Automatic gain control circuit for a modem receiver

[0002] This invention relates to the field of automatic gain control (AGC) circuits, and more particularly, for AGC circuits of modem receivers.

[0003] In communication systems a modem is used to convert (modulate) digital signals generated by a computer into analog signals suitable for transmission over telephone lines. Another modem, located at the receiving end of the transmission, converts (demodulates) the analog signals back into digital form. The transmission speed of digital subscriber loop (DSL) modems has exhibited a remarkable increase in recent years, and as the increase of the transmission speed progresses, it becomes increasingly necessary to reduce the occurrence of errors in data communication arising from disturbances of the circuit to as few as possible.

[0004] In addition, due to the large range of different twisted pair loops over which high speed modems operate and the varying amount of interference, the received signal at the analog-to-digital converter input (in the demodulation section of a modem) can present a dynamic range of over 70 dB.

[0005] An automatic gain control circuit is provided in the demodulation section of a modem to monitor the input signal level and to provide the appropriate gain to bring the signal to a desired level.

[0006] Traditional AGC circuits in modem receivers working over the necessary full dynamic range present significant differences in performance, convergence times, and gain variability at different regions of operation.

[0007] Consequently, there is a need for a AGC circuit in an modem receiver that provides gain tracking for a large range of twisted pair loops with over 70 dB range (for example), various signal bandwidths, and in the presence of radio frequency (RF) and other asynchronous digital subscriber line (ADSL) interference. Further, the AGC circuit must improve gain ripple after convergence to minimize its noise contribution to the system.

[0008] An object of the present invention is to provide an automatic gain control circuit for a modem receiver that improves gain tracking for a large range of twisted pair loops.

[0009] Another object of the present invention is to provide an automatic gain control circuit for a modem receiver that improves gain ripple after modem convergence to minimize noise contribution.

[0010] In accordance with an aspect of the present invention there is provided an automatic gain control circuit for a modem receiver, said receiver having an analog automatic gain control (AGC) amplifier to attenuate an input signal X prior to being digitized to a signal Y by an analog-to-digital converter, said automatic gain control circuit comprising: a programmable loop gain for scaling the signal Y to a first prescribed level during a first mode of operation and to a second prescribed level during a second mode of operation; and means for converting the scaled signal Y into an analog gain control signal for input to the analog AGC.

[0011] In accordance with another aspect of the present invention there is provided an automatic gain control circuit for a modem receiver, said receiver having an analog automatic gain control (AGC) amplifier to attenuate an input signal X prior to being digitized to a signal Y by an analog-to-digital converter, said automatic gain control circuit comprising: (a) means for obtaining an AGC signal from the signal Y; (b) means for subtracting the AGC signal from a prescribed reference signal to form a delta signal; (c) scaling means for reducing the delta signal by a prescribed value; (d) an integrator for integrating the scaled delta signal to form an AGC level control signal; and (e) conversion means for converting the AGC level control signal to an analog AGC control signal for the analog AGC amplifier to attenuate the input signal X.

[0012] In accordance with another aspect of the present invention there is provided an a method of attenuating an input signal X in an automatic gain control circuit of a modem receiver, said receiver having an analog automatic gain control (AGC) amplifier to attenuate the input signal X prior to being digitized to a signal Y by an analog-to-digital converter, the method comprising the steps of: (a) scaling the signal Y to a first prescribed level during a first mode of operation and to a second prescribed level during a second mode of operation; and (b) converting the scaled signal Y into an analog gain control signal for input to the analog AGC.

[0013] The present invention will be described in conjunction with the drawings in which:

[0014] FIG. 1 illustrates a block diagram of a section of a modem receiver;

[0015] FIG. 2 illustrates a block diagram of a first AGC control signal circuit shown in FIG. 1 according to an embodiment of the present invention;

[0016] FIG. 3 illustrates a block diagram of a second AGC control signal circuit shown in FIG. 1 according to another embodiment of the present invention; and

[0017] FIG. 4 illustrates a block diagram of a third AGC control signal circuit shown in FIG. 1 according to another embodiment of the present invention.

[0018] A portion of a receiver 10 of a modem is illustrated in FIG. 1. An input signal X is processed through analog filters 12 to condition the input signal X. The analog filters 12 reject and attenuate unwanted signals, such as other ADSL signals, feed-through signals from a transmitter through hybrid four-to-two wire interfaces, RF interference (e.g. from AM broadcasts), out-of-band noises, and the like.

[0019] The filtered input signal X is then processed through an analog automatic gain control (AGC) amplifier 14 to control the gain of the input signal X. After processing through the analog AGC 14 the signal is digitized by an analog-to-digital converter 16 to produce a digital signal Y.

[0020] The digital signal Y is processed by an AGC control signal circuit 24 to produce an analog control signal G, that is provided to the analog AGC 14 to dictate the amount of attenuation/gain required to bring the level of the input signal X to a desired level. Typically, it is desirable for the input signal X to be brought to near the full scale of the ADC 16, so that a majority of the ADC dynamic range can be exploited.

[0021] Due to the range of different telecommunications environments in which modems operate and the varying amounts of possible interference, the strength of the input signal X can vary considerably. To regulate the signal strength at the input to the ADC 16 the present invention provides the AGC control signal circuit 24 to provide the control signal G_cto the analog AGC 14 to modulate the gain of the input signal X. Three different implementations 24a, 24b, and 24cof the AGC control signal circuit 24 are described in detail below in conjunction with FIGS. 2 to 4.

[0022] An AGC control signal circuit 24aaccording to a first embodiment of the present invention is illustrated in FIG. 2. An absolute value filter 30 receives as input the digital signal Y to produce an output signal Vabs, which in the present embodiment is also a Vagc signal.

[0023] The Vagc signal is subtracted from a reference/threshold Vref signal to obtain a delta signal (delta=Vref−Vagc). The delta signal is processed by a programmable loop gain (PLG) 32 characterised by the following scaling factor equation:

[0024] 2^−k

[0025] where k₁represents a gain factor.

[0026] The loop gain 32 scales down the delta signal by a prescribed value to produce a signal x for use by an integrator (INT) 34. For example, for a 16QAM system the gain k₁of scaling factor defined by Eq. 1 varies over a range of k₁=15 to 18. The values of 2^−kare approximately inversely proportional of time constants of the AGC loop circuit (characterised by analog AGC 14, ADC 16 and control circuit 24). Therefore, for large k₁the AGC loop circuit is slow and is less noisy in steady state.

[0027] The scaled delta signal is used as a stimulus for the integrator 34. The INT 34 takes the form of an up/down counter with variable step size and is characterised by the following equation:

[0028] a(n)=b(n)+a(n−1)

[0029] where a represents input to the integrator, b represents output from the integrator (AGC level control signal), and n is a sample number (i.e. time).

[0030] The programmable loop gain 32 and the integrator 34 incorporate two main operating modes that can be programmably modified by the values of the controlling parameter k₁. For example, during initialization/start-up operation the PLG 32 has high gain 2^−k(i.e. small k₁), and during steady state operation the gain 2^−kis reduced (i.e. increase k₁) to minimize steady state noise introduced to the demodulator 10. In summary, the programmable loop gain 32 can be configured to provide fast acquisition time (with large gain ripple), and then after convergence, switch to a slow mode with small steady state gain ripple. The PLG 32 also supports a wide dynamic range.

[0031] The output of the integrator 34 is an agc level control signal that is converted into a single bit control signal by a sigma-delta modulator 36 known in the art. The 1-bit sigma-delta digital signal is output from the modulator 36 at a multiple of the ADC sampling rate and passed through an analog reconstruction filter 38 to produce the control signal G_c. The control signal G_cis used by the analog AGC 12 to control the gain of the input signal X as described above.

[0032] An AGC control signal circuit 24baccording to a second embodiment of the present invention is illustrated in FIG. 3. In circuit 24bthe Vabs signal is obtained taking the absolute value of the digital signal Y. The Vabs signal is processed by a leaky peak detector 50 to obtain a Vlpk signal which is then multiplexed with the Vabs signal by a multiplexer (MUX) 52 to obtain the Vagc signal. Therefore, the output from the MUX 52 (the Vagc signal) is either the Vabs signal, if the leaky peak detector 50 is bypassed, or the Vlpk signal. The Vagc signal that is then processed as discussed in conjunction with circuit 24aof FIG. 2 to produce the control signal G_cfor the analog AGC 14.

[0033] The peak detector 50 is used to detect a peak Px of the Vabs signal. The peak detector 50 incorporates a leak to ensure that a selected peak value tracks the level of the Vabs signal so that an originally estimated peak does not remain fixed at the same level. The detector 50 also reduces rapid fluctuations in the Vabs signal. The leaky peak calculation is characterised by the following equation:

[0034] Vlpk=Px−2^−kPx,

[0035] where Px is a selected peak and 2^−kis a leakage factor and is programmable over a range of k=4 to k=8.

[0036] The range of k allows control over leak speed (i.e. how quickly the detector 50 “forgets” the previous peak). The term “leak” refers to tracking the true peak of the Vabs signal, then allowing the peak to slowly reduce in time provided the following samples are smaller than the previous peak, otherwise a new peak will be declared and leaked.

[0037] An AGC control signal circuit 24caccording to a third embodiment of the present invention is illustrated in FIG. 4. The circuit 24cintroduces a low pass filter 54 and a multiplexer 56 to the circuit 24bof FIG. 3. The low pass filter 54 is characterised by the following equation: $\frac{1}{1-\left(1-2^{\mathrm{k2}}\right)Z^{-1}}$

[0038] where k₂represents the pole of the filter and Z⁻¹represents a phase shift

[0039] In the circuit 24cthe Vabs signal is processed by the leaky peak detector 50 to obtain the Vlpk signal which is then multiplexed with the Vabs signal by the multiplexer 52 to obtain a Vlpkm signal (i.e. Vlpkm is either Vabs or Vlpk).

[0040] The peak detector 50 is used to detect a peak x of the Vabs signal. The peak detector 50 incorporates a leak to ensure that the selected peak values track the agc level control signal, as discussed in conjunction with circuit 24b .

[0041] The Vlpkm signal is integrated by the low pass filter 54 to produce a Vlpf output signal that is multiplexed with the Vlpkm signal by the multiplexer 56 to produce the Vagc signal (i.e. Vagc is either (a) Vabs if detector 50 and LPF 54 are both bypassed, (b) Vlpk if Vabs is processed through detector 50, and LPF 54 is bypassed, or (c) Vlpf is Vabs processed through both the detector 50 and the LPF 54). The Vagc signal is then processed, as discussed in conjunction with circuit 24aof FIG. 2, to produce the control signal CS for the analog AGC 14.

[0042] In summary, the agc control signal circuits 24a-cprovide the following features:

[0043] (a) multiple settings for the gain (by changing k₁) of the programmable loop gain 32 are supported that change for initialization/start-up and converged states to allow for faster initial acquisition/convergence with higher gains and low steady-state ripple with small gains.

[0044] (b) the addition of a leaky peak detector 50 (circuit 24b) provides the possibility of adjusting the signal level with respect to its peak rather than to its absolute average value.

[0045] (c) the addition of a low pass filter 54 with the leaky peak detector 50 (circuit 24c) provides an improved estimate of the signal level with respect to a threshold.