Kfrequency tracking system using an analog correlator

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority under 35 U.S.C. §119(e) to U.S. Provisional Application Ser. No. 60/834,199, entitled KFM FREQUENCY TRACKING SYSTEM USING AN ANALOG CORRELATOR, filed Jul. 28, 2006, which is incorporated by reference herein for all purposes. This application is related to co-pending U.S. patent application Ser. No. 11/829,817 entitled K_FM PHASE TRACKING SYSTEM USING A DIGITAL CORRELATOR, filed on even date herewith.

FIELD OF THE INVENTION

The present application relates generally to phase/frequency modulators, and more particularly, to a system for K_FM frequency tracking using an analog correlator.

BACKGROUND OF THE INVENTION

Phase modulation schemes are very effective and are therefore widely used in communication systems. A simple example of a phase modulation scheme is quaternary phase shift keying (QPSK). FIG. 1 shows a constellation diagram that illustrates how QPSK maps two-bit digital data to one of four phase offsets. FIG. 2 shows a typical QPSK (or I/Q) modulator used to generate a phase-modulated signal. This technique relies on orthogonal signal vectors to realize the phase offsets—an inherently linear technique, since it depends solely on the matching of these orthogonal signals.

The I/Q modulator provides a straightforward approach to generating phase-modulated signals. However, it is also possible to generate the phase-modulated signals using a phase-locked loop. This approach offers reduced circuitry, lowers power consumption, and as a result, finds widespread use in narrowband systems. A variation of this approach, known as two-point modulation, introduces direct modulation of the VCO to support wideband phase modulation. Unfortunately, this requires the gain of the VCO to be accurately set—a difficult task since the VCO gain depends on multiple factors. It would therefore be advantageous to have a system that operates to accurately set the gain of a VCO.

SUMMARY OF THE INVENTION

In one or more embodiments, a very efficient system for wideband phase modulation is provided. The system operates to set the gain and offset parameters of a VCO within a phase modulation system, and then periodically adjust these parameters based on the operation of a frequency tracking loop.

One aspect of the invention relates to an apparatus providing a phase/frequency modulation system. The apparatus includes a first circuit configured to introduce an offset to center a signal applied to a VCO. The apparatus further includes a second circuit configured to set a gain of the VCO. A frequency tracking network is configured to dynamically adjust one or both of the offset and the gain.

In another aspect the invention pertains to an apparatus for tracking in a two point phase/frequency modulation system to adjust an offset and scale of a signal applied to a VCO of a phase-locked loop (PLL). The apparatus includes a correlator configured to project a PLL correction signal onto a reference modulation signal. The apparatus further includes an integrator for accumulating an output of the correlator. A window comparator determines a polarity of an output of the integrator. The apparatus also includes logic having a first counter configured to scale the signal applied to the VCO based on an output of the window comparator, and having a second counter configured to offset the signal applied to the VCO based on the output of the window comparator.

An aspect of the invention is also directed to an apparatus for tracking in a two point phase/frequency modulation system to adjust an offset and scale of a signal applied to a VCO. The apparatus includes a correlator operative to project a PLL correction signal onto a reference modulation signal and produce a correlation signal. An A/D converter produces a digital signal based upon the correlation signal. The apparatus also includes logic configured to compute a system error and to generate, based on the digital signal, a first output used in offsetting the signal applied to the VCO and a second output used in scaling the signal applied to the VCO.

In yet another aspect the invention relates to a phase/frequency modulation system. The system includes a phase-locked loop (PLL) having a VCO including a first port and a second port wherein during operation of the PLL an error signal is received by the first port. A correlator network is configured to track the error signal and produce an output signal. The system also includes a tracking loop which generates, in response to the output signal, a VCO control signal applied to the second port.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing aspects and the attendant advantages of the embodiments described herein will become more readily apparent by reference to the following description when taken in conjunction with the accompanying drawings wherein:

FIG. 1 shows a constellation diagram that illustrates how QPSK maps two-bit digital data to one of four offsets;

FIG. 2 shows a diagram of a typical I/Q modulator;

FIG. 3 shows a PLL that is used to synthesize a radio frequency carrier signal;

FIG. 4 shows a mathematical model of the PLL shown in FIG. 3;

FIG. 5 shows an integration filter;

FIG. 6a shows one embodiment of a fractional-N PLL using a AS modulator;

FIG. 6b illustrates the frequency response of the fractional-N PLL shown in FIG. 6a;

FIG. 7a illustrates one embodiment of a fractional-N PLL that supports direct frequency or phase modulation;

FIG. 7b illustrates the frequency response of the direct phase/frequency modulator shown in FIG. 7a;

FIG. 7c illustrates the effect on the frequency response of the two point frequency modulator shown in FIG. 7a with unmatched VCO gain K_FM;

FIG. 7d illustrates the time domain impact when the two point frequency modulator shown in FIG. 7a operates with unmatched VCO gain K_FM;

FIG. 8a shows a detailed view of a voltage-controlled oscillator;

FIG. 8b shows one embodiment of a VCO tank circuit that includes an auxiliary port to support linear phase/frequency modulation;

FIG. 9a shows the capacitance-voltage relationship for an accumulation-mode MOSFET device;

FIG. 9b shows the linear capacitance-voltage response from back to back MOSFET devices;

FIG. 10a shows the equivalent series capacitance of the back-to-back MOSFET devices within the VCO detailed in FIG. 8;

FIG. 10b shows the frequency modulation resulting from the control of the back-to-back MOSFET devices in the VCO of FIG. 8;

FIG. 10c shows the VCO gain K_FM sensitivity for the VCO in FIG. 8;

FIG. 11 illustrates the direct phase/frequency modulator with an adjustable VCO offset and gain K_FM;

FIG. 12 shows a block diagram of the system used to initially set the parameters of the direct phase/frequency modulator;

FIG. 13 shows a simplified block diagram of a correlator network included within the K_FM tracking system of the present invention;

FIG. 14a shows a diagram of an embodiment of a calibration system for calibrating the scaling parameter α and the VCO gain K_FM associated with the FM port of the VCO using the correlator network of FIG. 13;

FIG. 14b illustrates the timing associated with the calibration system of FIG. 14a;

FIG. 14c illustrates the signals generated by the PLL of FIG. 14a with different a scaling parameters;

FIG. 14d illustrates the convergence of the calibration system of FIG. 14a;

FIG. 15 shows a block diagram of an embodiment of the calibration system of FIG. 14a using an A/D converter instead of an analog window comparator;

FIG. 16 shows the effect of K_FM curvature on the scaling parameter a as determined by the calibration systems shown in FIG. 14a and FIG. 15;

FIG. 17a shows the operation of the tracking system for WCDMA RCC1 data with α set low;

FIG. 17b shows the operation of the tracking system for WCDMA RCC1 data with α set high;

FIG. 17c illustrates the convergence of the tracking system of FIG. 14a with WCDMA RCC1 data;

FIG. 18a shows the operation of the tracking system for WCDMA RCC3 high-speed data with α set low;

FIG. 18b shows the operation of the tracking system for WCDMA RCC3 high-speed data with a set high; and

FIG. 18c illustrates the convergence of the tracking system of FIG. 14a with WCDMA RCC3 high-speed data.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

A phase-locked loop (PLL) comprising a voltage-controlled oscillator (VCO), feedback counter (counter), phase/frequency detector (P/FD), charge pump (CP), and integration filter (LPF) is shown in FIG. 3. It is described by the mathematical model shown in FIG. 4.

The phase-locked loop uses feedback to minimize the phase difference between a very accurate reference signal and the PLL's output signal. As such, the PLL produces an output signal at a frequency given by;
f_VCO=Nf_REF

where f_vco is the frequency of the VCO output signal, N is the value of the feedback counter, and f_REF is the frequency of the reference signal.

The VCO produces an output signal at a frequency set by the control voltage v_ctrl according to;
v_out(t)=A cos(ω_ot+K_vco∫V_ctrl(t)dt)
where ω_o is the free-running frequency of the VCO and K_vco is its associated gain. The gain K_vco describes the relationship between the excess phase of the carrier Φ_out and the control voltage v_ctrl with;

Φ out ⁡ ( s ) v ctrl ⁡ ( s ) = K vco s
where K_vco is in radians/V. The VCO drives the feedback counter, which divides the output phase Φ_out by N. When the phase-locked loop is stable, the phase detector and charge pump circuits generate an output signal i_CP that is proportional to the phase difference Δθ between the two signals applied to the phase detector. The output signal i_CP can therefore be expressed as;

i CP ⁡ ( s ) = K pd ⁢ Δ ⁢ ⁢ θ ⁡ ( s ) 2 ⁢ ⁢ π
where K_pd is in A/radians and Δθ is in radians.

A simple integration filter, comprising resistor R₁ and capacitors C₁-C₂ as shown in FIG. 5, transforms the output signal i_CP to the control voltage v_ctrl as follows;

v ctrl ⁡ ( s ) = i CP ⁡ ( s ) ⁢ Z ⁡ ( s ) = i CP ⁡ ( s ) ⁢ ( sR 1 ⁢ C 1 + 1 s 2 ⁢ R 1 ⁢ C 1 ⁢ C 2 + s ⁡ ( C 1 + C 2 ) )
where a zero (at 1/R₁C₁) has been added to stabilize the second order system and the capacitor C₂ has been included to reduce any ripple on the control voltage. Combining the above relations yields the closed-loop response of the system to an input signal as follows.

T 1 ⁡ ( s ) = NK PD ⁢ K VCO ⁢ Z ⁡ ( s ) sN + K PD ⁢ K VCO ⁢ Z ⁡ ( s ) .

In a PLL, the feedback counter value N effectively sets the output frequency. In practice, its digital structure restricts N to integer numbers. As a result, the frequency resolution (or frequency step size) of an integer-N PLL is nominally set by f_REF. Fortunately, it is possible to dramatically decrease the effective frequency step by manipulating the value of N to yield a non-integer average value. This is the concept of the fractional-N PLL.

A fractional-N PLL that uses a ΔΣ modulator to develop non-integer values of N is shown in FIG. 6a. The ΔΣ modulator advantageously pushes spurious energy (created by the changing values of the feedback counter) to higher frequencies where it can be more effectively attenuated by the PLL's integration filter. It can be shown that the effective value of N is the average value described by;

N = ∑ x = 1 P ⁢ N ⁡ [ x ] P
where N[x] is the sequence of feedback counter values. This expands to;
N[x]=N_int+n[x]
where N_int is the integer part and n[x] is the fractional part of N[x]. The ΔΣ modulator generates the sequence n[x] that satisfies;

∑ x = 1 P ⁢ n ⁡ [ x ] P = k M
where k is the input to the ΔΣ modulator with resolution M.

The ΔΣ modulator introduces quantization noise that appears at the PLL output. The pseudo-random sequence n[x] possesses a quantization error equal to ±½ around N or;

Δ = 1 N
It follows that the quantization noise spectral density for this error, assuming a uniform distribution, is expressed by;

e rms 2 ⁡ ( f ) = 1 6 ⁢ N 2 ⁢ f REF
over the frequency range of dc to f_REF/2. This quantization noise is advantageously shaped by an L^th order ΔΣ modulator according to;
DS(z)=(1−z⁻¹)^L

In the PLL system, the feedback counter acts as a digital accumulator and reduces the effects of the ΔΣ modulator. That is, the output phase from the feedback counter depends on its previous output phase. As a result, the transfer function for the feedback counter (or prescaler as it is sometimes referred) is therefore.

P ⁡ ( z ) = 2 ⁢ ⁢ π ⁢ z 4 1 - z - 1
Combining the above terms shows that the noise at the output of the feedback counter is equal to;
n²(f)=e_rms²(f)[DS(f)]²[P(f)]²
which yields;

n 2 ⁡ ( f ) = 2 3 ⁢ π 2 N 2 ⁢ f REF ⁡ [ 2 ⁢ ⁢ sin ⁡ ( π ⁢ ⁢ f f REF ) ] 2 ⁢ ( L - 1 )
This noise seen at the output of the feedback counter is in turn shaped by the PLL transfer function T₁(s).

It is possible to use a fractional-N PLL as a very efficient phase/frequency modulator. That's because signals applied to the ΔΣ modulator's input control and actually modulate the frequency of the VCO according to;
f_VCO=f_c+Δf(t)=(N_int+n[x])f_REF
where Δf(t) is the frequency modulation equal to;

Δ ⁢ ⁢ f ⁡ ( t ) = ( n ⁡ [ x ] - k M ) ⁢ f REF = FMf REF
and FM is the applied modulation signal. In practice, the modulation is shaped by the PLL response described by transfer function T₁(s). The PLL's response generally limits the bandwidth of the system so as to attenuate the ΔΣ modulator's quantization noise. This is illustrated in FIG. 6b. Consequently, this phase/frequency modulation approach supports only narrowband signals.

To overcome this bandwidth limitation, a second high-frequency modulation path is added to the VCO of the phase-locked loop, as shown in FIG. 7a. The resulting two-point frequency modulation system now displays a second and complimentary transfer function given by;

T 2 ⁡ ( s ) = sNK FM sN + K PD ⁢ K VCO ⁢ Z ⁡ ( s ) ⁢ v FM
where K_FM is the gain of the VCO port at which the v_FM modulating signal is applied. Ideally, the two transfer function expressions (T₁ and T₂) combine to yield a flat and uniform response as illustrated in FIG. 7b. This occurs when;
FMf_REF=K_FMv_FM

The challenge with two-point modulation and specifically direct VCO modulation is that it may require near exact control of the VCO's frequency and consequently the product K_FMv_FM. This is illustrated in FIGS. 7c and 7d. Any frequency errors produce phase deviations that accumulate with time. Fortunately, the PLL's feedback helps to reduce these frequency errors. This occurs because the VCO output is driven by the PLL's feedback and can be expressed as;
f_VCO=Nf_REF+FM_REF
which is also substantially equal to;
f_VCO=K_VCOv_ctrl+K_FMv_FM
where v_ctrl is the error signal produced by the phase/frequency detector and v_FM is the FM signal applied to the VCO. Consequently, the error signal v_ctrl compensates for any VCO gain errors within the bandwidth of the PLL's integration filter. Outside the PLL's bandwidth, the effect of the feedback decreases. This may make setting the VCO's gain K_FM to its designed value an issue for wideband modulation.

The gain of the VCO depends heavily on the circuit structure. FIG. 8a shows a detailed view of a typical VCO. It oscillates at a frequency set by the resonance of the LC tank circuit;

f asc = 1 2 ⁢ ⁢ π ⁢ ( L 1 + L 2 ) ⁢ C eq
where C_eq is the equivalent shunt capacitance (comprised of capacitor C₁ and varactors C_2a-C_2b plus any parasitic capacitance). The equivalent capacitance C_eq may also include coarse-tuning capacitors (not shown) to subdivide the tuning range. The varactor C₂ (shown as C_2a and C_2b) allows the VCO—by way of the control signal v_ctrl—to be tuned to different radio frequencies. The varactor can be realized various ways.

A VCO tank circuit that includes an auxiliary port to support linear phase/frequency modulation (v_FM input) is shown in FIG. 8b. It uses the capacitance of accumulation-mode MOSFET devices N₃ and N₄ to achieve linear behavior even though these devices display an abrupt response as illustrated in FIG. 9a. The accumulation-mode MOSFET device presents a low capacitance C_min at applied gate-to-bulk voltages V_GB below the threshold voltage V_T while it displays a high capacitance C_maxat applied voltages above V_T. (In the graph, V_T equals 0) Capacitors C_4a and C_4b block the dc level present at the VCO output. Resistors Z₁-Z₃ provide some isolation between MOSFETs N₃-N₄ and the v_FM signal.

The gate-to-bulk voltage V_GB applied to each MOSFET device N₃-N₄ depends on the oscillator's output signal Asinωt, the modulation signal v_FM, and the common-mode voltage v_cm. The symmetric structure of the VCO means signals V_LO+ and V_LO− are differential with;
V_LO+=A sin ωt V_LO−=−A sin ωt
where A is the peak signal of each sinusoidal output and ω is the oscillation frequency. It follows then that;
V_C3=A sin ωt+v_FM−v_cm V_C4=−A sin ωt+v_FM−v_cm
which describe the gate-to-bulk voltages V_GB applied to MOSFET devices N₃ and N₄. The two MOSFET devices connect back-to-back in the VCO tank circuit, so their individual capacitances behave oppositely.

The modulation signal v_FM affects the MOSFET devices as follows. The devices nominally present a capacitance equal to;

C mid = C FM ⁡ ( v FM = 0 ) = C min ⁢ C max C min + C max
when v_FM equals zero. As the modulation signal v_FM moves positive, both MOSFET devices spend more time at their maximum capacitance values C_max, so that for a time (t) approximately equal to;

t = 1 ω ⁢ sin - 1 ⁡ ( - v FM A )
the structure presents a capacitance equal to C_max/2. A similar response occurs as the signal v_FM moves negative, where the structure spends an increasing time at C_min/2. As a result, this back-to-back structure linearizes the overall response of the accumulation-mode MOSFETs to yield the behavior shown in FIG. 9b.

In the above analysis, the common-mode voltage v_cm is assumed to be ac ground. This introduces some error as the signal, v_cm, although small, is actually non-zero. The differential voltage ΔV (which equals V_LO+−V_LO−) applied to the back-to-back MOSFET devices is simply 2A sin ωt and is independent of each device's capacitance, C₃ and C₄. It has already been shown that the voltages applied to each individual MOSFET device do not track and their capacitances change oppositely. This affects the common mode voltage Vcm according to the expression;

v cm ⁡ ( t ) = A ⁢ ⁢ sin ⁢ ⁢ ω ⁢ ⁢ t - Δ ⁢ ⁢ V ⁡ ( C 4 C 3 + C 3 ) which  simplifies  to ; v cm ⁡ ( t ) = A ⁢ ⁢ sin ⁢ ⁢ ω ⁢ ⁢ t ⁡ [ 1 - 2 ⁢ ( C 4 C 3 + C 4 ) ]
Note that the bracketed term possesses the same sign as A sin ωt. This is because C₄<C₃ when sin ωt is positive and C₄>C₃ when sin ωt is negative. As a result, the second harmonic of A sin ωt appears attenuated at the common-mode point. This tends to reduce positive values of the modulation signal v_FM and expand negative values of the modulation signal, further linearizing the behavior of the back-to-back structure.

In the above analysis, it is also assumed that capacitors C_4a and C_4b are much larger than C_max. This allows most of the VCO output signal 2 A sin ωt to appear across MOSFETs N₃ and N₄. Otherwise, ΔV would change with the capacitance of the MOSFET devices.

FIG. 10a shows a graph that illustrates the average capacitance C_FM of the back-to-back MOSFET devices for different values of modulation signal v_FM. As expected, it spans from C_min/2 to C_max/2 and equals C_mid at zero. This variable capacitance shifts the resonant frequency of the VCO's LC tank as shown in FIG. 10b. The frequency shift appears linear, but its derivative;

K FM = ⅆ ⅆ v FM ⁢ f VCO
shows otherwise. The derivative (which is equivalent to the VCO gain K_FM), shown in FIG. 10c, reveals two problems. First, the peak VCO gain lies off center at an offset voltage v_OFF. This is because C_mid lies closer to C_max/2 than C_min/2. This introduces asymmetry in the curve and potentially leads to a phase error that grows over time as illustrated in FIG. 7d. Second, the VCO gain K_FM decreases as the modulation signal v_FM increases. This is caused by the sin⁻¹ function in the formula for the time that each MOSFET device spends at C_min or C_max. It follows that the VCO gain can be approximated by the 2^nd order expression;
K_FM(v_FM)=K_pk−K₁(v_FM−v_OFF)²
where K_pk represents the peak VCO gain at v_OFF and K₁ describes the droop in the curve.

The nonlinear behavior of the VCO limits the usefulness for direct phase/frequency and two-point modulation architectures. Although the PLL removes some of the distortion generated by the VCO response, the distortion becomes untenable for wideband modulation. Fortunately, the system shown in FIG. 11 improves the VCO linearity. It aligns the midpoint of the signal v_FM to the peak VCO gain K_pk, shifting v_FM by the appropriate offset level v_OFF with;
v_MOD=αv_FM+v_OFF

as shown. It also introduces a scaling parameter α to compensate for variations in the VCO gain K_FM.

The offset level v_OFF can be determined by using a few different methods. One method uses the calibration system shown in FIG. 12. It disables the PLL feedback loop by setting the v_ctrl input to a fixed level V_set. (Connecting the VCO's v_ctrl input to a fixed level breaks the PLL feedback loop and allows the VCO to free run to a frequency that depends only on v_FM.) The system steps the FM level v_FM and measures the frequency step Δf until it finds the largest frequency step. (The parabolic shape of the VCO gain K_FM curve allows use of a simple algorithm—such as a binary search routine.) The largest frequency step max(Δf) occurs at the peak VCO gain K_pk where v_FM corresponds to the proper offset level v_OFF. It follows that the peak VCO gain K_pk ca be expressed as;

K pk = max ⁡ ( Δ ⁢ ⁢ f ) Δ ⁢ ⁢ v FM
The parameter a then becomes approximately equal to;

α = K FM ⁡ ( ideal ) K pk
where K_FM(ideal) is the desired VCO gain with;
FMf_REF=K_FM(ideal)v_modd=αK_FMv_mod

In practice, the droop in the VCO gain curve actually reduces the effective K_FM value since;
K_FM(eff)=ΣK_FM(v_mod)p(v_mod)
where p(·) is the probability associated with the FM signal v_mod. As such, it is better to evaluate the VCO gain at levels near the mean of the FM signal v_mod, with;

α = K FM ⁡ ( ideal ) K F ⁢ M  mean ⁡ ( v mod )

The calibration system counts the number of VCO cycles N (measured by the extended N counter) in a fixed time period (set by R/f_REF) with;

f VCO = N R ⁢ f REF
where R is the number of cycles of the reference signal. A zero-phase restart signal initiates the reference (R) and extended PLL feedback (N) counters at the same time. The accuracy of the frequency measurement depends on the system's ability to read the N counter at precisely time T. Fortunately, simply increasing the fixed time period T reduces any read errors.

The initial calibration operates with the PLL feedback loop disabled. It nominally sets both the offset level v_OFF and scaling parameter α. The offset level remains fairly constant since MOSFET device capacitances C_min and C_max, vary little with operating conditions. However, the value of the scaling parameter α varies with the circuit parameters and even the oscillation frequency of the VCO—as explained below. The VCO oscillates at the resonant frequency of the LC tank given by;

f osc = 1 2 ⁢ ⁢ π ⁢ ( L 1 + L 2 ⁢ ) ⁢ ( C T + Δ ⁢ ⁢ C ) = f c + Δ ⁢ ⁢ f
where C_T is the total tank capacitance less the variable capacitance ΔC. The frequency step Δf due to a change in the MOSFET capacitance AC is approximately equal to;

Δ ⁢ ⁢ f = f C ⁡ [ 1 - 1 2 ⁢ Δ ⁢ ⁢ C C T + 3 8 ⁢ ( Δ ⁢ ⁢ C C T ) 2 ]
for small values of ΔC. This simplifies to;

Δ ⁢ ⁢ f = f C ⁡ ( - 1 2 ⁢ Δ ⁢ ⁢ c C T )
which can then be rewritten as;
Δf=2π²Lf_C³ΔC
showing that Δf changes as the third power of f_C. Consequently, setting the parameter α and the VCO gain K_FM accurately is a challenging task.

It is possible to track any VCO gain errors due to improper a values using the correlator network shown in FIG. 13. It computes the projection of the correction signal v_ctrl onto the reference signal V_FM given by;

v corr = ∫ T ⁢ v ctrl × v FM
as seen at the output of the integrator. The integrator helps reduce the effects of noise and includes a periodic reset to prevent the build-up of errors. An analog window comparator determines the sense of the VCO gain error (and ultimately scaling parameter a).

The system shown in FIG. 14a uses the correlator network to track the frequency error signal v_ctrl and properly adjust the VCO gain in the direct phase/frequency modulator. The PLL feedback drives the VCO to produce an output where;
f_VCO=Nf_REF+FM_REF and f_VCO=K_VCOv_ctrl+αK_FMv_FM
Setting these two expressions equal and realizing that Nf_REF corresponds to the radio channel yields;

v ctrl = 1 K VCO ⁢ ( FMf REF - α ⁢ ⁢ K FM ⁢ v mod ) + v DC
where v_DC is the constant part of v_ctrl. This expression shows that the error term in parentheses moves in same direction as the FM data when the equivalent VCO gain αK_FM is set low, Similarly, the error term moves in the opposite direction of the FM data when the equivalent VCO gain αK_FM is set high.

The correlator responds like a simple analog multiplier. As such, it is desirable to remove or at least greatly reduce the constant term v_DC from the correction signal V_ctrl. Otherwise, the ref signal v_FM leaks to the correlator's output and degrades the result. This can be accomplished using a dc blocking capacitor, a high-pass filter, or auto-zero'ed correlator. The FM tracking loop comprises logic that generates a first control signal (V_off) and a second control signal (α). In an embodiment, the logic comprises counters, registers, or other logic suitable to increment or decrement based on the up/down signal received from the window comparator. The logic generates the control signal (V_off) to adjust the offset of a signal applied to the VCO. The logic generates the control signal (α) to scale the signal applied to the VCO.

FIG. 14b illustrates the signals produced when the equivalent VCO gain αK_FM is low. Identical sinusoidal signals are applied to the VCO modulation port (v_mod) and the ΔΣ modulator input (FM). Ideally, the feedback counter tracks and thereby removes the direct frequency modulation. However, the low equivalent VCO gain αK_FM produces a phase error Δφ that's detected by the phase/frequency detector (P/FD). This in turn directs the charge pump (CP) to produce current pulses that feed the integration filter and produce the correction signal v_ctrl. As expected, the low equivalent VCO gain αK_FM produces a correction signal v_ctrl that follows the applied v_FM signal.

The FM signal v_FM and correction signal v_ctrl for a few different values of scaling parameter a are shown in FIG. 14c. An intentional mismatch in the peak VCO gain K_pk and the droop in the K_FM curve pushes the correct value for a slightly above 1. 10. FIG. 14d shows that the system follows a linear response and therefore always converges to the correct α value.

It is also possible to refine the offset level V_OFF during the calibration of α. This is accomplished by separately observing the positive and negative peak FM deviations. If these differ, then the offset level needs to shift.

A variation of the K_FM tracking system is shown in FIG. 15. It uses an A/D converter to measure the output of the correlator. This provides a direct measure of the VCO gain error and potentially a faster, more accurate K_FM tracking loop. The FM tracking loop comprises logic that generates a first control signal (V_off) and a second control signal (a). In an embodiment, the logic comprises counters, registers, or other logic suitable to increment or decrement based on the output signal received from the A/D. The logic generates the control signal (V_off) to adjust the offset of a signal applied to the VCO. The logic generates the control signal (α) to scale the signal applied to the VCO.

As mentioned earlier, the scaling parameter α depends on the VCO's circuit elements and its operating frequency (f_c) But it also depends on the FM data applied to the two-point phase/frequency modulator. That's because the FM data maps to the VCO, which displays a curved K_FM response that droops at larger FM signals v_FM. As a result, the effective K_FM depends on the distribution of the FM data and the K_FM curvature. This is approximately equal to;

〈 K FM 〉 = 1 N ⁢ ∑ N ⁢ K FM ⁡ ( v MOD ⁢ )
where <K_FM> represents the ensemble average of the function K_FM, which varies with the adjusted FM signal v_MOD applied to the VCO. Fortunately, the K_FM tracking system shown in FIG. 14a properly adjusts the scaling parameter α in real time—that is, as the FM data drives the VCO.

The K_FM tracking system also compensates for changes in the VCO curve, in particular, the K_FM droop. This is important because the droop and the peak VCO gain K_pk both increase as the VCO amplitude decreases. In addition, circuit changes may also affect the droop in the K_FM curve. Fortunately, the tracking system resolves changes in the droop as well as other parameters and adjusts the parameter α accordingly as illustrated by FIG. 16.

FIG. 17 shows the operation of the tracking system with WCDMA RCC1 FM data applied. The effects of low and high a values are shown in FIGS. 17a and 17b respectively. As with the sinusoidal FM signal, the system displays a linear response in FIG. 17c which ensures convergence to the optimum a value.

FIGS. 18a-c illustrate the operation of the tracking system with WCDMA RCC3 high-speed FM data applied. As with the previous example, the behavior is predicted and the convergence is guaranteed. Notice in FIG. 18c that the higher-amplitude RCC3 FM data shifts the scaling factor as compared to the RCC1 FM data.

In one or more embodiments, the described closed-loop system advantageously tracks real-time errors in the two-point frequency modulator. The system provides methods and apparatus to compensate for various changes in the VCO amplitude, temperature shifts, and the distribution of the FM signal as it affects the VCO gain K_FM. As such, the system effectively addresses issues associated with two-point phase/frequency modulation systems. It should be noted that embodiments of the system are suitable for implementation using analog and digital technology including implementations in discrete hardware, gate arrays, programmable logic or other hardware implementations. The system is also suitable for implementations in software, for example, a computer program comprising program instructions that are executed by one or more processors.

The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that the specific details are not required in order to practice the invention. In other instances, well-known circuits and devices are shown in block diagram form in order to avoid unnecessary distraction from the underlying invention. Thus, the foregoing descriptions of specific embodiments of the present invention are presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed, obviously many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the following Claims and their equivalents define the scope of the invention.