A parallel-to-serial converter.

0001This invention relates to a parallel-to-serial converter for multiplexing a plurality of parallel input data into a serial data, comprising: <ul id="ul0001" list-style="none"><li>(a) means for multiplying an input clock synchronised with the parallel input data and generating a multiple clock synchronised with the serial data;</li><li>(b) means for dividing the multiple clock from said multiplying means and generating a divided clock synchronised with the parallel input data;</li><li>(c) means for converting the parallel input data into a serial output data in synchronism with a divided clock from said dividing means;</li><li>(d) means for shaping a waveform of the serial data output from said converting means using the multiple clock from said multiplying means and outputting multiplexed data.</li></ul>

0002Recently, transmission apparatus with high speed operation has been achieved. The current transmission rate in such an optical communication system has reached the order of gigabit/second.

0003To obtain such a high speed transmission rate, it is conventional practice to use multiplex techniques. Since a parallel-to-serial conversion is considered a simple method of multiplexing, a parallel-to-serial converter is often used for a multiplexer operated under high speed.

0004Moreover, it has also been known to employ a synchronous multiplexing network, such as Integrated Services for Digital Network ("ISDN") and Synchronous Optical Network ("SONET"), to easily connect the communication systems. The basic transmission rate is selected to several tens of megabit/second and a multiplexing pulse code modulation ("PCM") hierarchy is set to an integer multiple of this basic transmission rate.

0005In this synchronous multiplexing system, each network and each transmission line are synchronized with each other in order to easily connect one network to another and to form a simple and less costly multiplexer by parallel-to-serial conversion.

0006Therefore, a parallel-to-serial converter is now frequently employed for a multiplexer to realize a stable high speed data transmission yet less costly system.

0007As transmission speed has increased, however, converting a plurality of parallel input data into a serial data adversely reduces the phase margin in the parallel-to-serial converter, in view of a phase relation between a clock signal deciding an output timing of each of the parallel input data and the parallel input data themselves. Accordingly, in such a high speed operation, a rising time and falling time of a waveform of an input data, and a set up time and holding time of circuit elements in a parallel-to-serial converter cannot be neglected so that the phase margin of input data is occupied by the above rising and falling time and the set up and holding time.

0008When the phase margin becomes reduced, it is difficult to realize a stable operation and a higher speed operation of a parallel-to-serial converter.

0009It is an object of the present invention to improve the phase margin of a parallel-to-serial converter.

0010It is a further object of the present invention to realize a stable and faster parallel-to-serial converter.

0011Additional objects and advantages of the invention will be set forth in part of the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

0012According to the invention, the data converter initially defined is characterised by: <ul id="ul0002" list-style="none"><li>(e) means for comparing the phase relationship between the divided clock and the input clock that is synchronised with a set of said parallel input data; and</li><li>(f) means for adjusting the phase relationship between the set of parallel input data and the divided clock in accordance with said comparison so that the phase of the divided clock is adjusted to be in a predetermined relationship with the parallel input data.</li></ul>

<b>Brief Description of the Drawings</b>

:

0013The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention. <ul id="ul0003" list-style="none"><li>FIG. 1 is a circuit diagram of the prior art parallel-to-serial converter;</li><li>FIG. 2 is a timing chart illustrating signal waveforms at respective points of the circuit diagram of FIG. 1 for explaining the operation of the prior art converter of FIG. 1;</li><li>FIG. 3 is a block diagram of an embodiment of a parallel-to-serial converter in accordance with the present invention;</li><li>FIG. 4 is a block diagram of another embodiment of a parallel-to-serial converter in accordance with the present invention;</li><li>FIG. 5 is a timing chart illustrating signal waveforms at respective points of the parallel-to-serial converter of FIGS. 3 and 4;</li><li>FIG. 6 is a circuit diagram for the parallel-to-serial converter of FIG. 3;</li><li>FIG. 7 is a timing chart illustrating signal waveforms at respective points in the circuit diagram of FIG. 6;</li><li>FIG. 8 is another circuit diagram for the parallel-to-serial converter of FIG. 3;</li><li>FIG. 9 is a timing chart illustrating signal waveforms at respective points of the circuit diagram of FIG. 8;</li><li>FIG. 10 is a circuit diagram for the parallel-to-serial converter of FIG 4; and</li><li>FIG. 11 is a timing chart illustrating signal waveforms at respective points of the circuit diagram of FIG. 10.</li></ul>

<u style="single">Detailed Description of the Preferred Embodiments</u>:

0014Before describing the present invention, the prior art parallel-to-serial converter will be described and then the reason for and method of improving a phase margin will be explained.

0015FIG. 1 is a circuit diagram of the prior art parallel-to-serial converter. In FIG. 1, a multiplexing circuit 13 receives a pair of input data D₁, D₂ for parallel-to-serial conversion and outputs a serial multiplex data MD₁. These input data D₁, D₂ are synchronizing with each other. An input clock φ0 is also synchronizing with these input data D₁, D₂. A flip-flop circuit 15 is provided at the output of multiplexing circuit 13 for discriminating and shaping multiplex data MD₁ and for outputting a multiplexed data MD₂. A multiplier 17, connected to input clock φ₀, outputs multiple clock φ₃ by multiplying two times a frequency of the input clock φ₀. A dividing circuit 19 divides a frequency of multiple clock φ₃ into 1/2 and generates a couple of divided clocks φ₁ and φ₂ having reverse polarities respectively.

0016The multiplexing circuit 13 has three NOR circuits to convert the input data D₁, D₂ into serial multiplex data MD₁ by alternately selecting such input data with the phase of frequency divided clocks φ₁, φ₂ input from the dividing circuit 19. Namely, the input data D₁, D₂ are read out one-by-one when the divided clocks φ₁, φ₂ are in the "L" or "0" level and either this read out input data D₁ or D₂ is output as the multiplex data MD₁. The multiplex data MD₁ is then input to the flip-flop circuit 15, discriminated "0" or "1", and shaped therein by the multiple clock φ₃ from the multiplier 17 through reformation. This is then the output as the multiplex data MD₂.

0017A time chart of respective signals in the prior art parallel-to-serial converter of FIG. 1 is shown in FIG. 2. Operation of this parallel-to-serial converter will be further explained with reference to FIG. 2.

0018As shown in FIG. 2, the input data D₁, D₂ are mutually shifted in the phase by π (radian) to keep the sequence between the two input data D₁, D₂. Phase relations of the input data D₁, D₂ and input clock φ₀ are previously adjusted by the phase adjusting circuit not illustrated so that they are in a constant phase relation.

0019The multiplexing circuit 13 selects the parallel input data D₁, D₂ in the timing of divided clocks φ₁, φ₂ to convert these parallel data into the series data as the time series multiplex data MD₁. Namely, the multiplexing circuit 13 selects and outputs the input data D₁ when the divided clock φ₁ is in the "0" or "L" level and the input data D₂ when the divided clock φ2 is in the "0" or "L" level.

0020This multiplex data MD₁ is discriminated "0" or "1" and shaped by the flip-flop 15 in the rise timing of the clock φ₃ and is then output as the multiplex data MD₂ in the defined phase and time interval.

0021In this prior art parallel-to-serial converter, the multiple clock φ₃ output from the multiplier 17 is divided in the dividing circuit 19 and the divided clocks φ₁, φ₂ are used as the multiplexing clocks in the multiplexing circuit 13 in order to synchronize the divided clocks φ₁, φ₂ with the multiple clock φ₃.

0022In the prior art parallel-to-serial converter shown in FIG. 1, the polarity of divided clocks φ₁, φ₂ output from the dividing circuit 19 cannot be defined because of the nature of discrimination flip-flop 15 so that it is possible, for example, that the converter provides the divided clocks <o ostyle="single">φ₁</o>, <o ostyle="single">φ₂</o> having polarity inverted from divided clocks φ₁, φ₂. Therefore, in case the read timing defined by divided clocks φ₁, φ₂ is set to be located at the center of input data D₁, D₂ on the occasion of reading the input data D₁, D₂ by the multiplexing circuit 13, if the polarity of divided clocks φ₁, φ₂ is inverted and the divided clocks <o ostyle="single">φ₁</o>, <o ostyle="single">φ₂</o> appears, the data is read at the data converting point of the input data D₁, D₂ and read error may be generated.

0023Accordingly, in the prior art parallel-to-serial converter, the timing of reading the input data D₁, D₂ with the divided clocks φ₁, φ₂ is set to the front half or back half position of the input data D₁, D₂ and thereby read error is no longer generated if the polarity of divided clocks φ₁, φ₂ is converted.

0024However, using only the front half or back half position of input data D₁, D₂ for a read out timing substantially reduces the phase margin to a half. With recent advancement in transmission rates, a high transmission rate system cannot set the rising time t(r) or falling time t(f) of the input data D₁, D₂ to a sufficiently higher value relative to the transmission rate and also cannot neglect the setup time and hold time of the discrimination flip-flop 15. Accordingly, such a high transmission rate system would significantly reduce the phase margin in the prior art parallel-to-serial converter.

0025In accordance with the invention, it has been found that the phase margin must be improved by conducting discrimination at the center point of each of the input data in a parallel-to-serial converter. Block diagrams of two preferred embodiments of the present invention are shown in FIG. 3 and FIG. 4.

0026The first preferred embodiment of a parallel-to-serial converter of the present invention is shown in FIG. 3. A multiplier 33 generates a clock φ(1) having N-multiple frequency of an input clock φ(o). As used herein, N is an integer which is equal to 2 or larger than 2. A dividing circuit 34 generates N divided clocks φ(2)₁ ∼φ(2)_n by dividing the frequency of multiple clock φ(1) of the multiplier 33 into 1/N. A multiplexing circuit 31 receives in parallel the input data D₁ ∼ D_n to be converted into a series data MD₁ and multiplexes input data D₁ ∼ D_n in accordance with the phase of N divided clocks φ(2)₁ ∼ φ(2)_n from the dividing circuit 34. As previously described with reference to the converter of FIG. 1, the input data D₁ ∼ D_n are synchronizing with each other and the input clock φ(0) is also synchronizing with these input data D₁ ∼ D_n. A decision and shaping circuit 32 discriminates "0" or "1" of the multiplex data MD₁ received by circuit 32 from the multiplexing circuit 31 and shapes such data MD₁ with multiple clock φ(1) received from the multiplier 33. A phase control circuit 35 compares the phase of the input clock φ(0) with that of divided clock φ(2) of the dividing circuit 34 and controls the output phase of divided clock φ(2) of the dividing circuit 34 so that the divided clock φ(2)₁ ∼ φ(2)_n synchronizes with the input clock φ(0).

0027The above embodiment is based on the concept of controlling a phase of the divided clock φ(2)₁ ∼ φ(2)n for improving a phase margin. Namely, by controlling a phase of the divided clock φ(2)1 ∼φ(2)_n, a phase relationship between the parallel input data D₁ ∼ D_n and the divided clock φ(2)1 ∼φ(2)_n can be adjusted so that a phase of the divided clock is positioned at the center of a phase of the parallel input data D₁ ∼ D_n.

0028Another preferred embodiment of the parallel-to-serial converter of the present invention is shown in FIG. 4. A multiplier 33 generates a multiple clock φ(1) from an input clock φ(0). A dividing circuit 34 generates N divided clocks φ(2)₁ ∼ φ(2)_n by dividing the frequency of multiple clock φ(1) from the multiplier 33 into 1/N. A phase shift circuit 26₁ ∼ 26_n respectively shifts phases of input data D₁ ∼ Dn to be converted into d₁ ∼ d_n by a predetermined value. A selection circuit 27 selects either input data D₁ ∼ D_n or shifted input data d₁ ∼ d_n. A multiplexing circuit 31 receives in parallel the input data D′₁ ∼ D′_n from the selection circuit 27 and multiplexes D′₁ ∼ D′_n into MD₁ in accordance with the phase of N divided clock φ(2)₁ ∼ φ(2)_n from the dividing circuit 34.

0029As was explained with reference to the embodiment of FIG. 3, input data D₁ ∼ D_n and input clock φ(0) are synchronizing with each other. A decision and shaping circuit 32 discriminates "0" or "1" of the multiplex data MD₁ from the multiplexing circuit 31 and shapes such multiplex data MD₁ with the multiple clock φ(1). A selection control circuit 28 compares the phase of the divided clock φ(2) with that of input clock φ(0) and controls a changeover operation of the selection circuit 27.

0030This second preferred embodiment is based on the concept of controlling a phase of the input data D₁ ∼ D_n instead of the divided clock φ(2)₁ ∼φ(2)_n in order to improve a phase margin. Namely, by controlling a phase of the parallel input data D₁ ∼ D_n, a phase relationship between the parallel input data D₁ ∼ D_n and the divided clock φ(2)₁ ∼φ(2)_n can be adjusted so that a phase of the divided clock is positioned at the center of a phase of the parallel input data D′₁ ∼ D′_n.

0031A time chart of signals at respective points for the parallel-to-serial converters of FIGS. 3 and 4 is shown in FIG. 5. With respect to the converter of FIG. 3, phase control circuit 35 compares the phases of input clock φ(0) and divided clock φ(2) and thereby controls the output phase of the dividing circuit 34 so that data is read out at the center position of the data D₁ ∼D_n to be converted by the divided clocks φ(2)₁ ∼φ(2)_n in the multiplexing circuit 31.

0032In the parallel-to-serial converter of FIG. 4, if the phase of data D₁ ∼ D_n to be converted and the divided clocks φ(2)₁ ∼ φ(2)_n are mutually deviated in the multiplexing circuit 31, such deviation can be detected by comparing the input clock φ(0) and the divided clock φ(2) in the selection control circuit 28 because the input clock φ(0) is synchronized with data D₁ ∼ D_n. The changeover operation is carried out in the selection circuit 27 depending on such detection and thereby the data d₁ ∼ d_n to be converted which have been shifted a specified amount in the phase shifting circuit 26₁ ∼ 26_n are supplied to the multiplexing circuit 31. As a result, the data read out operation is carried out at the center position of data d₁ ∼ d_n to be converted by the divided clocks φ(2)₁ ∼ φ(2)_n in the multiplexing circuit 31 and thereby the phase margin can be improved.

0033Further details and explanation of the embodiments of FIGS. 3 and 4 will now be made with reference to FIGS. 6 to 11.

0034FIG. 6 is a circuit diagram for the parallel-to-serial converter of FIG. 3. This circuit includes a multiplexing circuit 31, a decision-shaping flip-flop 32, a multiplier 33, a dividing circuit 34 and a phase comparison circuit 35. The dividing circuit 34 includes a flip-flop having a set pulse input terminal. When the set pulse is applied to the set pulse input terminal, the polarity of output Q (namely, the divided clock φ₁) becomes "H" or "1".

0035Moreover, the phase comparison circuit 35 includes a NOR circuit with three inputs. The input clock φ₀, divided clock φ₁ from the dividing circuit 34 and the clock φ₃′ obtained by delaying the phase of the multiple clock φ₃ from the multiplier 33 with the predetermined degree are such three inputs to circuit 35. When the three inputs are all "0" or "L", the set pulse φ₄ is output to the set pulse input terminal of the dividing circuit 34.

0036Operation of this embodiment is explained by referring to FIG. 7 which is a time chart of signals thereof at respective points of the circuit of FIG. 6. As explained with reference to FIGS 1, 3 and 4, input data D₁, D₂ and input clock φ₀ are synchronizing with each other. In this embodiment, when the divided clocks φ₁, φ₂ are in the "L" level, corresponding NOR circuits are open and the input data D₁, D₂ are read out from a multiplexing circuit 31 via corresponding NOR circuits. Here, the input clock φ₀ has previously been adjusted in its phase so that the "H" level thereof is positioned at the center of input data D₁. In this embodiment the phases of divided clocks φ₁, φ₂ are adjusted with reference to the phase of this clock input φ₀.

0037Namely, the output phase of dividing circuit 34 is adjusted so that the polarities of input clock φ₀ and divided clock φ₁ have inverted polarities with each other. This polarity inversion can be achieved by outputting the set pulse φ₄ to invert the polarities of outputs Q, <o ostyle="single">Q</o> of the dividing circuit 34 when both input clock φ₀ and divided clock φ₁ are in the "L" level (refer to time t₁ in FIG. 7). Thereafter, the input data D₁, D₂ are thereby read out respectively by the divided clocks φ₁, φ₂ at the center position of these clocks.

0038The clock φ₃′ to be input to the phase comparison circuit 35 is used for narrowing the comparison period so that an erroneous set pulse is no longer output due to the existence of noise at the time of comparing the input clock φ₀ with the divided clock φ₁.

0039It can be readily seen that the present invention allows for a variety of modifications or changes in circuit design for the converter. For example, FIG. 8 is another alternative circuit diagram for the parallel-to-serial converter embodiment of FIG. 3. In this circuit diagram, the phase adjustment of divided clocks φ₁, φ₂ is carried out by inhibiting the multiple clock φ₃ to be input to the dividing circuit 34.

0040The circuit of FIG. 8 includes a multiplexing circuit 31, a flip-flop 32 as a decision and shaping circuit, a multiplier 33, a dividing circuit 34, a phase comparison circuit 35-1 and an inhibit circuit 35-2. The functioning of phase comparison circuit 35-1 and inhibit circuit 35-2 correspond to the phase control circuit 35 of FIG. 3. The phase comparison circuit 35-1 has a NOR circuit 39 which receives the input clock φ₀ and divided clock φ₁ and a flip-flop 38 which receives an output of the NOR circuit 39 at the data input terminal D and the clock φ₃′ having a predetermined phase delay of the multiple clock φ₃ at the clock input terminal C of flip-flop 38. The output pulse φ₅ of the flip-flop 38 is output to the inhibit circuit 35-2.

0041The inhibit circuit 35-2 is formed by a 2-input NOR circuit and receives the input pulse φ₅ and multiple clock φ₃ respectively and supplies output clock φ₆ to the dividing circuit 34.

0042Operation of this embodiment will be explained hereunder with reference to FIG. 9. FIG. 9 is a time chart of signals at respective points of the circuit of FIG. 8. In this embodiment, when the polarity of divided clock φ₁ is inverted from the normal polarity, it is detected (refer to the time t₁ in FIG. 7) in the phase comparison circuit 35-1 from the fact that both input clock φ₀ and divided clock φ₁ are in the "O" or "L" level at the rising edge of the clock φ₃′. Thereby the output pulse φ₅ is sent to the inhibit circuit 35-2 and one pulse of the multiple clock φ₃ inhibited by this inhibit circuit 35-2 is extracted and then supplied to the dividing circuit 34 as the clock φ₆.

0043Accordingly, the polarities of divided clocks φ₁, φ₂ output from the dividing circuit 34 are inverted and thereafter the input data D₁, D₂ are read out at the center position by the divided clocks φ₁, φ₂.

0044A circuit diagram for the converter embodiment of FIG. 4 is shown in FIG. 10. In this circuit diagram for the converter of FIG. 4, the phases of input data D₁, D₂ are shifted by π(radian) (instead of the FIG. 3 converter where the polarities of divided clocks φ₁, φ₂ are inverted) when the phases of divided clocks φ₁, φ₂ for reading out the input data D₁, D₂ are inverted with respect to the phase of the input clock φ₀.

0045The operation of the circuit of FIG. 10 is now described with reference to FIG. 11. FIG. 11 is a timing chart illustrating signal waveforms at respective points of the circuit of FIG. 10.

0046As described for the converter of FIG. 6 and the respective timing chart of FIG. 7, the input clock φ₀ has previously been adjusted in its phase so that the "H" or "1" level thereof is positioned at the center of input data D₁. See FIGS. 11(a) and (b).

0047This circuit of FIG. 10 includes phase delay circuits 26₁, 26₂ for delaying input data D₁, D₂ with π rad respectively. See FIGS. 11(b),(c) and (h), and (i)). That is, the polarities of input data D₁, D₂ are inverted. A selection circuit 27 selects either one of the input data D₁, D₂ or delayed data d₁, d₂ in accordance with selection signals SEL₁ and SEL₂ from a control circuit 28. A control circuit 28 compares a phase of the clock φ₀ with that of the divided clock φ₁ and controls changeover operation of selection circuit 27 depending on the result of the comparison.

0048As mentioned previously, the input clock φ₀ is adjusted so that the "H" level thereof is positioned at the center of an input data D₁ (FIGS. 11(a) and (b)). This means that the phase of the input clock φ₀ coincides with the phase of the divided clocks φ₁ and φ₂. However, since a control circuit comprises a flip-flop for comparing the phases of the input clock φ₀ and the divided clocks φ₁ and φ₂, this control circuit 28 cannot compare the phases if the phase of the input clock φ₀ coincides with the phase of the divided clocks φ₁ and φ₂, due to the nature of a flip flop. Therefore, the phases of the divided clock φ₁ and φ₂ from divided circuit 34 are adjusted to add πrad delay by buffer 54 and delay circuit 51(FIGS. 11 (a), (e) or (g)), so that a flip-flop of control circuit 28 can compare the phase relationship between the input clock φ₀ and a divided clock φ₁, that is, between the input data D₁ and a divided clock φ₁.

0049When the phase relationship between the input clock φ₀ and a divided clock φ₁ is normal, as shown in FIGS. 11(a), (d), (e), a selection signal SEL₁, which is an output from Q terminal of control circuit 28, is at an "L" level, and a selection signal SEL₂, which is an output from <o ostyle="single">Q</o> terminal of control circuit 28, is at an "H" level, so that a selection circuit 27 selects input data D₁ and D₂ for outputs D′₁ and D′₂ (FIGS. 11(b),(h)). Consequently, a multiplexer converts parallel input data D′₁ and D′₂ into a serial multiplexed data MD₁ as shown in FIG. 11(j).

0050To discriminate a multiplexed data MD₁ at the center thereof by a multiple clock φ₃ (FIG. 11(k)), a phase of a multiple clock is adjusted by buffer 52 as shown in FIG. 11(1).

0051On the other hand, when the phase relationship between the input clock φ₀ and a divided clock φ₁ is in reversed condition, that is, a polarity of a divided clock φ₁ from a divider 34 is inversed as shown in FIG. 11(f), selection signals SEL₁ and SEL₂ from control circuit 50 turn into "H" and "L" level, respectively. Then selection circuit 27 selects input data d₁ and d₂ for outputs D′₁ and D′₂ (FIG. 11(c),(i)). A multiplexer, then converts parallel input data D′₁ and D′₂ into a serial multiplexed data MD₁ as shown in FIG. 11(m). A multiple clock φ₃ is delayed by buffer 52 like FIG. 11(o).

0052Furthermore, a buffer 53 is just a timing interface to a next circuit and includes a flip-flop.

0053The embodiments of the converter have been described above as performing parallel-to-serial conversion of two input data. However, the present invention is not limited only to such input data. For example, when there are N parallel input data, a multiplier multiplying input clock frequency by N and a dividing circuit dividing a multiple clock frequency by 1/N can be used instead of the multiplier multiplying input clock frequency by 2 and the dividing circuit dividing a multiple clock frequency by 1/N as shown in FIGS. 3, 4, 6, 8 and 10. It will be apparent to those skilled in the art that modifications and variations could be made to the parallel-to-serial converter in accordance with the teachings of the invention.