Phase-locked loop circuit capable of adjusting the frequency range of spread spectrum clock

[0002] 1. Field of the Invention

[0003] The present invention relates to a phase-locked loop circuit, and in particular, to a phase-locked loop circuit capable of adjusting the frequency range of a spread spectrum clock.

[0004] 2. Description of the Prior Art

[0005] All electronic equipment inherently generate electromagnetic interference (EMI) of one sort or another. EMI can cause physical harm to humans, and can have a negative impact on the other components of the computer. For example, EMI can cause the other components to experience instantaneous high voltage. The clock generator on the motherboard of a computer is always the main source of EMI in the computer. As a result, in order to suppress the EMI, a phase-locked loop with a spread spectrum circuit is often used to suppress the EMI.

[0006] FIG. 1A is a primary block diagram showing a conventional phase-locked loop (PLL) circuit. The PLL circuit 100 includes a N-divider 102, a M-divider 104, a phase frequency detector 106, a charge pump 108, a loop filter 110, and a voltage-controlled oscillator 112.

[0007] The N-divider 102 receives the reference signal with frequency F_ref, divides the frequency F_refby N, and outputs the signal F_ref/N. The M-divider 104 receives the feedback signal F_vcowith the frequency F_vcooutputted from the voltage-controlled oscillator 112, divides the frequency F_vcoby M, and outputs the signal F_vco/M. F_ref/N is the external clock signal, while F_vco/M is the internal clock signal. The phase frequency detector 106 detects both the frequency difference and phase difference between the signals F_ref/N and F_vco/M, and outputs control signals “UP” and “DN” (i.e., down). When the phase of F_vco/M lags behind that of the F_ref/N, the phase frequency detector 106 will enable the control signal “UP”. Inversely, when the phase of signal F_ref/N lags behind that of the signal F_vco/M, the phase frequency detector 106 will enable the control signal “DN”. The charge pump 108 and the loop filter 110 are used for transforming the control signals “UP” and “DN” outputted from the phase frequency detector 106 into the control voltage V_c. When the control signal “UP” of the phase frequency detector 106 is enabled, the charge pump 108 generates a current I_chthat flows into the loop filter 110 so as to increase the voltage V_c. Inversely, when the control signal “DN” of the phase frequency detector 106 is enabled, current is drained from the charge pump 108, so that the voltage V_coutputted from the loop filter 110 will decrease. The voltage-controlled oscillator 112 generates the feedback signal F_vcowhich has a frequency F_vcothat varies according to the output voltage V_cfrom the loop filter 110. When the gain of the voltage-controlled oscillator is greater than 0 and V_cincreases, the frequency F_vcowill increase. On the other hand, when the gain of the voltage-controlled oscillator is smaller than 0 and V_cincreases, the frequency F_vcowill decrease. This process continues until, ideally, F_ref/N is equal to F_vco/M. In the following description, the gain of the voltage-controlled oscillator is supposed to be greater than 0. Therefore, when the phase-locked loop circuit is stable, the relationship between F_vcoand F_refthereof is represented by the following equation:

[0008] F_vco/M=F_ref/N

[0009] which translates to:

[0010] F_vco=F_ref*M/N  (1)

[0011] From equation (1), it is noted that only the values of the divisors, M and N, need to be controlled in order to control the value of F_vco. In the above-mentioned application, however, when the PLL is stable, F_vcois a constant frequency. When the constant frequency is transmitted on the trace in the motherboard, interference sources of this frequency and its harmonic component will be radiated. Therefore, the EMI specification (i.e., that the density of f_cand its harmonics be within a certain range) cannot be met. As shown in FIGS. 1B(i) and 1B(ii), the amplitude at the frequency F_cexceeds the EMI specification. The interference level can be decreased if the PLL structure can be modified to uniformly spread out the energy of the signal F_vcowithin a predetermined range.

[0012] FIG. 2A shows an ideal spread spectrum on time domain and frequency domain. In this graph, the frequency of the signal F_vcoperiodically rises and falls, with the frequency F_cas the center between the frequency ranges of f₁and f₂. FIG. 2B shows the relationship between the amplitude and frequency at the signal F_vcoof FIG. 2A, and the amplitude falls within the EMI specification. Usually, the oscillating frequency of the voltage-controlled oscillator may be expressed by the following equation:

[0013] F_vco=F₀+K_vco(V_c−V₀)  (2)

[0014] wherein F_vcois the oscillating frequency of the voltage-controlled oscillator, F₀is the free running frequency when V_c=V₀, K_vcois the slope of the voltage-controlled oscillator, which unit is Hz/V, V_cis the control voltage inputted to the voltage-controlled oscillator, and V_ois the threshold voltage of the voltage-controlled oscillator. The frequency of the voltage-controlled oscillator can be controlled only when V_cis greater than V_o. Therefore, if the PLL can control the V_cto vary linearly and periodically, the frequency F_vcowill vary linearly and periodically in correspondence with V_cto thereby achieve the effects of spectrum spreading.

[0015] To achieve the above effect, some conventional applications have added another loop to control the V_cin the PLL. In this case, the frequency F_vcocan vary linearly and periodically to achieve the effect of spectrum spreading. This method, however, has some disadvantages: (1) the circuit of the PLL becomes more complicated, (2) the entire structure tends to be unstable, and (3) the range of the spread spectrum is not easily controlled.

[0016] U.S. Pat. No. 5,610,955 discloses another attempt at achieving spread spectrum, which uses a method for periodically changing the M/N value in order to achieve the purpose of spectrum spreading. However, the disadvantage of this method is that the spread spectrum is not uniform. In other words, when the frequency reaches the upper and lower bounds, the frequency is constant within a period of time. This effect is illustrated in FIGS. 3A and 3B, where the outputted clocks F_vcomay jut at frequencies f₁and f₂(see FIG. 3B) and exceed the EMI specification.

[0017] In view of the above-mentioned problems, it is therefore an object of the present invention to provide a PLL circuit that is capable of adjusting the range of spread spectrum, eliminating the jutting phenomenon on the frequency spectrum, and spreading the energy over a wider range.

[0018] To accomplish the objectives of the present invention, there is provided a circuit that includes a first voltage-controlled oscillator (VCO) that generates a feedback clock. The circuit further includes a phase frequency detector, a charge pump and a loop filter that together receive a reference clock and the feedback clock, and in response thereto, generates a first voltage signal. The circuit further includes a timing generator that generates a control signal having varying periods, and a voltage modulator that receives the first voltage signal and the control signal, and in response thereto generates a second voltage signal and a third voltage signal. The second voltage signal is provided to the first VCO. The circuit also includes a second VCO that receives the third voltage signal and which outputs a spread spectrum clock.

[0019] Since the present invention utilizes an FM/AM timing generator to control the voltage modulator, the PLL of the present invention realizes several advantages:

[0020] (1) since only one loop is used, the PLL circuit is simple and easy to maintain stable;

[0021] (2) the frequency range of the spread spectrum clock can be quantified, and the frequency of the spread spectrum clock can be controlled in a programmable way;

[0022] (3) since the time of frequency variation can be quantified and is programmable, the energy can be spread over a much wider range; and

[0023] (4) since the frequency of the spread spectrum clock varies periodically, there will be no constant frequency, no surge will be generated, and the spread spectrum clock will be uniform.

[0024] These and other objects and advantages of the present invention will become apparent by reference to the following description and accompanying drawings wherein:

[0025] FIG. 1A is a primary block diagram showing a conventional phase-locked loop circuit.

[0026] FIG. 1B(i) shows the waveform of F_vcoon the time-domain.

[0027] FIG. 1B(ii) shows the relationship between the amplitude and the frequency of the waveform of F_vcoin FIG. 1B(i).

[0028] FIGS. 2A and 2B are schematic illustrations showing an idealized spread spectrum, wherein FIG. 2A shows the relationship between frequency and time, while FIG. 2B shows the relationship between frequency and amplitude.

[0029] FIGS. 3A and 3B are schematic illustrations showing the spread spectrum when changing the frequency-divided value of a frequency-divided counter of the PLL, wherein FIG. 3A shows the relationship between frequency and time, while FIG. 3B shows the relationship between frequency and amplitude.

[0030] FIG. 4 is a block diagram showing a phase-locked loop circuit in accordance with one embodiment of the present invention.

[0031] FIG. 5A is a circuit diagram of the voltage modulator in FIG. 4.

[0032] FIGS. 5B and 5C are circuit diagrams of the switches in the voltage modulator of FIG. 4.

[0033] FIG. 6A shows the waveform of F_vco2on the time-domain.

[0034] FIG. 6B shows the waveform of F_vco2on the frequency-domain.

[0035] FIG. 7A shows the waveform of the frequency spectrum of X₀(t)=A_ccos(2π*f_ct+Q_c)

[0036] FIG. 7B shows the waveform of the frequency spectrum of X_c(t)=A_ccos (2π*f_ct+Q_c) after narrowband modulation.

[0037] FIG. 7C shows the waveform of the frequency spectrum of X_c(t)=A_ccos (2π*f_ct+Q_c) after broadband modulation.

[0038] FIG. 7D shows the waveform of FM_CK.

[0039] FIG. 7E shows the waveform of the frequency spectrum after FM modulation.

[0040] FIG. 7F shows the waveform of the frequency spectrum after FM/AM modulation.

[0041] FIG. 8A is a circuit diagram of the FM controller of the FM/AM timing generator in FIG. 4.

[0042] FIG. 8B is a circuit diagram of the AM controller of the FM/AM timing generator in FIG. 4.

[0043] The following detailed description is of the best presently contemplated modes of carrying out the invention. This description is not to be taken in a limiting sense, but is made merely for the purpose of illustrating general principles of embodiments of the invention. The scope of the invention is best defined by the appended claims. In certain instances, detailed descriptions of well-known devices and mechanisms are omitted so as to not obscure the description of the present invention with unnecessary detail.

[0044] FIG. 4 is a circuit block diagram showing a PLL circuit capable of adjusting the frequency range of spread spectrum clock in accordance with the preferred embodiment of the invention. As shown in FIG. 4, the PLL circuit 10 of the invention includes a N-divider 12, a M-divider 14, a phase frequency detector 16, a charge pump 18, a loop filter 20, a voltage modulator 30, a FM/AM timing generator 40, and voltage-controlled oscillators (VCO) 22 and 24. The basic idea is to use the voltage modulator 30 to control or manipulate the control voltages V_c1and V_c2, as described in greater detail below. Since the capabilities and functions of the N-divider 12, M-divider 14, phase frequency detector 16, charge pump 18, loop filter 20, and voltage-controlled oscillators 22 and 24 are the same as those elements 102, 104, 106, 108 and 110 in FIG. 1, detailed description thereof will be omitted. The following detailed description will focus on the voltage modulator 30 and FM/AM timing generator 40.

[0045] The voltage modulator 30 in the PLL circuit 10 generates a first control voltage V_c1, and a second control voltage V_c2that varies periodically. The first VCO 22 receives the first control voltage V_c1and generates a stable feedback clock F_vco. The second VCO 24 receives the second control voltage V_c2and generates a spread spectrum clock F_vco2. The structure of the first VCO 22 and the second VCO 24 can be the same. A timing control signal is generated from the FM/AM timing generator 40 for controlling the voltage modulator 30 to generate the second control voltage V_c2that varies periodically.

[0046] The Voltage Modulator 30

[0047] FIG. 5A is a circuit diagram showing the voltage modulator 30. As shown in FIG. 5A, the voltage modulator 30 includes an operational (OP) amplifier 31, a reference current trace 32, a first current trace 33, a second current trace 34, and 2N sets of third current traces 35. The reference current trace 32 consists of a transistor mp0, a switch sw0, a transistor mn0, and a resistor R₀connected in series. The first current trace 33 consists of a transistor mp1, a switch sw1, a transistor mn1, and a resistor R₁connected in series. The second current trace 34 consists of a transistor mp2, a switch sw2, a transistor mn2, and a resistor R₂connected in series. In addition, each of the third current traces 35 consists of a transistor mp(i), a switch sw(i), and a transistor mn(i) connected in series, wherein i=1 to 2N. The ratios of the width over the length (W/L) of the transistor dimension and the ratio of the width over the length (W/L) of the switches of the reference current trace 32, first current trace 33, second current trace 34, and third current traces 35 are:

[0048] mp0:mp1:mp2:mp(i)=1:1:(1−Nδ): δ  (1)

[0049] mn0:mn1:mn2:mn(i)=1:1:(1−Nδ): δ  (2)

[0050] sw0:sw1:sw2:sw(i)=1:1:(1−Nδ): δ  (3)

[0051] wherein δ is a very small value, equivalent to the smallest adjustable unit (e.g., 0.01 to 0.001), and N is an adjustable variable. Furthermore, the values of the resistors R₀, R₁, and R₂, have to be the same. The voltage across the resistor R₁is defined as the first control voltage V_c1while the voltage across the resistor R₂is defined as the second control voltage V_c2.

[0052] As shown in FIG. 5A, the output voltage V_cfrom the loop filter 20 is connected to the positive terminal of the OP amplifier 31 of the voltage modulator 30, and the output of OP amplifier 31 is connected to the gate of the transistor mn0, with the negative terminal of the OP amplifier 31 connected to the source of the transistor mn0. Because the OP amplifier 31 is in negative feedback condition, the output voltage V_cshould be the same as V_X. Then, the current of the reference current trace 32 is mirrored to the first current trace 33, the second current trace 34, and the 2N sets of third current traces 35 using the known technique of current mirroring, and the voltages V_c1and V_c2are generated by the resistors R₁and R₂. Thus, the voltages of V_c1, V_Xand V_care the same, and the current flowing through resistor R1 can be defined by the equation I₁=V_c1/R₁. The current flowing through resistor R₂can be defined by the equation I₂=I₁(1−Nδ+Xδ), wherein “X” is the conductive trace number of the 2N sets of third current traces 35, and is between 1 and 2N. Therefore: $\begin{array}{c}{V}_{c2}=R_2{I}_2\\ =R_1{I}_1\left(1-N_{\delta }+X_{\delta }\right)\\ =V_{c1}\left(1-N_{\delta }+X_{\delta }\right)\\ =V_x\left(1-N_{\delta }+X_{\delta }\right)\end{array}$

[0053] Thus, the frequency difference between the clocks F_vco1and F_vco2are defined as F_vco1−F_vco2=K_vcoV_x(N−X) δ, and delta f=K_vcoV_xδ is defined to be the minimum programmable frequency step. Consequently, the frequency of clock F_vco2generated by VCO 24 can be periodically controlled between (F_vco1+X*delta f) and (F_vco1−X*delta f), as long as the “ON” state of each switch sw(i) can be periodically controlled. The clock F_vco2can be modulated by AM modulation or FM modulation or combined AM/FM modulation techniques, as described hereinbelow.

[0054] The switch sw(i) can be embodied in one of two different circuits. FIGS. 5B and 5C each illustrates a different circuit for the switch sw(i) in FIG. 5A. The switches from sw(1) to sw(N) are shown in FIG. 5B, and the switches from sw(N+1) to sw(2N) are shown in FIG. 5C. Operation of the switches sw(1) to sw(2N) will be described in greater detail below.

[0055] The switches sw0, sw1 and sw2 in current traces 32, 33 and 34 are always kept in the “ON” state, for matching with the switches sw(i). Each switch sw(i) is controlled by two control signals FM_CK and EN(i) that are outputted from the AM/FM timing generator 40. If only AM modulation is to be used to control the clock F_vco2, then the signal FM_CK is set as a timing signal with a constant frequency. If only FM modulation is to be used to control the clock F_vco2, then EN(1)-EN(N) is set to a constant value. If both AM and FM modulation are to be used to control the clock F_vco2, then the signal FM_CK is set as a timing signal with varying frequencies.

[0056] The switch circuit in FIG. 5B has a transmission gate 60 having an NMOS gate that is coupled to the output of an inverter 62. A NOR gate 64 has one input coupled to the FM_CK signal, and another input coupled to the output of an inverter 66, whose input is coupled to the signal EN(i). The output of the NOR gate 64 is coupled to the PMOS gate of the transmission gate 60 and the input of inverter 62. According to the circuit in FIG. 5B, when the signal FM_CK is 1, all the switches from sw(1) to sw(N) are turned on. However, when the signal FM_{—CK is}0, only the switches from among sw(1) to sw(N) in which the control signal EN(i) is 0 are turned on.

[0057] The switch circuit in FIG. 5C also has a transmission gate 60 having an NMOS gate that is coupled to the output of an inverter 62. A NAND gate 68 has one input coupled to the FM_CK signal, and another input coupled to the signal EN(i). The output of the NAND gate 68 is coupled to the PMOS gate of the transmission gate 60 and the input of inverter 62. According to the circuit in FIG. 5C, when the signal FM_CK is 0, all the switches from sw(N+1) to sw(2N) are turned off. However, when the signal FM₁₃CK is 1, only the switches from among sw(N+1) to sw(2N) in which the control signal EN(i) is 1 are turned on. In light of the operation of the switch circuits illustrated in FIGS. 5B and 5C, the number of switches sw(1) to sw(2N) that are turned on can be controlled periodically by varying the signals FM_CK and EN(i).

[0058] Control of the Clock F_vco2by AM/FM

[0059] There are various ways for implementing the present invention, such as the use of AM (amplitude modulation) control, FM (frequency modulation) control, and AM/FM control.

[0060] The use of AM modulation to control the clock F_vco2is described as follows. In this case, the signal FM₁₃CK in FIGS. 5B and 5C is set as a timing signal with a constant frequency. The FM₁₃CK signal is also connected as signal AM-CK in FIG. 8B. Therefore, in the case that there are X bits of 1 among the signal EN(i), when the signal FM₁₃CK is 1, it will be known that the frequency F_vco2=(F_vco1+X*delta f). On the other hand, when the signal FM₁₃CK is 0, it will be known that the frequency F_vco2=(F_vco1−X*delta f). This can be obtained by understanding the following. There are X bits of “1” among the signal EN(i), with i=1 to N. Thus, when the signal FM_CK is “0” (i.e., low), switches sw(1) to sw(N) will only have (N−X) switches “on”, and switches sw(N+1) to sw(2N) are all “off”. Similarly, when the signal FM_CK is “1” (i.e., high), switches sw(1) to sw(N) are all “on”, and switches sw(N+1) to sw(2N) will only have X switches “on”. Consequently, as long as X (i.e., the number of bits of the signal EN(i) that have a 1 value) is periodically modulated, the frequency F_vco2is also periodically varied within a predetermined frequency range, thereby eliminating the jutting phenomenon shown in FIG. 3B. FIG. 6A illustrates the waveform of the frequency F_vco2varying periodically in the time domain, and FIG. 6B illustrates the waveform of the frequency F_vco2varying periodically in the frequency domain. FIGS. 6A and 6B only illustrate one non-limiting embodiment. Thus, the effect of a spread spectrum clock can be achieved as long as X is periodically varied.

[0061] The use of FM modulation to control the clock F_vco2is described as follows. In this case, the signal FM₁₃CK in FIGS. 5B and 5C is set as a timing signal with varying frequencies. The signal FM₁₃CK can be generated by using the circuit illustrated in FIG. 8A, where CK_F is a reference clock signal. There are X bits of “1” from among the signals EN(i), with i=1 to N. Assume that there is a sinusoidal signal X_c(t)=A_ccos(27π*f_ct+Q_c), wherein A_cis the amplitude, f₀is the center frequency, and Q_cis the phase angle (assumed to be a constant value). The graph shown in FIG. 7A is the representation of X_c(t) in the frequency spectrum. If the center frequency f_cof X_c(t) is changed by a modulation signal and the modulation frequency for modulating f_cis f_m, then when f_mis much smaller than f_c, this case is called a narrowband FM modulation. FIG. 7B shows the narrowband modulation of X_c(t), where there are some signals generated at f_c, f_c+f_m, and f_c−f_m, and the amplitude of each of the signals is smaller than A_c. Other high-order harmonic components can be omitted. When f_mis not much smaller than f_c, this case is called a broadband FM modulation. FIG. 7C shows the broadband modulation of X_c(t). In FIG. 7C, there are also some signals generated at f_c+nf_m, f_c−nf_m, and the amplitude of each of the signals is also smaller than A_c, wherein n is an integer. The control principle of FM modulation is to make some changes to the period of the signal FM₁₃CK, so as to modulate the staying time of clock F_vco2staying between the frequencies (F_vco1+X*delta f) and (F_vco1−X*delta f). As shown in FIG. 7D, the period of the signal FM₁₃CK varies from the minimum to the maximum, and from the maximum to the minimum. In other words, the period of the signal FM₁₃CK varies periodically. Consequently, as shown in FIG. 7E, the frequency F_vco2is distributed at each of the frequency spectrums of F_vco1+nfm and F_vco1−nf_m. However, the amplitude of each signal is smaller than the upper bound of EMI. Since the frequency F_vco2still stays between the frequencies (F_vco1+X*delta f) and (F_vco1−X*delta f) for a period of time, the jutting phenomenon of FIG. 3B still exists at the position of frequencies F_vco1+nf_mand F_vco1−nf_m.

[0062] As a result, AM modulation control can be used if it is desired to effectively eliminate the jutting phenomenon on the frequency spectrum, while the FM modulation control can be used if it is primarily desired to spread the energy over a much wider range. FIG. 7F shows the result obtained by combining AM and FM modulation control, where the jutting phenomenon is eliminated, and the energy can also be spread over a much wider range.

[0063] Given the above description, it can be understood as to why there are a total of 2N switches sw(i). From FIG. 5A, if half of the switches sw(i) are opened or closed, the currents I₂=I₂=I₀because of the existence of the current mirror.

[0064] The FM/AM Timing Generator

[0065] FIGS. 8A and 8B are circuit diagrams illustrating the FM/AM timing generator 40. As shown in FIGS. 8A and 8B, the timing generator 40 includes two parts, one of which is a FM generator 41 (FIG. 8A) for generating a FM timing signal FM₁₃CK, and the other of which is an AM generator 42 (FIG. 8B) for generating an AM timing signal AM_CK. If FM and AM modulation are combined, the signal FM₁₃CK in FIG. 8A will be connected to the signal AM CK in FIG. 8B.

[0066] In the FM generator 41 as shown in FIG. 8A, a frequency divider 411 is provided for generating a frequency-divided signal CK_O from a reference frequency CK_F that has a fixed period. A flip-flop 412 is triggered by the frequency-divided signal CK_O, and a FM timing signal FM₁₃CK whose duty cycle is 50% is then generated from the Q output of the flip-flop 412. The flip-flop 412 essentially operates as a duty-cycle controller to ensure the signal “high” time is equal to the signal “low” time. The FM timing signal FM₁₃CK is a modulated clock, and can be used for controlling the switching (ON/OFF) frequency of each of the switches sw(1) to sw(2N). A counter 413 has an input coupled to the FM timing signal FM₁₃CK, and outputs a divisor (D) to the frequency divider 411 and two comparators 414 and 415. This divisor (D) can be incremented during each iteration, so that the division by, for example 2, becomes division by 3 during the next iteration. Since the counter 413 is triggered by the positive-edge of the FM timing signal FM₁₃CK to change its counting value, and the divisor D of the frequency divider 411 is provided by the counter 413, the FM generator 41 is capable of generating FM timing signals FM₁₃CK having different periods as the output value of the counter 413 changes. In addition, the comparators 414 and 415 are provided for comparing the counting value (i.e., the divisor D) with the counting bounds Hb_FM and Lb_FM, and for determining whether the counting value exceeds the counting bounds or not. The output signals of the comparators 414 and 415 are transformed into UP/DOWN control signals (i.e., to increase or decrease the bounds) for the counter 413 by way of an OR gate 417 and a flip-flop 416. The flip-flop 416 functions as a time delay to prevent glitch, and controls the direction of the counter 413. Thus, the desired FM timing signal FM₁₃CK can be generated by the FM generator 41 as long as the desired UP/DOWN counting bounds Hb_FM and Lb_M can be programmed. The maximum period T_maxand minimum period T_minof FM₁₃CK are defined by the values of Hb_FM and LbFM, and the frequency defined from T_maxto T_minand from T_minto T_maxis f_m.

[0067] In the AM generator 42 shown in FIG. 8B, a decoder 421 is used for decoding the output value of a counter 422 and generating N sets of the signal EN(i). In other words, the decoder 421 determines which range of switches sw(i) to open. When the output value of the counter 422 is j, there should be j bits of 1 in the signal EN(i). As described above, the signals EN(i) are used for controlling the number of switching operations of the switches sw(1) to sw(2N). The counter 422 is triggered by the Q output from the flip-flop 428, and the flip-flop 428 is triggered by a clock AM_CK. The flip-flop 428 also functions as a duty-cycle controller. The frequency of the clock AM_CK is constant if only AM modulation is used (and the circuit of FIG. 8A will not be used during AM modulation only), and can be coupled to FM₁₃CK from the FM generator 41 in FIG. 8A if AM and FM modulation is combined. In addition, comparators 424 and 425 are provided for comparing the counting value (outputted from the counter 422) with the UP/DOWN counting bounds Hb_AM and Lb_AM, and for determining whether the counting value exceeds the counting bounds or not. The output signals of the comparators 424 and 425 are transformed into UP/DOWN control signals for the counter 422 by way of an OR gate 427 and a flip-flop 426. Thus, the desired AM timing signals EN(1) to EN(N) can be generated by the AM generator 42 as long as the desired UP/DOWN counting bounds Hb_AM and Lb_M can be programmed. The maximum frequency F_maxand minimum frequency F_minof F_vco2are defined by the values of Hb_Am and Lb_Am, and the frequency defined from F_maxto F_minand from F_minto F_maxis f_a. Thus, N sets of signal EN(i) generated by the AM generator 42 can be outputted to two sets (i.e., sw(1)-sw(N) and sw(N+1)-sw(2N)) of switches sw(i) of the voltage modulator 30.

[0068] Thus, the so-called AM control is to set the timing signal FM₁₃CK in FIGS. 5B and 5C to be a signal with a constant frequency and to connect FM₁₃CK to the clock signal AM_CK in FIG. 8B, so as to periodically change the number of 1's in EN(i). In essence, AM control will determine the range of switches sw(i) to be controlled (open or close) by FM₁₃CK. The so-called FM control is to use the FM generator 41 to output the timing signal FM₁₃CK with varying periods, and to make one set of signal EN(i) constant. The signal FM₁₃CK in FIGS. 5B and 5C is a period-varying signal, and EN(i) has a fixed number of 1's. In essence, FM control will control or determine the time of the opening or closing of switches sw(i) in a selected range, as well as other parameters. The so-called AM/FM control is to combine the AM control with the FM control, output the timing signal FM₁₃CK with varying periods by the FM generator 41, and output the EN(i) signal with varying periods by the AM generator 42. At the same time, the timing signal FM₁₃CK can serve as the clock AM_CK of the AM generator 42 for AM/FM modulation control. Consequently, the jutting phenomenon of the spread spectrum clock can be eliminated, and the energy of the spread spectrum clock can also be spread over a much wider range.

[0069] The data of Hb_FM, Lb_FM, Hb_AM, and Lb_AM can be stored in the memory on the motherboard, and the desired data can be outputted to the FM/AM timing generator 40 using program software, so that the present invention is capable of controlling spread spectrum in a programmable way.

[0070] While the description above refers to particular embodiments of the present invention, it will be understood that many modifications may be made without departing from the spirit thereof. The accompanying claims are intended to cover such modifications as would fall within the true scope and spirit of the present invention.