Multichannel Sine Synthesizer

This invention relates to multichannel sine synthesizer for primary signal production in measuring signal generators comprising a phase accumulator, a low-pass filter, an interposed synthesis channel and a clock control unit for converting a phase increment applied to the phase accumulator into a sine wave signal.

Synthesizers of this kind, i.e. sine wave generators based on the principle of digital signal synthesis, are known. From the magazine "Elektronik 1976", part Il, pp. 106 to 110, principles of digital synthesizers are known. Such synthesizers have a digital phase accumulator, a synthesis channel and an analog low-pass filter. The synthesis channel can be composed of a series arrangement of a sawtooth-to-sine converter and of a digital-to-analog converter (DAC). An enumerated phase increment is converted by the phase accumulator into a sequence of phase values. The advancement rate is then the clock frequency of a clock control unit. On account of the finite range of values of the phase accumulator, according to value a sequence of sawtooth-shaped phase values is obtained at the output of the phase accumulator. With a given clock frequency, the phase increment therefore determines the output frequency of the sine signal. Consequently, with sine synthesizers variable in frequency the output frequency, which is identical to the fundamental frequency of the sawtooth-shaped sequence of phase values, is controlled proportionally through the phase increment. The sawtooth-to-sine converter converts the individual elements of the sequence of phase values occurring on the input side to corresponding sine function values. The digital-to-analog converter converts these sine function values into linearly assigned voltage or current values. The resulting output signal represents a sample-and-hold function of a sinusoidal signal because the individual supporting values are kept constant within the sampling period. Efficaciously, this is realized by means of a clocked transfer memory of the digital-to-analog converter. For a phase increment .phi. larger than or equal to unity, i.e. the least significant bit (LSB) of the phase accumulator output, sampling and clocking periods are identical. The effective amplitude of the sinusoidal signals then depends according to sin(.pi.f.sub.N /f.sub.T)/(.pi.f.sub.N /f.sub.T) upon the ratio between the effective frequency f.sub.N and the clock frequency f.sub.T. The low-pass filter at the output of the synthesizer serves to attenuate the alias frequency components contained in the output signal of the digital-to-analog converter. The component both highest and lowest in frequency lies at a frequency formed from the difference between the clock frequency f.sub.T and the effective frequency f.sub.N. In order to achieve sufficient attenuation with the use of a reasonable amount of filtering means, as a maximum effective frequency f.sub.Nmax of synthesizers variable in frequency, in general a quarter of the clock frequency f.sub.T is fixed so that the critical alias frequency is at least twice the maximum effective frequency f.sub.Nmax and therefore lies a frequency octave above the maximum frequency. The relative effective amplitude is then, according to the aforementioned formula, about 91% of the effective amplitude with low frequencies. An amplitude compensation of up to 100% can be effected with synthesizers variable in frequency by an increase in height of the transmission band of the low-pass filter. An increase of the effective frequency f.sub.Nmax through, for example, f.sub.T /4 with an additional amplitude compensation is limited because of the approximation to the critical alias frequency f.sub.T -f.sub.Nmax.

From European Patent Specification 0 078 588 (corresponding to U.S. Pat. No. 4,454,486), a device and a method for synthesizing a signal waveform are known. In this variation of a multichannel synthesizer operating by means of a combination of digital signals shifted in time, in the end a high-frequency output signal is produced in which the digital signals shifted in time are supplied to a multiplexer and are then finally converted via a digital-to-analog converter into the desired signal. In such a digital synthesizer, with respect to the dynamic range, more stringent requirements are imposed on the multiplexer and on the digital-to-analog converter as the number of sampled channels is increased. More particularly, the time of conversion of the digital-to-analog converter must be adapted to the sampling period of the multiplexer.

The invention has for its object to provide a multichannel sine synthesizer which, with approximately unchanged dynamic requirements with respect to the synthesis channel and with a given clock frequency, guarantees a substantial increase of the maximum frequency of the sinusoidal signal to be produced.

According to the invention, this object is achieved in that parallel to the interposed synthesis channel at least one further modified synthesis channel, is connected. This further channel is connected on the output side through a device, which operates so as to be shifted in time and to which on the input side moreover a frequency-dependent constant phase value is applied, which receives through the clock control unit a clock signal adapted in time, and in that the sinusoidal signal has a maximum frequency increased in dependence upon the parallel synthesis channel with respect to the monochannel synthesis channel.

The multichannel sine synthesizer according to the invention permits a substantial increase of the maximum effective frequency f.sub.Nmax with respect to a quarter of the clock frequency f.sub.T because the critical lowest alias frequency is increased from f.sub.T -f.sub.Nmax to Kf.sub.T -f.sub.Nmax, K representing the number of the parallel-connected synthesis channels. The practically utilized increase of the effective frequency f.sub.N depends upon the still acceptable degradation of the effective amplitude with f.sub.N according to sin(.pi.f.sub.N /f.sub.T)/(.pi.f.sub.N /f.sub.T). With synthesizers variable in frequency, a doubling of the maximum frequency with K=2 or K=3 and at most a tripling with K=3 or K=4 can suffice so that in the first case with a maximum effective frequency f.sub.Nmax corresponding to half the clock frequency f.sub.T an amplitude drop to about 64% and in the second case with a maximum effective frequency f.sub.Nmax of three quarters of the clock frequency f.sub.T an amplitude drop to 30% must be compensated.

According to the invention, each synthesis channel has a sawtooth-to-sine converter and a digital-to-analog converter comprising a clocked transfer memory. Each of the parallel-connected synthesis channels has a sawtooth-to-sine converter, which is preceded by an adding member. The parallel-connected synthesis channels operate by samplings delayed with different times.

According to a preferred embodiment of the invention, an individual phase value .phi. according to (i-1) .phi./K is applied to the input of each parallel synthesis channel, .phi. being the phase increment applied to phase accumulator and i=1, 2, 3 . . . ,K represents a continuous enumeration of the synthesis channels. The low-pass filter is connected on the input side through a summing member to the outputs of the synthesis channels. The clocking of the transfer memories of the digital-to-analog converters takes place with a time shift according to (i=1)T.sub.T /K, T.sub.T =1/f.sub.T being the clock period. Advantageously, further collecting memories or transfer memories may be arranged in the synthesis channels in order to guarantee a perfect clocking.

Advantageously, it may further be ensured that the sawtooth-to-sine converters may be constructed as read-only memories (ROM's) or random access memories (RAM's), as a result of which also other arbitrary signal-time functions also may be synthesized as sinusoidal functions.

According to a particular embodiment of the invention, the clock control unit clocks the phase accumulator and the synthesis channels and supplies to the transfer memories of the digital-to-analog converters a clock signal delayed by (i-1)T.sub.T /K for each i.sup.th synthesis channel.

Advantageously, it may be ensured that the adding member has a subtracting function. In that case the clock control unit clocks in the end the i.sup.th synthesis channel with a delay of (K-i)T.sub.T /K.

According to a further alternative preferred embodiment of the invention, the clock control unit supplies the phase accumulator with a clock signal and supplies the synthesis channels with time clock signals. For the delay in time of the synthesis channels with respect to each other an analog dead time member having the clock time (i-1)T.sub.T /K is assigned to each of the parallel synthesis channels.

Advantageously, each dead time member may be constructed as a delay line with distributed passive network elements dL/dx and dC/dx, where L represents inductance, C represents capacitance and dx represents a differential unit length.

Advantageously, the rise and fall times of the digital-to-analog converters may be chosen so as to be short with respect to the quotient of the clock signal period T.sub.T and the factor K in order to avoid a reduction of the yield in effective amplitude.

Further preferred embodiments of the invention appear in the dependent claims.

An embodiment of the invention will now be described more fully with reference to the accompanying drawing, in which:

FIG. 1 shows a two-channel sine synthesizer according to the invention comprising a clock control unit which produces for the delayed channel a lagging clock signal,

FIG. 2 shows a two-channel sine synthesizer according to the invention comprising a dead time member for a time delay,

FIG. 3 shows two sample-and-hold functions of a summing member at the output of the synthesizers shown in FIGS. 1 and 2 and the sum of these sample-and-hold functions, and

FIG. 4 shows a general K-channel sine synthesizer according to the invention.

FIG. 1 shows a two-channel sine synthesizer 10 comprising a first synthesis channel 13, which is connected between a phase accumulator 11 and a low-pass filter 12 and which receives at its input side the output signal of the phase accumulator 11. A second synthesis channel 14 is arranged parallel to the first synthesis channel 13. It is apparent from the parallel arrangement that the second synthesis channel 14 also receives the input signal of the synthesis channel 13 and is combined on the output side with the output signal of the first synthesis channel 13. The outputs of the synthesis channels 13, 14 are combined through a summation member 15.

The output signal of the summation member 15 is connected to the input of the low-pass filter 12 whose output the smoothed sinusoidal signal is produced. The input signal of the phase accumulator is an enumerated phase increment .phi., which is designated in the Figures by reference numeral 16. The input side of the second synthesis channel 14 additionally receives a frequency-dependent constant phase value, which in the Figures is designated by reference numeral 17. With several additional parallel-connected synthesis channels, the phase values 17 of the synthesis channels are different, as will be explained more fully hereinafter.

The first synthesis channel 13 essentially comprises a series arrangement of a sawtooth-to-sine converter 18 and a digital-to-analog converter 19 which may include at the input side a transfer memory 20.

The sawtooth-to-sine converter 18 converts the signals or signal values supplied to it, i.e. the sequence of phase values, into corresponding sine function values. It may be internally constructed as a read-only memory (ROM) or as a random access memory (RAM), into which the sine function values are loaded. The phase values supplied to it represent the reading addresses of these memories. By storing other function values, arbitrary signal-time functions other than sinusoidal functions also may be synthesized.

The digital-to-analog converter 19 converts the signals supplied to it into linearly assigned voltage or current values, which are each time kept constant within a sampling period T.sub.T. The resulting output signal s.sub.1 (t) represents a sample-and-hold function of a sine oscillation, s.sub.0 (t)=A.sub.0 sin(.omega..sub.N t). A holding member is required to accomplish this end and is constituted by the transfer memory 20, which is arranged at the input of the digital-to-analog converter 19.

The second parallel-connected synthesis channel 14 also comprises a sawtooth-to-sine converter 21 and a following digital-to-analog converter 22 having a transfer memory 23. These members, which may correspond to the members 18, 19 and 20, are preceded by an adding member 24, which is supplied on the input side with the input signals already described.

The two-channel synthesizer 10 according to the invention also has a clock control unit 25, which supplies on the one hand the sawtooth-to-sine converters 18 and 21 and on the other hand the transfer memories 20 and 23 and the phase accumulator 11 with clock signals.

At the output of the second synthesis channel 14, a second sample-and-hold function s.sub.2 (t) of the sine oscillation is therefore produced. The clock control unit 25 controls the samplings in a manner such that the sample-and-hold function s.sub.2 (t) is delayed with respect to the sample-and-hold function s.sub.1 (t) or its sampling instants by half a clock period T.sub.T. This delay is obtained in that the transfer edges of the clock signal s.sub.T (t-T.sub.T /2) for in the end the digital-to-analog converter 22 lag by half a clock period, i.e. T.sub.T /2, with respect to the transfer edges of the clock signal s.sub.T (t) for in the end the digital-to-analog converter 19. The two clock signals s.sub.T (t) and s.sub.T (t-T.sub.T /2) and the clock signals for the phase accumulator 11 and, as the case may be, the sawtooth-to-sine converters 18 and 21 are produced in the clock control unit. Since the samplings, i.e. the sampling instants of the synthesis channel 14 lag behind those of the synthesis channel 13 by T.sub.T /2, the individual phase values supplied to the sawtooth-to-sine converter 21 must be increased each time by .phi./2, i.e. by half the phase increment .phi.. This takes place in the adding member 24. In the summation member 15, the two sample-and-hold functions s.sub.T (t) and s.sub.2 (t) are joined. The desired sine signal having the frequency f.sub.N is filtered out of the sum signal by a low-pass filtering in the low-pass filter 12, N representing the effective frequency.

FIG. 2 shows a two-channel sine wave synthesizer 28, which is a variation of the two-channel sine wave synthesizer 10 shown in FIG. 1. This advantageous variation of the multichannel synthesizer according to the invention is essentially distinguished only with respect to the method of delaying in time the samplings or the sampling instants of the second synthesis channel 26, whose construction otherwise corresponds to that of the synthesis channel 14, in comparison with the first synthesis channel 13. The delay in this embodiment takes place with coinciding clock signals for in the end the digital-to-analog converters 19 and 22 by an analog dead time member 27, which follows the digital-to-analog converter 22. Otherwise, the basic modes of operation are identical.

Advantages of this synthesizer 28 according to FIG. 2 as compared with that of FIG. 1 consist in that the effective delay time, which should be as exactly equal as possible to T.sub.T /2, is due substantially exclusively to passive network elements. As a result, the delay time is more stable especially with respect to temperature influences. This is the case, for example, if the dead time member 27 is constructed as a delay line 31 with distributed passive network elements dL/dx and dC/dx.

Further, the perfect timing of the two synthesis channels 13 and 26 is simplified, it being advantageous that as to the propagation delay times the two synthesis channels 13 and 26 differ from each other only with respect to the additional propagation delay time of the adding member 24 in the synthesis channel 26. It should be noted that, when a highest possible maximum frequency f.sub.N of the sinusoidal signal to be produced is indicated, the digital switching circuits of the two synthesis channels 13 and 26 are operated tendentially approximately to the clock frequency limit fixed by propagation delay times and transition times. In some cases, it may therefore be required to provide, besides the transfer memories 20 and 23 already present, additional collecting memories 30 connected to transfer memories.

FIG. 3 shows the two sample-and-hold functions s.sub.1 (t) and s.sub.2 (t) for f.sub.T /f.sub.N =8, for example, one below the other. Below the function s.sub.2 (t) the sample-and-hold function s(t) is shown, which is obtained for an imaginary sine synthesizer with double clock and sample frequency. The signal-time areas indicated by cross-hatched lines of s.sub.1 (t) and s.sub.2 (t) are composed in the envelope to s(t). The remaining signal-time areas of s.sub.1 (t) and s.sub.2 (t) are identical to the signal-time area of s(t-T.sub.T /2), i.e. a sample-and-hold function T.sub.T /2 of double the sample frequency delayed by s(t).

Thus, it holds that

The two synthesis channels 13 and 14 and 26, respectively, of the two-channel sine synthesizer, which act in time multiplex at the summation member 15, therefore increase the effective sampling rate by the factor 2. In the sum signal, the side frequencies of the odd-numbered clock frequency multiples are neutralized by compensation in the summation member 15 for the synthesis channels 13 and 14 and 26, respectively. The critical alias frequency is therefore 2f.sub.T -f.sub.N instead of f.sub.T -f.sub.N, as is the case with the synthesizer according to the prior art, and the maximum frequency can be increased with respect to the prior art to double the value, i.e. to f.sub.N max=f.sub.T /2.

The smaller yield in amplitude resulting according to (.pi.f.sub.N /f.sub.T)/(.pi.f.sub.N /f.sub.T) with the maximum effective frequency f.sub.Nmax of 64% with respect to 91% with f.sub.Nmax =f.sub.T /4 is obtained in synthesizers variable in frequency by an increase in height of the transmission band of the low-pass filter (12).

In order that no substantial additional amplitude degradation occurs, the rise and fall times of the digital-to-analog converters must be short with respect to T.sub.T /2 and short with respect to T.sub.T /K, respectively, with the K-channel synthesizer. However, the relevant requirements are considerably less stringent than in the multichannel synthesizer according to the aforementioned European Patent Specification 0 078 588 because the individual sample value of each channel is operative through T.sub.T.

FIG. 4 shows the multichannel sine wave synthesizer 30 according to the invention, which has an internal construction according to the variation in FIG. 2. The phase value, which is added to the i.sup.th synthesis channel through the respective adding member 24, is (i-1).phi./K if the undelayed interposed synthesis channel 13 is indicated by i=1. The dead times of the corresponding dead time members 27 are (i-1)T.sub.T /K. The effective sample frequency is increased with respect to the monochannel synthesizer from f.sub.T to Kf.sub.T , so that the critical lowest alias frequency lies at Kf.sub.T -f.sub.Nmax.

The features of the invention revealed in the above description, in the Figures and in the claims may be utilized both separately and in arbitrary combinations in order to realize the invention in its various embodiments.