FI63646C - FOERFARANDE FOER MINSKNING AV KVANTISERINGSBRUSET I PULSKODMODULATIONSANLAEGGNINGAR - Google Patents

Description

I- -1 r_, .... ANNOUNCEMENT

W <11> uTLAGGNINGSSKRIFT 6 36 4 6 c Patent granted 11 07 1923

Patent oeddelat ^ τ ^ (51) K ».ik? /Int.a.3 H 03 K 13/01! SUOM I —FI N LAN D (21) Fwwttihekemwi —F * www «nilns 76028? (22) Hak «miiptlvl - AMÖknlnffdtg 06.02.76 (23) AlkupUvi — GIMfhatadac 06.02.76 (41) Become Public - Blhrlt offtntllg 07.11.76

PaUntti- ja registalialite · (44) Date of Nlhdvikslpanon Ja kiMiL | ulkaiau. -

Patent · och registerstyrelsen Anaekan utlagd oeh utlskrifun puMkwsd 31.03.83 (32) (33) (31) «Vr * · tty« uoikwM — Begird prtorK · * 06.05.75 Ο6.Ο5.75 Switzerland-Switzerland (CH) 5807/75 , 5808/75 (71) Siemens-Albis Aktiengesellschaft, Albisriederstrasse 2 ** 5, 8ol + 7 Zurich, Switzerland-Switzerland (CH) (72) Josef Fluhr, Arni, Jean-Maurice Lutz, Hietikon, Nikolaus Molnar,

The present invention relates to a method for reducing the quantization noise in a pulse coding mode or for expansion piecewise with a linear characteristic curve.

Converting continuous analog signals to a non-continuous digital format is known to generate quantization noise, which must be considered to be due to the fact that an infinite number of possible analog signal amplitude values must be processed by a limited number of quantization steps. Thus, the reduction of quantization noise can only be achieved by increasing the number of quantization steps. On the other hand, since each quantization result must be represented by a unique codeword, the number of bits required for each codeword depends directly on the number of quantization steps, i.e., the larger the number of quantization steps, the greater the number of bits required for each codeword. The requirement of low basic noise in non-speaking channels (speech pauses) and the requirement of good intelligibility even at low speech level require a sufficiently large number of quantization steps, especially in the range of low amplitude values. In order to achieve this with the smallest possible number of bits for each codeword, various quantization principles using a non-linear quantization characteristic have already been proposed, using larger quantization steps for increasing analog signal amplitude values. For this purpose, reference is made to the article "AD and DA converters, methods and their use" in "Neue Technik", no: 3/1972, p. 80-84 and no: 4/1972, p. 103-111.

Non-linear quantization can be particularly simply implemented in the form of a weighted linear characteristic. An example of a non-linear quantization characteristic is the so-called A-characteristic curve (CCITT recommendation G711), in which the ratio of the analog signal amplitude to the quantization noise remains approximately constant over a large part of the dynamic range. Such a quantization characteristic curve can consist of, for example, thirteen blocks, in which case a linear and thus simple quantization can be performed in each block.

Figure 1 shows such a quantization characteristic, with the X-axis showing the instantaneous values of the analog signal and the Y-axis showing the parts of the orifice curve. Analog / digital conversion by non-linear quantization corresponds to the instantaneous value reduction of the instantaneous value of the respective analog signal, and the resulting digital signals must be expanded accordingly when converting back to analog signals. It is obvious that additional signal distortions can only be avoided when the quantization characteristic curve of the analog / digital converter, hereinafter also abbreviated A / D converter11a, and the reverse conversion of the digital / analog converter, hereinafter also abbreviated D / A converter, characteristic, correspond to each other as mirror images.

In the following text, the term "block part" is used instead of "block". For example, the middle block of a characteristic curve consisting of thirteen blocks extending partly into the positive and partly into the negative characteristic curve range is divided into two characteristic curve parts, positive and negative, which together form a straight line. The thirteen-block characteristic curve then consists of parts of 2 x 7 = 14 characteristic curves. The point where the two parts of the characteristic curve contact each other is called the limit value of the parts of the characteristic curve and the corresponding amplitude value is called the limit value of the characteristic curve. It is further pointed out that in PCM transmission technology, analog-to-digital converters or digital-to-analog converters are often also referred to as encoders and decoders, respectively.

3 6 3 6 4 6

Figure 2 shows the near-ideal path between the quantization stages of the A / D converter and the corresponding amplitude stages of the D / A converter, with the D-A converter stages shown on the X-axis and the A / D converter stages on the Y-axis. The smallest deviations obviously occur when the averages of all quantization steps are on the setting lines a passing through the zero point of the coordinate system. But it can also be seen from this representation that either the A / D converter or the D / A converter must correspond to the first positive and first negative amplitude steps from zero half the next amplitude step after the part of the characteristic curve. As further shown in Figure 2, similar additional corrections are necessary for each transition from one part of the characteristic to another part of the characteristic. Without these corrections, systematic errors occur, which must be attributed to the fact that, as shown in Figure 3, the averages of the individual amplitude stages depending on the instantaneous value of the analog signal and the polarity of the respective characteristic curve deviate sharply in the positive or negative direction from the setting line a.

To improve this process and thereby reduce quantization noise pulssikoodimodulointilaitteistoissa have so far become known for performing the correction only proposals vastaanottopuolisessa the D / A converter by carrying out there in each of the quantization characteristic of each side of the repair. The correction can be performed relatively simply in the ΰ / Α converter. This can be deduced from the fact that the A / D converters used in the pulse code modulation equipment normally output a digital value corresponding to the amplitude value represented by an integer number of quantization steps for each instantaneous value of the input analog signal.

But conceivable systems are now conceivable in which a single transmitter has to take care of a large number of receivers, for example in radio networks, etc. In such cases, there is a single transmitter A / D converter against a large number of receive-side D / A converters. It is further conceivable to use coding methods in which for each instantaneous value of the input analog signal from the transmitting A / D converter, a next higher digital value corresponding to the amplitude value represented by an integer number of quantization steps is output to the D / A converters. In such cases, a correction to reduce quantization noise is preferably performed only in the transmission A / D converter.

4,63646

It is an object of the invention to provide a method for an analog-to-digital converter which, without high cost, can result in a reduction in quantization noise in PCM equipment in which encoding or decoding is combined in pieces with a linear characteristic to reduce or expand.

For analog-to-digital conversion, the current value of the analog signal is fed to the capacitor in the analog-to-digital converter and the capacitor charge is compared (weighing method) with reference values and / or at least approximately a linear capacitor discharge is performed. .

Block-based linear quantization assumes that in order to convert the instantaneous value of an analog signal to analog / digital, it is first determined by comparing one part of the reduction curve characteristic to the first quality reference values corresponding to the limit values of the characteristic part of the curve the number of quantization steps corresponding to the value is indicated by comparing for each part of the characteristic with individual reference values of another quality, each consisting of a basic value characteristic for that part of the curve and for each quantization step number in the part of the characteristic. Of course, in the case of analog signals, in which signals both polarities occur, the polarity of each instantaneous value of the analog signal must be determined, whereby the result is added, for example, the so-called in the form of a polarity bit to a codeword corresponding to the absolute value of the instantaneous value of the respective analog signal. Determining the number of quantization steps corresponding to the instantaneous value of the analog signal in the above-mentioned part of the characteristic curve is also expressed by the term fine coding.

The method according to the invention is characterized in that in the transmitting analog-to-digital converter, except for the limit values of the characteristic part corresponding to the zero point of the quantization characteristic, each characteristic part limit value is reduced in the receiving digital / analog converter part of the number of quantization steps of the part of the characteristic curve bounded by the limit value, that further in the analog-to-digital converter a unique basic value is determined for each part of the characteristic curve as the output value within the part of the characteristic curve of the analog signal

II

5 6 3 6 4 6 to indicate the instantaneous number of quantization steps, and that the basic value is set either in the digital / analogue repeater to match the value of the corresponding characteristic part of the characteristic part of that characteristic or quantified half of the quantization step of that characteristic part of the characteristic part of that characteristic. value different from.

In connection with the drawing, the method according to the invention will be explained in more detail below with examples. In this case, Fig. 1 shows the positive area of the so-called A-characteristic curve, Fig. 2 shows the ideal course of a quantization characteristic curve with periodically different quantization step heights, Fig. 3 shows two possible uncorrected quantization characteristics used in the A / D converter part of the characteristic curve and binary staggering.

Fig. 5 shows a switching device using an iteration method for defining and fine coding a characteristic part, Figs. 6 and 7 each show a variation of a further embodiment of this switching device, and Fig. 8 shows a switching device using an iteration method for defining a characteristic curve part and calculations.

Figures 1, 2 and 3 have already been explained. In Fig. 4, for the sake of clarity, only the three parts of the positive curve in the first quarter, S1, S2 and S3, and the part of the negative curve in the third quarter of the quantification curve are shown as examples. The amplitude levels corresponding to the quantization stages of the A / D converter are marked on the y-axis and the amplitude stages corresponding to the decoded digital values of the D / A converter are marked on the x-axis. The dashed line represents the uncorrected quantization characteristic and the solid line represents the quantized characteristic curve corrected according to the invention of the transmission-side A / D converter. The individual quantization stages Q are both the parts of the characteristic curve SI and -SI bounded to the zero point of the quantization curve, each being twice as large as the quantization stages Q of the adjacent characteristic part in the opposite polarity direction. as a binary staggered quantization characteristic. But, of course, the method according to the invention can be used equally well with non-binary porcine dichedomy characteristic curves.

The message of both characteristic curves shown in Fig. 4 is illustrated in connection with a calculation example. The X-axis shows the amplitude steps of the D / A converter and the Y-axis shows the amplitude steps of the A / D converter. In order for this receive uncorrected quantization characteristic curve, an amplitude value equal to or greater than the value of the twelve amplitude base stages and less than the value of the fourteen amplitude base stages must be input to the transmission-side A / D converter in order for the receive-side D / A converter to output the amplitude value 12 the amplitude base step corresponds to the amount of the smallest quantization step Q present in the entire characteristic curve. In contrast, in the case of a corrected quantization characteristic corresponding to the method according to the invention, an amplitude value equal to or greater than the value of eleven amplitude base stages and less than the value of thirteen amplitude base stages must be input to the transmission-side A / D converter.

The steps required to implement an ideal quantization characteristic in a transmit-side analog-to-digital converter will now be explained. In the first step, the instantaneous value of the analog signal input to the A / D converter is determined to belong to a part S of the quantization characteristic. For this purpose, the iteration method from the reference values URS until the reference values URS of the first quality limiting the current value of the respective analog signal have been determined.

In the following, the values denoted by the theoretical characteristic part limit values USG correspond to the actual characteristic part limit values of the receiving D / A converter, and with the exception of the characteristic part limit values forming the zero point of the curve, they do not match the values used in the transmitting A / D converter. values.

The reference values of the first quality URS of the characteristic curve parts SI and -SI touching the zero point of the quantization characteristic curve correspond to the amplitude value zero. Each one of the other components of the characteristic -Sn - -S2, S2 + - + Sn integral to the reference value, which is the first quality, provides the opposite polarity direction of the closest, in the direction of the opposite polarity of the adjacent part of the characteristic curve of the S-side quantization of the Q value of the characteristic of the theoretical amount in volume of the reduced portion using the threshold USG. As shown in Fig. 4, in the example, the theoretical characteristic part S3 is limited by the amplitude values 16 and 32. In the direction of opposite polarity of the characteristic part S2, the nearest theoretical characteristic

II

63646 female part 7 of the curve limit value USG is thus amplitude value 16 of the opposite polarity direction of the side S2 of the adjacent quantizing characteristic Q value increases amplitudiperusportaaseen. The first quality reference value URS3 of the characteristic part S3 is obtained by quantitatively reducing the amplitude value 16 corresponding to the theoretical characteristic part limit value USG “16, which gives the reference value URS3 of the first quality with an amplitude value 15. Correspondingly, the amplitude value corresponding to the characteristic part S2 first quality benchmark URS2.

After determining the part S of the characteristic curve corresponding to the instantaneous value of the provided analog signal, in the second step the number corresponding to this instantaneous value of the analog signal is quantized in the predefined part S of the characteristic curve S. ., which are of a second quality until the instantaneous values of each quality of the analog signal are determined by the reference values UQ of the second quality. Each quantization characteristic zero reference point UQ, which is an integral part of the quantization characteristic curve, is integral for that part of the characteristic curve.

Each fundamental value URB of the characteristic part S is determined by quantifying by using the iteration method and quantitatively increasing or decreasing by using the calculation method by increasing the limit value correction value.

Thus, for example, in Fig. 4, the theoretical characteristic portion S3 is limited by amplitude values 16 and 32, and in the opposite polarity direction, the nearest theoretical characteristic portion limit value USG is thus in amplitude value 16. The basic value URB31 or URB 32 of the characteristic curve portion S3 is obtained amplitude 16 to enlarge or reduce a quantitative characteristic of the quantization step S3 side of the Q value, that is, a value of 2, the basic value of 31 or URB URB 32 for amplitudiar will result in RX-14 or 18.

Correspondingly, for the characteristic part S2, the basic value URB21 or URB 22 corresponding to the amplitude value 7 or 9 is obtained, the characteristic part S1 of the characteristic curve

63646 8 for the amplitude value +0.5 or -0.5 corresponding to the base value -UBB1.

If a calculation method is used for fine-tuning instead of an iteration method, then the values of the basic values URB must be determined in each case in accordance with the execution of the first calculation step. If the first calculation step is not included in each case, then the same basic values URB1, URB21, URB31, UBB41 ... etc. are required as with the iteration method. If, on the other hand, each first calculation step is included, then the basic values URB22, URB32, URB42, which are quantitatively larger than the corresponding theoretical limit values of the part of the characteristic curve USG, are necessary. If the duration of each first descent step is shortened, basic values URB identical to the corresponding limit values USG of the characteristic part of the curve are required. The shortening of each first descent step can take place, for example, by taking the synchronous signal from the regular time step sensor of the basic time step sensor for the first descent step after the delay time and only then starting to compare the voltage of the already discharged capacitor with the reference values.

Figure 5 shows the principle of a switching device using an iteration method (weighing method) for defining and fine-coding a part of the characteristic curve. The current instantaneous value of the PAM analog signal, regardless of its polarity, is output to one input of the reference device V connected to the control logic SL. The second input of the comparator V is connected to a plurality of positive and negative polarity sources G which can be individually switched on and off by the control logic SL. The number of these sources G and their magnitudes and polarities adapt to the number of characteristic curves in the quantization curve used for a given dynamic range. and their limit values and the number and size of the quantization steps Q in the parts S of the characteristic curve.

Thus, either a) for each part of the characteristic curve not limited to the zero point of the quantization characteristic -SN ...- S2, + S2 ... + Sn, a source G with a magnitude representing the limit value USG of the theoretical characteristic part in question is required and for all parts of the characteristic always a source G, which has a quantity representing half the value of the quantization step Q of the part S of the characteristic curve in question.

Or b) one positive and negative polarity source G is required, each having half the value of the smallest quantization step Q present in the quantization characteristic and for each part of the characteristic curve -Sn ...- S2, + S2 ... + Sn, which are not limited to the quantization characteristic 9 63646 to zero, source G having a magnitude indicative of that first quality reference value URS, and source G having a magnitude indicative of the difference between said first quality reference value URS and said base value URB.

Or c) one positive and negative polarity source G is required, each having half the value of the smallest quantization step Q present in the quantization characteristic curve and for each part of the characteristic curve -Sn ...- S2, + S2 ... + Sn not limited to the zero point of the quantization characteristic curve , source G having a magnitude indicative of that first quality reference value URS, and source G having a magnitude indicative of that base value URB.

In addition to the sources G mentioned in a) to c), sources G are required for fine coding, in which a number of quantization steps corresponding to the instantaneous value of the PAM analog signal are defined within the part S of the characteristic curve indicated above for the instantaneous value of the analog signal.

Thus, for example, according to Figure 4, for a binary stepped quantization characteristic curve with seven characteristic parts + S1 ... + S7 for a positive and seven characteristic parts -SI ... -S7 for a negative amplitude range, a total of the following sources G are required:

Eleven sources G for the positive amplitude range at -0.5 / + 1 / + 2 / + 4 / + 7 / + 14 / + 28 / + 56 / + 112 / + 224 / + 448 and eleven sources G for the negative amplitude range at + 0.5 / -1 / -2 / -4 / -7 / -14 / -28 / -56 / -112 / -224 / -448.

With the exception of the source G of -0.5, which is necessary for the positive characteristic range, all positive reference values can only be generated with positive polarity sources G. With the exception of the source G of +0.5 necessary for the negative characteristic range, all negative reference values can be generated only with polarity .Benchmarks of first quality URS can be represented by these quantities as follows. For the positive characteristic range: UHS1-0; URS2- + 7 + 1-0.5- + 7 * 5! URS3- + 14 + 1- +15 URS4- + 28 + 2- + 30j URS5— + 56 + 4— +60} etc.

For the area of the negative characteristic: URS1-0; URS2- -7-1 + 0.5- -7.5; URS3- -14-1- -15; And so on.

Of course, all the sources mentioned here for sources G, the first quality URS reference values and the basic value URB are only a matter of giving examples. However, these show that the corresponding selection and combination of individual quantities makes it possible to reduce the number of sources 0 required. All that needs to be taken into account is that the sources required to produce the base value URB are no longer available for fine coding within the part of the characteristic curve belonging to this base value.

The basic values URB can be represented when using the iteration method for fine coding with the quantities indicated in the above example. For the positive characteristic range: URB1 »-0.5; URB21- +7; URB31- +14; URB41- +28? and so on.

For the negative characteristic range: URB1 to +0.5; URB21 »-7; URB31- -14; URB41- -28; and so on.

The second quality reference values required for fine-tuning according to the iteration method can be expressed as for the positive characteristic range given in the example given below: - Part SI characteristic: basic value URB1 -0.5 and combination of + 1 / + 2 / + 4 , 5 / + 1,5 / + 2,5 / + 5,5 / + 4,5 / + 5,5 / + 6,5 - Part S2 of the characteristic curve: the basic value TJRB21 +7 and the combination of variables + 2 / + 4 give as second quality reference values: + 9 / + 11 / + 13 - Part of the characteristic curve SJ: basic value URB31 + 14 and a combination of variables + 1 / + 4 / + 7 give second quality reference values »+ 18 / + 22 / + 26 etc.

If a calculation method is used instead of the iteration method (weighing method) for fine coding, then the values of the basic values URB must be determined in each case in accordance with the execution of the first calculation step. If the first calculation step is not included in each case and the start of the capacitor C discharge utilization is postponed, then the same basic values URB1, URB21, URB3I, URB4I ... etc. Are necessary as in the iteration method. If, on the other hand, the first calculation step is included in each case, then the basic values URB22, URB32, URB42 ... are quantitatively larger than the corresponding theoretical characteristic part limit values USG, etc. If the duration of the first calculation step is reduced by half, the corresponding characteristic part limit values USG basic values identical to URB. In the case where the first calculation step is included in each case, for example, for the quantization characteristic curve according to Fig. 4, the following basic values for the positive characteristic curve range are obtained: TJRB1 - +0.5; URB22 - +9; URB32 - +18? URB42 - +36? and so on.

And for the negative characteristic range: URB1 - -0.5; URB22 - -9? URB32 - -18? URB42 - -36; and so on.

Il 11 6 364 6

Figure 6 shows a first further embodiment of a switching device according to the invention using an iteration method for defining parts of a characteristic curve and for fine coding. The current value of the respective PAM analog signal is also fed here, regardless of its polarity, to the first input of the reference device V connected to the control logic SL on the output side. Several sources G with positive and negative polarity, individually switched on and off by the control logic SL, are connected for all sources G to the second input of the common summing amplifier SV and via the resistor BZ to the output of the summing amplifier SV and the second input of the comparator V. Each source G includes a DC resistor R which determines its value and a switch K connected in series therewith. Each of these switches K individually switched on and off by the control logic SL connects said first input of the summing amplifier SV in case the source G has a positive polarity, a resistor R through a positive polarity of the first voltage source UA common to all positive polarity sources G and, in the case where the source G is negative polarity, through a resistor R to the negative terminal of the second voltage source UB common to all negative polarity sources G. The negative terminal of the first voltage source UA and the positive terminal of the second voltage source UB are fixedly connected to the second input of the summing amplifier SV and to a common potential identical to the operating potential PAM of the instantaneous value of the analog signal.

Fig. 7 shows another further embodiment of a switching device according to the invention using an iteration method for defining and fine-coding the parts of the characteristic curve. Here, too, the current instantaneous value of the PAM analog signal, regardless of its polarity, is externally connected to the first input of the comparator V connected externally to the control logic SL and several positive and negative polarity sources G with positive and negative polarity are connected to the second input of the comparator V. Each source G has a DC resistor R which determines its value, which in turn is connected to the emitter of the transistor Ta or Tb in each source G and to the control logic SL via a locking diode D in each source G. The collectors of the transistors Ta, Tb of all sources G are connected on the one hand via the common collector resistor RK to the reference potential of the instantaneous value of the PAM analog signal and on the other hand to the second input of the reference device V. In sources of negative polarity 12 6 3 6 4 6, their resistors H are, on the other hand, fixedly connected to the negative terminal of the same common first voltage source and its positive terminal to the bases of transistors Ta and to the negative terminal of the second voltage source Ug common to its negative polarity and its positive terminal. is fixed to the reference value of the instantaneous value of the PAM analog signal. In sources G with positive polarity, their resistors H are in turn fixedly connected to the positive terminal of the third voltage source common to these sources and its negative terminal to the bases of transistors Tb, and to the positive terminal of the fourth voltage source common to positive sources G and its negative terminal to the instantaneous reference value of the PAM analog signal. For sources G with a positive polarity, for example, PNP-type transistors Tb can be used, and for sources G with a negative polarity, NPN-type transistors Ta can be used. For this case, the shut-off diodes D are arranged between their sources G and the control logic SL with respect to their transmission direction, so that control pulses of positive polarity can pass from the control logic SL to the emitters of the PNP transistors Tb and negative control poles from the control logic SL to the emitters of the NPN transistors Ta. This structure of the source G shown in Fig. 7 is also called a constant current source.

Fig. 8 shows an additional switching device in which an iteration method (weighing method) and a calculation method for fine coding are used to define the parts of the characteristic curve. Each instantaneous value of the PAM analog signal, regardless of its polarity, is applied on the output side to the first input of the reference device V connected to the control logic SL. This first input is connected to a plurality of positive and negative polarity current sources Gi with positive and negative polarity and a capacitor C by means of a control logic SL, and a second input of the comparator V is connected to a plurality of sources G individually controlled by the control logic SL. positive and negative polarity. The sources G are used to generate a unique reference value of first quality for each part S of the characteristic curve to indicate the part S of the characteristic curve, and a unique basic value URB for each part of the characteristic curve S for fine coding. Each current source Gi has a direct current resistor Re which determines its value, which in turn is connected to the emitter of a transistor Ta or Tb belonging to each current source Gi and to a control logic SL via a closing diode D belonging to each current source Gi. The collectors of the transistors Ta, Tb of all current sources Gi are connected on the one hand to the first input of the comparator V and on the other hand to the capacitor C.

Il 13 63646

In negative polarity current sources Gi, their resistors Re are fixedly connected to the negative terminal of the same common first voltage source and its positive terminal to the bases of transistors Ta, and to negative polarity current sources Gi to the negative terminal of its common second voltage source and its positive terminal is fixedly connected to the PAM analog signal . Positive polarity power supplies Gi have their resistors Re on the other hand connected to these positive poles of the common third voltage source TJ ^ and its negative poles to the bases of transistors Tb. For positive-polarity current sources Gi, for example, PNP-type transistors Tb can be used, and for negative-polarity current sources Gi, NPN-type transistors Ta can be used. For this case, the closing diodes D are arranged with respect to their transmission direction between the current sources Gi and the control logic S1 so that positive polarity control pulses can pass from the control logic SL to the emitters of the PNP transistors Tb and negative polarity control pulses from the control logic to the NP NPN transistors Ta. With the aid of each of these current sources Gi, a different, at least approximately linear discharge of the capacitor C can take place, whereby the time required for complete discharge in each case is calculated in basic time steps.

Each source G also contains here a DC resistor R which determines its value and a switch K connected in series with each other. Each of these switches K individually switched on and off by the control logic SL connects the second input of the comparator V in case the source G has a positive polarity, the resistor Via R to the positive terminal of the fifth voltage source UA common to all positive polarity sources G and in the case where the source G is negative polarity, through the resistor R to the negative terminal of the sixth voltage source yhteisen common to all negative polarity sources G. The negative terminal of the fifth voltage source TJA and the positive terminal of the sixth voltage source UB are fixedly connected to the reference potential of the instantaneous value of the PAM analog signal. Furthermore, the second input of the comparator V through the DC resistor RK and the second terminal of the capacitor C are fixedly connected to the reference potential of the instantaneous value of the PAM analog signal.

For the quantization characteristic curve corrected according to Fig. 4, in order to reduce the quantization coefficient, in the switching devices according to Figs. an additional source, the amount of which coincides with the side of the smallest amount of quantization kvantisointiominaiskäyrään included.

FIG Θ switching device, wherein hienokoodaus execute a calculation method is required which case an analog / digital acetal-conversion for addition to the sources required for both polarities one additional source, the amount of which coincides with the side of the smallest kvantisointiominaiskäyrään number of quantization included, and / or device in each case of the first basic time to shorten the step. In each case, the shortening of the first step can take place, for example, by taking the clock signal from the regular time step sensor of the basic time stepper for the first step only after the delay time and only after it Comparison of the already discharged capacitor voltage with reference values.

Il

Claims (5)

1. A method for reducing quantization noise in pulse code modulation systems, in which coding and decoding are associated with compression and decoding respectively. expansion with sectional linear characteristic, characterized in that an analog / digital converter arranged on a transmission side, with the exception of the curve section limit value (URS), which coincides with the zero point of the quantization curve, with respect to its corresponding curve section limit value, provided digital / analog converter is reduced in amount by half of the amount of a quantization step (Q) of the curve section boundary value (URS) bounding the section section in the direction of origin; furthermore, in the analogue / digital converter for each curve section, an individual base value (URB) is determined as the output value for determining the number of quantization steps (Q) corresponding to a quantization step (Q) within a curve section (S); and that, as a base value (URB), either a value corresponding to the curve section limit value (USG) for the curve section corresponding to the digital / analog converter or an amount by half of the amount of a quantization step (Q) in the relevant curve section USG) deviating value is determined. Method according to claim 1, wherein the determination of the number of quantization steps corresponding to an analogue signal torque within a curve section is carried out according to an iteration method, characterized in that for each curve section (S), a curve section limit value corresponding to the digital / analogue converter (S) USG) amount reduced base value (URB) is determined. Method according to claim 1, wherein the determination of the number of quantization steps corresponding to an analogue signal moment value within a curve section is carried out according to a calculation method, characterized in that the duration of the first counting step is shortened with respect to the subsequent section (S). with the relevant curve section limit value (USG) for the corresponding curve section (S) corresponding to the curve section (S) in the digital / analogue converter is determined. A method according to claim 1, wherein the determination of the number of quantization steps corresponding to an analog signal torque value within a curve section is carried out according to a calculation method, in which the first counting step is taken into account, respectively. not counted, characterized by that for each