feedback digital-to-analog converter;  Adjusting the second analog filter to a digital loop filter, adjusting the device connection as follows to construct a second modulator: the input of the digital loop filter is connected to the output of the multi-bit quantizer, and the output The terminal is connected to an input end of the multi-bit feedback digital-to-analog converter, and an output end of the multi-bit feedback digital-to-analog converter is connected to an input end of the adder.  7. A signal shunting method for a modulator, characterized in that the modulator is used for analog to digital conversion, including an analog filter, an analog adder, a multi-bit quantizer, a digital loop filter, and a multi-bit feedback digital-to-analog conversion The analog filter is configured to pre-modulate an input analog signal and output it to an input of the analog adder, the analog adder is used for summing analog signals, and outputting to the multi-bit quantizer, The multi-bit quantizer is configured to convert an analog signal and a multi-bit digital signal output; the digital loop filter is configured to filter a multi-bit digital signal output by the multi-bit quantizer, and output the multi-bit feedback digital mode a converter, the multi-bit feedback digital-to-analog converter is configured to convert the filtered multi-bit digital signal into a feedback analog signal, and output to another input end of the analog adder; the signal shunting method includes the following steps:  Determining at least one signal shunt node of the digital loop filter;  And providing at least one shunt feedback branch, wherein the input end of each of the shunt feedback branches is connected to one of the signal shunt nodes, and a signal shunt, a shunt feedback digital-to-analog converter, and an adder are sequentially disposed from the input end;  An adder of each of the shunt feedback branches is inserted into a corresponding shunt signal feed node of the analog filter.  8. The signal shunting method according to claim 7, wherein each of said parts is right ^ 夂^ X Ύ 4ι 1 λχι !^ - ^ /^ powder ^ ^^τ