DE69814565T2 - RECEIVER WITH A SIMPLIFIED SAMPLE RATE CONVERTER - Google Patents

Description

The present invention relates focus on a recipient with a first filter for deriving a first signal with a first sampling frequency from an input signal, and with a second Filter for deriving a second signal at a second sampling rate from the first signal.

The invention also relates to a sampling rate converter for converting a first signal with a first sampling rate into a second signal with a second Sampling rate.

A recipient according to the introduction is from the U.S. Patent No. 5,058,107 known.

Such receivers can be used to receive (digital) Broadcast signals, e.g. B. digital audio and video signals (DAB and DVB) can be used. In such a receiver after converting the RF signal at the input into a digital signal it was converted down to a signal with a frequency that is significantly lower than the frequency of the RF signal. The implementation the analog signal into a digital signal is done by a Analog-to-digital converter of the analog signal with a first Sampling rate and samples it into a digital signal with the said implemented the first sampling rate. Generally the first sampling frequency is firm, and so chosen their value is that the sampling theorem is met.

In digital signal receivers, it is convenient to process the digital signal at a second sampling rate, which is a multiple of the character speed of the received signal is. This processing can be done by filtering with a so-called half-Nyquist filter include. To do this to be able the signal with the first sampling rate must be converted into a signal with a second sampling rate can be converted. So-called aliasing To avoid, the first filter is used to filter the first signal arranged in such a way that the sampling theorem for the second Signal still met becomes. If the second sampling rate is significantly lower than that first sampling rate, the first filter has a very narrow transition range, in which the weakening of the first filter from the low value in the pass band the high value in the restricted area increases. Such a narrow transition area requires a fairly complex filter.

The present invention has Task, a recipient according to the introduction to create the complexity of the first filter significantly was reduced.

The receiver according to the invention is therefore characterized in that that the upper limit of the transition area of the first filter is larger than half the second sampling rate and less than the difference between the second sampling rate and the mentioned cutoff frequency of the second filter.

The present invention is based on the realization that the second filter does not have signals with a Frequency between the cutoff frequency of the second filter and the Difference between the second sampling rate and the mentioned cutoff frequency suppress must, because no signals from the second Filters are forwarded. It is therefore possible that the upper limit the transition area is bigger than half the second sampling rate without aliasing.

One embodiment of the invention is characterized in that the second sampling rate is a variety of possible Sampling rates include, and that the upper limit of the transition range of the first filter is larger than half of at least one of the plurality of sample rates, and less than the difference between the smallest of the possible values of the second sampling rate and the mentioned cutoff frequency of the second filter for the smallest the possible Second sampling rate values.

In some receivers, the speed of the characters have different values. Examples are voice band modems for fixed networks or in digital television technology (digital video broadcasting, DVB) used receivers. In the latter the drawing speed can be between 1 Mbaud and 45 Mbaud vary. The second sampling rate must therefore be appropriate be adjusted. By choosing the transition area accordingly the above concept of the invention gives a single filter, that can handle all drawing speeds.

A further embodiment of the invention is characterized in that the second sampling rate comprises a further plurality of sampling rates with values that are higher than the values of the plurality of sampling frequencies; that the transfer function of the first filter has a value of H (f) when the sampling rate has one of the plurality of sampling rates, and that the transfer function of the second filter has a value of H (0) - ½ {H (f - f _{SAMPLE / 2} ) + H (f + f _{SAMPLE / 2} )} is switched when the actual sampling frequency is equal to one of the further plurality of sampling rates.

The greater the number of possible Values of the second sampling rate, the steeper the first filter must be, which is an increased complexity of the filter. It can therefore be advantageous to use different filter coefficients for the to use a wide variety of sampling rates. A practical way to get these different filter coefficients is to to invert some of the filter coefficients of the first filter and subtract the average coefficient of the filter from 1.

A further embodiment of the invention is characterized in that the first filter is a time sharing filter means for performing a plurality of successive filtering operations and includes sample rate reductions, each of said successive filtering and sample rate reduction operations being performed based on the result of the previous filtering and sample rate reduction operation.

A practical way of realizing it the first filter is a cascade of filter sections and decimators. As the computational effort required decreases at each level it possible to use a filter in time multiplex to do all filtering operations perform. If e.g. B. decimators with a decimation factor of 2 are used the required computing effort for each level is a factor of 2 less than for the previous stage. If a filter with a computing speed is used, which is twice that required for the first filter stage Computing speed is any number of filter stages can be implemented.

The invention is described below Described with reference to the drawing. Show it:

1 a receiver in which the invention can be applied;

2 an embodiment of the sample rate converter 14 out 1 ;

3 Curves of the transfer function of the first and the second filter according to the invention;

4 the transfer functions of a switched first filter according to an embodiment of the invention;

5 an alternative embodiment of the first filter according to the invention.

In the receiver 2 out 1 becomes a first input of a tuner 4 an input signal supplied. A control signal carrying a signal representing the selected channel becomes a second input of the tuner 4 fed. An IF output of the tuner is connected to an input of a SAW filter (Surface Acoustic Wave Filter, surface wave filter) 6 connected. An output of the SAW filter 6 is with an input of a demodulator 8th connected.

A first output of the demodulator 8th , which carries an in-phase signal I, is with an input of an analog-to-digital converter 10 connected. A second output of the demodulator, which carries a quadrature signal Q, is connected to an input of an analog-to-digital converter 12 connected. The output of the analog-digital converter 10 is with a first input of a sample rate converter 14 connected. The output of the analog-to-digital converter 12 is with a second input of the sampling rate converter 14 connected. The sampling rate converter provides an in-phase signal I _B and a quadrature signal Q _B at its output.

The tuner 4 converts the selected input signal into an IF signal with a nominal frequency of 480 MHz. In the case of a receiver that is intended for DBS reception, the IF frequency may deviate from the nominal frequency by 5 MHz due to the drift of the local oscillator in the LNC of the outdoor unit.

The SAW filter 6 ensures the selectivity between the adjacent channels. Its bandwidth is chosen to be 10 MHz wider than the maximum bandwidth of the signal to be received. The additional 10 MHz is chosen to prevent the desired signal from falling out of the transmission range of the SAW filter due to tolerances and drift of the LNC.

In the demodulator 8th the input signal is mixed with a quadrator oscillator signal with a fixed frequency of 480 MHz. The demodulated signals I and Q are available at the output of the demodulator. These quadrature signals are sampled at the first sampling rate and by the digital-to-analog converter 10 and 12 converted into digital signals. In the case of a receiver for DVB, a suitable choice for the first sampling rate mentioned is 64 MHz.

The sample rate converter 14 converts the quadrature signals sampled at the first sampling rate into quadrature signals that are sampled at a sampling rate that is twice the actual character speed. The character speed can vary between 12 and 24 MHz. The sample rate converter 14 is also seen before to prevent frequency and phase shift of the signal at its input.

In the sample rate converter 14 out 2 the in-phase signal I becomes a first input of a mixer 16 fed and the quadrature signal Q to a first input of a mixer 20 , A first output of a digitally tunable oscillator (DTO) 18 is with a second input of the mixer 16 connected and a second output of a digitally tunable oscillator 18 is with a second input of the mixer 20 connected. An outlet of the mixer 16 is connected to an input of the first (in-phase) filter, which is a low-pass filter 22 is. An outlet of the mixer 20 is connected to an input of the first (quadrature) filter, which is a low-pass filter 24 is. An output of the low pass filter 22 is with an input of a decimator 26 connected, and an output of the low pass filter 24 is with an input of a decimator 28 connected. The output of the decimator 26 is with an input of a square root Nyquist filter 30 connected and the output of the decimator 28 is with an input of a square root Nyquist filter 32 connected. The combination of decimator 26 , Square root Nyquist filter 30 , Decimator 28 and square root Nyquist filter 32 represents the second (in-phase and quadrature) filter. The decimators 26 and 28 are controlled by a clock signal that by a clock signal recovery circuit 31 is produced. The Frequency of said clock signal corresponds to twice the character _speed F _CHARACTERS

The output of the filter 30 is with an input of an AGC amplifier 34 connected, and the output of the filter 32 is with an input of an AGC amplifier 36 connected. An output of the AGC amplifier 34 , which carries the output signal I _A , is with a first input of a mixer 42 and with a first input of an AGC controller 38 connected. An output of the AGC amplifier 36 , which carries the output signal Q _A , is with a first input of a mixer 44 and with a second input of the AGC controller 38 connected. An output of the AGC controller 38 is the AGC amplifier with a control input 34 and 36 connected.

A first output of a digital oscillator 40 is with a second input of the mixer 42 connected, and a second output of the digital oscillator 40 is with a second input of the mixer 44 connected. The output of the mixer 42 is with a first input of the digital oscillator 40 connected, and an output of the mixer 44 is with a second input of the digital oscillator 40 connected.

At the mixer exits 42 and 44 the output signals I _B and Q _B are also available. The purpose of combining the mixers 16 and 20 and the digitally tunable oscillator 18 consists of a signal for the first (in-phase and quadrature) filters 22 and 24 to deliver that is free of frequency offsets. This is done by measuring the frequency of the signals I _A and Q _A and the oscillator 18 is adjusted if the frequency of the signals mentioned deviates from 0. The output signals of the mixers 16 and 20 are through the first filter 22 and 24 filtered for aliasing due to the decrease in the sampling rate in the decimators 26 and 28 to prevent.

The decimators 26 and 28 convert their input signals at the first sampling rate (e.g. 64 MHz) into an output signal at a second sampling rate that corresponds to twice the actual symbol speed of the received signal. This second sampling rate can e.g. B. vary from 24 to 48 MHz. The decimators 26 and 28 are known to the expert and z. B. in the article "Interpolation in Digital Modems - Part I: Fundamentals" by FM Gardner in IEEE Transactions on Communications, Volume 41, No. 3, March 1993, pp. 501-507, and in the article "Interpolation in Digital Modems - Part II: Implementation and performance "by L. Erup, FM Gardner and RA Harns in IEEE Transactions on Communications, Volume 41, No. 6, June 1993, pp. 998-1008. An alternative embodiment of the decimators 26 and 28 is described in U.S. Patent No. 5,349,548.

The output signals of the decimators 26 and 28 are using the filters 30 and 32 filtered that have a square root Nyquist transfer function. This transfer function can be written as follows:

In (1) is | H (f) | ² the square of the module of the transfer function of the filter 30 and 32 , f _CHARACTER is the actual character _frequency , and α is the so-called damping _factor .

The amplifiers 34 and 36 deliver together with the AGC controller 38 an output signal with a constant amplitude. The mixer 42 and 44 are provided to phase errors from the output signals of the amplifier 34 and 36 to eliminate. The output signals of the amplifiers are therefore combined with a signal from the controlled oscillator 40 mixed, the output signals I _B and Q _{B of} the mixer 42 and 44 used for comparison with a reference value to determine the phase error. The signals I _B and Q _B represent the output signals of the sampling rate converter 14 represents.

In 3 is the spectrum of the input signal of the sample rate converter 14 with the letter C designated. It is assumed that the spectrum of the input signal in the transmitter is shaped according to a square root Nyquist filter with the same value a as that used in the receiver. This spectrum has non-zero components for f <f _B , where f _B = (1 + α) f _CHARACTER / 2. Due to the non-ideal filtering and the introduction of noise and interfering signals on the transmission medium, the input signal of the sampling rate converter can 14 have significant values even at frequencies lower than f _B. Out 3 it can be seen that all signals with a frequency b between f _B and 2f _CHARACTERS - f _B through the square root Nyquist filter 30 and 32 be suppressed, such as using the curve B in 3 can be seen. The transition area of the anti-aliasing filter 22 and 24 can therefore range from f _B to 2f _CHARACTERS - f _B like the curve A in 3 represents. Using the relationship between f _B and f _CHARACTER set out above, one finds for the area of the transition area:

The above value for f _MIN is determined on the assumption that the first filter does not filter for frequencies below f _B. If filtering below f _{B is} tolerated, the value of f _{MIN may be} lower.

The table below shows the relationship between the symbol frequency and the frequencies f _MIN and f _MAX for α = 0.35.

The requirements according to the table above cannot be met by a single filter with fixed values for f _MIN and f _MAX , because the largest value of f _{MIN is} greater than the smallest value of f _MAX . One solution to this problem is to use a filter with two sets of coefficients that are switched depending on the character frequency. A possible solution is to use a first coefficient set for character frequencies from 12 to 16 MHz with a transition range from 10.8 to 15.9 MHz, and a second coefficient set for character frequencies from 16 to 24 MHz with a transition range from 16.2 to 21, 2 MHz.

According to a further aspect of the present invention, these two sets of coefficients can be implemented in a very simple manner if the transfer functions which correspond to the two filter sets are made symmetrical around 16 MHz. In 3 is this symmetry with the curves A and B shown. The relationship between the transfer functions H '(f) and H (f), which correspond to the two sets of coefficients, is given by: H '(f) = 1 - ½ {H (f - f CHARACTER / 2) + H (f + f CHARACTER / 2)} (3)

The coefficients of the filter H '(f) can be set to easily determined from the coefficients of the filter H (f). The average coefficient of the filter H '(f) can be determined by subtracted the average coefficient of the filter H (f) from 1. The remaining coefficients of the filter H '(f) can be determined by alternating the sign of the coefficients of the filter H (f) inverted starting with the middle coefficient. If the coefficients the filter H (f) is 0.1, 0.2, 0.4, 0.2, 0.6, 0.2, 0.4, 0.2, 0.1 , the coefficients of the filter H '(f) are -0.1, 0.2, -0.4, 0.2, 0.4, 0.2, -0.4, 0.2, -0 ,1.

In the first filters 22 . 24 out 5 is the input with an input of a low pass filter 50 connected. The output of the low pass filter 50 is with an input of a decimator 52 connected. The output of the decimator 52 is with an input of a low pass filter 54 connected. The output of the low pass filter 54 is with an input of a decimator 56 connected.

In order to be able to implement an anti-aliasing filter over a frequency range of several octaves, a large number of filters with decimators can be used, as in 5 shown. Each of the fil ter implements an anti-aliasing operation that is sufficient to prevent aliasing after the corresponding decimation operation has taken place. According to a further aspect of the invention, the multiplicity of anti-aliasing filters can be implemented on a single filter by means of time sharing.

If the decimation by a factor used by two is needed each subsequent filter only half the time window for the previous one Filters are required. If three filters with the numbers 1, 2, 3 are available, the allocation of the time window can be made accordingly in the table below.

The table shows that some time slots are still not used. In these could be unused time slots a fourth filter that works at half the sampling rate of the third Filter is running, can be realized, which still leaves some time slots unused stay in a fifth Filters could be implemented. This way, any number of decimation filters can be used can be realized with a single filter in time sharing mode.

Text in the drawing

1

tuner

tuner

SAW

SAW filter (Surface acoustic wave filters)

THE

demodulator

A / D

A / D converter

SRC

sampling rate converter

Claims (8)

Receiver with a first filter for deriving a first signal with a first sampling frequency from an input signal, and with a second filter for deriving a second signal with a second sampling rate from the first signal, characterized in that the upper limit of the transition range of the first filter is larger is than half the second sampling rate and less than the difference between the second sampling rate and the mentioned cutoff frequency of the second filter. receiver according to claim 1, characterized in that the second sampling rate a variety of possible Sampling rates include, and that the upper limit of the transition range of the first filter is larger than half of at least one of the plurality of sample rates, and less than the difference between the smallest of the possible values of the second sampling rate and the mentioned cutoff frequency of the second filter for the smallest the possible Second sampling rate values. Receiver according to claim 2, characterized in that the second sampling rate comprises a further plurality of sampling rates with values which are higher than the values of the plurality of sampling frequencies; that the transfer function of the first filter has a value of H (f) when the sampling rate has one of the plurality of sampling rates, and that the transfer function of the second filter has a value of H (0) - ½ {H (f - f _{SAMPLE / 2} ) + H (f + f _{SAMPLE / 2} )} is switched when the actual sampling frequency is equal to one of the further plurality of sampling rates. receiver according to one of the preceding claims, characterized in that the first filter is a time sharing filtering means for performing one Variety of sequential operations for filtering and Includes sample rate reductions, each of said successive filtering operations and sample rate reduction based on the result of the previous ones Operation for filtering and sampling rate reduction is performed. Sampling rate converter with a first filter for derivation a first signal with a first sampling frequency from an input signal, and with a second filter for deriving a second signal with a second sampling rate of the first signal, characterized in that that the upper limit of the transition area of the first filter is larger than half the second sampling rate and less than the difference between the second sampling rate and the mentioned cutoff frequency of the second filter. Sampling rate converter according to claim 5, characterized in that the second sampling rate comprises a large number of possible sampling rates, and that the upper limit of the transition area of the first filter is larger than half of at least one of the plurality of sample rates, and less than the difference between the smallest of the possible values of the second sampling rate and the mentioned cutoff frequency of the second filter for the smallest the possible Second sampling rate values. Sampling rate converter according to claim 6, characterized in that the second sampling rate comprises a further plurality of sampling rates with values which are higher than the values of the plurality of sampling frequencies; that the transfer function of the first filter has a value of H (f) when the sampling rate has one of the plurality of sampling rates, and that the transfer function of the second filter has a value of H (0) - ½ {H (f - f _{SAMPLE / 2} ) + H (f + f _{SAMPLE / 2} )} is switched when the actual sampling frequency is equal to one of the further plurality of sampling rates. Sampling rate converter according to claims 5, 6 or 7, characterized in that the first filter is a time sharing filter means to perform a variety of sequential filtering operations and includes sample rate reductions, each of the above successive operations for filtering and sampling rate reduction based on the result of the previous filtering operation and sample rate reduction is performed.