WO1997006604A1 - Universal rf receiver - Google Patents

Abstract

A universal receiver (10) for processing RF signals modulated by various analog and digital modulation techniques. The receiver (10) includes a down converter (52) for converting received RF signals to in-phase and quadrature zero IF signals, and active low pass filter (106, 108) for channel selection and an up converter (132, 134) for converting the channel selected in-phase and quadrature zero IF signals to an IF signal. Only moderate system gain is introduced at zero IF in order to avoid increasing any DC offsets. The receiver (10) further includes DC offset compensation circuitry and a period-to-digital (P/D) converter (260) for quantizing the IF signal. A signal processor (35) equalizes and demodulates the IF signal to recover the transmitted signal.

Description

UNIVERSAL RF RECEIVER

RELATED CASE INFORMATION This application claims the benefit of U.S. Provisional

Application Serial No. 60/001,907, filed August 4, 1995 and U.S. Provisional Application Serial No. 60/010,568, filed January 25, 1996.

FIELD OF THE INVENTION

This invention relates generally to RF receivers and, more particularly, to a highly integrated universal direct conversion receiver for receiving RF signals modulated by any of various analog and digital techniques.

BACKGROUND OF THE INVENTION Superheterodyne RF receivers which operate by mixing an incoming RF signal with a local oscillator (LO) signal are known. The output of the mixer is an intermediate frequency (IF) signal which is filtered at IF with the use of passive bandpass filters in order to select a particular channel (i.e., frequency) of interest. Such filters generally consist of a resonant element in which the physical properties of the material determine filter characteristics, including filter size. Use of such passive bandpass filters in superheterodyne receivers has precluded size reduction and integration of such receivers.

Another type of receiver that has been considered is the zero IF receiver which, like the superhederodyne receiver, down converts received RF signals. However, instead of down converting to some IF frequency in the manner of a super¬ heterodyne receiver, zero IF receivers down convert to zero IF (i.e., to the modulating frequency, by removing the carrier frequency) . While down conversion to zero IF advantageously permits the use of active filters for purposes of channel selection, zero IF receivers have not been practical to implement due to the inherent DC offsets which result from the introduction of significant gain at zero IF. The advent of superheterodyne receivers coincided with the use of analog modulation techniques for RF transmission, such as amplitude modulation (AM) and frequency modulation (FM) techniques. Thus, superheterodyne receivers were developed based on analog modulation standards. With the more recent popularity of wireless transmissions however, including cellular communications, personal communications services and wireless area networks, digital modulation techniques have proliferated. As examples, North America has developed two major standards, including the Code Division Multiple Access (CDMA) IS-95 standard and the North American Digital Cellular (NADC) IS-54 standard. European communities have, in large part, adopted the pan-European digital cellular radio (GSM) standard and Japan has adopted a variation of the NADC IS-54 standard entitled the Personal Digital Cellular (PDC) RCR-27 standard.

SUMMARY OF THE INVENTION

A highly integrated universal RF receiver capable of processing RF signal modulated by various analog and digital modulation techniques is described. The receiver includes a down converter for converting received RF signals to in- phase and quadrature zero IF signals (i.e., baseband signals at the frequency of the modulating signal with the carrier frequency removed) . Active channel selection filters pass signals within a desired frequency band and an up converter converts the zero IF signals to an IF signal. Channel selection filtering at zero IF permits the use of readily integratable active filters for this purpose, enabling the receiver to be implemented on one or two application specific integrated circuits (ASICs) .

Only moderate gain is introduced at zero IF. Additional gain is introduced to the up converted IF signal. DC offsets are minimized by avoiding significant gain at zero IF. DC offset compensation circuitry is also provided.

The amplified IF signal is quantized by a Period-to- Digital (P/D) converter which provides a count value signal to a signal processor. In general, the P/D converter is capable of quantizing signals that contain information in the form of phase or frequency modulation (i.e. , constant envelope modulation) . In such constant envelope modulation schemes, signal information is contained in the zero crossings. The count value signal is representative of the period between consecutive zero crossings of the IF signal and thus, is also representative of the instantaneous frequency of the modulated signal. The P/D converter advantageously uses digital only construction, has low gate complexity and can be fabricated by any of a variety of semiconductor processes.

In one embodiment, a phase comparator compares the amplified IF signal to a reference voltage in order to provide a signal containing zero crossing information to the P/D converter. The phase comparator also serves to amplitude limit the IF signal, thereby eliminating performance degradation due to amplitude fading and the need to otherwise compensate for amplitude fading. Amplitude limiting in this manner is possible due to the use of the P/D converter for equalization, since the P/D converter relies on zero crossing information, as opposed to amplitude information.

The signal processor demodulates the count value signal provided by the P/D converter. Additional functionality of the signal processor includes intersymbol interference (ISI) equalization and additionally, in mobile receiver applications, time dispersion equalization. Clock recovery may also be performed by the signal processor. In one embodiment, the signal processor is a digital signal processor (DSP) . The receiver is compatible with various modulation standards, both analog and digital. In particular, different modulation standards are accommodated by modifying the bandwidth of an RF input bandpass filter of the receiver, the down converter LO frequency, the IF frequency to which the zero IF signal is up converted and the demodulation technique employed by the signal processor.

BRIEF DESCRIPTION OF THE DRAWINGS The foregoing features of this invention, as well as the invention itself, may be more fully understood from the following detailed description of the drawings in which:

Fig. 1 is a block diagram of a universal RF receiver in accordance with the invention;

Fig. 2 is a detailed block diagram of the receiver of Fig. 1; Fig. 3 is a block diagram of the P/D converter of Fig.

2;

Fig. 4 is a schematic of a first portion of the RF/analog circuitry of the receiver of Fig. 2;

Fig. 5 is a schematic of a second portion of the RF/analog circuitry of the receiver of Fig. 2;

Fig. 6 is a schematic of a third portion of the RF/analog circuitry of the receiver of Fig. 2;

Fig. 7 is a schematic of a fourth portion of the RF/analog circuitry of the receiver of Fig. 2; Figs. 8 and 8A are a schematic of the digital portion of the receiver of Fig. 2;

Fig. 9 is a graph illustrating peak-to-peak signal levels over the receiver's dynamic range;

Fig. 10 is a functional block diagram of an embodiment of the a signal processor for use in non-mobile receiver applications;

Fig. 11 is a flow diagram of the steps performed by the signal processor of Fig. 10;

Fig. 12 is a functional block diagram of an embodiment of the signal processor for use in mobile receiver applications; and Fig. 13 is a flow diagram of the steps performed by the signal processor of Fig. 12.

, 5 DESCRIPTION OF THE PREFERRED EMBODIMENT

Referring to Fig. 1, a universal RF receiver 10 includes an RF bandpass filter (BPF) 18 responsive to transmitted RF signals 16 which are received by an antenna 14. The RF BPF 18 attenuates signals having frequencies outside a particular

10 frequency band of interest.

The filtered output signal 20 of the RF BPF 18 is coupled to a low noise amplifier (LNA) 22 which provides gain to the received signals. The amount of gain provided by the LNA 22 is selected to ensure a satisfactory receiver noise

15 figure (NF) , since the LNA 22 is predominantly responsible for establishing the overall system NF. The bandwidth of the LNA is wide enough to pass signals within the frequency band of interest.

A circuit 26 operates to down convert, filter and up

20 convert output signals 24 of the LNA 22. The down conversion is achieved with an image reject mixer (Fig. 2) which converts RF signals to in-phase (I) and quadrature (Q) zero IF signals (i.e., direct conversion). Zero IF signals are baseband signals having the frequency of the modulating

25 signal, with the carrier frequency components removed. The I and Q zero IF signals are then filtered to select a particular channel within the frequency band of interest for reception. In the preferred embodiment, the filter is an active filter. The filtered I and Q zero IF signals are then

30 up converted to respective IF signals and summed to provide an IF output signal 28. In the preferred embodiment, the up conversion is achieved with an image canceling mixer (Fig. 2) .

The resulting IF signal 28 is then processed by an IF

35 low pass filter (LPF) 30 which passes the lowest frequency signal components and rejects the odd harmonic components. The filtered output signal 32 of the IF LPF 30 is further processed by an IF amplifier 33 which provides the additional system gain. The system gain is adjusted over the dynamic range of the receiver, as described in conjunction with Fig. 9 below.

A quantizer 34 is responsive to the IF output signal 37 of the IF amplifier 33 for generating a count value signal 38 for coupling to a signal processor 35. The count value signal 38 is representative of the period between consecutive zero crossings of the IF signal 37 and thus, the instantaneous frequency of the IF signal. In the preferred embodiment, the quantizer 34 is a Period-to-Digital (P/D) converter described in U.S. Patent Nos. 5,159,281, 5,239,273 and 5,272,448, which are assigned to the assignee of the subject application and incorporated herein by reference.

Referring also to Fig. 2, the RF antenna 14 is coupled to the RF BPF 18 which filters out-of-band received signals 16. In one embodiment designed to accommodate GMSK modulated signals and described further below in conjunction with Figs. 4-8A (i.e., the GSM embodiment), the frequency band of interest is between 935Mhz and 960Mhz and the RF BPF 18 has a bandwidth of 25MHz. One suitable type of RF BPF 18 is a duplexer, in which two filters are provided in a single resonant element. The filtered output signal 20 of the RF BPF 18 is coupled to the LNA 22 for amplification. The bandwidth of the LNA 22 is at least the same as that of the RF BPF 18. The output signal 24 of the LNA 22 may be coupled to an optional additional RF BPF 40 which provides additional out- of-band filtering. Use of the RF BPF 40 may be particularly advantageous in cellular applications.

In the illustrative embodiment, the down converter/channel selector/up converter circuit 26 includes an integrated down converter sub-circuit 52 having an automatic gain control (AGC) amplifier 48 which introduces a selectable gain, such as -3dB or 22dB of gain, to the input

SUBSTπM SEE! (RULE 2β) signal 42 in accordance with an AGC signal 49. The AGC amplifier 48 further serves to convert the single-ended input signal 42 to a differential output signal 50. The output signal 50 is coupled to an image reject mixer including an in-phase (I) mixer 54, a quadrature (Q) mixer 58, an I low pass filter (LPF) 80 and a Q LPF 90. Each mixer 54 and 58 receives a respective local oscillator (LO) signal 60 and 62, with the LO signal 62 being 90° out-of-phase with respect to the LO signal 60. A reference clock oscillator 72, such as a crystal, generates a reference clock signal 65 having a precise frequency, such aε 13.0MHz in the illustrative GSM embodiment. A frequency synthesizer 64, such as a phase locked loop (PLL) and a voltage controlled oscillator, converts the reference clock signal 65 into a higher frequency LO signal 67 which is phase-locked to the reference clock signal 65. In particular, the LO signal 67 generated by the frequency synthesizer 64 is matched to the carrier frequency of the RF input signal 16 which, in the GSM embodiment, is between 935MHz and 960MHz. A circuit 68 is responsive to the LO signal 67 for generating the in-phase LO signal 60 and the quadrature LO signal 62 which are thus, phase-locked to the reference clock signal 65.

In order to illustrate the operation of the image reject mixer, consider the case where the differential signal 50 coupled to the mixers 54, 58 is given by cos(ω.t) , the I LO signal 60 is given by cos(ω₀t) and the Q LO signal 62 is given by cos(ω₀t -90°), where the LO frequency ω₀ is equal to the carrier frequency ω_c. The output signal I(t) 78 of I mixer 54 is then given by:

(1)

J( t) =cos (ω_ct) -cos (ω₀t) =l/2cos (ω_c+ω_σ) fc+l/2cos (ω_c-ω₀) t This I(t) signal is then filtered by the LPF 80 which eliminates the summation term. Thus, the I_f(t) output signal 82 of the LPF 80 is given by:

I_f(fc) =l/2cos (ω_c-ω₀) t (2)

Since the LO frequency ω₀ is matched to the carrier frequency ω_c, the signal I_f(t) can be expressed as the zero IF signal:

I_f ( t) =l/2cos(0) t ⁽³⁾

The output signal Q(t) 88 generated by the Q mixer 58 is given by:

Q( t) =cos(ω_cfc) ^•cos(ω_ot-90°) (4)

=l/2cos(ω_cfc+ω_ot-90°) +l/2cos(ω_ct-ω_ot+90^o)

The Q(t) signal 88 is then filtered by the LPF 90 to generate an output signal Q_f(t) 92 which contains only the frequency difference term as follows:

Again, since the LO frequency ω₀ is matched to the carrier frequency ω_c, the signal Q_f(t) can be expressed as the zero IF signal:

Q_f { fc) =l/2sin(ω-,-ω₀) t=0 (6)

Moderate gain may be provided at zero IF in both the I and Q signal paths by respective instrumentation amplifiers 94 and 96. The optional additional linear gain provided by the amplifiers 94, 96 helps to establish the desired overall system NF. However, most of the system gain is not provided at zero IF in order to minimize DC offsets. Additionally, the instrumentation amplifiers 94, 96 convert the respective differential input signals 82, 92 into single-ended output signals 98, 100, respectively. In the illustrative GSM embodiment, each of the instrumentation amplifiers 94, 96 provides a fixed 6dB of gain. The single-ended output signals 98, 100 of the instrumentation amplifiers 94, 96, respectively, are coupled to respective active LPFs 106, 108 which are active filters operable to pass the desired frequency spectrum (i.e., channel selection) while attenuating noise and interference outside of the desired signal bandwidth. Since channel selection and interference rejection is performed at zero IF, active LPFs 106, 108 are suitable for providing such functionality. Advantageously, active LPFs 106, 108 are readily integratable. In the illustrative embodiment, each of the LPFs 106, 108 is implemented as an eight pole Bessel filter, as shown and described below in conjunction with Fig. 5. It will be appreciated by those of ordinary skill in the art however, that the LPFs 106, 108 may be provided by any suitable active filter, such as switched capacitor filters. In the GSM embodiment, LPFs 106, 108 provide matched filter response for the receive channel. Additionally, any filter structure which can approximate the required gaussian filter response would suffice for LPFs 106, 108, provided they feature minimal group (phase) delay. Additional gain is introduced to the output signals 110,

112 of the LPFs 106, 108, respectively, by IF amplifiers 116, 118. The IF amplifiers 116, 118 provide a selectable gain in accordance with an AGC signal 120 in order to establish the total system gain and the overall system NF. For example, in the GSM embodiment, each of the amplifiers provides a gain of OdB or 20dB.

The amplifiers 116 and 118 are additionally responsive to DC offset compensation circuitry, as indicated by respective DC offset control signals 102, 104. The DC offset compensation circuitry will be described in conjunction with Fig. 5 below. Suffice it to say that the DC offset circuitry serves to minimize DC offsets at the outputs of the IF amplifiers 116 and 118, such as may be caused by differential DC offsets at the outputs of the image reject mixers 54, 58. The output signals 126, 128 of the IF amplifiers 116, 118 are up-converted by an image canceling mixer, including an I mixer 132 and a Q mixer 134 responsive to respective LO signals 138, 140. In particular, a circuit 142 is responsive to a divided version 139 of the reference clock signal 65 for generating the in-phase LO signal 138 and the quadrature LO signal 140, which is 90° phase-shifted relative to the signal 138. In the illustrative GSM embodiment, the LO signals 138, 140 have a frequency which meets the Nyquist criterion for the data rate of the transmitted GSM signal or 203.125KHz, where the GSM data rate is equal to 270.833KHZ. In order to illustrate the operation of the up converting mixers 132, 134, consider the case where the single-ended I signal 126 is given by cosω_mt, the single- ended Q signal 128 is given by cos(ω_mt-90°) , the LO signal 138 is given by cos(ω₀t) and the LO signal 140 is given by cos(ω₀t-90°) , where the LO frequency ω„ is equal to the data rate of the transmitted RF signal. In this case, the output signal I_p(t) 146 of mixer 132 is given by: (7)

and the output signal Q_p(t) 148 of mixer 134 is given by:

The I_p(t) and Q_p(t) signals 146 and 148 are then summed by a summation circuit 147 to provide an IF signal S(t) 28 as follows: S ( t) =I_p ( t) +Q_p ( t)

=l/2cos (ω₀+ω t+l/2cos (ω₀-ω fc-l/2cos ( ω_σ+ω_m) fc+l/2cos (ω₀-ω_m) t

=cos (ω_σ-ω_jn) t

(9 )

Consideration of equation (9) reveals that only the lower sideband signal remains, thereby converting the zero IF signals 126 and 128 into an IF signal 28 centered about the LO frequency ω₀. The IF signal 28 is coupled to the IF LPF 30 which passes the lowest frequency signal components and rejects the odd harmonic response. In the illustrative embodiment described below, the LPF 30 is two cascaded six pole Butterworth filters (Figs. 6 and 7) , but may alternatively be implemented with any suitable filter.

The output signal 32 of the LPF 30 is coupled to the IF amplifier 33 which introduces additional gain to the IF signal 32. Amplifier 33 introduces a selectable gain, such as OdB, lOdB or 20dB, in accordance with an AGC signal 182. Amplifier 33 additionally implements a DC offset correction feature to minimize DC offsets introduced prior to the amplifier inputs in response to a DC offset control signal

39, as will be described below in conjunction with Fig. 7.

The output signal 37 of the amplifier 33 is coupled to a phase comparator 250 and to a Received Signal Strength Indicator (RSSI) circuit 38. The phase comparator 250 configured as a zero crossing detector has a non-inverting input responsive to the output signal 37 of the IF amplifier 33 and an inverting input receiving a reference voltage, such as 2.5 volts. The phase comparator 250 provides a pair of output signals 252, 254 indicative of the zero crossings of the IF signal 37 and which are inverted versions of one another.

More particularly, the phase comparator output signals 252 and 254 are hard amplitude limited pulse trains having transitions at the zero crossings of the IF signal 37. The hard amplitude limiting of the output signals 252 and 254 reduces amplitude flat fading of the received RF signals, a phenomena which can occur as a mobile receiver moves. In conventional receivers which do not utilize zero crossing detection (i.e., time domain sampling systems utilizing A/D converters as the digitizer) , in which the linearity of the processed signals is relied on to determine the phase of the incoming signal, amplitude limiting is not possible; rather, complex equalization techniques have been employed to address amplitude fading. In the present receiver 10 however, it is the zero crossings of the processed signals, rather than their amplitude, that is used by the quantizer 34 to determine the instantaneous frequency of the IF signal, thereby enabling the advantageous technique of hard amplitude limiting to be used to address amplitude fading.

The output signals 252 and 254 of the phase comparator 250 are coupled to the quantizer 34 which is implemented with a P/D converter 260. The P/D converter 260 will be described further in conjunction with Fig. 3 and is the subject of the above incorporated issued U.S. Patents. Additionally, the P/D converter 260 is described in a Numa Technologies data sheet entitled Period to Digital (P/D) Converter NT304 dated 3QTR/94, which is incorporated herein by reference.

The P/D converter 260 integrates the signals 252, 254 with high resolution so as to provide a digital output signal 264 indicative of the instantaneous frequency of the IF signal 37. More particularly, the output of the P/D converter 260 is a sixteen bit count value signal 264 having a value equal to the number of clock cycles which occur between consecutive zero crossings of the IF signal 37. Since frequency is inversely proportional to period, the count value signal 264 can be expressed as:

i ll f) 12 <¹⁰⁾ where f is the instantaneous frequency of the IF signal 37. The count value signal 264 of the P/D converter 260 is equalized and demodulated by the signal processor 35, as will be described below in conjunction with Figs. 10-13. As noted, the output signal 37 of the IF amplifier 33 is further coupled to the RSSI circuit 38. The RSSI circuit 38 shown in Fig. 2 is a linear slope integrating analog-to- digital (A/D) converter, as described in a National Semiconductor Application Note 260 entitled "A 20-Bit (1 ppm) Linear Slope-Integrating A/D Converter" which is incorporated herein by reference. However, it will be appreciated that various A/D converter configurations are suitable for determining received signal strength, such as the successive approximation A/D converter 282 shown in the GSM embodiment below (Figs. 8 and 8A) .

The RSSI circuit 38 includes an RMS to DC converter 270 which converts the RMS value of the IF signal 37 to a DC signal 318 for further coupling to the inverting input of a comparator 274. A non-inverting input of the comparator 274 is responsive to a ramp signal 276 generated by a ramp generator 278. The comparator 274 compares the DC signal 318 to the ramp signal 276 and provides an output signal 278 to a control circuit 280, as shown.

The ramp signal 276 is further coupled to a comparator 284 which compares the ramp signal to a reference signal 286. The reference signal 286 is selectively provided by either a reference voltage source 288 or ground in accordance with the position of a switch 290 controlled by the control circuit 280 via a control signal line 294. The output signal 296 of the reference comparator 284 is a reference output signal coupled to the control circuit 280, as shown.

The control circuit 280 provides Q and ^~Q output signals

300 and 302, respectively, to a second P/D converter 306.

A twelve bit count value signal 310 is provided by the P/D converter 306 to the signal processor 35. With these signal widths provided by the P/D converter 306, the signal processor 35 is able to calculate the RSSI as follows:

Cvw-C_{ZERO Kμ y ( ll )}

^FSREF ^ZERO

where C_VIN is the count provided by the P/D converter 306 in response to the comparator output signal 278, C^_Q is the count provided by the P/D converter 306 in response to the reference comparator output signal 296 when its inverting input is connected to ground, C_FSREF is the count provided by the P/D converter 306 in response to the reference comparator output signal 296 when its inverting input is connected to the reference voltage source 288 and K is a constant, such as IO⁷. Note that the P/D converter is available in a sixteen bit output version, as in the case of the quantizer P/D converter 260, or a twelve bit output version, as in the case of the RSSI P/D converter 306.

As should now be apparent, baseband signals are processed from the instrumentation amplifiers 94, 96 through the quantizer 34. This baseband processing portion of the receiver may be fabricated as an ASIC, by either a bipolar, CMOS or BiCMOS process. Although both the RF and baseband receiver portions may be integrated on a single die using a BiCMOS process, isolation considerations may render it desirable to provide separate RF and baseband devices.

Referring also to Fig. 3, a block diagram of an illustrative P/D converter 260 is shown. In general, the P/D converter is capable of quantizing signals that contain information in the form of phase or frequency modulation (i.e. , constant envelope modulation) , such as Frequency Shift Keying (FSK) , Gaussian Minimum Shift Keying (GMSK) , Binary Phase Shift Keying (BPSK) and Quadrature Phase Shift Keying (QPSK) modulation techniques. In such constant envelope modulation schemes, signal information is contained in the zero crossings. The P/D converter 260 includes a pair of gates 350, 352, each responsive to a respective one of the phase comparator output signals 252 and 254 (Fig. 2) and to a CLK signal 356. Control signals 360 and 362 are further coupled to gates 350 and 352 by a timing and control circuit 370. The control signals 360 and 362 operate to alternatingly enable the gates 350 and 352. The gates 350, 352 are coupled to respective sixteen bit counters 372, 374.

When the input signal 252, 254 to one of the gates 350, 352 is high, the associated counter 372, 374, respectively, advances one count for each cycle of the CLK signal 356. The counters 372 and 374 thus provide respective digital count value signals 376, 378 representative of the period of a half cycle of the IF signal 37 (Fig. 2) . The count value signals 376 and 378 are further coupled to respective data latches 380 and 382, as shown.

A FIFO 390 alternatingly receives the latched count value signals from latches 380 and 382 in accordance with enabling control signals 384 and 386 which control the data latches 380 and 382, respectively. The output of the FIFO 390 is in the form of a sixteen bit DATA OUTPUT count value signal 264.

An interface and control circuit 404 provides the control interface to the signal processor 35 (Fig. 2) . Asynchronous operation of the P/D converter 260 iε accomplished by reading the DATA OUTPUT signal 264 at a rate greater than two times the frequency of the input signals 252, 254. Synchronouε operation, on the other hand, iε accomplished by using the ^~DA output signal 406 of the interface and control circuit 404. A low ^~DA signal 406 indicates the availability of data at the output of the FIFO 390.

A RESET input signal 408 is applied upon power up in order to ensure proper initialization. In particular, the RESET εignal, generated by reset circuitry 450 (Fig. 8) causes the FIFO 390 and counters 372, 374 to be cleared. A WR input signal 410, in conjunction with a CSO signal 414 and a CS1

418 signal, causes the same action as the RESET signal, except that the ^~WR reset is issued by the signal processor

35. An -RZ? input signal 412, in conjunction with theCSO signal 414 and the CS1 signal 418, causes the DATA OUTPUT signal 264 to change from a tri-state condition to enabled. Data is shifted out of the FIFO 390 on the falling edge of the RD signal. The CSO signal 414 and the CS1 signal 418 are chip select input signals. When the CSO signal 414 is low and the CS1 signal 418 is high, theεe inputs indicate that the control and data lines of the P/D converter 260 are valid and that the operation specified by the ^"RD andltfR inputs should be performed. Additional circuitry of the P/D converter 260 includes a divider 360 which is responsive to the CLK signal 356 for generating various frequency sub-multiples of the CLK signal, such as four, eight, sixteen, thirty-two and sixty-four at signal lines 392, 394, 396, 398 and 400, respectively. The clock signals 392-400 may be used as a clock source for other devices, such as the signal processor 35.

Referring also to Figs. 4, 5, 6 and 7 RF/analog portions 10a (Fig. 2) of the receiver 10 are shown. In particular, Fig. 4 shows the receiver 10 from the RF input 16 through the instrumentation amplifiers 94, 96, Fig. 5 shows the receiver 10 from the LPFs 106, 108 through the IF amplifiers 116 and 118, Fig. 6 shows the receiver 10 from the up converting mixers 132, 134 partially through the IF LPF 30, and Fig. 7 shows the remainder of the RF/analog portion of the receiver through the IF amplifier 33. Figs. 8 and 8A show the digital portion 10b (Fig. 2) of the receiver. The circuit of Figs. 4-8A represents an embodiment of the receiver adapted for receiving signals in a GSM cellular system (i.e., the GSM embodiment) . The modulation technique used in GSM systems is .3Bt Gaussian Minimum Shift Keying (GMSK). Pertinent

SDBSΠTIΠΈ SHEET (RULE 26) parameters of a GSM system are a data rate of 270.833KHZ, received signal frequencies between 935.2MHz and 959.8MHz and channel spacing of 200KHz.

With particular reference to Fig. 4, the RF input 16 is coupled to the RF BPF 18, which, in the illustrative embodiment may be of the type sold by Integrated Microwave under the part number 917745. The output signal 20 of the RF BPF 18 is coupled to the LNA 22 which may be of the type sold by Amplifonix under the part number AX0686. The optional RF BPF 40 may be coupled to the output of the LNA 22, as shown.

As noted above, DC offset compensation circuitry compensates for DC offsets at the I and Q IF amplifiers 116, 118 as well as at the IF amplifier 33. The DC offset compensation circuitry associated with the I and Q IF amplifiers 116, 118 operates, generally, by measuring the DC offset during intervals when RF signals 16 are not being received and using this measured DC offset to compensate for the actual DC offset occurring during operating intervals when RF signals 16 are received. More particularly, most digital modulation techniques, including GMSK, utilize Time Division Multiple Access (TDMA) by which RF energy is received in burεtε. Because the receiver 10 is not continuously receiving RF signalε, the receiver iε "powered down" during intervals when no RF signals are received. In particular, the LNA 22 is powered down and the antenna 14 (Fig. 2) is decoupled from the receiver 10.

To accomplish this power down/blocking operation, a switch 36 is coupled between a reference voltage, such aε +5 voltε, and εignal line 24 via a reεiεtor Rl, an inductor Ll, a capacitor Cl and a capacitor C3, aε shown. The switch 36 is responsive to an LNA control εignal 41 provided by the digital portion 10b of the receiver (Figε. 8 and 8A) via a connector Jl. With this arrangement, when the LNA control signal 41 iε low, the εwitch 36 is closed εo as to bias the output of the LNA 22; whereas, when the LNA control signal 41 is high, the switch 36 iε open, thereby preventing the output of the LNA 22 from being biased.

The LNA control signal 41 is further coupled to a driver 66 which provides output signals 70, 74 to a switch circuit 84, as shown. The switch circuit 84 is operable to isolate the RF input 16 from down stream portions of the receiver 10. That is, during operating intervals when RF signals are received, the output signals 70, 74 of the driver 66 cause the switch 84 to serially connect its RFI input to its RFC output. Alternatively however, during operating intervals when RF signals are not being received, the output signals 70, 74 of the driver 66 open the switch 84 so as to decouple the RFI input from the RFC output. Additionally, RFC is connected to RF2 in order to terminate the impedance matching network comprising C150, L3 and L6 into a 50 ohm terminating impedance connected to RF2 output, thereby decoupling the RF input 16 from down εtream portions of the receiver 10.

The RFC output of the switch 84 is coupled to the down converεion sub-circuit 52 (Fig. 2) . In particular, the RFC output is coupled to an R^~F input of εub-circuit 52. Recall from Fig. 2 that the εub-circuit 52 includeε the AGC amplifier 48, the I and Q down converεion mixerε 54 and 58, and the I and Q LPFε 80 and 90, reεpectively. Thus, the outputs of the sub-circuit 52 are two pairs of differential signal lines 82 and 92, aε εhown. Signal 82 provides the differential I output of the sub-circuit 52 and εignal 92 provides the differential Q output of the sub-circuit 52. In the illustrative embodiment, the down conversion sub¬ circuit 52 is of the type sold by Temic Telefunken Semiconductors under the part number U2791B and described in a Temic data sheet entitled 1000MHz Quadrature Demodulator U2791B, dated 08.06.1995 and incorporated herein by reference.

The down conversion sub-circuit 52 is additionally responεive to AGC εignal 49 for causing the internal AGC amplifier 48 to selectively introduce -3dB or 22dB of gain.

SUBSTlIlϋl In particular, the AGC εignal 49, labelled GCO, iε provided by the signal procesεor 35 (Fig. 8A) . The GCO εignal controls a switch Ql such that, when the GCO signal is high, the GC pin of the sub-circuit 52 at the collector of switch Ql is pulled low, causing 22dB of gain to be introduced by the internal AGC amplifier 48. When the GCO εignal is low on the other hand, the GC pin of the sub-circuit 52 is high, causing -3dB of gain to be introduced by the AGC amplifier 48.

As noted above, most digital modulation techniques employ TDMA access. However, the receiver 10 is equally operable in the case of continuous modulation techniqueε, whereby RF signals are continuously received by the receiver 10. In such continuous applications, the DC offset correction provided by the DC offset circuitry described herein may be enhanced by including optional capacitors C73, C74, C120 and C125. In particular, the AC coupling provided by these capacitors blocks any DC offεet at the outputs of the down conversion mixers 54, 58 within εub-circuit 52. However, use of the capacitorε C73, C74, C120 and C125 in diεcontinuouε receiver applications may be detrimental to the DC offset, since the resulting RF level shifts generate a DC component which is subject to integration by the capacitors, thereby aggravating any DC offset introduced by the mixerε. Thuε, in diεcontinuouε receiver applications, capacitor C73, C74, C120 and C125 are jumpered or otherwise removed.

The reference clock oscillator 72 provides the reference clock signal 65 (OSCOUT) having a predetermined frequency. In the GSM embodiment, the reference clock oscillator 72 is a crystal operating at 13.0MHz. The OSCOUT signal 65 is coupled to the frequency syntheεizer 64 converts the OSCOUT signal 65 to a signal 67 having a frequency matched to the carrier frequency. More particularly, the syntheεizer 64 iε reεponsive to control signalε 75 via a connector 73 which set the frequency of the LO signal 67 provided by the syntheεizer. The control εignalε 75 may be provided by any suitable user interface, such aε a microprocessor. The down

SUSΠΠΠE SHEET (RULE 26) conversion sub-circuit 52 includeε the circuit 68 which provides the in-phase LO signal 60 and the quadrature LO signal 62 to the I and Q mixers, respectively. In the illustrative embodiment, the OSCOUT signal 65 is coupled to the signal processor 35 (Fig. 8A) , where it is divided by sixteen to generate an 812.5KHZ LO signal 139 for use in up conversion, as will be described.

Downstream of the sub-circuit 52 are the instrumentation amplifiers 94 and 96 which may be of the type sold by Burr- Brown under the part number INA118. In one embodiment, each of the instrumentation amplifierε 94 and 96 introduceε a fixed 6dB gain.

Referring alεo to Fig. 5, the matched LPFε 106, 108 (Fig. 2) are shown. In the illustrated embodiment, the LPFs 106 and 108 are implemented as eight pole Bessel filters, each having four stages 106a, 106b, 106c, I06d and 108a, 108b, 108c, 108d, respectively. Recall that the LPFs 106 and 108 provide matched filtering for channel selection. No additional gain is introduced by the LPFs 106 and 108. The I and Q frequency selected output signals 110, 112 of the LPFs 106, 108, reεpectively, are coupled to the inverting input terminalε of reεpective IF amplifiers 116, 118, as shown. The IF amplifiers 116 and 118 have a selectable gain asεociated therewith. In the illustrative embodiment, amplifiers 116, 118 provide either OdB of gain or 20dB of gain. To thiε end, each of the amplifierε 116, 118 has a switch circuit 122, 124 coupled in feedback relationship therewith. Switch circuitε 122, 124 have an internal εwitch between the Sl and S2 pinε. The AGC εignal 120 (Fig. 2), labelled GC1, iε generated by the signal processor 35 and controls the εwitcheε 122, 124 in order to selectively open and close the respective internal switch between the Sl and S2 pins. In the event that the GC1 signal 120 causes the internal switch of switch circuit 122 to close, then the feedback resistor of amplifier 116 is provided by the parallel combination of resistorε R73 and R87; whereas if the GCl signal 120 causes the internal switch to open, then the feedback resistor of the amplifier 116 is provided by resiεtor R87. Similarly, if the GCl signal 120 causeε the internal εwitch of εwitch circuit 124 to open, then the feedback resistor of the amplifier 118 is provided by the parallel combination of resistorε R58 and R79; whereas, if the GCl signal 120 cauεeε the internal switch to open, then the feedback resistor of the amplifier 118 is provided by resistor R79. In the illustrative embodiment, when the GCl signal is high, each of the amplifiers 116, 118 introduces 20dB of gain and, when the GCl signal iε low, each of the amplifierε 116, 118 introduceε OdB of gain.

Recall that in TDMA receiver applicationε, when RF εignalε are not received, the LNA 22 iε powered down and the RF input 16 blocked from the receiver 10. When the LNA 22 iε powered down and the RF input 16 iε blocked from downstream portions of the receiver 10, the DC offset compensation circuitry associated with IF amplifiers 116 and 118 is operable to measure the DC offset asεociated with feeding a εignal through the receiver 10. This is achieved by feeding a signal through I and Q inverting operational amplifiers 130, 136 and applying the output of such amplifiers to the non-inverting input of respective sample and hold circuits 150, 152. A εample and hold control εignal 151 generated by the εignal proceεsor (Fig. 8A) is coupled to the sample and hold circuitε 150 and 152 for cauεing the reεpective output εignals 131, 137 of the inverting amplifiers 130, 136 to be sampled. The output εignalε 153, 155 of the sample and hold circuits 150, 152 are coupled to the non-inverting inputs of the IF amplifiers 116, 118, respectively, as shown.

With thiε arrangement, the DC offεet attributable to the componentε of the receiver 10 up εtream of the IF amplifierε 116, 118 iε measured, inverted and subtracted from the procesεed IF εignals 110, 112. In this way, any such DC offset appearing at the IF amplifiers 116, 118 iε compensated in the respective output signals 126, 128.

Referring also to Fig. 6, the output signals 126 and 128 of the I and Q IF amplifiers 116 and 118 are coupled to unity gain inverting amplifiers 164 and 166. The output signals 168 and 170 of the inverting amplifiers 164 and 166, respectively, are coupled to inputs of a quad FET switch 174, such as the type sold by Maxim under the part number MAX393. The quad FET switch 174 provides the up conversion mixer functionality of mixers 132, 134 (Fig. 2). The switch 174 includes four internal FETs (FET1, FET2, FET3 and FET4) , each having a drain coupled to a respective pin Dl, D2, D3, D4, a source coupled to a respective pin Sl, S2, S3, S4 and a gate coupled to a respective pin INI, IN2, IN3, IN4. The switch 174 is further responsive to the LO signalε

138, 140. In particular, The LO εignalε 138 and 140 are generated in response to an LO signal 139 from the εignal proceεεor 35 (Fig. 8A) by a flip-flop circuit 142, aε εhown. In particular, a flip-flop 142a divideε the εignal 139 by four and phase-shifts the signal by 90° to provide the LO signal 140 to the Q mixer 134 (i.e., to the IN3 and IN4 inputs to switch 174) and a flip-flop 142b divides the signal 139 by four to provide the LO εignal 138 to the I mixer 132 (i.e., to the INI and IN2 inputε to εwitch 174). Since the LO εignal 139 has a frequency of 812.5KHZ in the GSM embodiment, the LO signalε 138, 140 have a frequency of 203.125KHZ to satisfy the minimum Nyquist frequency required for the data rate of the transmitted GSM signals.

Considering the I channel for example, input signal 126 is coupled to the drain of FET1 and 90° phaεe-εhifted εignal 168 iε coupled to the drain of FET2. The sources of FET1 and FET2 are externally coupled together at the Sl and S2 pins, as shown. The gates of FET1 and FET2 are likewise externally coupled together at the INI and IN2 pins. The Q channel is similarly mixed by a pair of FET switcheε, FET3 and FET4, one of which, FET3, haε a drain terminal coupled to signal 128

SβBSIITUϊI SHEET (RULE 26) and the second of which, FET4, has a drain terminal coupled to 90° phase-shifted signal 170. The sources of FET3 and FET4 are commonly connected external to the switch 174 at pins S3 and S4. The gate terminals of FET3 and FET4 are likewise commonly connected at the IN3 and IN4 pins, as shown.

The I output signal 146 of the up conversion εwitch 174 iε coupled to an inverting input terminal of a summing amplifier 176. The Q output signal 148 of the εwitch 174 is likewise coupled to the inverting input of the εumming amplifier 176. The output εignal 28 of the εumming amplifier 176 iε thus, the sum of the εignalε 146 and 148 and, in particular, iε given by equation (9) above.

The εummed IF εignal 28 is coupled to the LPF 30, aε shown. In the illustrative embodiment, the LPF 30 is implemented with two six pole Butterworth filters, four two pole stageε of which are shown in Fig. 6 (30a, 30b, 30c and 30d) and the last two two pole εtages of which are εhown in Fig. 7 (30e and 30f) . Each of the operational amplifierε in the Butterworth filter 30 may be of the type εold by Linear Technologieε under the part number LT1214, for example.

The output signal 32 of the IF LPF 30 is coupled to the IF amplifier 33 which, in the illustrative embodiment, is implemented in two stageε 33a, 33b. The IF amplifierε 33a and 33b are reεponsive to a AGC signals 182 for selecting the gain to be introduced by such amplifiers. In particular, each of the IF amplifierε 33a and 33b provides OdB or lOdB of gain. The gain provided by each such amplifier stage is selected by a quad FET switch 180, of the type described above in conjunction with the up conversion mixers, in reεponεe to control εignalε 182. FET1 and FET2 control the gain of the firεt amplifier εtage 33a. Control εignalε 182 alternatingly actuate FET1 and FET2 εuch that, at any given time, one of FET1 and FET2 iε cloεed and the other iε open. In particular, if FET1 is on (i.e., cloεed), then feedback resistor R72 is connected between the inverting input and output of amplifier 33a and if FET2 is closed, then the feedback resistor R55 is connected between the inverting input and output of amplifier 33a. In a similar manner, FET3 and FET4 of switch 180 control the gain of the amplifier stage 33b. Control signalε 182 alternatingly control FET3 and FET4 such that only one of such FETs is on at any given time. If FET3 iε closed, then feedback resiεtor R149 is connected between the inverting input and output of amplifier 33b and if FET4 is closed, then the feedback resistor R126 is connected between the inverting input and output of amplifier 33b. In the illustrative embodiment, control εignalε 182 are provided by signal processor generated GC2 and GC3 signalε 181 (Fig. 8A) . Specifically, when GC2 and GC3 are both low, stages 33a and 33b introduce OdB of gain and when GC2 is high and GC3 is low, each of stageε 33a and 33b introduceε lOdB of gain.

Additional DC offset compensation circuitry is provided for the gain stage 33 by a servo amplifier 200 which compares the output signal 37 of amplifier stage 33b to a reference voltage. The output signal 204 of the servo amplifier 200 iε fed back to the non-inverting inputs of the amplifiers 33a and 33b, aε shown. With this arrangement, the DC offset associated with the amplifiers 33a, 33b is compensated. In the illustrative embodiment, the servo amplifier 200 is of the type εold by Maxim under the part number MAX400.

The phase comparator 250 receives the IF output signal 37 of the amplifier εtage 33b at itε non-inverting input and a reference voltage at itε inverting input, such as 2.5 volts. The Q and ^~Q output signals 252, 254 of the phase comparator 250 are indicative of the zero crossings of the IF signal 37. Phaεe comparator output εignalε 252, 254 are coupled to the quantizer 34 (Fig. 8) . In the illuεtrative embodiment, the phaεe comparator iε of the type sold by Maxim under the part number MAX913. The IF signal 37 is additionally coupled to the RSSI circuit 38, a portion of which is shown in Fig. 7. In

SUBSTlTUli; SHEET E 26) particular, the IF signal 37 iε coupled to a 2X inverting amplifier 258, the output of which is coupled to the RMS to DC converter 270. The output signal 262 of the RMS to DC converter 270 is further procesεed by an operational amplifier 268 which provideε an RSSI signal 310 to an A/D converter 282 (Fig. 8) .

Referring also to Figs. 8-8A, the digital portion 10b of the receiver 10 is shown to include the quantizer 34, RSSI circuit 38 and signal processor 35. In the GSM embodiment, the RSSI circuit includes a successive approximation A/D converter 282 of the type sold by Maxim under part number MAX153. The output of the A/D converter 282 is an eight bit digital signal 292 representative of the RMS amplitude of the received RF signal. The A/D output signal 292 is coupled to the signal procesεor 35, as shown. In particular, the εignal processor 35 εelectε between the P/D converter output εignal 264 and the A/D converter output signal 292 for receipt. Chip select logic 308 iε provided for thiε purpose.

The Q output signal 252 and the ^~Q output signal 254 of the phase comparator 250 are coupled to the P/D converter 260. The CLK signal 356 coupled to the P/D converter is provided by a clock generator 420, such as the illustrated 100MHz crystal oscillator. While the illustrated signal processor 35 is a digital signal processor (DSP) , other types of εignal proceεεorε may be εuitable, particularly for non- mobile receiver applicationε, as described below. The sixteen bit DATA OUTPUT signal 264 and the ^~DA signal 406 are coupled from the P/D converter 260 to the DSP 35 and the P/D converter 260 receives the CSO signal 414, the CS1 signal 418, the R^~D εignal 412, and the ^"WR signal 410 from the DSP

35. More particularly, WR , RD and DMS output εignalε of the DSP 35 are buffered by a buffer 484 to provide the ^"WR , ^"RD and CSO input εignalε to the P/D converter 260 which are also coupled to IfR and ^~RD inputs of the A/D converter 282 and to the chip select circuit 308, aε εhown. Alεo provided in the digital portion 10b of the receiver 10 is power up/reset circuitry 450 for the DSP 35. An EPROM 454 stores the code executed by the DSP 35. Upon power up, code stored in the EPROM 454 is read and executed by the DSP 35. A one bit latch 458 is coupled to the DSP 35. The AO,

Al and A2 control inputs of the latch 458 (collectively labelled 460) receive control signalε from the DSP by which one of the eight latch outputε Q0-Q7 iε selected for coupling to the latched data input signal 462. In the illuεtrative embodiment, only seven of the latch outputs Q0, Ql, Q2, Q3, Q4, Q6 and Q7 are used. In particular, latch output Q0 provides the GCO εignal 49, latch output Ql provideε the GCl signal 120, latch output Q2 provides the GC2 signal, latch output Q3 provideε the GC3 εignal, latch output Q4 provides the LNA control signal 41, latch output Q6 provides the sample and hold control εignal 151 and latch output Q7 provideε the receiver data output signal 470. Enable circuitry 464 provides an enable εignal ^~E to the one bit latch 458. The GC2 and GC3 εignals together (labelled 181) provide gain control signalε 182 (Fig. 7) for selecting the gain of amplifier stages 33a and 33b, as discuεεed above.

The receiver data output εignal 470 iε repreεentative of the tranεmitted bit and iε coupled to a flip-flop 486 and to a clock recovery circuit 488. A divider/clock source 480 divides the frequency of the OSCOUT signal 65 to generate the LO signal 139 for use in up converεion (Fig. 4) and a εecond εignal 490 for uεe by the clock recovery circuit 488. In the GSM embodiment, the OSCOUT εignal 65 is a 13.0MHz signal which is divided by four to generate the signal 490 at 3.25MHz and iε divided by sixteen to generate the LO signal 139 at 812.5KHZ.

The clock recovery circuit 488 provides a recovered clock signal 482 at the data rate and in-phase with the transmitted RF signalε. The recovered clock εignal 482 is coupled to DSP 35 and the flip-flop 486, as shown. The flip- flop 486 clocks the receiver data output signal 470 in accordance with the recovered clock signal 482 in order to provide an RXDATA signal which iε repreεentative of the recovered tranεmitted bit.

The RSSI circuit output 292 is used by the DSP 35 to adjust the overall system gain via the AGC control signals 49, 120 and 182 (Fig. 2). More particularly, in operation, the gain stageε provided by the AGC amplifier 48, the IF amplifiers 116 and 118, and the IF amplifier 33 are sequentially "turned off" (i.e., εwitched to a gain of OdB), from the rear of the receiver 10 proximal to the εignal proceεsor 35, forward toward the antenna 14 as the received signal strength increaseε.

Referring to Fig. 9, a graph illuεtrates peak-to-peak signal level over the dynamic range of the receiver 10. Curve 500 illustrates the signal level of a received RF signal. The other curveε 502, 504, 506, 508 and 510 represent gain at various εtageε of the receiver. Specifically, curve 502 represents the gain at the output of the IF amplifier 33, curve 504 representε the gain at the outputs of the IF amplifiers 116 and 118, curve 506 repreεentε the gain at the output of the up converter mixerε 132, 134 and curve 508 repreεentε the gain at the output of the down converεion εub-circuit 52.

The εtepε in each of theεe curveε represent the "turning off" of the particular gain stage aε the received signal strength increaεeε. Thiε operation is performed by the signal processor 35 and, specifically, by signals GCO 49, GCl 120, GC2 and GC3 181 (Fig. 8A) . Step 512 represents the gain of amplifier stage 33b being switched from lOdB to OdB when the received εignal εtrength reaches approximately -85dBm, step 514 represents the gain of amplifier stage 33a being switched from lOdB to OdB when the received signal strength reaches approximately -75dBm, εtep 516 repreεentε the gain of amplifiers 116 and 118 being switched from 2OdB to OdB when the received εignal εtrength reacheε approximately - 65dBm and εtep 518 repreεentε the gain of amplifier εtage 48 being switched from 22dB to -3dB when the received signal strength reaches approximately -45dBm. The choice of when to turn off each of the gain stageε is baεed on a compromise of SNR ratio and signal (i.e., gain) compression, as may occur when exceεεive signal strengths cauεe the amplifiers to saturate.

The signal procesεor 35 demodulateε and equalizeε the output 264 of the P/D converter 260. Recall that amplitude fading is addressed by the hard amplitude limiting of the phase comparator 250. Thus, the signal procesεor 35 need not provide equalization to addresε amplitude fading. Intersymbol interference (ISI) is a phenomena which is introduced deliberately in GSM syεtems, by the modulator at the transmitter and is produced by each bit of energy being spread over several bit periods, both before and after the currently transmitted bit. Thiε energy spreading is the result of the impulse responεe of the Gauεsian filter at the modulator. The equalizer of the signal procesεor 35 addreεses ISI in order to prevent degradation in the bit error rate (BER) caused by a reduction in the SNR of the receiver. In the illustrative GSM embodiment, ISI equalization is achieved with a decision-feedback equalization (DFE) approach.

Referring to Fig. 10, a functional block diagram of an embodiment of the signal proceεεor 35 for use in non-mobile receiver applications is shown. Demodulation, or εignal detection, iε achieved with a one-bit differential detector. To this end, the P/D converter 260 provides the count value signal 264 repreεentative of the period between conεecutive zero crossingε of the IF εignal 37 to an integrator 580 of the εignal proceεεor 35. The count value signal 264 is integrated over a one bit time interval T_b as provided to the integrator by a signal 584.

ISI equalization is performed by summing the integrated signal 581 with a DFE εignal 583 generated in a manner described below. A data slicer 586 determineε whether the εummation signal 585 represents the transmission of a one or a zero over the bit time T_b. The output signal 587 of the data slicer 586 representε the recovered, tranεmitted bit. For purposes of linearization, a scale factor α is determined in accordance with whether the output signal 587 is a one or zero. That is, the P/D count value signal generated in responεe to a mark (logic 1) or a εpace (logic 0) condition at the modulator (i.e., tranε itter) haε a non¬ linear transfer function due to the l/f or period measurement. For example, in the case where the center IF frequency is approximately 203KHz corresponding to a P/D count of approximately 246, a +67KHz deviation equals an IF frequency of 270KHz and a P/D count value of approximately 184, whereas a -67KHz deviation at the transmitter results in an IF frequency of 135KHZ and a P/D count value of approximately 369. In one embodiment, if the transmitted bit is a one, then α is set to -0.67 and, if the transmitted bit is a zero, then α is set to +1.

After a one bit delay 588, the scale factor a is set equal to the previous scale factor value .,. The previous scale factor ct_, is then summed with a DFE constant e 592 by a multiplier 594 to provide the DFE signal 583. The DFE constant e is derived from the transfer ratio of the l/f P/D process (i.e., the non-linear transfer function). The ISI equalization provided by multiplying the scale factor ._j and the DFE constant e implements a feedback filter 596, which may be referred to as a postcurεor equalizer. Aε εuch, the feedback filter 596 eliminateε the postcursor portion of the ISI (i.e., the interference from past data symbols) . Referring also to Fig. 11, a flow diagram of stepε performed by the εignal processor of Fig. 10 for demodulation and ISI equalization is εhown. The signal proceεεor 35 demodulateε received εignalε by integrating the difference between a currently detected P/D count value signal and a previously detected P/D count value signal over a single bit time. To this end, the signal procesεor 35 waitε for an interrupt IRQ2 to occur in step 600. Interrupt IRQ2 is generated by the ^"DA output signal 406 of the P/D converter 260 at each detected zero crossing.

In step 604, a timer is read to determine the time ET₂ that has lapsed since a prior IRQ2 interrupt was received and the time ET₂ stored. Also in step 604, the timer is loaded with a count of 100 and restarted. Note that in the illustrative embodiment, one bit time is equal to 80 machine cycles, so the timer is loaded with a value greater than one bit time.

The most recently computed difference value d(x) is stored as a previouε difference value d(x)., in εtep 608. In a εubsequent step 612, the current count value signal 264 from the P/D converter 260 (i.e., N) is εtored as a previous count value N.j. The current count value N is then provided by the count value signal 264 from the P/D converter 260 in step 616.

Stepε 620-648 collectively correεpond to the operation of the integrator 580 in Fig. 10. In εtep 620, the difference value d(x) iε calculated by εubtracting 246 from the value N. An intermediate integrator value f(x)' iε then calculated in step 624 by multiplying the previous difference value d(x)__! by the timer value ET₂. The intermediate integrator value f(x)' is then updated in step 628 by incrementing the intermediate integrator value f(x)' by a previouε intermediate integrator value f(x)'.,. In step 632, the intermediate integrator value f(x)' is stored as the previous intermediate integrator value f(x) '., . Note that the Mask Off IRQ0 box 634 indicates that the serieε of sequential steps 604 through 632 is fully executed once entered.

The εignal proceεεor 35 then idles in step 636 and waits for a subsequent interrupt IRQ0 in step 640. Interrupt IRQ0 is provided by the 270.833Khz recovered clock signal 482 (Fig. 8A) . Once an IRQ0 interrupt is received, εtep 644 iε performed in which the timer iε again read and the lapεed time ET₀ since the prior IRQO interrupt stored. The timer is also loaded with a count of 100 and restarted.

Thereafter, in step 648, the integrator value f(x) is updated by adding the intermediate integrator value f(x)' to the product of the difference value d(x) and the timer value ET₀. In step 652, the previous intermediate integrator value f(x)'_., iε reset to zero. The Mask Off IRQ2 box 654 indicates that sequential stepε 640 through 652 are completed once the sequence is entered. In εtep 656, the integrator value f(x) is stored as the value I.

Step 660 implements the one bit adaptive DFE equalization discussed above in conjunction with Fig. 10 by which the integrated signal 581 (Fig. 10) is summed with the DFE signal 583. In particular, the value I is summed with the product of the DFE constant e and a previous scale factor α.,.

The data slicer 586 of Fig. 10 iε implemented in εtepε 664 and 680. In particular, in εtep 664, it is determined whether the value I is greater than or equal to zero. In the event that I is greater than or equal to zero, then the tranεmitted bit iε determined to be a zero. Alternatively, if it is determined in step 680 that I is leεε than zero, then the tranεmitted bit is determined to be a one.

In the event that the transmitted bit is determined to be a zero, then step 668 is next performed, in which the scale factor α is set to -.67. Alternatively, if it iε determined that the tranεmitted bit iε a one, then εtep 684 iε performed in which the εcale factor α of set to +1. After steps 668 and 684, step 672 is performed in which the data signal 462 (Fig. 8A) iε provided to the one bit latch 458. The previouε scale factor is then set to the current scale factor in step 676, following which the signal procesεor 35 idles again in step 636.

In mobile receiver applicationε, in addition to ISI equalization and demodulation, the εignal proceεεor 35 additionally equalizes to addreεε time diεperεion. Time dispersion is a phenomena whereby changeε in the impulεe response of the mobile channel cause an increased BER. To this end, in mobile receiver applications, an adaptive multipath equalization technique is used to address both time dispersion and ISI equalization.

In order to implement adaptive multipath equalization, in mobile receiver applications, it may be desirable to implement the signal processor 35 with a DSP. In non-mobile receiver applications however, the εignal processor 35 may be implemented with hardware logic. This is the because the adaptive equalization used to address time disperεion is not necesεary, thereby greatly εimplifying the functionality of the εignal proceεεor 35.

Referring to Fig. 12, a functional block diagram of a signal procesεor 35 for use in mobile receiver applications iε εhown to include an adaptive multipath equalization scheme. The P/D converter 260 provides the count value signal 264 representative of the period between consecutive zero crossings of the IF signal to an integrator 702. Integrator 702 integrateε the count value signal 264 over a one bit time interval T_b as provided by a signal 704. The integrated signal 708 is procesεed by a feedforward filter 710. The feedforward filter 710 may be referred to aε a precurεor equalizer and performε the function of a whitening matched filter. The feedforward filter 710 alεo equalizeε the precursor portion of the ISI (i.e., interference from future data symbols) .

The feedforward filter 710 includes a plurality of delay circuits 712₀, 712,, 712₂, ...712_n and corresponding feedforward coefficients g₀, g,, g₂, ...g_n.,. Each of the coefficients g₀, g,, g₂, ...g_n., is multiplied in a multiplier 714₀, 714,, 714₂, ...714_n., by a reεpective output εignal 718₀, 718,, 718₂, ...718-,-_! of a correεponding delay circuit 712_0/ 712,, 712₂, ...712_n. The εignalε 720₀, 720,, 720₂, ...720_n., provided by multipliers 714₀, 714,, 714₂, ...714_n., are coupled to a εummation circuit 724 where they are summed with output signals 714₀, 714,, 714₂, ...714_n_, from a feedback filter 750. The output of the summation circuit 724 is coupled to a data slicer 728 for determination of the tranεmitted bit at each bit time T_b and to a summation circuit 730. Summation circuit 730, in conjunction with a coefficient adaptation circuit 739 and a stored training sequence code 738, adaptively determines optimum values for the feedforward coefficients g₀, g,, g₂, g₃...g_n., and feedback DFE coefficientε e₀, e,, e₂, e₃...e_m. The training sequence code 738 is specified by the GSM specification for the received time slot and, in particular, is specified by the 1994 European Telecommunications Standards Inεtitute (ETSI) Standard GSM 05.02 specification which is incorporated herein by reference.

More particularly, every received GSM burst contains 156.25 bits, of which 26 bits comprise a training sequence. The summation circuit 730 strips the training sequence from the input to the data εlicer 728 and correlateε the training sequence to the stored training εequence code 738. The reεult of thiε correlation iε provided to the coefficient adaptation circuit 739 which iteratively εets the feedforward coefficients g₀, g,, g₂, g₃...g„-, and feedback DFE coefficients ^o, ei, e₂, e₃...e_m ^and compares the training sequence stripped from the resulting input to the data slicer 728 to the known training εequence code 738 until an optimum match is achieved.

The feedback filter 750, like the feedback filter 596 of Fig. 10, eliminates the postcurεor portion of ISI, albeit in an adaptive manner. To thiε end, the feedback filter 750 includeε a plurality of delay circuitε 740₀, 740,, 740₂, ...740_n, the outputε of which provide reεpective scale factor valueε α_.,, α_.2, α_.3, ... ._n. The εcale factorε α.,, α_₂, α_₃, ...α__n are multiplied in multipliers 742,, 742₂, 742₃, ...742_m by corresponding feedback coefficients e,, e₂, e₃, ...e_m. The signals 744,, 744₂, 744₃, ...744_m provided by multipliers 742,, 742₂, 742₃, ...742_m are coupled to the summation circuit 724 where they are summed with output signals 714_0/ 714,, 714₂, ...714.,., from the feedforward filter 710.

Referring also to Fig. 13, a flow diagram of steps performed by the signal processor of Fig. 12 is shown. It is noted that the stepε associated with adaptive coefficient determination are omitted from Fig. 13 for simplicity. In step 800, the signal processor 35 waits for an interrupt IRQ2 which is generated by the ~DA output signal 406 of the P/D converter 260. In step 804, a timer is read to determine the time ET₂ that has lapsed since a prior IRQ2 interrupt was received. The time ET₂ is stored and the timer iε loaded with a count of 100 and reεtarted.

In step 808, the current difference value d(x) is stored as the previous difference value d(x).,. Thereafter, the current count value signal 264 from the P/D converter 260 is stored as a value N. A new difference value is then computed in step 816 by subtracting 246 from the value N. An intermediate integrator value f(x)' iε then computed in εtep 820 by multiplying the previouε difference value d(x)., by the timer value ET . In a εubεequent εtep 824, the intermediate integrator value f(x)' iε incremented by the previouε intermediate integrator value f(x)'.,. In εtep 828, the current intermediate integrator value f(x)' is set to the previouε intermediate integrator value f(x)'.,. The Mask Off IRQ0 box 832 indicates sequential steps 804 through 828 are completed once the sequence is entered.

In a subεequent εtep 836, the signal procesεor 35 idleε, after which it waitε for an interrupt IRQ0, provided by the 270.833KHZ recovered clock εignal 482 (Fig. 8A) . Once an IRQ0 interrupt iε received, step 844 is next performed in which the timer iε again read and the lapεed time since the last IRQ0 interrupt stored as timer value ET₀. The timer is also loaded with a value of 100 and restarted in step 844. In a step 848, a value I is set equal to the sum of the current integrator value f(x)' and the product of the difference value d(x) and the timer value ET₀. Steps 816-848 collectively represent the operation of the integrator 702 of Fig. 12. Thereafter, the previous intermediate integrator value f(x)'., is set to zero. The operation of the feedforward filter 710 and summation circuit 724 is then performed in step 856, in which a value I' is computed by summing the products of the delayed εignals I_n., 718₀, I., 718,, ... I 718_n_, with the corresponding feedforward coefficient g₀, g,, ... g_n_,. Note that in the illustrative flow diagram of Fig. 13, n=3. Thereafter, in a step 860, the operation of the feedback filter 750 and the summation circuit 724 is performed by computing a value I" which is equal to the summation of the value I' and the product of each of the scale factors .,, α.₂, ...α_.n with the corresponding DFE constant e,, e₂, ..e_m. In the illustrative embodiment, the value m=2. The value I ' ' representε the output εignal of the εummation circuit 724 provided to the data slicer 728 (Fig. 12) .

The operation of the data slicer 728 is performed in steps 864 and 872. In particular, in step 864, if it iε determined that the value I' ' is greater than or equal to zero, then the transmitted bit is a zero. In εtep 872, if it iε determined that the value I" iε leεε than zero, then the tranεmitted bit iε a one. The εcale factor value α iε then computed in steps 868 and 876. In particular, if the transmitted bit is a zero, then the scale factor value α iε εet to -1 in step 868. Alternatively, if the transmitted bit is a one, then the scale factor value α is set to +1 in εtep 876. Thereafter, in εtep 890, the data εignal 462 (Fig. 8A) iε provided to the one bit latch 458. The scale factor values α.,, _.2, ...α_.n are then computed in εteps 894 and 896, for n=2. That is, in step 894, the α_ εcale factor value is εet equal to the a., εcale factor value and, in step 896, the α., scale factor value iε εet equal to the a εcale factor value. The Mask Off

SIΪEST-TUH SHEET 10LE 26) IRQO box 900 indicateε that the εequence of steps 840 through 896 is completed once entered.

As noted above, the receiver 10 described herein is readily adaptable for use in receiving RF signals modulated by various analog and digital schemeε. In particular, the required modificationε to the receiver include modifying the bandwidth of the RF BPF 18, the down converter frequency (i.e., the frequency of the LO εignalε 60 and 62 in Fig. 2), the bandwidth of the IF LPF 30 and the demodulation technique employed by the signal procesεor 35.

Consider for example, modification of the receiver 10 to procesε received RF εignalε which are analog FM modulated in the FM broadcaεt band. In thiε case, the bandwidth of the RF BPF 18 is changed from between 935MHz and 960MHz to between 88MHz and 108MHz. The frequency of the down converεion LO signals 60, 62 is reduced from between 935.2MHz and 959.8MHz to between 88MHz and 108MHz, depending on the channel of interest, and the IF bandwidth remains unchanged at 200KHz. In proceεsing FM signalε, the demodulation performed by the signal proceεsor 35 generates a complex signal. A conventional stereo decoder iε uεed to decode L and R channelε.

Having described the preferred embodiments of the invention, it will now become apparent to one of skill in the art that other embodiments incorporating their conceptε may be used. It is felt therefore that theεe embodiments should not be limited to disclosed embodiments but rather should be limited only by the spirit and εcope of the appended claimε.

Claims

1. An RF receiver comprising: a down converter for converting a received RF εignal to in-phaεe and quadrature zero IF signals; a channel selection filter for paεεing a εelected frequency component of εaid in-phase and quadrature zero IF signals; and an up converter for converting said selected frequency component of said in-phase and quadrature zero IF signals to an IF signal.

2. The RF receiver recited in claim 1 wherein said channel selection filter comprises an in-phase active filter and a quadrature active filter.

3. The RF receiver recited in claim 1 wherein said down converter is an image reject filter.

4. The RF receiver recited in claim 1 wherein said up converter is an image canceling filter.

5. The RF receiver recited in claim 1 further comprising an amplifier for providing gain to εaid IF εignal.

6. The RF receiver recited in claim 5 further compriεing a phase comparator responsive to said IF εignal for providing a zero croεεing εignal indicative of zero croεεingε of εaid IF εignal.

7. The RF receiver recited in claim 6 further compriεing a quantizer for quantizing said zero croεsing signal.

8. The RF receiver recited in claim 7 wherein εaid quantizer compriεeε a period-to-digital converter responsive to said zero crosεing εignal for providing a count signal representative of the period between consecutive zero crossings of said IF signal and the instantaneouε frequency of said IF signal.

9. The RF receiver recited in claim 8 further comprising a signal processor for processing said count signal.

10. The RF receiver recited in claim 9 wherein εaid signal processor compriseε: a demodulator for demodulating εaid count εignal; and an equalizer for removing the effects of intersymbol interference on said count signal.

11. The RF receiver recited in claim 10 wherein εaid εignal proceεεor further comprises a second equalizer for removing the effects of time disperεion on said count εignal.

12. The RF receiver recited in claim 11 wherein εaid εignal proceεsor is a digital signal procesεor.

13. A method for proceεεing a received RF εignal compriεing the εteps of: converting said RF εignal to in-phaεe and quadrature zero IF signalε; paεsing a εelected frequency component of εaid in-phase and quadrature zero IF signals; and converting said pasεed frequency εelected component of said in-phase and quadrature zero IF signals to an IF signal.

14. The method recited in claim 13 further compriεing the εtep of providing gain to εaid IF signal.

15. The method recited in claim 13 further comprising the step of comparing said IF signal to a reference signal to provide a zero croεεing εignal indicative of zero crossingε of εaid IF signal.

16. The method recited in claim 15 further comprising the step of quantizing said zero crossing signal to provide a count signal representative of the period between consecutive zero croεεings of said IF signal, and the inεtantaneouε frequency of said IF εignal.

17. The method recited in claim 16 further comprising the steps of: equalizing said count signal to remove the effect of intersymbol interference on said count signal; and demodulating said count εignal.

18. A RF receiver compriεing: a converter and filter for converting a received RF εignal to an IF εignal having a εelected frequency component; a phaεe comparator for comparing said IF signal to a reference signal to provide an amplitude limited output signal indicative of zero croεsings of said IF εignal; and a quantizer for quantizing εaid output εignal of said phase comparator so as to provide a count signal representative of the period between consecutive zero crosεingε of εaid IF signal and the instantaneouε frequency of εaid IF εignal.

19. The RF receiver recited in claim 18 wherein said converter and filter comprises: a down converter for converting said received RF signal to in-phase and quadrature zero IF signals; a channel selection filter for pasεing εaid εelected frequency component of said in-phase and quadrature zero IF signals; and an up converter for converting εaid selected frequency component of said in-phase and quadrature zero IF signals to an IF signal.

20. The RF receiver recited in claim 18 further comprising a signal processor for demodulating and equalizing said count signal.

21. An RF receiver receiving a transmitted RF signal modulated in accordance with a predetermined modulation scheme, said RF receiver comprising: an RF bandpass filter for filtering said received RF signals to pass RF signals within a predetermined frequency band; a down converter for converting an RF εignals within said predetermined frequency band to in-phase and quadrature zero IF signals, said down converter compriεing a mixer for mixing said RF signalε within εaid predetermined frequency band with a local oscillator signal having a first local oscillator frequency; a channel selection filter responεive to said in-phase and quadrature zero IF signalε fbr paεsing a selected frequency component of said in-phase and quadrature zero IF signals; an up converter for converting said selected frequency component of said in-phase and quadrature zero IF signals to an IF signal having an IF frequency, said up converter compriεing a mixer for mixing εaid selected frequency component of said in-phase and quadrature zero IF εignalε with a local oεcillator signal having a second local oscillator frequency; a quantizer for quantizing said IF signal to provide a count signal indicative of the period between conεecutive zero crossingε of said IF signal; and a demodulator for demodulating said count signal in accordance with a demodulation technique in order to recover εaid transmitted RF signal, wherein said predetermined frequency band, said first local oscillator frequency, IF frequency and said demodulation technique are selectable in accordance with said predetermined modulation scheme of said RF εignals.