CA2201753A1 - Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion - Google Patents
Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversionInfo
- Publication number
- CA2201753A1 CA2201753A1 CA 2201753 CA2201753A CA2201753A1 CA 2201753 A1 CA2201753 A1 CA 2201753A1 CA 2201753 CA2201753 CA 2201753 CA 2201753 A CA2201753 A CA 2201753A CA 2201753 A1 CA2201753 A1 CA 2201753A1
- Authority
- CA
- Canada
- Prior art keywords
- frequency
- signal
- discrete
- sampling
- channel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Landscapes
- Analogue/Digital Conversion (AREA)
Abstract
An RF communications receiver permits greater integration on standard silicon chips and consumes less power than previous receivers. Sub-sampling and discrete-time signal processing techniques are used to frequency down-convert, filter, amplify, and select a desired analog RF channel. A sample-and-hold circuit (81) sub-samples the desired analog RF channel of carrier frequency fc, thereby down-converting it to a discrete-time image signal of frequency fi. Successive down-sampling, (99. 1) anti-alias filtering, (95.1) and amplification of the discrete-time image signal yields a low-frequency discrete-time signal containing a down-converted channel of frequency fk. The low-frequency discrete-time signal is then digitized, (87) filtered, (91) and demodulated (93) to reveal its baseband information content.
Description
2~0 175~
W096/39750 PCT~S96/09311 Radio Freguency Signal Reception Usin~ Fre~uency Shifting by Discrete-Time Sub-Samplin~ Down-~onversion 5 Bac~o~d - Technical Field of Invention The present invention relates to radio receivers and methods for the reception of RF (radio frequency) communications signals. In particular i~ relates to radio receivers usirlg high-speed discrete-time electronic circuits and methods for RF signal 20 reception using sub-sampling for frequency down-conversion.
~ckground of the Invention and Discussion of Prior Art At the present time, the vast majority of RF communications receivers are of the traditional superheterodyne type. This type 25 of receiver uses one or more IF (intermediate frequency) stages for filtering and amplifying signals at a fixed fre~uency within an IF chain. This radio architecture has the advantage that fixed filters may be used in the local oscillator (LO) chain.
30 A block diagram of a typical superheterodyne radio receiver is shown in Figure 1. An RF signal arriving at an antenna 11 passes through a bandpass RF filter 13, an LNA (low noise amplifier) 15 and into an image filter 17 which produces a band-limited RF
signal. This band-limited RF signal then enters a first mixer 19 35 which translates the RF signal down to an intermediate frequency by mixing it with the signal produced by a first LO 21. The undesired mixer products in the IF signal are rejected by an IF
filter 23. The filtered IF signal then enters a second mixer 25 which translates it down to yet another intermediate fre~uency by 2~0 ~75`3 ~ ~ W096~97~0 PCT~S96/09311 mixing it with the signal produced by a second LO 27. The signal finally passes through an IF amplifier 29 and an IF filter 31, resulting in a channelized IF signal, i.e., a signal cont~;n;ng a particular channel isolated from the other channels present within the original band-limited RF signal. An analog demodulator 33 demodulates the channelized IF signal into a baseband information signal. Tuning into a particular channel within the band-limited RF signal is accomplished by varying the L0 frequency of LO 21.
In order to reduce size, power consumption, and cost, it would be advantageous to integrate the electronic components of radio receivers onto a single silicon chip. The superheterodyne design, however, requires high-quality, narrowband IF bandpass filters operating at high frequencies, and these components can not be built on-chip with present technology. Instead, off-chip components such as SAW (surface acoustic wave) filters or tuned circuits requiring resonating components such as inductors or cavities must be used. These external filtering components impose a lower limit to the size, materials cost, assembly cost, and power consumption of receivers built using the superheterodyne design. Moreover, the necessity for mixer and local oscillator circuits operating at high frequencies contributes greatly to the power consumption and general complexity of the superheterodyne receiver. In particular, high-frequency analog mixers often require a large amount of power to maintain linear operation.
Although many variations of the superheterodyne design exist, they all share the limitations of the particular design just described.
The growing demand for portable communications has motivated attempts to design radio receivers that permit the integration of more components onto a single chip. For example, a digital IF
receiver design, as shown in Figure 2, can be integrated further than the superheterodyne design and therefore requires fewer external components. As in the superheterodyne design, an antenna 35 couples an RF signal through an RF filter 37 into an LNA 39 and then through an image filter 41. The band-limited RF signal mixes in a mixer 43 with a L0 signal from a L0 45 to produce an IF
signal. The IF signal then is amplified in an IF amplifier 47 before being channelized in an IF filter 49. An ADC (analog-to-2 ~ O ~ 7 5 3 - ~ W096~9750 PCT~S96/09311 digital converter) 51, operating at a conversion rate determined by a clock circuit 53, digitizes the channelized IF signal into a digital IF signal. This digital IF signal is filtered and demodulated by a digital demodulator 55 and is passed on to the 5 r~m~ er of the communications system as a digital baseband information signal. Tuning is accomplished by varying the frequency of LO 45. Multiple techniques for performing the digital filtering and demodulation have been developed for this receiver design such as those described in U.S. Patents No.
lC 4,888,557 and No. 4,902,979.
Compared to the superheterodyne design, the digital IF design can cost less to manufacture and assemble because it has fewer off-chip components. Nevertheless, the digital IF design still requires several off-chip components, namely, external IF filters, a high-frequency mixer, and a high-frequency LO. U.S. Patent No.
5,339,459 describes a technique that eliminates the necessity for external high-frequency mixer and LO circuits by using a sample-and-hold circuit. By operating at a sampling rate lower than the 20 band-limited RF signal, i.e. by sub-sampling, the sample-and-hold circuit acts as a mixer to translate the RF signal into an IF
signal. Although this technique allows further integration and lower power operation, it still requires the use of off-chip components for the high-quality, narrowband IF bandpass filters 25 nee~e~ for channelization.
The digital IF design has other disadvantages as well. In order to faithfully digitize the analog signal, the ADC is required to operate at a conversion rate of at least twice the intermediate 30 frequency. Such a high-frequency ADC consumes a large amount of power to maintain the required linearity and dynamic range. This becomes even more problematic for receiver designs in which the A/D conversion is performed at the front end of the receiver, e.g., as in U.S. Patent No. 4,893,316.
A third receiver design is the direct-conversion, or zero-IF, receiver shown in Figure 3. An antenna 57 couples an RF signal through a first bandpass RF filter 59 into an LNA 61. The signal then proceeds through a second RF filter 63, yielding a band-~ ~ W096/39750 PCT~S96/09311 .
limited RF signal, which then enters a mixer 65 and mixes with anLO frequency produced by an LO 67. Up to this point, the direct-conversion receiver design is essentially the same as the two previous receiver designs.
Unlike the previous designs, however, the LO frequency is set to the carrier frequency of the RF channel of interest. The resulting mixer product is a zero-frequency IF signal-a modulated signal at baseband frequency. This signal couples into a lowpass analog filter 69 before proceeding into an analog demodulator 71 to yield the demodulated analog baseband information signal for use by the r~m~;n~er of the co~ml~nications system. Alternatively, an analog-to-digital converter can be inserted after mixer 65 and the filtering and demodulation can be done digitally, producing a 15 digital baseband information signal for use by the remainder of the com~l1nications system. In either case, tuning is accomplished by varying the frequency of LO 67, thereby converting different RF
channels to zero-frequency IF signals.
20 Because the direct-conversion receiver design produces a zero-frequency IF signal, its filter requirements are greatly simplified-no external IF filter components are needed since the zero-IF signal is an audio frequency signal that can be filtered by a low-quality lowpass filter. This allows the receiver to be 25 integrated in a st~n~rd silicon process from mixer 65 onwards, making the direct-conversion receiver design potentially attractive for portable applications.
The direct-conversion design, however, has several problems, some 30 of which are quite serious. As with the other designs described above, the RF and image filters required in the direct-conversion design must be high-quality narrowband filters that must remain off-chip. Moreover, this design requires the use of high-frequency mixer and LO circuits that require large amounts of 35 power. Additionally, radiated power from LO 67 can couple into antenna 57, producing a DC offset at the output of mixer 65. This DC offset can be much greater than the desired zero-IF signal, making signal reception difficult. Radiated power from LO 67 can also affect other nearby direct-conversion receivers tuned to the 2 2 0 ~ 7 5 3 - ~ WO 96/39750 PCT/US96/09311 same radio frequency. Furth~rm~re, to receive signals transmitted using modulation techniques (such as FM) in which access to both the lower and upper sidebands is required, two mixers and two LOs are required to produce both an in-phase and a quadrature baseband 5 signal. Not only does this increase the power required by the receiver, but also the phase between the two LO signals must be precisely maintained at 90 degrees to prevent demodulation distortion. This can be difficult to accomplish with variations in temperature and other operational parameters.
In summary, although the prior art includes various receiver designs that achieve successively greater integration, each one has significant disadvantages including one or more of the following: the necessity for several external circuit components, 5 the consumption of large amounts of power, poor signal reception, interference with other receivers, distortion, and limited dynamic range.
Objects a~d Advanta~es of the Invention 20 Accordingly, it is a primary object of the present invention to provide a radio receiver design which has increased integration, decreased materials and assembly costs, and decreased power consumption without the operational problems associated with previous receiver designs. It is a further object of the 25 invention to provide a method of frequency down-converting an RF
channel without requiring any off-chip IF filter components, multiple LOs, high frequency LOs operating at frequencies near the RF signal frequency of interest, in-phase and quadrature LOs, analog mixers, or high-speed, high-power analog-to-digital 30 converters. It is another object of the present invention to provide a method for filtering and selecting a particular channel within a channel allocation band of interest without requiring any off-chip IF components, especially components for filtering.
Further objects and advantages will become apparent from a 3s consideration of the ensuing description and drawings.
Summary of the Invention The present invention achieves the above objects and advantages by providing a new method for RF comm-~n;cations signal reception and - 2~01753 - ~ W096/39750 PCT~S96/09311 a new receiver design that incorporates this method. This method includes filtering an RF signal cont~;n;~g a channel allocation band of bandwidth B to yield a band-limited RF signal of bandwidth W. An RF channel of carrier frequency fc is included within this channel allocation band. The method further includes a sample-and-hold means to sub-sample the band-limited RF signal at a sampling rate fs ~ fc in order to obtain an image signal of the RF
channel. The image signal frequency fi satisfies the conditions fi = nfs + fc and fs - 2fi ' (W + B)/2, where n is an integer chosen to minimize Ifil Furthermore, the method includes a discrete-time down-converting means to successively down-convert, filter, and amplify the image signal of the RF channel to a low-frequency signal of frequency fk. In contrast with digital signal processing techniques, the down-converting method of the invention employs discrete-time signal processing techniques that do not require digital conversion and processing of the signal.
Description of Drawings Figure l is a block diagram of a superheterodyne receiver considered as prior art.
Figure 2 is a block diagram of a digital IF receiver considered as prior art.
Figure 3 is a block diagram of a direct-conversion receiver considered as prior art.
Figure 4 is a block diagram of a receiver constructed with the principles of the invention.
Figure 5 is a block diagram of a discrete-time signal processing block, which is a key component of a receiver constructed according to the invention.
Figure 6 is a graph showing a typical arrangement of RF channels within a channel allocation band used for common radio c~mmlln;cations.
Figure 7 is a graph showing the band-limited RF signal after passing through the noise filter of Figure 4 according to the invention.
Figure 8 is a graph showing the spectrum of image signals after the band-limited RF signal is sub-sampled according to the invention.
=
2 ~ 0 ~ 7 ~ 3 WO 96/397s0 PCT/US96/09311 Figures 9a and 9b are graphs illustrating the necessary relationships between fs, fi, B, and W to permit sub-sampling without destructive aliasing.
Figure lO pictorially illustrates the aliasing of spectral components due to 2:l down-sampling.
Figure ll is a graph showing the normalized frequency response of the discrete-time anti-aliasing filters used in the preferred embodiment.
o Description of the Preferred Embodiment - Physical Attributes Figure 4 is a block diagram of an RF communications receiver constructed in accordance with the principles of the present invention. It includes an antenna 73 for coupling an RF signal .15 into the input of an analog bandpass RF filter 75. The output of analog bandpass RF filter 75 connects to the input of an LNA 77 whose output couples to the input of an analog noise filter 79.
The output of analog RF noise filter 79 connects to the input of a sample-and-hold circuit 81, which is built using switched-capacitor circuit techniques.
In addition to its signal input which connects to the output of RF
noise filter 79, sample-and-hold circuit 81 has a clock input and a discrete-time signal output. Its discrete-time signal output connects at a node 82 to the input of a discrete-time signal processing block 85. Discrete-time signal processing block 85 is a key component of the preferred embodiment of the invention and is described in greater detail below in connection with Figure 5.
The output of discrete-time signal processing block 85 couples through node 86 to the input of an ADC 87 whose output connects to the input of a digital filter 9l. The output of digital filter 9l couples to the input of a digital demodulation circuit 93 whose digital baseband output feeds into the remainder of the cnmmllni cations system.
In Figure 4, there are also connections between a clock circuit 83 and five circuit blocks that require synchronizing clock signals for proper operation. Clock circuit 83 has j+l output lines which 2 ~ O ~ 7 $ 3 ~ W096~9750 PCT~S96/09311 carry j+l clock signals. One of these output lines from clock circuit 83 connects to sample-and-hold circuit 81 while another of these output lines from clock circuit 83 connects to ADC 87, digital filter 91, and digital demodulation circuit 93.
Discrete-time signal processing block 85 also connects to clock circuit 83. The connection between these two circuits, however, is a multiple-line connection that couples through node 84 and is composed of all j+l clock signal lines. The relationships among the clock signals delivered from clock circuit 83 to these five circuit blocks are discussed later in this description.
Figure 5 illustrates the details of discrete-time signal processing block 85. This block is composed of j serially cascaded down-sampling stages 99.1 to 99.j. The input to down-sampling stage 99.1 serves as the input to discrete-time signal processing block 85 as a whole and connects through node 82 to the output of sample-and-hold circuit 81 of Figure 4. Similarly, the output of down-sampling stage 99.j serves as the output to discrete-time signal processing block 85 as a whole and connects through node 86 to the input of ADC 87 of Figure 4.
Each down-sampling stage connects to two clock signal lines.
Except for the first and the last clock signal line in discrete-time signal processing block 85, each of the clock signal lines 25 connect to two adjacent down-sampling stages. The first clock signal line tha~ connects to discrete-time signal processing block 85 also connects to sample-and-hold circuit 81, and similarly, the last clock signal line that connects to discrete-time signal processing block 85 also connects to ADC 87, digital filter 91, 30 and digital demodulation circuit 93. Accordingly, j+l clock signal lines connect from the down-sampling stages through node 84 to clock circuit 83 of Figure 4.
Because the down-sampling stages share a common structure, it is 3s sufficient to explain the contents and connections of just one.
Down-sampling stage 99.1 is composed of two circuits, a discrete-time anti-aliasing filter 95.1 and a N:l discrete-time sampling down-converter 97.1. In the preferred embodiment, these two circuits are built using switched-capacitor circuit techni~ues.
2~0 ~753 - W096~97~0 PCT~S96/09311 Within down-sampling stage 99.l, the output of discrete-time anti-aliasing filter 95.1 couples to the input of N:l discrete-time sampling down-converter 97.1. Discret~-tim~ anti-aliasing filter 95.l and N:l discrete-time sampling down-converter 97.l each 5 connect to a separate clock signal line from clock circuit 83.
Note that the clock signal line that connects to the N:l discrete-time sampling down-converter 97.l also connects to discrete-time anti-aliasing filter 95.2. The other down-sampllng stages 99.2 to 99.j have similar connections.
Referring back to Figure 4, all circuit blocks in the preferred embodiment from sample-and-hold circuit 81 through digital demodulation circuit 93 as well as LNA 77 are integrated on a st~n~rd silicon chip. Because the analog bandpass RF filter 75 and the noise filter 79 are high-quality, narrowband RF filters, they are built using off-chip components. Nevertheless, this embodiment displays a higher level of integration than possible with previous receiver designs, resulting in decreased manufacture and assembly costs. Moreover, the design consumes less power than 20 previous receiver designs since the discrete-time signal processing techniques for down-conversion require no external IF
filtering components, no high-frequency mixers or LOs, and no high-speed analog-to-digital converters.
25 Figure 6 illustrates the characteristics of the RF com~l-n;cation signal of interest for the preferred embodiment. The following description illustrates the reception (down-conversion, filtering, and amplification) of a single analog voice channel l0la chosen from a set of analog voice channels l0l contained in a channel 30 allocation band of bandwidth B. For the sake of definiteness, it is assumed that B = l0 MHz and that voice channel l0la is positioned in the center of the band with carrier frequency fc = 400 MHz. It is also assumed that each channel in the band has a l0 kHz information bandwidth and is separated from adjacent channels by 30 kHz. These values are specially chosen merely to help elucidate the relationships involved in the down-conversion and filtering, and those skilled in the art will understand that the present method of RF signal reception does not depend on these specific values.
2~0~753 - ~ W096/39750 PCT~S96/09311 Descriptio~ of the Preferred Embodiment - Operation The operation of an RF communications receiver constructed in accordance with the principles of the present invention is as 5 follows.
Referring to Figure 4, an antenna 73 couples RF signal energy into the input of an analog bandpass RF filter 75. Analog hAn~p~cs RF
filter 75 is used to selectively pass a channel allocation band of bandwidth B while attenuating RF energy at frequencies outside this band, thereby producing a band-limited RF signal which includes a specific channel having carrier frequency fc.
The band-limited RF signal then passes from the output of analog 15 bandpass RF filter 75 to the input of LNA 77 which amplifies the signal while contributing mln;m~l noise. The amplified RF signal from the output of LNA 77 is then coupled to RF noise filter 79.
This RF noise filter 79 has a response similar to that of analog bandpass RF filter 75. It passes a bandwidth B of signal 20 frequencies in the channel allocation band while attenuating any RF energy that lies outside this bandwidth, such as energy that may have been produced or amplified by LNA 77. This noise filter improves the performance of the receiver by preventing excessive noise from folding into the channel of interest when down-25 conversion is performed further along in the receiver.
Although filters 75 and 79 efficiently pass signals within thepassband of bandwidth B they also pass non-negligible signals in a larger, anti-aliasing bandwidth W including the passband of 30 interest. (See Figures 9a and 9b.) This anti-aliasing bandwidth W arises from the combined roll-off of the two filter responses determined by their quality factors. To m;n;m; ze W, analog bandpass RF filter 75 and noise filter 79 must be high-quality filters built using off-chip components. LNA 77, however, can be 35 integrated along with the rest of the receiver. The passband gain and noise performance of the front-end components from antenna 73 through sample-and-hold circuit 81 determine the noise figure and sensitivity of the receiver.
220~753 ' W096/39750 PCT~S96/09311 After passing through RF noise filter 79, the band-limited RF
signal couples into the input of sample-and-hold circuit 81 which also receives a clock signal of frequency fs from clock generator circuit 83. This clock signal determines the sampling rate of s sample-and-hold circuit 81. Because the sampling rate f5 is less than the RF carrier frequency fc, sub-sampling will occur. In accordance with the principles of Fourier theory, the discrete-time signal produced at the output of sample-and-hold circuit 81 will contain infinite replicas of the band-limited RF signal, each centered at an image signal frequency fi = nfs + fc, where n is an integer. The original signal is shown in Figure 7 and its replication is shown in Figure 8. By selecting the lowest frequency image signal, i.e., by choosing the integer n that minimizes ¦fi¦, this process effectively down-converts, or frequency translates, the RF signal to an image signal at a lower frequenCY fi-The sampling rate, fs, must be set above a min;mllm rate to avoid destructive aliasing during the sub-sampling process. This min; mum rate is determined by both the bandwidth B of the allocation band and the bandwidth W of the front-end RF filters.
To avoid aliasing an image channel of frequency fi in the image signal band, the image channel must not overlap with the outer tail of the envelope of the adjacent image signal band which contains the corresponding image channel of frequency fs ~ fi.
The lower bound on the difference between fs ~ fi and fi depends on the position of the ~hAnnel within the band. Figures 9a and 9b show the overlap between adjàcent image bands in the best case and in the worst case, respectively. In the worst case, there is no overlap whenever fs - 2fi > (W ~ B) /2. In the preferred embodiment, B = lO MHz and W = 80 MHz, so we require fs - 2fi ~ 45 MHz. For reasons discussed below, the preferred embodiment selects fi = fs/3, so the constraint fs - 2fi ~ 45 MHz implies fs > 135 MHz, and the constraint fi = nfs i fc implies fs = i 3fc/(3n-l) = 1200 MHz, 600 MHz, 300 MHZ, 240 MHz, 171 MHz, 150 MHz, 120 MHz, 109 MHz, 92 MHz, ... Thus the lowest frequency image signal that avoids destructive aliasing is f5 = l50 MHz (n = 3). Consequently, fi = 50 MHz.
2 ~ O ~ 7 ~ ~
- ~ W096~9750 PCT~S96/09311 After sampling the band-limited RF signal at rate fs, sample-and-hold circuit 81 couples the discrete-time signal through node 82 into the input of discrete-time signal processing block 85. This discrete-time signal processing block simultaneously performs s anti-alias filtering, channel-select filtering, amplification, and frequency down-conversion of the signal in j successive steps, ultimately selecting and isolating analog voice channel lOla from within the chAnnel allocation band and translating it in frequency down to a low-frequency signal of frequency fk.
The selection of different channels from within the RF channel allocation band is accomplished by adjusting the sample rate fs produced by clock circuit 83 of Figure 4. Since all clock lines leading from clock circuit 83 deliver a coherent synchronizing 15 signal of frequency fs or an integral fraction thereof, all the discrete-time and digital circuit blocks 99 in Figure 5 which are connected to clock circuit 83 through node 84 will track the changes in fs. Accordingly, as the value of f5 is varied, different RF channels will be tuned by the same down-conversion, 20 filtering, and amplification process as for the RF channel having fc = 400 MHz.
Referring to Figure 5, upon coupling through node 82 into discrete-time signal processing block 85, the discrete-time signal 25 enters down-sampling stage 99.l. Inside down-sampling stage 99.1, the discrete-time signal passes through anti-aliasing filter 95.l, which is driven by a clock signal coupled through node 83 to operate at a rate f5 equal to that of sample and hold circuit 81 of Figure 4. Anti-aliasing filter 95.l removes from the 30 discrete-time signal all frequencies that would otherwise alias onto the desired channel during down-sampling and amplifies the r~m~;n;ng desired frequencies. The frequency response of anti-aliasing filter 95.l used in the preferred embodiment is shown in Figure ll where the frequency axis is normalized to half its 35 150 MHZ clock rate.
The filtered and amplified discrete-time signal then couples into an N:l discrete-time sampling down-converter 97.l. The N:l discrete-time sampling down-converter 97.l is a sample-and-hold 2~0 ~75 ~
W096~9750 PCT~S96/09311 circuit with a sampling rate of fs/N, where N, the down-sampling ratio, is an integer greater than one. After being down-sampled by a factor of N, the signal then passes into down-sampling stage 99.2.
In the preferred embodiment N = 2 since this results in the simplest response for the anti-aliasing filters. The use of a higher value for the down-sampling ratio N would require the use of a more complex anti-aliasing filter with more than one notch in its frequency response for removing the additional interfering signals that could alias onto the desired channel during down-sampling. The higher the value of N, the greater the number of normalized frequencies that would have tO be notched out.
Accordingly, as N increased, the order of the discrete-time anti-lS aliasing filters would also have to increase. In order tomaintain simplicity, a lower down-sampling ratio is preferable, especially for the down-sampling stages in the high-speed front end. This ensures that an integrated filter is realizable.
Because the other down-sampling stages 99.2 through 99.j function similarly to down-sampling stage 99.l, their operation on the discrete-time signal need not be individually explained. It is important to understand, however, that each discrete-time anti-aliasing filter 95.i is driven by clock circuit 83 to operate at a clock rate equal to the sampling rate of the N:l discrete-time sampling down-converter 97.i-l that preceded it. Additionally, each successive down-sampling stage 99 is driven by clock circuit 83 to operate at a clock or sampling rate N times lower than the previous down-sampling stage. Thus, in the preferred embodiment where N = 2, each successive down-sampling stage operates at half the rate of the previous down-sampling stage. In this way, the discrete-time signal is successively down-converted in frequency.
Additionally, the desired channel is successively amplified and filtered in order to separate it from the other channels originally present in the band-limited RF signal.
The ratio of signal frequency to sample rate is critical in the design of the anti-aliasing filters in the down-sampling stages.
In order to perform anti-alias filtering in the discrete-time ~ ~ O ~ 7 5 3 ~ ~ W096/39750 PCT~S96/09311 ~nm~;n, it is necessary to ensure that the desired channel and aliasing interfering signals are separated as much as possible in frequency. This allows the separation between the stop band and pass band to be sufficiently large so that a low-order filter can 5 be feasibly designed. Additionally, since many down-sampling stages are used, it is necessary to ensure that the separation between the stop band and pass band r~;n~ reasonable after many stages. As the bandwidth of the desired channel becomes a significant fraction of the sample rate, a low-order filter is no o longer sufficient for anci-aliasing because to properly prevent diLa~iny, ~h~ ~t~p ~ IIU~C ~ W1~r ~ldll C~l~ ~d~l~wld~h or th~
desired channel. At this point, however, the signal is shifted sufficiently down in frequency to make digital processing practical.
A frequency map showing the effect of 2:1 down-sampling is shown in Figure 10. The upper frequency axis shows the signal frequency location of discrete-time signals just before being sampled by a 2:1 discrete-time sampling down-converter. The signal frequencies are normalized to the sampling rate of the 2:1 discrete-time sampling down-converter. The lower frequency axis shows the frequency spectrum just after down-sampling, normalized to the sampling rate of the next 2:1 discrete-time sampling down-converter. The arrows indicate the mapping of normalized 25 frequency as a result of down-sampling. Each value of frequency aliases with exactly one other value of frequency that is located symmetrically about the point f = 0.5.
For the preferred embodiment in which N = 2, there are two special 30 locations for the channel frequency: a normalized frequency of 4/5 and a normalized frequency of 2/3. For the chAnnel frequency located at a normalized frequency of 4/S, the component that aliases onto the ch~nnel is located at a frequency of 1/5. After 2:1 down-sampling, the signal moves to a frequency of 2/5, and the 35 aliasing frequency for the next stage of down-sampling is located at 3/5. After one more stage of down-sampling, the desired channel moves back up to a normalized frequency of 4/5, and the cycle can be continued until the signal bandwidth reaches the range 3/10 to 1/2 (0.3 to O.S). At this point, it becomes ~ 2 0 ~ 7 5 3 W096/39750 PCT~S96/09311 impossible to remove the aliasing components in the next stage of down-sampling with a discrete-time filter. The use of the normalized frequency of 4/5 therefore requires the alternating use of two different discrete-time anti-aliasing filter designs.
For the case in which the desired channel is located at a normalized frequency of 2/3, the aliasing frequency is located at a frequency of l/3. After 2:l down-sampling, the desired channel r~mA; n.~ at a normalized frequency of 2/3 with respect to the lC sampling rate of the next down-sampling stage. Thus, the desired ~h~nnel never changes location relative to the sampling frequency.
This is a particularly advantageous result because anti-aliasing filters with similar responses can be used in all the down-sampling stages. This down-sampling can be continued until the 15 channel bandwidth spans the range l/2 to 5/6 (0.5 to 0.833) after which it becomes impossible to anti-alias filter in the discrete-time domain. For these reasons, the preferred embodiment uses this value of 2/3 for the location of the desired channel in normalized frequency. Consequently, the preferred embodiment 20 selects fi ~ 2/3(fS/2) = fs/3. In other embodiments it may be desirable to relax this requirement slightly to obtain, for example, fs/4 ' fi ' fs/2-Referring now back to Figure 4, after the desired channel 25 progresses through all the down-sampling stages 99 (Figure 5), it appears at the output of discrete-time signal processing block 85 as a low-frequency discrete-time signal of frequency fk and couples through node 86 into the input of ADC 87. ADC 87 samples and digitizes the signal at a conversion rate determined by clock 30 circuit 83. In the preferred embodiment, this clock rate is the same as the sampling rate of the final 2:l discrete-time sampling down-converter in discrete-time signal processing block 85, and is therefore low enough that ADC 87 does not consume large amounts of power. The digital representation of the desired channel is now 35 passed into a digital filter 9l which performs the final ~hAnnelization filtering of the desired channel before passing it on to a digital demodulator 93. Digital demodulator 93 demodulates the desired channel, extracting its information ~ , = = ~ ~
:
~20 ~753 - ~ W096/39750 PCT~S96/09311 content onto a digital baseband output signal which it passes on to the rest of the commllnications system.
The present invention is described through a specific preferred 5 embodiment. This merely serves the purpose of facilitating the description of the principles of the present invention and in no way is meant to limit its scope. Those skilled in the art will realize that many changes and modifications can readily be made to the preferred embodiment without departing from the principles of o the invention.
For example, instead of using switched-capacitor techniques to build the sample-and-hold circuits, the discrete-time anti-aliasing filters, and the N:1 discrete-time sampling down-15 converters, these components may be built using switched-current circuits or other techniques.
Although the preferred embodiment performs the analog-to-digital conversion after the down-conversion, it is possible to perform 20 the analog-to-digital conversion earlier in the receiver, such as just after sample-and-hold circuit 81. The successive down-conversion and filtering stages would then be implemented in the digital ~m~;n using digital filtering and decimation techniques.
This option will become increasingly attractive as analog-to-25 digital conversion technology improves.
Those skilled in the art will also understand the applicability ofthe present invention to RF channels with different characteristics than those chosen for the preferred embodiment.
30 The present invention is applicable to a large range of RF carrier frequencies, limited only by technology. Also, the present invention is applicable to many other channel allocation band configurations containing any number of channels, each of arbitrary bandwidth. Moreover, the method may be used with any 35 type of modulation including, for example, AM, FM, phase modulation, and digital forms of modulation such as quadrature amplitude modulation, phase shift keying, Gaussian m;n;mllm shift keying, and frequency shift keying. Additionally, those skilled in the art will recognize the choices available in the selection 2 ~ Q ~ 7 5 3 W096/39750 PCT~S96/09311 of the sampling frequency, the selection of the down-converting ratio, the optional use of LNA 77, the trade-offs in frequency response characteristics between analog bandpass RF filter 75 and RF noise filter 79, and the selection of the number of down-5 sampling stages. These and other modifications which are obviousto those skilled in the art are intended to be included within the scope of the present invention. Accordingly, the scope of the invention should be determined not by the embodiment described, but by the appended claims and their legal equivalents.
W096/39750 PCT~S96/09311 Radio Freguency Signal Reception Usin~ Fre~uency Shifting by Discrete-Time Sub-Samplin~ Down-~onversion 5 Bac~o~d - Technical Field of Invention The present invention relates to radio receivers and methods for the reception of RF (radio frequency) communications signals. In particular i~ relates to radio receivers usirlg high-speed discrete-time electronic circuits and methods for RF signal 20 reception using sub-sampling for frequency down-conversion.
~ckground of the Invention and Discussion of Prior Art At the present time, the vast majority of RF communications receivers are of the traditional superheterodyne type. This type 25 of receiver uses one or more IF (intermediate frequency) stages for filtering and amplifying signals at a fixed fre~uency within an IF chain. This radio architecture has the advantage that fixed filters may be used in the local oscillator (LO) chain.
30 A block diagram of a typical superheterodyne radio receiver is shown in Figure 1. An RF signal arriving at an antenna 11 passes through a bandpass RF filter 13, an LNA (low noise amplifier) 15 and into an image filter 17 which produces a band-limited RF
signal. This band-limited RF signal then enters a first mixer 19 35 which translates the RF signal down to an intermediate frequency by mixing it with the signal produced by a first LO 21. The undesired mixer products in the IF signal are rejected by an IF
filter 23. The filtered IF signal then enters a second mixer 25 which translates it down to yet another intermediate fre~uency by 2~0 ~75`3 ~ ~ W096~97~0 PCT~S96/09311 mixing it with the signal produced by a second LO 27. The signal finally passes through an IF amplifier 29 and an IF filter 31, resulting in a channelized IF signal, i.e., a signal cont~;n;ng a particular channel isolated from the other channels present within the original band-limited RF signal. An analog demodulator 33 demodulates the channelized IF signal into a baseband information signal. Tuning into a particular channel within the band-limited RF signal is accomplished by varying the L0 frequency of LO 21.
In order to reduce size, power consumption, and cost, it would be advantageous to integrate the electronic components of radio receivers onto a single silicon chip. The superheterodyne design, however, requires high-quality, narrowband IF bandpass filters operating at high frequencies, and these components can not be built on-chip with present technology. Instead, off-chip components such as SAW (surface acoustic wave) filters or tuned circuits requiring resonating components such as inductors or cavities must be used. These external filtering components impose a lower limit to the size, materials cost, assembly cost, and power consumption of receivers built using the superheterodyne design. Moreover, the necessity for mixer and local oscillator circuits operating at high frequencies contributes greatly to the power consumption and general complexity of the superheterodyne receiver. In particular, high-frequency analog mixers often require a large amount of power to maintain linear operation.
Although many variations of the superheterodyne design exist, they all share the limitations of the particular design just described.
The growing demand for portable communications has motivated attempts to design radio receivers that permit the integration of more components onto a single chip. For example, a digital IF
receiver design, as shown in Figure 2, can be integrated further than the superheterodyne design and therefore requires fewer external components. As in the superheterodyne design, an antenna 35 couples an RF signal through an RF filter 37 into an LNA 39 and then through an image filter 41. The band-limited RF signal mixes in a mixer 43 with a L0 signal from a L0 45 to produce an IF
signal. The IF signal then is amplified in an IF amplifier 47 before being channelized in an IF filter 49. An ADC (analog-to-2 ~ O ~ 7 5 3 - ~ W096~9750 PCT~S96/09311 digital converter) 51, operating at a conversion rate determined by a clock circuit 53, digitizes the channelized IF signal into a digital IF signal. This digital IF signal is filtered and demodulated by a digital demodulator 55 and is passed on to the 5 r~m~ er of the communications system as a digital baseband information signal. Tuning is accomplished by varying the frequency of LO 45. Multiple techniques for performing the digital filtering and demodulation have been developed for this receiver design such as those described in U.S. Patents No.
lC 4,888,557 and No. 4,902,979.
Compared to the superheterodyne design, the digital IF design can cost less to manufacture and assemble because it has fewer off-chip components. Nevertheless, the digital IF design still requires several off-chip components, namely, external IF filters, a high-frequency mixer, and a high-frequency LO. U.S. Patent No.
5,339,459 describes a technique that eliminates the necessity for external high-frequency mixer and LO circuits by using a sample-and-hold circuit. By operating at a sampling rate lower than the 20 band-limited RF signal, i.e. by sub-sampling, the sample-and-hold circuit acts as a mixer to translate the RF signal into an IF
signal. Although this technique allows further integration and lower power operation, it still requires the use of off-chip components for the high-quality, narrowband IF bandpass filters 25 nee~e~ for channelization.
The digital IF design has other disadvantages as well. In order to faithfully digitize the analog signal, the ADC is required to operate at a conversion rate of at least twice the intermediate 30 frequency. Such a high-frequency ADC consumes a large amount of power to maintain the required linearity and dynamic range. This becomes even more problematic for receiver designs in which the A/D conversion is performed at the front end of the receiver, e.g., as in U.S. Patent No. 4,893,316.
A third receiver design is the direct-conversion, or zero-IF, receiver shown in Figure 3. An antenna 57 couples an RF signal through a first bandpass RF filter 59 into an LNA 61. The signal then proceeds through a second RF filter 63, yielding a band-~ ~ W096/39750 PCT~S96/09311 .
limited RF signal, which then enters a mixer 65 and mixes with anLO frequency produced by an LO 67. Up to this point, the direct-conversion receiver design is essentially the same as the two previous receiver designs.
Unlike the previous designs, however, the LO frequency is set to the carrier frequency of the RF channel of interest. The resulting mixer product is a zero-frequency IF signal-a modulated signal at baseband frequency. This signal couples into a lowpass analog filter 69 before proceeding into an analog demodulator 71 to yield the demodulated analog baseband information signal for use by the r~m~;n~er of the co~ml~nications system. Alternatively, an analog-to-digital converter can be inserted after mixer 65 and the filtering and demodulation can be done digitally, producing a 15 digital baseband information signal for use by the remainder of the com~l1nications system. In either case, tuning is accomplished by varying the frequency of LO 67, thereby converting different RF
channels to zero-frequency IF signals.
20 Because the direct-conversion receiver design produces a zero-frequency IF signal, its filter requirements are greatly simplified-no external IF filter components are needed since the zero-IF signal is an audio frequency signal that can be filtered by a low-quality lowpass filter. This allows the receiver to be 25 integrated in a st~n~rd silicon process from mixer 65 onwards, making the direct-conversion receiver design potentially attractive for portable applications.
The direct-conversion design, however, has several problems, some 30 of which are quite serious. As with the other designs described above, the RF and image filters required in the direct-conversion design must be high-quality narrowband filters that must remain off-chip. Moreover, this design requires the use of high-frequency mixer and LO circuits that require large amounts of 35 power. Additionally, radiated power from LO 67 can couple into antenna 57, producing a DC offset at the output of mixer 65. This DC offset can be much greater than the desired zero-IF signal, making signal reception difficult. Radiated power from LO 67 can also affect other nearby direct-conversion receivers tuned to the 2 2 0 ~ 7 5 3 - ~ WO 96/39750 PCT/US96/09311 same radio frequency. Furth~rm~re, to receive signals transmitted using modulation techniques (such as FM) in which access to both the lower and upper sidebands is required, two mixers and two LOs are required to produce both an in-phase and a quadrature baseband 5 signal. Not only does this increase the power required by the receiver, but also the phase between the two LO signals must be precisely maintained at 90 degrees to prevent demodulation distortion. This can be difficult to accomplish with variations in temperature and other operational parameters.
In summary, although the prior art includes various receiver designs that achieve successively greater integration, each one has significant disadvantages including one or more of the following: the necessity for several external circuit components, 5 the consumption of large amounts of power, poor signal reception, interference with other receivers, distortion, and limited dynamic range.
Objects a~d Advanta~es of the Invention 20 Accordingly, it is a primary object of the present invention to provide a radio receiver design which has increased integration, decreased materials and assembly costs, and decreased power consumption without the operational problems associated with previous receiver designs. It is a further object of the 25 invention to provide a method of frequency down-converting an RF
channel without requiring any off-chip IF filter components, multiple LOs, high frequency LOs operating at frequencies near the RF signal frequency of interest, in-phase and quadrature LOs, analog mixers, or high-speed, high-power analog-to-digital 30 converters. It is another object of the present invention to provide a method for filtering and selecting a particular channel within a channel allocation band of interest without requiring any off-chip IF components, especially components for filtering.
Further objects and advantages will become apparent from a 3s consideration of the ensuing description and drawings.
Summary of the Invention The present invention achieves the above objects and advantages by providing a new method for RF comm-~n;cations signal reception and - 2~01753 - ~ W096/39750 PCT~S96/09311 a new receiver design that incorporates this method. This method includes filtering an RF signal cont~;n;~g a channel allocation band of bandwidth B to yield a band-limited RF signal of bandwidth W. An RF channel of carrier frequency fc is included within this channel allocation band. The method further includes a sample-and-hold means to sub-sample the band-limited RF signal at a sampling rate fs ~ fc in order to obtain an image signal of the RF
channel. The image signal frequency fi satisfies the conditions fi = nfs + fc and fs - 2fi ' (W + B)/2, where n is an integer chosen to minimize Ifil Furthermore, the method includes a discrete-time down-converting means to successively down-convert, filter, and amplify the image signal of the RF channel to a low-frequency signal of frequency fk. In contrast with digital signal processing techniques, the down-converting method of the invention employs discrete-time signal processing techniques that do not require digital conversion and processing of the signal.
Description of Drawings Figure l is a block diagram of a superheterodyne receiver considered as prior art.
Figure 2 is a block diagram of a digital IF receiver considered as prior art.
Figure 3 is a block diagram of a direct-conversion receiver considered as prior art.
Figure 4 is a block diagram of a receiver constructed with the principles of the invention.
Figure 5 is a block diagram of a discrete-time signal processing block, which is a key component of a receiver constructed according to the invention.
Figure 6 is a graph showing a typical arrangement of RF channels within a channel allocation band used for common radio c~mmlln;cations.
Figure 7 is a graph showing the band-limited RF signal after passing through the noise filter of Figure 4 according to the invention.
Figure 8 is a graph showing the spectrum of image signals after the band-limited RF signal is sub-sampled according to the invention.
=
2 ~ 0 ~ 7 ~ 3 WO 96/397s0 PCT/US96/09311 Figures 9a and 9b are graphs illustrating the necessary relationships between fs, fi, B, and W to permit sub-sampling without destructive aliasing.
Figure lO pictorially illustrates the aliasing of spectral components due to 2:l down-sampling.
Figure ll is a graph showing the normalized frequency response of the discrete-time anti-aliasing filters used in the preferred embodiment.
o Description of the Preferred Embodiment - Physical Attributes Figure 4 is a block diagram of an RF communications receiver constructed in accordance with the principles of the present invention. It includes an antenna 73 for coupling an RF signal .15 into the input of an analog bandpass RF filter 75. The output of analog bandpass RF filter 75 connects to the input of an LNA 77 whose output couples to the input of an analog noise filter 79.
The output of analog RF noise filter 79 connects to the input of a sample-and-hold circuit 81, which is built using switched-capacitor circuit techniques.
In addition to its signal input which connects to the output of RF
noise filter 79, sample-and-hold circuit 81 has a clock input and a discrete-time signal output. Its discrete-time signal output connects at a node 82 to the input of a discrete-time signal processing block 85. Discrete-time signal processing block 85 is a key component of the preferred embodiment of the invention and is described in greater detail below in connection with Figure 5.
The output of discrete-time signal processing block 85 couples through node 86 to the input of an ADC 87 whose output connects to the input of a digital filter 9l. The output of digital filter 9l couples to the input of a digital demodulation circuit 93 whose digital baseband output feeds into the remainder of the cnmmllni cations system.
In Figure 4, there are also connections between a clock circuit 83 and five circuit blocks that require synchronizing clock signals for proper operation. Clock circuit 83 has j+l output lines which 2 ~ O ~ 7 $ 3 ~ W096~9750 PCT~S96/09311 carry j+l clock signals. One of these output lines from clock circuit 83 connects to sample-and-hold circuit 81 while another of these output lines from clock circuit 83 connects to ADC 87, digital filter 91, and digital demodulation circuit 93.
Discrete-time signal processing block 85 also connects to clock circuit 83. The connection between these two circuits, however, is a multiple-line connection that couples through node 84 and is composed of all j+l clock signal lines. The relationships among the clock signals delivered from clock circuit 83 to these five circuit blocks are discussed later in this description.
Figure 5 illustrates the details of discrete-time signal processing block 85. This block is composed of j serially cascaded down-sampling stages 99.1 to 99.j. The input to down-sampling stage 99.1 serves as the input to discrete-time signal processing block 85 as a whole and connects through node 82 to the output of sample-and-hold circuit 81 of Figure 4. Similarly, the output of down-sampling stage 99.j serves as the output to discrete-time signal processing block 85 as a whole and connects through node 86 to the input of ADC 87 of Figure 4.
Each down-sampling stage connects to two clock signal lines.
Except for the first and the last clock signal line in discrete-time signal processing block 85, each of the clock signal lines 25 connect to two adjacent down-sampling stages. The first clock signal line tha~ connects to discrete-time signal processing block 85 also connects to sample-and-hold circuit 81, and similarly, the last clock signal line that connects to discrete-time signal processing block 85 also connects to ADC 87, digital filter 91, 30 and digital demodulation circuit 93. Accordingly, j+l clock signal lines connect from the down-sampling stages through node 84 to clock circuit 83 of Figure 4.
Because the down-sampling stages share a common structure, it is 3s sufficient to explain the contents and connections of just one.
Down-sampling stage 99.1 is composed of two circuits, a discrete-time anti-aliasing filter 95.1 and a N:l discrete-time sampling down-converter 97.1. In the preferred embodiment, these two circuits are built using switched-capacitor circuit techni~ues.
2~0 ~753 - W096~97~0 PCT~S96/09311 Within down-sampling stage 99.l, the output of discrete-time anti-aliasing filter 95.1 couples to the input of N:l discrete-time sampling down-converter 97.1. Discret~-tim~ anti-aliasing filter 95.l and N:l discrete-time sampling down-converter 97.l each 5 connect to a separate clock signal line from clock circuit 83.
Note that the clock signal line that connects to the N:l discrete-time sampling down-converter 97.l also connects to discrete-time anti-aliasing filter 95.2. The other down-sampllng stages 99.2 to 99.j have similar connections.
Referring back to Figure 4, all circuit blocks in the preferred embodiment from sample-and-hold circuit 81 through digital demodulation circuit 93 as well as LNA 77 are integrated on a st~n~rd silicon chip. Because the analog bandpass RF filter 75 and the noise filter 79 are high-quality, narrowband RF filters, they are built using off-chip components. Nevertheless, this embodiment displays a higher level of integration than possible with previous receiver designs, resulting in decreased manufacture and assembly costs. Moreover, the design consumes less power than 20 previous receiver designs since the discrete-time signal processing techniques for down-conversion require no external IF
filtering components, no high-frequency mixers or LOs, and no high-speed analog-to-digital converters.
25 Figure 6 illustrates the characteristics of the RF com~l-n;cation signal of interest for the preferred embodiment. The following description illustrates the reception (down-conversion, filtering, and amplification) of a single analog voice channel l0la chosen from a set of analog voice channels l0l contained in a channel 30 allocation band of bandwidth B. For the sake of definiteness, it is assumed that B = l0 MHz and that voice channel l0la is positioned in the center of the band with carrier frequency fc = 400 MHz. It is also assumed that each channel in the band has a l0 kHz information bandwidth and is separated from adjacent channels by 30 kHz. These values are specially chosen merely to help elucidate the relationships involved in the down-conversion and filtering, and those skilled in the art will understand that the present method of RF signal reception does not depend on these specific values.
2~0~753 - ~ W096/39750 PCT~S96/09311 Descriptio~ of the Preferred Embodiment - Operation The operation of an RF communications receiver constructed in accordance with the principles of the present invention is as 5 follows.
Referring to Figure 4, an antenna 73 couples RF signal energy into the input of an analog bandpass RF filter 75. Analog hAn~p~cs RF
filter 75 is used to selectively pass a channel allocation band of bandwidth B while attenuating RF energy at frequencies outside this band, thereby producing a band-limited RF signal which includes a specific channel having carrier frequency fc.
The band-limited RF signal then passes from the output of analog 15 bandpass RF filter 75 to the input of LNA 77 which amplifies the signal while contributing mln;m~l noise. The amplified RF signal from the output of LNA 77 is then coupled to RF noise filter 79.
This RF noise filter 79 has a response similar to that of analog bandpass RF filter 75. It passes a bandwidth B of signal 20 frequencies in the channel allocation band while attenuating any RF energy that lies outside this bandwidth, such as energy that may have been produced or amplified by LNA 77. This noise filter improves the performance of the receiver by preventing excessive noise from folding into the channel of interest when down-25 conversion is performed further along in the receiver.
Although filters 75 and 79 efficiently pass signals within thepassband of bandwidth B they also pass non-negligible signals in a larger, anti-aliasing bandwidth W including the passband of 30 interest. (See Figures 9a and 9b.) This anti-aliasing bandwidth W arises from the combined roll-off of the two filter responses determined by their quality factors. To m;n;m; ze W, analog bandpass RF filter 75 and noise filter 79 must be high-quality filters built using off-chip components. LNA 77, however, can be 35 integrated along with the rest of the receiver. The passband gain and noise performance of the front-end components from antenna 73 through sample-and-hold circuit 81 determine the noise figure and sensitivity of the receiver.
220~753 ' W096/39750 PCT~S96/09311 After passing through RF noise filter 79, the band-limited RF
signal couples into the input of sample-and-hold circuit 81 which also receives a clock signal of frequency fs from clock generator circuit 83. This clock signal determines the sampling rate of s sample-and-hold circuit 81. Because the sampling rate f5 is less than the RF carrier frequency fc, sub-sampling will occur. In accordance with the principles of Fourier theory, the discrete-time signal produced at the output of sample-and-hold circuit 81 will contain infinite replicas of the band-limited RF signal, each centered at an image signal frequency fi = nfs + fc, where n is an integer. The original signal is shown in Figure 7 and its replication is shown in Figure 8. By selecting the lowest frequency image signal, i.e., by choosing the integer n that minimizes ¦fi¦, this process effectively down-converts, or frequency translates, the RF signal to an image signal at a lower frequenCY fi-The sampling rate, fs, must be set above a min;mllm rate to avoid destructive aliasing during the sub-sampling process. This min; mum rate is determined by both the bandwidth B of the allocation band and the bandwidth W of the front-end RF filters.
To avoid aliasing an image channel of frequency fi in the image signal band, the image channel must not overlap with the outer tail of the envelope of the adjacent image signal band which contains the corresponding image channel of frequency fs ~ fi.
The lower bound on the difference between fs ~ fi and fi depends on the position of the ~hAnnel within the band. Figures 9a and 9b show the overlap between adjàcent image bands in the best case and in the worst case, respectively. In the worst case, there is no overlap whenever fs - 2fi > (W ~ B) /2. In the preferred embodiment, B = lO MHz and W = 80 MHz, so we require fs - 2fi ~ 45 MHz. For reasons discussed below, the preferred embodiment selects fi = fs/3, so the constraint fs - 2fi ~ 45 MHz implies fs > 135 MHz, and the constraint fi = nfs i fc implies fs = i 3fc/(3n-l) = 1200 MHz, 600 MHz, 300 MHZ, 240 MHz, 171 MHz, 150 MHz, 120 MHz, 109 MHz, 92 MHz, ... Thus the lowest frequency image signal that avoids destructive aliasing is f5 = l50 MHz (n = 3). Consequently, fi = 50 MHz.
2 ~ O ~ 7 ~ ~
- ~ W096~9750 PCT~S96/09311 After sampling the band-limited RF signal at rate fs, sample-and-hold circuit 81 couples the discrete-time signal through node 82 into the input of discrete-time signal processing block 85. This discrete-time signal processing block simultaneously performs s anti-alias filtering, channel-select filtering, amplification, and frequency down-conversion of the signal in j successive steps, ultimately selecting and isolating analog voice channel lOla from within the chAnnel allocation band and translating it in frequency down to a low-frequency signal of frequency fk.
The selection of different channels from within the RF channel allocation band is accomplished by adjusting the sample rate fs produced by clock circuit 83 of Figure 4. Since all clock lines leading from clock circuit 83 deliver a coherent synchronizing 15 signal of frequency fs or an integral fraction thereof, all the discrete-time and digital circuit blocks 99 in Figure 5 which are connected to clock circuit 83 through node 84 will track the changes in fs. Accordingly, as the value of f5 is varied, different RF channels will be tuned by the same down-conversion, 20 filtering, and amplification process as for the RF channel having fc = 400 MHz.
Referring to Figure 5, upon coupling through node 82 into discrete-time signal processing block 85, the discrete-time signal 25 enters down-sampling stage 99.l. Inside down-sampling stage 99.1, the discrete-time signal passes through anti-aliasing filter 95.l, which is driven by a clock signal coupled through node 83 to operate at a rate f5 equal to that of sample and hold circuit 81 of Figure 4. Anti-aliasing filter 95.l removes from the 30 discrete-time signal all frequencies that would otherwise alias onto the desired channel during down-sampling and amplifies the r~m~;n;ng desired frequencies. The frequency response of anti-aliasing filter 95.l used in the preferred embodiment is shown in Figure ll where the frequency axis is normalized to half its 35 150 MHZ clock rate.
The filtered and amplified discrete-time signal then couples into an N:l discrete-time sampling down-converter 97.l. The N:l discrete-time sampling down-converter 97.l is a sample-and-hold 2~0 ~75 ~
W096~9750 PCT~S96/09311 circuit with a sampling rate of fs/N, where N, the down-sampling ratio, is an integer greater than one. After being down-sampled by a factor of N, the signal then passes into down-sampling stage 99.2.
In the preferred embodiment N = 2 since this results in the simplest response for the anti-aliasing filters. The use of a higher value for the down-sampling ratio N would require the use of a more complex anti-aliasing filter with more than one notch in its frequency response for removing the additional interfering signals that could alias onto the desired channel during down-sampling. The higher the value of N, the greater the number of normalized frequencies that would have tO be notched out.
Accordingly, as N increased, the order of the discrete-time anti-lS aliasing filters would also have to increase. In order tomaintain simplicity, a lower down-sampling ratio is preferable, especially for the down-sampling stages in the high-speed front end. This ensures that an integrated filter is realizable.
Because the other down-sampling stages 99.2 through 99.j function similarly to down-sampling stage 99.l, their operation on the discrete-time signal need not be individually explained. It is important to understand, however, that each discrete-time anti-aliasing filter 95.i is driven by clock circuit 83 to operate at a clock rate equal to the sampling rate of the N:l discrete-time sampling down-converter 97.i-l that preceded it. Additionally, each successive down-sampling stage 99 is driven by clock circuit 83 to operate at a clock or sampling rate N times lower than the previous down-sampling stage. Thus, in the preferred embodiment where N = 2, each successive down-sampling stage operates at half the rate of the previous down-sampling stage. In this way, the discrete-time signal is successively down-converted in frequency.
Additionally, the desired channel is successively amplified and filtered in order to separate it from the other channels originally present in the band-limited RF signal.
The ratio of signal frequency to sample rate is critical in the design of the anti-aliasing filters in the down-sampling stages.
In order to perform anti-alias filtering in the discrete-time ~ ~ O ~ 7 5 3 ~ ~ W096/39750 PCT~S96/09311 ~nm~;n, it is necessary to ensure that the desired channel and aliasing interfering signals are separated as much as possible in frequency. This allows the separation between the stop band and pass band to be sufficiently large so that a low-order filter can 5 be feasibly designed. Additionally, since many down-sampling stages are used, it is necessary to ensure that the separation between the stop band and pass band r~;n~ reasonable after many stages. As the bandwidth of the desired channel becomes a significant fraction of the sample rate, a low-order filter is no o longer sufficient for anci-aliasing because to properly prevent diLa~iny, ~h~ ~t~p ~ IIU~C ~ W1~r ~ldll C~l~ ~d~l~wld~h or th~
desired channel. At this point, however, the signal is shifted sufficiently down in frequency to make digital processing practical.
A frequency map showing the effect of 2:1 down-sampling is shown in Figure 10. The upper frequency axis shows the signal frequency location of discrete-time signals just before being sampled by a 2:1 discrete-time sampling down-converter. The signal frequencies are normalized to the sampling rate of the 2:1 discrete-time sampling down-converter. The lower frequency axis shows the frequency spectrum just after down-sampling, normalized to the sampling rate of the next 2:1 discrete-time sampling down-converter. The arrows indicate the mapping of normalized 25 frequency as a result of down-sampling. Each value of frequency aliases with exactly one other value of frequency that is located symmetrically about the point f = 0.5.
For the preferred embodiment in which N = 2, there are two special 30 locations for the channel frequency: a normalized frequency of 4/5 and a normalized frequency of 2/3. For the chAnnel frequency located at a normalized frequency of 4/S, the component that aliases onto the ch~nnel is located at a frequency of 1/5. After 2:1 down-sampling, the signal moves to a frequency of 2/5, and the 35 aliasing frequency for the next stage of down-sampling is located at 3/5. After one more stage of down-sampling, the desired channel moves back up to a normalized frequency of 4/5, and the cycle can be continued until the signal bandwidth reaches the range 3/10 to 1/2 (0.3 to O.S). At this point, it becomes ~ 2 0 ~ 7 5 3 W096/39750 PCT~S96/09311 impossible to remove the aliasing components in the next stage of down-sampling with a discrete-time filter. The use of the normalized frequency of 4/5 therefore requires the alternating use of two different discrete-time anti-aliasing filter designs.
For the case in which the desired channel is located at a normalized frequency of 2/3, the aliasing frequency is located at a frequency of l/3. After 2:l down-sampling, the desired channel r~mA; n.~ at a normalized frequency of 2/3 with respect to the lC sampling rate of the next down-sampling stage. Thus, the desired ~h~nnel never changes location relative to the sampling frequency.
This is a particularly advantageous result because anti-aliasing filters with similar responses can be used in all the down-sampling stages. This down-sampling can be continued until the 15 channel bandwidth spans the range l/2 to 5/6 (0.5 to 0.833) after which it becomes impossible to anti-alias filter in the discrete-time domain. For these reasons, the preferred embodiment uses this value of 2/3 for the location of the desired channel in normalized frequency. Consequently, the preferred embodiment 20 selects fi ~ 2/3(fS/2) = fs/3. In other embodiments it may be desirable to relax this requirement slightly to obtain, for example, fs/4 ' fi ' fs/2-Referring now back to Figure 4, after the desired channel 25 progresses through all the down-sampling stages 99 (Figure 5), it appears at the output of discrete-time signal processing block 85 as a low-frequency discrete-time signal of frequency fk and couples through node 86 into the input of ADC 87. ADC 87 samples and digitizes the signal at a conversion rate determined by clock 30 circuit 83. In the preferred embodiment, this clock rate is the same as the sampling rate of the final 2:l discrete-time sampling down-converter in discrete-time signal processing block 85, and is therefore low enough that ADC 87 does not consume large amounts of power. The digital representation of the desired channel is now 35 passed into a digital filter 9l which performs the final ~hAnnelization filtering of the desired channel before passing it on to a digital demodulator 93. Digital demodulator 93 demodulates the desired channel, extracting its information ~ , = = ~ ~
:
~20 ~753 - ~ W096/39750 PCT~S96/09311 content onto a digital baseband output signal which it passes on to the rest of the commllnications system.
The present invention is described through a specific preferred 5 embodiment. This merely serves the purpose of facilitating the description of the principles of the present invention and in no way is meant to limit its scope. Those skilled in the art will realize that many changes and modifications can readily be made to the preferred embodiment without departing from the principles of o the invention.
For example, instead of using switched-capacitor techniques to build the sample-and-hold circuits, the discrete-time anti-aliasing filters, and the N:1 discrete-time sampling down-15 converters, these components may be built using switched-current circuits or other techniques.
Although the preferred embodiment performs the analog-to-digital conversion after the down-conversion, it is possible to perform 20 the analog-to-digital conversion earlier in the receiver, such as just after sample-and-hold circuit 81. The successive down-conversion and filtering stages would then be implemented in the digital ~m~;n using digital filtering and decimation techniques.
This option will become increasingly attractive as analog-to-25 digital conversion technology improves.
Those skilled in the art will also understand the applicability ofthe present invention to RF channels with different characteristics than those chosen for the preferred embodiment.
30 The present invention is applicable to a large range of RF carrier frequencies, limited only by technology. Also, the present invention is applicable to many other channel allocation band configurations containing any number of channels, each of arbitrary bandwidth. Moreover, the method may be used with any 35 type of modulation including, for example, AM, FM, phase modulation, and digital forms of modulation such as quadrature amplitude modulation, phase shift keying, Gaussian m;n;mllm shift keying, and frequency shift keying. Additionally, those skilled in the art will recognize the choices available in the selection 2 ~ Q ~ 7 5 3 W096/39750 PCT~S96/09311 of the sampling frequency, the selection of the down-converting ratio, the optional use of LNA 77, the trade-offs in frequency response characteristics between analog bandpass RF filter 75 and RF noise filter 79, and the selection of the number of down-5 sampling stages. These and other modifications which are obviousto those skilled in the art are intended to be included within the scope of the present invention. Accordingly, the scope of the invention should be determined not by the embodiment described, but by the appended claims and their legal equivalents.
Claims (18)
1. A method for filtering and frequency translating an RF signal containing an RF channel of carrier frequency fc to yield a low-frequency signal containing a down-converted channel of frequency fk, the method comprising:
filtering the RF signal to yield a band-limited signal of bandwidth W, the band-limited signal containing a channel allocation band of bandwidth B, where the channel allocation band contains the RF channel;
discrete-time sampling the band-limited signal at a sampling frequency fs to yield an image signal that contains an image channel of frequency fi, where fi < fs < fc;
discrete-time down-converting the image signal to yield the low-frequency signal containing the down-converted channel of frequency fk.
filtering the RF signal to yield a band-limited signal of bandwidth W, the band-limited signal containing a channel allocation band of bandwidth B, where the channel allocation band contains the RF channel;
discrete-time sampling the band-limited signal at a sampling frequency fs to yield an image signal that contains an image channel of frequency fi, where fi < fs < fc;
discrete-time down-converting the image signal to yield the low-frequency signal containing the down-converted channel of frequency fk.
2. The method of claim 1 wherein fi = mfk for a rational number m, fs - 2fi > (W+B)/2, and fi = nfs fc for an integer n which minimizes fi.
3. The method of claim 1 wherein fs/4 < fi < fs/2.
4. The method of claim 3 wherein fi = fs/3.
5. The method of claim 1 wherein the down-converting step comprises:
discrete-time filtering the image signal to prevent aliasing; and down-sampling the image signal to shift the image signal down in frequency.
discrete-time filtering the image signal to prevent aliasing; and down-sampling the image signal to shift the image signal down in frequency.
6. The method of claim 5 wherein the down-sampling step comprises successively down-sampling the image signal by factors of two.
7. The method of claim 5 wherein the discrete-time filtering step comprises amplifying the signal.
8. The method of claim 1 further comprising digitizing the band-limited signal.
9. The method of claim 1 further comprising digitizing the image signal.
10. A radio receiver comprising:
a receiving means for receiving an RF signal containing an RF
channel of carrier frequency fc within a channel allocation band having bandwidth B;
a filtering means for filtering the RF signal to a bandwidth W to yield a band-limited signal, the band-limited signal containing the channel allocation band;
a discrete-time sampling means for sampling the band-limited signal at a sampling frequency fs to yield an image signal containing an image channel of frequency fi, where fi < fs < fc;
a discrete-time down-converting means for down-converting the image signal to yield a low-frequency signal containing a down-converted channel of frequency fk, where fk < fi;
a discrete-time channel-select filtering means for removing adjacent-channel interference from the low-frequency signal to isolate the down-converted channel; and a demodulation means for demodulating the down-converted channel to yield a baseband signal.
a receiving means for receiving an RF signal containing an RF
channel of carrier frequency fc within a channel allocation band having bandwidth B;
a filtering means for filtering the RF signal to a bandwidth W to yield a band-limited signal, the band-limited signal containing the channel allocation band;
a discrete-time sampling means for sampling the band-limited signal at a sampling frequency fs to yield an image signal containing an image channel of frequency fi, where fi < fs < fc;
a discrete-time down-converting means for down-converting the image signal to yield a low-frequency signal containing a down-converted channel of frequency fk, where fk < fi;
a discrete-time channel-select filtering means for removing adjacent-channel interference from the low-frequency signal to isolate the down-converted channel; and a demodulation means for demodulating the down-converted channel to yield a baseband signal.
11. The radio receiver of claim 10 wherein fi = mfk for a rational number m, fs - 2fi > (W+B)/2, and fi = nfs fc for an integer n which minimizes fi¦.
12. The radio receiver of claim 10 wherein fs/4 fi < fs/2.
13. The radio receiver of claim 12 wherein fi = fs/3.
14. The radio receiver of claim 10 wherein the down-converting means comprises:
a discrete-time filtering means for anti-alias filtering the image signal; and a down-sampling means for shifting the image signal down in frequency.
a discrete-time filtering means for anti-alias filtering the image signal; and a down-sampling means for shifting the image signal down in frequency.
15. The radio receiver of claim 14 wherein the down-sampling means comprises a 2:1 discrete-time sampling down-converter.
16. The radio receiver of claim 10 further comprising a digitizing means for digitizing the band-limited signal.
17. The radio receiver of claim 10 further comprising a digitizing means for digitizing the image signal.
18. The radio receiver of claim 10 wherein the demodulation means comprises a demodulator chosen from the group consisting of an AM demodulator, an FM demodulator, a phase modulation demodulator, a quadrature amplitude modulation demodulator, a phase shift keying demodulator, a Gaussian minimum shift keying demodulator, and a frequency shift keying demodulator.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US08/468,280 | 1995-06-06 | ||
| US08/468,280 US5640698A (en) | 1995-06-06 | 1995-06-06 | Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion |
| PCT/US1996/009311 WO1996039750A1 (en) | 1995-06-06 | 1996-06-06 | Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2201753A1 true CA2201753A1 (en) | 1996-12-12 |
Family
ID=29405924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA 2201753 Abandoned CA2201753A1 (en) | 1995-06-06 | 1996-06-06 | Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion |
Country Status (1)
| Country | Link |
|---|---|
| CA (1) | CA2201753A1 (en) |
-
1996
- 1996-06-06 CA CA 2201753 patent/CA2201753A1/en not_active Abandoned
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US5640698A (en) | Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion | |
| Mirabbasi et al. | Classical and modern receiver architectures | |
| US6498819B1 (en) | Integrated multi-mode bandpass sigma-delta receiver subsystem with interference mitigation and method of using same | |
| US7173980B2 (en) | Complex-IF digital receiver | |
| US20060153284A1 (en) | Method and apparatus for a frequency agile variable bandwidth transceiver | |
| EP0766409B1 (en) | Multiband downconverter for digital receivers | |
| US7529322B2 (en) | Two-step channel selection for wireless receiver front-ends | |
| EP0935851B1 (en) | A digital communication device and a mixer | |
| EP1336246B1 (en) | Direct conversion receiver | |
| US6373422B1 (en) | Method and apparatus employing decimation filter for down conversion in a receiver | |
| US20150214926A1 (en) | Discrete-time filter | |
| US20070060077A1 (en) | Receiver architecture for wireless communication | |
| US20070099588A1 (en) | Method and apparatus for receiving and/or down converting high frequency signals in multi mode/ multi band applications, using mixer and sampler | |
| Abidi | Evolution of a software-defined radio receiver's RF front-end | |
| US7123892B2 (en) | Architecture for an AM/FM digital intermediate frequency radio | |
| US7634247B2 (en) | Method of sampling an analogue radiofrequency signal | |
| EP0629040A1 (en) | Method and apparatus for IF demodulation of a high frequency input signal | |
| CA2201753A1 (en) | Radio frequency signal reception using frequency shifting by discrete-time sub-sampling down-conversion | |
| JP3996053B2 (en) | Receiver circuit suitable for mobile radio | |
| JP2000092021A (en) | Digital broadcast receiver | |
| Luy | Software configurable receivers | |
| US20190190533A1 (en) | Wideband sigma delta modulator receiver for fm signal reception | |
| Coy et al. | HF-band radio receiver design based on digital signal processing | |
| Ernou et al. | Software radio receiver for seismic data processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| FZDE | Dead |