Digital signal generating device.

The present invention relates to a digital signal generating device for outputting a digital signal having, when converted into an analog signal, a desired frequency or frequencies. For example, it relates to a digital signal generating device for providing a digital frequency-shift keying (FSK) signal or for providing a digital signal having, when converted into an analog signal, two mixed frequencies.

The transmission of analog FSK signals is disclosed in, for example, U.S. Patent No. 4,368,439.

In the field of telephone-line systems, however, wide use is made of PCM channels through which a so-called digital FSK signal is transmitted. The digital FSK signal consists of a sequence of binary coded signals expressing a logic "1" and a logic "0". Each of the binary coded signals in the digital FSK signal consists of a PCM signal obtained by sampling a sine wave data. That is, the logic "1" in the digital FSK signal is formed by a PCM signal having, when converted into an analog signal, a certain frequency, and the logic "0" in the digital FSK signal is formed by a PCM-signal having, when converted into an analog signal, another frequency, different from the above certain frequency. For example, according to CCITT, V21, the two frequencies are 980 Hz and 1180 Hz, respectively.

In a conventional digital FSK signal generating device, two PCM signal sources are employed for generating the above-mentioned two PCM signals. These two PCM signals are always generated from the two PCM signal sources, and a selection unit switches from one of the two PCM signals to the other PCM signal to be output from the device.

There are, however, disadvantages in the above conventional digital FSK signal generating device. That is, it needs two independent PCM signal sources, making it expensive, and at a switching point, these two PCM signals are not always in the same phase. Therefore, the phases of the PCM signals or analog signals obtained from the PCM signal, at the switching point, usually become discontinuous, as later described in more detail. Because of this discontinuity, the digital FSK signal at the switching point has an impulse noise and, therefore, may be-erroneously received at a receiver as noise or as another error signal.

Apart from the digital FSK signal generating device, a device for generating two mixed signals having different frequencies is useful for various applications. For example, in testing a modem, by applying a signal having two mixed frequencies, the characteristics for respective frequency components can be simultaneously examined. Also, a tone signal composed of two mixed frequencies is often used.

In order to generate such a signal having two mixed frequencies, a conventional technique is to provide two oscillators respectively generating frequencies different from each other. In this conventional case also, the need for two oscillators makes the device expensive.

- A read only memory (ROM) may be used to obtain two signals having different frequencies by storing mixed data obtained by sampling two sine wave data having different frequencies. By reading the ROM with a sampling period step by step sequentially, the two mixed data can be read. In this case, however, it is impossible to change the frequencies of the output two signals because the stored mixed data are fixed data and are read step by step sequentially. Also, because the ROM must store the two mixed data having different frequencies, the memory size of the ROM must be large.

An object of the present invention is to provide a digital signal generating device which can generate a digital FSK signal with no discontinuity in phase and amplitude.

Another object of the present invention is to provide a digital FSK signal generating device which can change the frequency of the output without a discontinuity in the phase of the PCM signal, thereby decreasing the possibility of erroneous detection at the receiver side.

Still another·object of the present invention is to provide a digital signal generating device which can output two mixed signals having different frequencies without the necessity for using two oscillators but, instead, using only a single ROM.

A still further object of the present invention is to provide a digital signal generating device which can output two mixed signals having different frequencies, and in which the frequencies are easily changed.

To achieve the above objects, there is provided,-- according to the present invention, a digital signal generating device comprising a sampled data storing means for sequentially storing, in its continuous addresses, a plurality of digital sampled data obtained by sampling, with a predetermined sampling frequency, a fundamental sine wave data having a fundamental frequency; an extracting step number storing means for storing at least one extracting step number; and a data extracting means for extracting, with the predetermined sampled frequency, a part of the digital sampled data from the sampled data storing means by accessing extracting addresses in the continuous addresses of the sampled data storing means. The extracting addresses are determined in such a way as to add the extracting step number to a previous extracting address to form a new extracting address.

The sampled data storing means stores the data of a predetermined cycle of the fundamental since wave. This predetermined cycle is at least equal to a multiple of one quarter of one cycle of the fundamental sine wave data.

The data extracting means comprises an extracting address latching means, and an extracting address calculating means (ALU).

The extracting address latching means operatively latches the previous extracting address until a first new extracting address is determined.

The extracting address calculating means operatively calculates the first new extracting address by adding the extracting step number to the previous extracting address output from the extracting address latching means.

According to one aspect of the present invention, the predetermined cycle is equal to, for example, a half cycle or even one cycle of the fundamental sine wave data.

The device further comprises a control means for providing, when the first new extracting address is greater than or equal to the number of sampling steps in the one cycle, a first control signal to the calculating means. The calculating means operatively calculates, in response to the first control signal, a second new extracting address by subtracting the number of sampling steps from the first new extracting address.

The extracting address latching means operatively updates the previous extracting address by the first new extracting address when the first control signal is not applied to the calculating means or by the second new extracting address when the first control signal is applied to the calculating means.

According to another aspect of the present invention, the device further comprises means for changing every alternate first new extracting address to a predetermined extracting address. In said predetermined address of the sampled data storing means, the data "0" is stored.

The above objects and features of the present invention will be more apparent from the following description of the preferred embodiments with reference to the accompanying drawings, wherein: <ul id="ul0001" list-style="none"><li>Figure 1 shows a waveform diagram for explaining a well-known FSK signal;</li><li>Fig. 2 shows a well-known telephone-communication line system employing PCM channels for which the present invention is applicable;</li><li>Fig. 3 shows an example of a conventional digital FSK signal generating device;</li><li>Fig. 4 shows a waveform diagram for explaining the problems in the conventional device;</li><li>Fig. 5 is a block diagram of a digital FSK signal generating device, according to a first embodiment of the present invention;</li><li>Fig. 6 shows a plurality of digital sampled data obtained from a fundamental sine wave data for its one cycle; ―</li><li>Fig. 7 shows a flowchart explaining the operation of the device shown in Fig. 5;</li><li>Fig. 8 shows an example of the digital FSK signal extracted from the device shown in Fig. 5;</li><li>Fig. 9 shows the digital sampled data obtained from the 1/4 cycle of the fundamental sine wave data;</li><li>Fig. 10 is a block diagram of a digital FSK signal generating device according to a third embodiment of the present invention;</li><li>Fig. 11 shows a flowchart explaining the operation of the device shown in Fig. 10;</li><li>Fig. 12 is a block diagram illustrating a digital signal generating device according to a third embodiment of the present invention;</li><li>Fig. 13A is a waveform diagram illustrating a sine wave data obtained from the device shown in Fig. 12;</li><li>Fig. 13B is a waveform diagram illustrating an analog signal including two frequencies which are also obtained from the device shown in Fig. 12;</li><li>Fig. 14 shows a frequency spectrum of the two signals included in the analog signal shown in Fig. 13B;</li><li>Fig. 15 is a diagram explaining the operation of the device shown in Fig. 12; and</li><li>Fig. 16 is a block diagram of a digital signal generating device according to a fourth embodiment of the present invention.</li></ul>

Before describing the embodiments of the present invention, the background of the invention and the prior art and problems therein will first be described with reference to Figs. 1 through 4.

Figure 1 shows a waveform diagram for explaining a well known FSK signal. As shown in Fig. 1, the FSK signal consists of a sequence of logical states, "1" and "0", of digital data to be sent. Each of the logical states of "1" consists, when the FSK signal is formed by PCM signals, .of a.sampled data which has, when converted into an analog sine wave data, a first frequency fa equal to 980 Hz in this example. Also, each of the logical states "0" in the digital FSK signal consists of a sampled data having a second frequency fz equal to 1180 Hz in this example. Hereinafter, the term "the frequency of the digital sampled data" will refer to the frequency of the sine wave data corresponding to the digital sampled data. These data samples are PCM signals. And thus, when the FSK signal is formed by the PCM signals, the FSK signal is referred as "digital" to distinguish it from the usual FSK sign.il whose logical values consist of analog sine wave data. The digital or analog FSK signal has advantages over a simple binary coded signal having high and low constant voltages of "1" and "0", in that the frequencies of the sine wave data can be easily detected instead of the voltages, which may drop over a long distance.

Figure 2 shows a well-known telephone-communication line system employing PCM channels for which the present invention can be applied, and through which PCM channels the above-mentioned digital FSK signal is transmitted. In this system, a modulator - demodulator (modem) 1 in a 'toll switching system 2 should preferably generate a PCM signal such as the above-mentioned digital FSK signal, rather than the analog FSK signal. In Fig. 2, the modem (MOD) 1 provides a.digital FSK signal to a toll switch (TS) 3. The toll switch (TS) 3 selects a digital terminal (DT) 4 to transmit the digital FSK signal through a PCM line 5 to a digital terminal (DT) 6 at a local switching system 7. In the local switching system 7, the digital FSK signal from the digital terminal (DT) 6 is time division switched by a local switch (LS) 8. The output of the local switch (LS) 8 is converted into an analog signal by a coder-decoder (CODEC) 9. The analog signal is then applied to a subscriber 10. The system shown in Fig. 2 is applied to, for example, a detailed charging system for general telephone services in a hotel. In this case, the digital FSK signal on the PCM line 5 is transmitted at a low speed of 300 baud.

Figure 3 shows an example of a conventional digital FSK signal generating device used in prior art. In Fig. 3, a selecting unit (SEL) 30 selects, in repsonse to a logical state of a command C consisting of a binary code applied to an input terminal (IT) 31, one of two PCM signals generated by two signal sources (SGa) 32 and (Gz) 33, respectively. The signal source (SGa) 32 always generates a PCM signal having a frequency fa, and the signal source SGz 33 always generates a PCM signal having a frequency fz. The selected PCM signal is output from an output terminal (OT) 34. Thus, by changing the command C, a digital FSK signal such as shown in Fig. 1 can be obtained. However, since the conventional device includes the two signal sources (SGa) 32 and (SGz) 33, it has a disadvantage in that it is too expensive, and further, since the phases of the two signals generated by the two signal sources 32 and 33 do not alwasy coincide, the digital FSK signal includes, at its switching points from "0" to "1" or from "1" to "0", impulse noises. This will be explained in more detail with reference to Fig. 4.

Figure 4 shows a waveform diagram illustrating a sequence of two sine waves which are obtained by converting the digital FSK signal output from the selecting unit (SEL) 30 shown in Fig. 3. The first sine wave S₁ has the high frequency fz, and the second sine wave S₂ has the low frequency fa. Between a time t0 and a time t₁, the first sine wave S₁is switched t0 the second sine wave S₂. This switching is accomplished by changing the binary code of the command C applied to the input terminal (IT) 31 shown in Fig. 3. Since the phase and amplitude of the first sine wave S₁does not coincide, at the time to and of course at the time t₁, with the phase and amplitude of the second sine wave S₂, the wave forms S₁and S₂are discontinuous between the time to and the time t_l. This discontinuity causes the impulse noises at the receiver side, such as the subscriber 10 shown in Fig. 2.

Figure 5 is a block diagram of a digital FSK signal generating device, according to a first embodiment of the present invention. In Fig. 5, a sampled data storing unit (SD) 51 is a read only memory (ROM) for sequentially storing, in its continuous addresses, a plurality of digital sampled data. These digital sampled data are obtained by sampling a fundamental sine wave data for a predetermined cycle. The fundamental sine wave data has a fundamental frequency f₀ which is a common factor of the two frequencies fa and fz in the desired digital FSK signal. For example, when the frequencies fa and fz of the digital FSK signal are 980 Hz and 1180 Hz, respectively, a common factor of, for example, 2.5 Hz, is selected as the fundamental frequency f₀ of the fundamental sine wave data. The fundamental frequency f₀ may, of course, be the other common factor such as 10 Hz, 20 Hz, etc. The sampling frequency f of the PCM signal in a telephone switching system is usually equal to 8 kHz which is twice the maximum voice frequency. In a simplest model for explanation, the predetermined cycle is equal to one cycle C₀of the fundamental sine wave data. The SD 51 in Fig. 5 stores the one cycle C_o of the fundamental sine wave data.

Figure 6 shows the plurality of digital sampled data obtained from the fundametnal sine wave data for the one cycle C₀. In Fig. 6, the fundamental sine wave data FS has the fundamental frequency f0 of 2.5 Hz. The sampling frequency f is 8 kHz. Therefore, the number of sampling steps in the one cycle C_o of the sine wave data FS is equal to f/f₀, i.e., 3200 steps. Also, the sampling period is 1/8000 second, i.e., 125 us. Sampled values D₀ , D₁ , D₂ , D₃, ..., and ^D₃₁₉₉ obtained by the sampling are stored sequentially as the digital sampled data in the sampled data storing unit (SD) 51 shown in Fig. 5.

Referring back to Fig. 5, the digital FSK signal generating device further includes an extracting step number storing unit (EN) 52, an extracting address calculating unit (ALV) 53, an extracting address latching unit (ADL) 54, and a control unit (CTL) 55.

The EN 52 stores two extracting step numbers d_la and d_lz which are, in this embodiment, 392 and 472, respectively. The extracting step number d_1a or d_1z is represented by a notation "d₁" shown in Fig. 6. The extracting step numbers d_1a and d_1z are determined by dividing the frequencies fa and fz in the desired digital FSK signal with the fundamental frequency f₀. That is, in this embodiment, d_1a = fa/f₀ = 980/2.5 = 392 and; d_lz= fz/f₀= 1180/2.5 = 472.

Figure 7 shows a flowchart explaining the operation of the device shown in Fig. 5. Referring to Figs. 5 and 7, at step S₁, the CTL 55 discriminates whether the command C applied to the input terminal (IT) 56 is a command "a" or "z" for specifying the first frequency fa or the second frequency fz to be output. The commands "a" and "z" are, for example, "1" and "0", respectively which are equal to the binary data to be sent from the device shown in Fig. 5 to receiver side.

If the binary code C is the command "a", the CTL 55 provides a control signal C_1a to the EN 52 which, at step S₂, outputs the first extracting step number d_la as the extracting step number d₁, in response to the control signal C_la. If the command C is the command "z", the CTL 55 provides a control signal C_lz and the EN 52 outputs, at step S₃, the second extracting step number d_lz as the extracting step number d_l.

At step S₄, the ALU 53 calculates a next extracting address d₂ by adding the extracting step number d₁ with a previous extracting address d₃ which has been latched in the ADL 54.

At step S₅, the CTL 55 discriminates whether the calculated extracting address d₂ output from the ALU 53 is greater than or equal to the number of sampling steps in the one cycle, i.e., the number of sampling steps in this embodiment is 3200.

If d₂ ≧ 3200, then, at step S₆, the CTL 55 provides another control signal C₂ to the ALU 53, and the ALU 53 converts the extracting address d₂ into another new extracting address by subtracting the number of sampling steps (3200) from the new extracting address d₂. That is, d₂ is updated to d₂- 3200.

After step S₆, or if d₂ &lt; 3200 at step S₅, then the CTL 55 provides still another control signal C₃ to the ADL 54 so that, at step S₇, the ADL 54 latches the updated or original new extracting address d₂. Thus, the previous extracting address d₃ is updated to the updated or original new extracting address d₂, to form a current extracting address d₃ which is equal to d₂.

Then, at step S₈, the SD 51 is accessed by the current extracting address d₃ so that a sampled data D_i stored in the address d₃ of the SD 51 is read and output.

At step S₉, the CTL 55 discriminates whether the transmission from the SD 51 should be ended. Until the transmission ends, the above mentioned steps S₁ through S₈are repeated with the sampling frequency of 8 kHz. As a result, a digital FSK signal is obtained at the output terminal (OTa) 57.

Figure 8 shows an example of the digital FSK signal extracted from the SD 51 shown in Fig. 5. As shown in Fig. 8, data D_1a . D_2a , D_3a , ..., and D4a are those extracted from the SD 51 when the extracting step number is d_la equal to 392. In this case, the extracting addresses d₁ + d₃ are sequentially calculated as 0, 392 + 0 = 392, 392 + 392 = 784, ..., and 3528. The extracting address 3528 for extracting the data D_9a is greater than 3200. Therefore, the address 3528 is converted into 3528 - 3200, i.e., 328. On the other hand, data D_{1z .} D_2z ' D_{3z '} ... are those extracted from the SD 51 when the extracting step number is d_1z equal to 472. In this case, each of the extracting addresses for extracting the data D_1z , D_2z , ... is calculated by adding the extracting step number 472 to the previous extracting address. For example, the extracting address for extracting the data D_1z is equal to 328 + 472 = 800. By the same way, the data d_2z= 800 + 472 = 1272, and so forth. These data D_1a , D_{2a '} ..., D_4a, D_lz, D_2z , ... are extracted from the SD51 with a frequency equal to the sampling frequency f of the fundamental sine wave data FS. That is, the extracting period for extracting these data is equal to the sampling period, i.e., 1/f = 125 ^ps. Since the data D_1a , D_2a , ..., and D_9a are extracted from each d_1a = 392 step in the fundamental sine wave data stored in the SD 51, and since the fundamental sine wave data has the fundametal frequency f₀ = 2.5 Hz, the extracted PCM signal constituting the extracted data D_1a, D_{2a '} ..., and D9 a a has a frequency equal to d_1a x f₀ = 392 x 2.5 = 980 hz. Thus, the PCM signal has the first frequency fa equal to 980 Hz, corresponding to, for example, "1" of the binary data "1" and "0" in the digital FSK signal to be sent to the receiver side. Similarly, the extracted PCM signal constituting the extracted data D_1z , D_2z , D_3z, ... has the second frequency fz equal to 472 x 2.5 = 1180 Hz, corresponding to, for example, "0" of the binary data to be sent. It should be noted that the term "digital" in the digital FSK signal does not relate to the binary data but relates to the PCM signal.

Instead of storing the one cycle C_o (2π in phase) of the fundamental sine wave data FS in the SD 51, it is possible to store only a 1/2 cycle or even a 1/4 cycle (π/2 in phase) of the fundamental sine wave data RS.

Figure 9 shows the digital sampled data obtained from the 1/4 cycle of the fundamental sine wave data FS. As shown in Fig. 9, 801 data samples D₀, D₁ , D₂ , D₃ , ..., and D₈₀₀ are obtained from the 1/4 cycle of the fundamental sine wave data FS. The sampling period for obtaining the data samples D₀ , D₁ , ..., and ^D₈₀₀ is the same as that shown in Fig. 6, i.e., 125 µs.

Figure 10 is a block diagram of a digital FSK signal generating device according to a second embodiment of the present invention. In Fig. 10, the device comprises a sampled data storing unit (SD) 51a, an extracting number storing unit (EN) 52a, an extracting address calculating unit (ALU) 53a, an extracting address latching unit (ADL) 54a, and a control unit (CTL) 55a, which are similar to those shown in Fig. 5 of the first embodiment.

The SD 51a stores the sampled data D₀ , D₁ , D₂ , ..., and D₈₀₀ obtained from the 1/4 cycle of the fundamental sine wave data FS shown in Fig. 9. Because of this, the operation of the device shown in Fig. 10 is slightly different from that of the first embodiment.

Figure 11 shows a flowchart explaining the operation of the device shown in Fig. 10. Referring to Figs. 10 and 11, steps S_la through S_4a are the same as the steps S₁ through S₄ shown in Fig. 7. At step S_5a, the CTL 55a discriminates whether the calculated extracting address d₂is greater than or equal to the number of sampling steps in the 1/4 cycle C_O/4, i.e., whether d₂is at a phase more than π/2.

If d₂&gt; 800, i.e., if d₂ is at a phase more than π/2, then, at step S_6a, the CTL 55a provides a control signal C_2a to the ALU 53a, which then converts the extracting address d₂into another new extracting address by calculating 1600 - d₂. Also, in response to the control signal C_2a, the ALU 53a inverts the sign of the extracting step number d₁ output from the EN 52a.

After step S_6a or if d₂^&lt; 800 at step S_5a , the CTL 55 discriminates, at step S_7a , whether the calculated extracting address d₂ is equal to or smaller than zero.

If d₂≦ 0 at step S_7a, then, at step S_8a, the CTL 55a provides still another control signal C_3a to the ADL 54a. In response to the control signal C_3a, the ADL 54a inverts the sign of the thus calculated extracting address d₂ and latches it as an extracting address d₃. Also, in reponse to the control signal C_3a, the ADL 54a changes a sign code d₄ of the extracted sampled data D_i. Accordingly, at step S_8a, the operations d₃= -d₂, and d₄= -d₄ are carried out in the ADL 54a. After step S_8a, the CTL 55a provides a still further control signal C₄ to the ALU 53a, which then inverts the sign of the extracting step number d₁.

If d₂^&gt; 0 at step S_7a , then, at step S_10a , the ADL 54a latches the thus calculated extracting address d₂as the extracting address d₃, and outputs the noninverted sign code d₄.

At step Slla ' the SD 51a is accessed by the extracting address d₃ so that a sampled data D_i stored in the address d₃is output from the SD 51a to an output terminal OT_bl. Also, the sign code d₄ is output from the ADL 54a to another output terminal OT_b2. In practice, the data Di x d₄ is is output from the device shown in Fig. 10.

The above-mentioned steps S_la through 5lla are repeated with the sampling frequency of 8 kHz until transmission end is detected by the CTL 55a at step 12a.

According to the above-described first and second embodiments, because the frequency of the digital FSK signal is changed by changing the extracting step number, the phases of the PCM signals or analog signals obtained from the PCM signals are not discontinuous. Therefore, at a receiver side, the digital FSK signal is received with decreased noise or with decreased or no error.

Figure 12 is a block diagram illustrating a digital signal generating device according to a third embodiment of the present invention. The device shown in Fig. 12 is for generating a digital signal which has, when converted into an analog signal, two mixed frequencies. In Fig. 12, the digital signal generating device 120 comprises an SD 51b, an EN 52b, an ALU 53b, an ADL 54b, and a CTL 55b which are similar to those shown in Fig. 10. The output of the SD 51b is connected to a D/A converter 56. The main differences from the second embodiment are that, in Fig. 12, the EN 52b stores only a single extracting step number d₁, and that the CTL 55b controls the ADL 54b so that the ADL 54b outputs a fixed address alternately in the output addresses. In the fixed address of the DK 51b, the data "0" is stored.

Thus, from the output of the SD 51b, the data "0" is output at alternate outputs. When the D/A converter 56 converts the digital data including "0" in alternate data, the analog signal obtained at the output of the D/A converter 56 includes two frequencies.

Assume that the SD 51b stores sampled data D_{1b '} D_{2b '} D_{3b '} ... obtained from the 1/4 cycle of a sine wave data having a frequency f₀, as shown in _Fig. 9. When the sampled data D_1b , D_2b , D_3b, ... are sequentially read, with the extracting step number d₁, from the SD 51b in the way similar to that mentioned with reference to Fig. 11 of the second embodiment, and when the read sampled data are converted into analog signal by the D/A converter 56, a sine wave data SW_A having a frequency equal to f_o x d₁ can be obtained as shown in Fig. 13A.

Conversely, when alternate outputs from the SD 51b are made "0", that is, when the data "0" is output from the SD 51b instead of outputting the sampled data D_{2b '} D_{4b '} D_{6b '} ..., the D/A converter converts the digital data D_1b, "0", D_3b "0", D_5b "0", ... into an analog signal SW_B, as shown in Fig. 13B. The analog signal SW_B contains two frequencies f₁ and f₂. When the sampling frequency f is 8 kHz, then the two frequencies f₁ and f₂ are f₀ x d₁ and 4 kHz - f₀x d_l. For example, assume that the fundamental frequency f₀ of the fundamental sine wave data is 2.5 Hz, and that the extracting step number d₁ is 500, then the two frequencies are 1250 Hz and 2750 Hz.

Figure 14 shows a frequency spectrum of the two signals included in the analog signal SW_B. As shown in Fig. 14, the two frequencies f₁ and f₂ = 4 kHz - f₁are placed symmetrically with respect to 2 kHz, and the two frequencies are present within 4 kHz.

This can be theoretically proven as follows. Assume again that the sampling frequency f is 8 kHz, and that the fundamental frequency of the fundamental sine wave is f₀: Then the digital sampled data D_nb is

<maths id="math0001" num=""><img file="EP0142847A2_D0001.tif" /></maths>Consider another fundamental sine wave having a fundamental frequency equal to 4000 - f_l. Then the digital sampled data D_nc is<maths id="math0002" num=""><img file="EP0142847A2_D0002.tif" /></maths>

For the sake of simplicity, assume 2πf₀d₁/8000 = θ. Then, <ul id="ul0002" list-style="none"><li>D_nb+ D_nc= sinen + sin(π - θ)n</li><li>= sinθn + sinπn-cosθn -sinθn·cosπn</li><li>= (1 - cosπn) sinθn</li></ul>

If n is an odd number, Dnb + Dnc = 2sinθn

If n is an even number, Dnb + Dnc = 0

Accordingly, by making alternate extracted data from the SD 51b to be "0", the analog signal at the output of the D/A converter 56 becomes the superimposition of two sine waves having the frequencies f₀d₁ and 4 kHz - f₀d₁ , respectively.

Referring back to Fig. 12, the operation of the device 120 is almost the same as the operation of the device shown in Fig. 10. Therefore, the control signals C_{2a '} C_3a, and C₄, and the extracting addresses d₂ and d₃, are the same as those in Fig. 10. The differences between the devices in Figs. 10 and 12 are that, in Fig. 12, the EN 52b is not controlled by the CTL 55b but provides the predetermined single extracting step number d₁, and that the CTL 55b provides a still further control signal C₅ to the ALU 53b and to the ADL 54b whenever the calculated extracting addresses d₂ are output twice.

Figure 15 is'a diagram explaining the operation of the device shown in Fig. 12. In Fig. 15, the SD 51b outputs digital data D_0c , D_1c, D_2c , ... at times t₀, t₁ , t₂, ..., respectively. Alternate data D_1c. D_3c , D_5c ... are "0". The extracting period is equal to the sampling period, i.e., 125 us. The extracting number d₁ is assumed to be 300. The control signal C₅ becomes "1" at alternate times t₀ , t₂ , t₄ , t₆ , .... Because of this, the extracting address d₃ output from the ADL 54b is made to be zero at alternate times t₀, t₂ , t₄ , t₆, .... The control signal C₄ is turned from "0" to "1" when the calculated address d₂ output from ALU 53b becomes greater than the number of sampling steps in the range between 0 and π/2 in phase, i.e., when d₂becomes greater than 800. The control signal C₄ is turned from "1" to "0" when d₂ becomes negative, and the control signal C_3a is turned from "0" to "1" when the control signal C₄ changes its logic level, and the "1" level of the control signal C_3a continues for the sampling period of 125 µs. The level of the sign code d₄ changes when the control signal C₄ changes from "1" to "0".

At the time t₀, d₂ and d₃ are both preset values equal to zero in this example. Since C₅ is "1", d_3a for accessing the SD 51b is zero.

At the time t_l, d₂ is calculated as d₁ + d₃, i.e., d₂ = 300. Also, d₃ = d₂ = 300. Since C₅is "0", d_3a= d₃= 300.

At the time t₂, d₂ is d₁+ d₃= 600, and d_{3 =}d₂⁼ 600. Since C₅ = "1", d_3a= zero.

At the time t₃, d₂= d_l+ d₃= 300 + 600 = 900. Since d₂ &gt; 800, the control signal C_2a rises. Thus, d₃⁼ 1600 - d₂= 700. d_3a⁼ d₃ because C₅ is "0". Also, since d₂ &gt; 800, the control signal C₄ rises. While C₄ is "1", the extracting step number d₁ = -300.

At the time t₄, C2a turns to "0", and C₅ turns to "1"; d₂ is d₁+ d₃= -300 + 700 = 400; d₃ is updated to 400; and since C₅ is "I", d_3a is zero.

At the time t₅ d₂ is calculated as d₁+ d₃= -300 + 400 = 100. Since C₅ is "0", d₃ and d_3a are equal to d₂, i.e., 100.

At the time t₆, d₂ is calculated as d_l⁺ d₃⁼ -300 + 100 = -200. Since d₂ is negative, C_3a rises and C₄falls. In response to the rise of C_3a , d₄ falls and the negative sign is applied to the D/A converter 56 (Fig. 12). Because C₄ is "0", d₁ is _'changed from -300 to +300, and d₃ is updated to -200. Then because C_3a is "1", d₃ is further changed to +200, and because C₅ is "1", C_3a is zero.

At the time t₇, d₂ is calculated as d₁⁺ d₃⁼ 300 + 200 = 500. Thus d₃is updated to 500, as since C₅ is "0", d_3a is d₃= 500.

At the times t₈, t₉, t₁₀ , t₁₁ , ..., the extracting addresses d₂ , d₃, and d_3a can be calculated in a way similar to the above.

Where the extracting step number d₁ is 300, two frequencies, i.e., 300 x 2.5 = 750 Hz and 4000 - 750 = 3250 Hz, are mixed in the output analog signal from the D/A converter 56.

The extracting step number d₁ can be arbitrarily set in the EN 52b. Also, the sampling frequency f can be arbitrarily selected. Therefore, the two frequencies f₀ x d₁ and <maths id="math0003" num=""><img file="EP0142847A2_D0003.tif" /></maths> - f₀ x d₁ can be arbitrarily changed.

Figure 16 is a block diagram of a digital signal generating device according to a fourth embodiment of the present invention. The digital signal generating device 160 shown in Fig. 16 is very similar to the device 120 shown in Fig. 12. The same portions, signals, and extracting addresses are represented by the same reference symbols in both Figs. 12 and 16. The differences between Figs. 12 and 16 are that, instead of accessing the address zero in alternate accessing timings, the device shown in Fig. 16 has a data storing unit (DS) 57 for storing the fixed data "0", and a switching circuit 58 for selecting the output of the SD 51c, or the output of the DS 57, in response to the control signal C₅ which turns to "1" at alternate extracting timings, as shown in Fig. 15. It is apparent that the device shown in Fig. 16 can also provide a digital signal including the arbitrarily determined two frequencies f₀ x d₁ and 2 - f_O x ^d_l.

The present invention is not restricted to the above described embodiments. Various changes and modifications are possible. For example, in place of changing the alternate extracting addresses d₃ to be "0", as described with reference to Fig. 12, an address generating unit for generating the fixed address for accessing the "0" data may be provided between the ADL 54b and the SD 51b so that, instead of the fixed address from the ADL 54b, the fixed address .from the address generating unit may be applied to the SD 51b.

From the foregoing description, according to the present invention, a digital signal generating device can generate a digital FSK signal with no discontinuity in phase and amplitude. Therefore, at a receiver side, the digital FSK signal is received with decreased noise and with a decrease in the number of errors detected.

Further, the digital signal generating device, according to the present invention, can output two mixed digital signals from a single ROM so that the cost of the device is decreased in comparison with the conventional device employing two oscillators.

Still further, the frequency of the digital FSK signal or the frequencies of the two mixed frequencies can be easily changed.

Accordingly, the present invention is advantageous in, for example, telephone-communication line systems employing PCM channels.