Wide bandwidth phase-locked loop circuit

This invention relates to a phase-locked loop circuit having a wide bandwidth and a method of implementing the same.

The invention described herein was made by employee(s) of the United States Government and may be made and used by or for the Government of the United States of America for governmental purposes without the payment of any royalties thereon or therefor.

The quality of a clock signal tends to degrade as the signal travels down a transmission line due to attenuation, interference, noise, and a number of other factors. One method of compensating for this problem at the receiving end is by using a phase-locked loop (“PLL”) to synthesize a signal having the same frequency and wave form as the original clock signal. FIG. 1 illustrates a functional block diagram of a basic PLL. A phase comparator 10 produces a DC or low frequency signal having a voltage proportional to the phase difference between an input signal and an output signal (which also acts as the reference signal for the phase comparator 10 ). The DC or low frequency signal from the phase comparator 10 is filtered by a low frequency loop filter 12 and applied to the input of a voltage-controlled oscillator (“VCO”) 14 . The VCO increases or decreases the frequency of the output signal based on the voltage of the signal from the phase comparator 10 . The output signal is then fed back to the phase comparator 10 for phase comparison with the input signal. If the frequencies of the two signals are different, resulting in a phase difference, the phase comparator 10 adjusts the control signal voltage until the VCO 14 has increased or decreased the output signal frequency sufficiently to eliminate the phase difference. In this way, the output signal can be phase-locked with the input signal.

PLLs generally have narrow bandwidths and, therefore, are limited in the range of frequencies the PLLs can phase lock. A PLL designed for low frequency will not generally phase lock a high frequency signal, and vice versa. This makes the basic PLL unsuitable for use in applications where broad bandwidths are required.

One attempt to address this shortcoming is to provide the PLL with multiple frequency ranges such as in Cypress Semiconductor's programmable skew clock buffer (“PSCB”). This integrated circuit timing generator includes, among other functions, an internal PLL with three user-selectable VCO frequency ranges: 15-30, 25-50, and 40-80 MHz. A user may configure the PSCB VCO to oscillate in any one of the three ranges by setting the appropriate input pins. Furthermore, the output of the PSCB VCO may be divided by up to a factor of four by appropriate selection of internal dividers. Division of the VCO output allows the PSCB PLL to phase lock signals having frequencies as low as 3.75 MHz (15 MHz÷4=3.75 MHz) even though such frequencies are technically outside the lowest PSCB VCO frequency range. Data sheets and application notes for the PSCB may be obtained from Cypress Semiconductor's home page, http://www.cypress.com, and are incorporated herein by reference.

Although the multiple frequency ranges of the PSCB PLL allow it to phase lock either high or low frequencies, the bandwidth of each frequency range is still relatively narrow. In other words, once the PSCB is configured to operate in a particular frequency range, it is effectively dedicated to that frequency range and will be unsuitable for use in applications requiring different frequency ranges or broader bandwidths.

The PLL circuit of the present invention uses a PLL having multiple frequency ranges to phase lock an input signal having a wide range of frequencies. The circuit selects one of the multiple frequency ranges based on the frequency of an input signal and configures the PLL to operate in the selected frequency range. A new frequency range may be subsequently selected as necessary to track changes in the input signal frequency.

In general, in one aspect, the invention relates to a wide bandwidth phase-lock loop circuit. The circuit comprises a frequency detector, a frequency range selector connected to the frequency detector, and a PLL connected to the frequency range selector. The PLL can be configured to operate in a number of frequency ranges, and the frequency range selector configures the PLL to operate in one of these frequency ranges based on information about the input signal frequency obtained by the frequency detector.

In general, in another aspect, the invention relates to a method of phase locking a signal having a wide range of frequencies. The method comprises obtaining information on an input signal frequency, selecting one of a plurality of frequency ranges based on the input signal frequency information, and generating an output signal having a frequency within the selected frequency range.

In general, in another aspect, the invention relates to a phase-lock loop circuit having a wide bandwidth. The circuit comprises means for obtaining information on an input signal frequency, and means for selecting one out of a plurality of frequency ranges based on the input signal frequency information. The circuit also comprises means for generating an output signal having a frequency within the selected frequency range.

Advantages of the invention include a PLL circuit that can phase lock signals having a wide range of frequencies. Other advantages of the invention will become apparent from the following description and the claims.

FIG. 1 is a block diagram of a basic PLL.

FIG. 2 is a block diagram of the invention.

FIG. 3 is a schematic diagram of a zero-crossing synchronizer.

FIG. 4 is a schematic diagram of a zero-crossing synchronizer including a voltage comparator.

FIG. 5 is a flow diagram for a zero-crossing counter.

FIG. 6 is a flow diagram for a range selector.

FIG. 7 is a flow diagram for a range detector.

FIG. 8 is a block diagram of a multiple frequency range PLL.

FIG. 9 is a block diagram of a multiple frequency range PLL including an external divider.

Referring to FIG. 2 , a PLL circuit 20 has a frequency detector 22 , a frequency range selector 24 connected to the frequency detector 22 , and a multiple frequency range PLL 26 connected to the frequency range selector 24 . The frequency detector 22 obtains the frequency of an input signal and provides this information to the frequency range selector 24 . The frequency range selector 24 determines which one of a plurality of frequency ranges the input signal frequency falls within and selects an appropriate frequency range for the multiple frequency range PLL 26 . The multiple frequency range PLL 26 generates an output signal having a frequency within the selected frequency range and corresponding to the input signal frequency. If the input signal frequency changes such that a different PLL frequency range is required, the frequency range selector 24 selects a new frequency range and configures the PLL 26 accordingly.

For purposes of this description, the input signal can have, but is not limited to, any frequency ranging approximately from 2-48 MHz. The frequency detector 22 , frequency range selector 24 , and multiple frequency range PLL 26 will now be discussed in detail.

Frequency Detector

The frequency detector 22 obtains the frequency of the input signal, or more accurately, obtains the number of cycles (which is the inverse of frequency) in the input signal by counting the number of “zero-crossings” in the input signal over a predetermined time interval. Zero-crossing refers to the transition of a signal from a logic HIGH to a logic LOW state, and vice versa. Two consecutive zero-crossings constitute one cycle.

In one embodiment, the frequency detector 22 includes two functional components, a zero-crossing synchronizer 22 a and a zero-crossing counter 22 b . The function of the zero-crossing synchronizer 22 a is to synchronize the input signal to a global clock signal so that the zero-crossing counter 22 b can accurately count the number of cycles in the input signal. The global clock signal may be provided, for example, by a 50 MHz crystal oscillator (not shown) on-board the PLL circuit 20 . Both the zero-crossing synchronizer 22 a and zero-crossing counter 22 b may be implemented using a single Altera EPM7064LC44-7 programmable logic device (“PLD”). Data sheets for the Altera PLD are available from the company's home page, http://www.altera.com, and are incorporated herein by reference.

The zero-crossing synchronizer 22 a may be implemented by programming three D-type flip-flops 30 , 32 , 34 , an AND-gate 36 , and a cycle counter 38 into the PLD, all of which are graphically depicted in FIG. 3 . The flip-flop 30 receives the input signal and is configured to divide the frequency of the input signal by two. As should be well known to those having ordinary skill in the art, a 2-48 MHz input signal needs to be synchronized with a clock signal having at least twice the maximum frequency of the input signal, or 96 MHz to prevent aliasing of the input signal. Because the Altera PLD has a maximum operating frequency of only 83.3 MHz, it is necessary to reduce the maximum frequency of the input signal. Therefore, the flip-flop 30 has been configured to reduce the maximum input signal frequency by half (to 24 MHz), and the 50 MHz clock signal is used as the synchronizing clock signal. The flip-flop 32 receives the output of the flip-flop 30 and synchronizes the output of the flip-flop 30 to the 50 MHz clock signal which also clocks the flip-flop 32 . Because the flip-flop 32 is clocked at more than twice the maximum output frequency of the flip-flop 30 , aliasing should not occur, and all outputs from the flip-flop 30 should be latched correctly by the flip-flop 32 .

The output of the flip-flop 32 is applied to the input of the flip-flop 34 and also to one input of the AND-gate 36 . The output of the flip-flop 34 is then applied to the other, inverted input of the AND-gate 36 . This arrangement prevents any one cycle of the output of the flip-flop 30 from being counted more than once because the AND-gate 36 turns on only when both the output of the flip-flop 32 is HIGH and the output of the flip-flop 34 is LOW. However, if a HIGH appears at the input of the flip-flop 32 for two or more consecutive clock cycles, the output of the flip-flop 34 will go HIGH, thus turning off the AND-gate 36 and stopping the count.

The cycle counter 38 is also clocked by the 50 MHz clock signal and is incremented only when the AND-gate 36 is turned on. Again, this ensures each cycle of the output of the flip-flop 30 is counted only once.

The zero-crossing counter 22 b is implemented by programming the PLD to include a main counter 40 which is clocked by the 50 MHz clock signal. The main counter 40 is used to time the interval during which the cycle counter 38 counts the cycles of the input signal.

In one embodiment, referring to FIG. 4 , the input signal may be applied to the input of a voltage comparator 42 instead of directly to the input of the flip-flop 30 to condition and shape the wave form of the input signal. The comparator 42 outputs a LOW when the voltage of the input signal drops below a reference voltage of the comparator 42 and a HIGH otherwise. The output of the comparator 42 is then applied to the input of the flip-flop 30 , thus providing clear, well-shaped square waves to the flip-flop 30 . The reference voltage of the comparator 42 is preferably set at halfway between the minimum and maximum voltage of the input signal to optimize the accuracy of the comparator 42 . If necessary, a voltage divider (not shown) may be used to reduce large swings in the input signal voltage.

Operation of the frequency detector 22 is shown in FIG. 5 (with reference to FIG. 3 ). Upon receiving a pre-selected signal from the frequency range selector 24 (ST 50 ) via one of the PLD input pins, indicating that the frequency range selector 24 is ready for a new count, the frequency detector 22 clears both the cycle counter 38 and the main counter 40 (ST 51 ). The frequency detector 22 then starts both counters 38 , 40 (ST 52 , ST 53 ) and, in some embodiments, sends a signal to the frequency range selector 24 via one of the PLD output pins to indicate a count is in progress. In some embodiments, the main counter 40 counts from 0 to 2047 by one, which takes 40.94 μs to complete using the 50 MHz clock. During this time, the cycle counter 38 counts the cycles of the input signal which are propagating through the flip-flops 30 , 32 , 34 . When the main counter 40 reaches 2047 (ST 54 ), the frequency detector 22 stops both counters 38 , 40 (ST 55 ) and, in some embodiments, sends a signal to the frequency detector 24 to indicate the count is finished. The frequency detector 22 may then provide the number of cycles counted, or the “count,” to the frequency detector 24 for frequency range determination, as will be described further herein.

Frequency Range Selector

Referring again to FIG. 3 , after receiving the count from the frequency detector 22 , the frequency range selector 24 determines an appropriate frequency range based on the count and configures the multiple frequency range PLL 26 to operate in that frequency range accordingly. In some embodiments, the frequency range selector 24 includes two functional components, a range detector 24 a which determines an appropriate frequency range for the count received, and a range selector 24 b which confirms the frequency range and configures the multiple frequency range PLL 26 to operate in that frequency range. Both functional components may be implemented also using an Altera PLD.

In one embodiment, the PLD for the frequency range selector 24 uses a 12.5 MHz instead of the 50 MHz clock signal. The reason for this is, generally, the larger the amount of logic implemented by the PLD, the slower the PLD must be clocked. Therefore, because the frequency range selector 24 requires more logic to implement relative to the frequency detector 22 , the PLD for the frequency range selector 24 must be clocked at a slower frequency. The 12.5 MHz clock signal may be a separate clock signal or it may be the same 50 MHz clock signal divided by four. Division by four of the 50 MHz clock signal may be implemented, for example, by configuring two T-type flip-flops (not shown) to divide by two each.

The function of the range selector 24 b as implemented by the PLD is shown in FIG. 6 (with further reference to FIG. 3 ). Initially, the value of the currently selected frequency range is set to zero (ST 60 ). Next, the range selector 24 b obtains a new frequency range (ST 61 ) from the range detector 24 a . The range selector 24 b then confirms the new frequency range is correct (ST 62 ) in order to prevent propagation of spurious counts that may be due to noise or spikes on the input signal. In some embodiments, confirmation may be done by obtaining and comparing two or more consecutive new frequency ranges to see if they are the same. If the second new frequency range is different from the first, the first is discarded and the range selector 24 b attempts to confirm the second. Once a new frequency range is confirmed, the range selector 24 b compares it to the current frequency range (ST 63 ). If they are the same, no change takes place and the current range is maintained. If they are different, the range selector 24 b configures the multiple frequency range PLL 26 to operate in the new frequency range (ST 64 ). The range selector 24 b uses combinatorial logic programmed in the PLD to assert the appropriate input pins of the PLL 26 in order to configure the PLL 26 .

As previously mentioned, the range detector 24 a determines an appropriate frequency range based on the count received from the frequency detector 22 . In one embodiment, the count falls within one of seven possible frequency ranges, the boundaries for which are listed in TABLE 1. The count in this embodiment has been divided by a factor of two by ignoring the least significant bit from each count received from the frequency detector 22 . The resulting smaller counts allow the overall amount of logic required for processing the counts to be reduced.

As a side note, the counts listed in TABLE 1 may be converted into a using the following relationship:

Frequency=(Count×50 MHz×DF)÷2047  (Equation 1)

where DF is a Division Factor, or a factor by which the input signal was divided at any point, for example, at the flip-flop 30 . For the counts listed in TABLE 1, the Division Factor DF equals four to compensate for division by the flip-flop 30 and by the range detector 24 a.

The boundaries of the frequency ranges in TABLE 1 have been defined to overlap so that frequency range hysteresis may be implemented. In general, the term “hysteresis” means the delaying of slowing down of an event such as the switching between the different frequency ranges. With frequency range hysteresis, a count falling within a boundary overlap will not cause a change of frequency range. Only a count falling squarely within a different frequency range outside any boundary overlaps will result in a change of frequency range. Further explanation of frequency range hysteresis is provided by the following description of the range detector 24 a function.

As shown in FIG. 7 (and with reference to FIG. 3 ), to determine the correct frequency range for the PLL 26 , the range detector 24 a first obtains a count from the frequency detector 22 (ST 70 ). The range detector 24 a then assigns the value of the currently selected range number to a temporary placeholder (ST 71 ), which will be referred to herein as “temp range” for purposes of clarity of the description. Next, if temp range equals Range 0 and the count is greater than the maximum count for Range 0, the range detector 24 a assigns the value of Range 1 to temp range (ST 72 ). If either of these conditions is not satisfied, the value of temp range remains unchanged. The range detector 24 a then checks to see if temp range equals Range 1 and the count is greater than the maximum count for that range, and assigns the value of Range 2 to temp range if both conditions are met (ST 73 ). The range detector 24 a continues this manner of comparison for each range up to Range 6 (ST 77 ), then reverses the process by comparing in descending order the minimum count for each range against the count obtained from the frequency detector 22 (ST 78 -ST 83 ). After checking the count against each range in both ascending and descending order, the range detector 24 a assigns the final value of temp range as the new range (ST 84 ). This value may then be provided to the range selector 24 b for configuration of the PLL 26 as needed.

The foregoing selection process uses the overlaps between the frequency range boundaries to implement frequency range hysteresis, that is, counts falling within a boundary overlap will not cause a change of frequency range. This avoids the situation where an input signal frequency falls very near the boundary between two adjacent frequency ranges and causes the PLL 26 to oscillate between the two frequency ranges. If there were no boundary overlaps, slight decreases and increases in the input signal frequency may cause the PLL 26 to cross back and forth over the boundary, resulting in continuous switching back and forth between the two adjacent frequency ranges.

Multiple Frequency Range PLL

Referring now to FIG. 8 , the multiple frequency range PLL 26 operates essentially like the basic PLL discussed in the background section with the exception that it has a plurality of possible frequency ranges. The components of the multiple frequency range PLL 26 include a phase comparator 80 , a loop filter 82 , a VCO 84 , and a divider bank 86 , all connected as shown. The phase comparator 80 produces a DC or low frequency signal having a voltage proportional to the phase difference between the input signal and the output (or reference) signal. The loop filter 82 filters out any high frequency components that may be present in the DC or low frequency signal from the phase comparator 80 and passes the DC or low frequency signal to the VCO 84 . The VCO 84 has multiple user-selectable frequency ranges and produces a signal having a frequency within one of these frequency ranges. The frequency of the signal produced by the VCO 84 is proportional to the voltage of the signal from the phase comparator 80 (via the loop filter 82 ). The divider bank 86 is independently configurable to divide the frequency of the signal from the VCO 84 and to output this divided signal as an output signal. In one embodiment, the VCO 84 and the divider bank 86 are controlled by the frequency range selector 24 (as shown in FIG. 3 ). The combination of the VCO 84 and the divider bank 86 allows the PLL 26 to operate in a plurality of frequency ranges.

In some embodiments, the multiple frequency range PLL 26 may be implemented using the PSCB from Cypress Semiconductor (described in the background section). The operating ranges of the PSCB VCO are selectable via a three-level Function Select (FS) input pin as follows: FS=LOW (15-30 MHz), FS=MID (25-50 MHz), and FS=HIGH (40-80 MHz). Two additional, three-level Function Select input pins (3F0, 3F1) may be used to configure the PSCB internal dividers to divide the output of the PSCB VCO by up to a factor of four and output this signal at the PSCB output pins (3Q0, 3Q1).

In another embodiment, referring to FIG. 9 , the frequency of the output signal from the PLL 26 may be further divided by an external T-type flip-flop 90 . As shown, the flip-flop 90 is configured to divide the output signal frequency by a factor of two. This is useful, for example, in an embodiment where the PLL 26 is implemented using the Cypress Semiconductor PSCB. Recall from the background section the PSCB internal dividers can be configured to divide by up to a factor of four, thus allowing the PSCB PLL to phase lock signals having frequencies as low as 3.75 MHz. Adding the external flip-flop 90 to the PSCB increases that factor to eight, thus allowing the PSCB PLL to phase lock signals having frequencies as low as 1.875 MHz. In order to phase lock an input signal having a frequency of 2 MHz, for example, the PSCB VCO would only have to generate a 16 MHz output signal (16 MHz÷8=2 MHz), which is well within the lowest frequency range of the PSCB VCO (15-30 MHz). Additional flip-flops may, of course, be added should it be necessary to phase lock signals having even lower frequencies.

TABLE 2 lists the frequency ranges and associated pin settings to configure the PSCB PLL according to one embodiment of the invention. The information contained in this table may be used by one having ordinary skill in the art to develop the combinatorial logic of the frequency range selector 24 (shown in FIG. 3 ) for configuring the PSCB PLL.

It is to be understood the embodiments described herein are illustrative only, and other embodiments may be derived by one having ordinary skill in the art without departing from the scope of the invention. For example, TABLE 1 lists seven frequency ranges for a 248 MHz input signal, but other bandwidths and numbers of frequency ranges may certainly be defined as required for a particular application. Also, although the frequency detector 22 , frequency range selector 24 , and multiple frequency range PLL 26 have been described herein as separate components, in other embodiments they may all be combined in a single semiconductor integrated circuit. In still other embodiments, they may be implemented using discrete logic components, or a combination of discrete logic components and integrated circuits. Accordingly, the scope of the invention should be limited only by the following claims.