Continuously Tuneable Frequency Steerable Frequency Synthesizer Having Frequency Lock for Precision Synthesis

This invention relates to freqency synthesizers and more particularly to a frequency steerable frequency synthesizer which is continuously tuneable and which employs frequency lock for precision synthesis.

Frequency synthesis refers to the generation of a signal of precise frequency by the use of one or more reference frequencies. A commercially successful method employs a control system known as a phase locked loop (PLL). This method utilizes negative feedback to match the phase of the frequency divided output of a controlled oscillator to that of a reference frequency. The output is an adjustable multiple of the reference frequency. This multible is usually an integer or ratio of simple integers, and only a finite number of output frequencies normally are available.

The method and apparatus provides frequency accuracy proportional to the accuracy of the reference sources, usually crystal oscillators. While the output frequencies are changeable only in discrete steps, the modern phase locked synthesizers are capable of so many different frequencies that they can simulate, but not attain, the function of continuous tunable oscillators. In many applications this stepwise tuning is acceptable, but in others the ability to generate an arbitrary frequency is desirable. One of those applications is that of a local oscillator for radio receivers and transmitters, where phase locked synthesis has recently supplanted free running tunable oscillators as the primary tuning technique. Phase lock has the advantages of precise control of frequency, extremely low drift and compatibility with computer control.

Despite these advantages, phase lock does have the disadvantages of discrete frequency stepping and also considerable difficulty in attaining accurate frequency modulation. These weaknesses motivated the development of the precision Frequency Locked Loop of the present invention.

The frequency locked loop (FLL) is similar, in concept, to the PLL. Like the PLL, the system is a negative feedback control system that drives a voltage controlled oscillator (VCO) frequency to be N times a reference frequency. For this to hold exactly, the loop filter must have infinite DC gain. The PLL did not require this condition because of the integrating action of the VCO with respect to phase gives infinite DC gain. So long as the PLL has nonzero DC gain, the loop has infinite DC gain. With frequency as the control variable, infinite gain must be supplied by the filter to drive the error to zero.

An advantage of the FLL is that it inherently has 90 degrees less loop phase shift than the PLL. This is due to the VCO being merely a gain block, rather than an integrator. Consequently 90 degrees more phase shift is available for filtering. This allows better suppression of noise in the detector output than the PLL.

A FLL frequency syntheziser, despite its advantages, has not found widespread use because of the requirement of a linear frequecy detector. This requirement has been provided by the delta sigma frequency detector disclosed in U.S. Pat. No. 4,758,821. Accordingly, it is an object of the present invention to provide a highly accurate, continuously tuneable frequency synthesizer.

It is another object of the present invention to provide a linear frequency detector in a FLL thus making available in industry a frequency synthesizer of widespread application.

This invention provides a continuously tuneable frequency synthesizer comprising a FLL including a linear digital phase lock loop frequency detector having an input connected to an output of a voltage controlled oscillator. Another input of the detector has applied to it a reference frequency to produce at its output a digital word or voltage whose value is a function of the difference in the frequencies applied to its inputs. The voltage controlled oscillator has applied to its input the output of the detector to produce at the output of the oscillator a signal of desired frequency.

In a preferred embodiment, a low pass filter is included between the output of the detector and the input of the oscillator to remove high frequency noise that may be present in the the signal from the detector. The low pass filter may be an operational amplifier type. The value of freqency to be produced is under control of a steering voltage to be applied to an input of the low pass filter.

Further utilized in the preferred embodiment, as a frequency detector, is a .DELTA..SIGMA. phase lock loop, and a divide by N counter is connected in series between the output of the oscillator and the input to the frequency detector.

The invention, with further features, aspects and advantages thereof, will be better understood from a consideration of the following detailed description in conjunction with the accompanying drawings showing the best mode currently known to the inventor for the practice of the invention.

FIG. 1 is a block schematic representation of the frequency synthesizer of the present invention;

FIG. 2 is a time-waveform diagram helpful in the understanding of the present invention;

FIG. 3 is a circuit schematic of the frequency synthesizer of the present invention; and

FIG. 4 is a circuit schematic of a frequency detector useful in the practice of the present invention.

FIG. 5A illustrates a basic phase locked loop;

FIG. 5B illustrates a basic phase locked loop with variable assignments and transfer functions;

FIG. 5C illustrates a type 3 active loop filter having a single ended input;

FIG. 5D illustrates a type 3 active loop filter having a differential input.

FIG. 5E illustrates a magnitude and phase response of the type 3 active loop filter;

FIG. 5F illustrates a curve for H(s) using normalized form;

FIG. 5G illustrates a curve for H.sub.e (s) using normalized form;

FIG. 5H illustrates a curve for FM PLL with modulating voltage V.sub.m ;

FIG. 5I illustrates an FM PLL small signal model;

FIG. 5J illustrates a signal to distortion function in normalized form;

FIG. 5K illustrates an FM PLL with integrator error correction;

FIG. 6A illustrates a bypassed integrator;

FIG. 6B illustrates a bode plot of the bypassed integrator.

FIG. 6C illustrates a generalized error corrected FM PLL;

FIG. 6D illustrates a type 4 digital phase detector output;

FIG. 7A illustrates a basic frequency locked loop;

FIG. 7B illustrates an error corrected FM FLL;

FIG. 7C illustrates a reference modulated integrating loop filter;

FIG. 7D illustrates a reference modulated FM FLL;

FIG. 8A illustrates a conceptual .DELTA. modulation encoder;

FIG. 8B illustrates a conceptual .DELTA..SIGMA. modulation encoder;

FIG. 8C illustrates a conceptual .DELTA..SIGMA. frequency detector;

FIG. 8D illustrates a .DELTA..SIGMA. digital PLL FD model with small signal definitions;

FIG. 8E illustrates a digital filter implementation of the reference modulated integrating loop filter;

FIG. 8F illustrates a second order .DELTA..SIGMA. digital PLL frequency detector with PLL phase filtering;

FIG. 8G illustrates a predicted and Timeback Systems supplied first order .DELTA..SIGMA. noise spectrums;

FIG. 9A illustrates a PLL with input referred oscillator noise and its loop modified version V.sub.tncl ;

FIG. 9B illustrates an FLL with input referred oscillator noise and its loop modified version V.sub.tncl ;

FIG. 9C illustrates a quantization noise model of the .DELTA..SIGMA. FLL;

FIG. 9D illustrates a noise performance in the example VHF .DELTA..SIGMA. FLL;

FIG. 9E illustrates a noise performance plot of the example VHF D.SIGMA. PLL.

FIG. 9F illustrates a V.sub.qcl and V.sub.tncl for the example system with .alpha..sub.e =2E-14;

FIG. 9G illustrates a total .phi.(f.sub.m) for the example system with .alpha..sub.e =2E-14;

FIG. 9H illustrates a V.sub.qcl and V.sub.tncl for the example system with .alpha..sub.e =0.

FIG. 9I illustrates a total .phi.(f.sub.m) for the example system with .alpha..sub.e =0;

FIG. 10A illustrates a multivibrator VCO used in the experimental system;

FIG. 10B illustrates a .DELTA..SIGMA. model test supply;

FIG. 10C illustrates a .DELTA..SIGMA. frequency detector frequency domain behavior;

FIG. 10D illustrates a 43 Hz loop bandwidth .DELTA..SIGMA. FLL experimental system;

FIG. 10E illustrates an output frequency vs. tune voltage for the experimental .DELTA..SIGMA. FLL;

FIG. 10F illustrates an open loop, closed loop measured, and closed loop predicted phase noise for the 43 Hz loop bandwidth system;

FIG. 10G illustrates an open loop, closed loop measured, and closed loop predicted phase noise for the 900 Hz loop bandwidth system; and

FIG. 10H illustrates a phase noise in the 900 Hz bandwidth system;

FIG. 11A illustrates a graph of normalized magnitude noise;

FIG. 11B illustrates a simplified schematic diagram of a .DELTA..SIGMA. frequency detector;

FIG. 11C illustrates a graph of open loop phase noise;

FIG. 11D illustrates a graph of closed loop phase noise of the 43 Hz loop bandwidth .DELTA..SIGMA. FLL; and

FIG. 11E illustrates a graph of closed loop phase noise of the 900 Hz loop bandwidth .DELTA..SIGMA. FLL.

In the description which follows, like elements will be designated with the same reference characters throughout the various FIGURES. Referring now to FIG. 1, there is illustrated in block schematic form, a frequency synthesizer 100 embodying features of the present invention. The synthesizer 100 comprises a voltage controlled oscillator 102 having an output .DELTA..omega..sub.out applied to a divide by N counter 106, which may be provided where typically the frequency output 25 from the voltage controlled oscillator is larger than a reference frequency. The output .DELTA..omega..sub.2 (s) is applied to an input of a frequency detector 104. A reference signal .omega..sub.ref is applied to another input to the frequency detector 104 from a source (not shown), but which, preferably is a crystal controlled oscillator. The result is a correction signal .DELTA.U.sub.d (s) which is fed to a low pass loop filter 108 to remove high frequency noise. The filtered output .DELTA.U.sub.f (s) is applied to the voltage controlled oscillator to produce the desired frequency output. The selection of frequency is implemented by injecting a voltage .DELTA.V.sub.m (s) into the loop. As illustrated, the selection voltage is applied to an input to the loop filter 108.

The frequency detector 104 is a .DELTA..SIGMA. digital phase locked loop. This detector is of the type described in U.S. Pat. No. 4,758,821 granted to Robert Nelson and Onkar Modgil. The circuit operates to keep the edges of the reference .omega..sub.ref and the output .DELTA..omega..sub.2 (s) lined up. The loop filter 108 is an operational amplifier having infinite gain.

FIG. 3 illustrates in more detail the frequency synthesizer of the present invention. The frequency synthesizer 200 was built and tested in a low radio frequency CMOS implementation. The system was designed for a 5 volt power supply. The system clock (f.sub.ref) was 800 KHZ. The frequency detector 204 had M=-8, and .DELTA.M=1. This led to a system with a frequency tuning range of 8 times f.sub.ref to 9 times f.sub.ref, or 6.4 to 7.2 MHz. The oscillator or VCO 202 was implemented using a Motorola MC4024 multivibrator integrated circuit. This approach had the advantage of circuit simplicity, and a high enough phase noise content to clearly view the effect of the .DELTA..SIGMA. Frequency Lock Loop on the noise spectrum.

The .DELTA..SIGMA. frequency detector 204 and the VCO 202 are connected in a loop with the loop filter 208 to complete the frequency lock loop design. The loop filter was provided by a Texas Instruments TLC 271 CMOS operational amplifier, primarily because of its ability to run in a single supply low voltage system. The loop filter 208 was powered from a 7 volt supply, and the rest of the circuit from a 5 volt supply. The integrating loop filter 208 had a unity gain bandwidth of 43 Hz. The 800 KHz signal f.sub.ref was generated by clock 210. The clock was a 3.2 MHz crystal oscillator followed by a divide by 4 circuit.

The system 200 was found to function in a very acceptable manner. When powered up it instantly acquired frequency lock at the frequency specified by the tuning voltage V.sub.m from source 212. There was no sign of instability.

When the system of FIG. 3 was frequency tuned by varying V.sub.m it tuned precisely and with perfect linearity from 6.4 to 7.2 MHz as the tuning voltage was varied from 0 to 5 volts.

Referring now to FIG. 2 the waveforms initially illustrate that the frequency synthesizer of FIG. 3 is in a stable state. Now upon the desire to change the frequency output the steering or modulating voltage .DELTA.V.sub.m (s) is changed in value, for example, from 1 volt to 2.5 volts. The .DELTA..SIGMA. signal produced within the frequency detector is a series of one bits. When the voltage is changed the bit density increases and there is produced at the output of the frequency detector a filtered .DELTA..SIGMA. signal .DELTA.U.sub.d (s). Note however the presence of high frequency noise. The low pass loop filter 108 smooths the signal by filtering out the high frequency noise to produce the signal .DELTA.U.sub.f (s). This signal, free of high frequency noise, is applied to the input of the voltage controlled oscillator.

For more details, supplementing the above description, reference may be had to the Thesis entitled A NEW APPROACH TO FREQUENCY MODULATED FREQUENCY SYNTHESIS authored by the inventor and presented to the Faculty of the Graduate School of The University of Texas at Arlington. This thesis is included in this specification in its entirety at Appendix A attached hereto.

Referring now to FIG. 4 where there is illustrated in greater detail the .DELTA..SIGMA. PLL frequency detector described and claimed in U.S. Pat. No. 4,758,821.

The following discussion is extracted essentially verbatim from columns 4 through 8 of U.S. Pat. No. 4,758,821 as a full description of the digital phase lock loop detector that is herein referred to as the delta sigma phase lock loop (PLL) frequency detector. The delta sigma PLL frequency detector permits (1) generating two high frequency signals, (2) causing a shift in the phase of one of two signals in response to an analog condition, (3) comparing the phase of the signals and (4) adjusting the phase of the phase-shifted signal toward an in-phase condition. More specifically, one of the signals is at a fixed frequency that a clock may produce. The other signal also essentially has a fixed frequency and is produced by dividing the output of a variable frequency oscillator (whose frequency is a function of an analog condition) by a ratio that is determined by the phase relationship of the two fixed frequency signals. The division ratio will either be N or (N-M), where N and M are whole integers and M is less than N. A comparison of the phase relationship between the signals yields a measure that is utilized to select the division ratio to bring the signals to an in-phase relationship. The output or measure of phase difference is in the form of a single-bit digital function. This function is filtered to produce a binary word representative of the instantaneous value of the analog condition.

Referring now to FIG. 4, a digital implementation of the analog design illustrated in FIG. 3 is shown wherein delta sigma phase lock 10 of frequency detector 204 of the present invention is shown to include two sources of high-frequency signals, TCLK from clock 12 corresponding to f.sub.ref from clock 210 of FIG. 3 and FCLK from VCO 14 corresponding to f.sub.in from VCO 202 of FIG. 3. Low pass filter 20 corresponds to analog loop filter 208 for the digital implementation. The first source includes clock 12 that has a constant frequency output. The second source includes variable frequency oscillator 14 and counter-divider 16. In accordance with the present invention, whereas variable frequency oscillator 14 has an output whose frequency varies as a function of applied digital condition received from low pass filter 20, counter-divider 16 has a frequency output that is constant and harmonically related to the frequency of clock 12. In the preferred embodiment, the frequencies at the outputs of counter-divider 16 and clock 12 are equal.

Counter-divider 16 has an output DIN and clock 12 has an output TCLK. Both outputs are applied to phase-comparator 18, a D flip-flop, that produces an output DATA whose character depends on whether or not the transition of the output DIN of counter-divider 16 leads or lags the transition of the output TCLK from clock 12. The selected transitions may be either positive or negative going. The present specific embodiment uses the negative-going transition of the output DIN from counter-divider 16 and the positive-going transition of the output TCLK from clock 12.

The output DATA from comparator 18, a single-bit digital function, will either be a logic level 0 if the negative transition of the output of counter-divider 16 leads the onset of the output from clock 12 or a logic level 1 if the negative transition of the output of counter-divider 16 lags the onset from clock 12. This single-bit digital function is applied to low-pass digital filter 20 which produces at its output a binary word. The binary word is then applied to a VCO 14.

Counter-divider 16 is controlled such that it either divides by a value N or by a smaller value (N-M). These values or dividing ratios, N and (N-M), are selected on the basis of a criterion of whether the negative transition of the output of counter-divider 16 occurs early or late with respect to the positive going transition from clock 12. If the negative transition of the output DIN (FIG. 4) of counter-divider 16 occurs early with respect to the positive transition of clock 12 output TCLK, the inverted output DATA of comparator 18, a D flip-flop is fed back via NAND gate 24 to the P input of counter-divider 16 via NAND gate 24 to cause counter-divider 16 to divide by N-M. If the value of the digital condition is such that the output FCLK from variable frequency oscillator 14 is at a higher frequency, then counter-divider 16 will count faster, counter-divider's 16 output DIN will occur early more often with respect to the output TCLK of clock 12 and inverted output DATA from comparator 18 will cause counter-divider 16 to divide by N more often. On the other hand, if the value of the digital condition is such as to cause variable frequency oscillator (VFO) 14 to output FCLK at a lower frequency, counter-divider 16 will count slower, the negative transition of the output DIN of counter-divider 16 will be late more often, and inverted output DATA of comparator 18 will cause counter-divider 16 to divide by (N-M) more often. Accordingly, the phase of the negative transition of output DIN of counter-divider 16 with respect to clock 12 output TCLK is controlled by the frequency of variable frequency oscillator 14. The phase is measured with phase comparator 18 and the negative transition of output DIN of counter-divider 16 is brought in phase with the signal TCLK by adjusting the dividing ratio of counter-divider 16.

Control logic 30 (including NAND gate 24) assures that the dividing ratio of counter-divider 16 may be changed only once for every TCLK and in conjunction with TCLK. TCLK controls comparator 18 so that only a single binary value is produced for each TCLK pulse. This single binary value is processed by the decimating digital low-pass filter 20 in which binary ones and zeros are multiplied with the finite impulse response filter coefficients to provide the binary word at a lower sample rate. Digital low-pass filters are described in "Digital Filters" by R. W. Hamming, 2nd Edition published by Prentice Hall, 1983.

Control logic 30 includes two D flip-flops designated as 32 and 34, three NAND gates 36, 38 and 24, and one AND gate 40. Control logic 30 had three logic states: logic state zero, logic state one (represented by signal logic ST1 at logic level 1), and logic state two (represented by signal ST2 at a logic level 1). Logic state zero occurs in the absence of logic states one and two. Logic state one is entered when TCLK goes to logic level 1 and upon the next positive transition of FCLK. More specifically, there are three inputs to AND gate 40. These inputs are TCLK from the clock 12 applied by way of conductor 42, ST1 and ST2. With all three inputs at logic level 1 at time t.sub.1, there appears at the output of AND gate 40 a positive going pulse D1 applied to the D input of flip-flop 32. With the next positive going transition of FCLK at time t.sub.2, pulse ST1 representative of logic state one is generated and latched up for one cycle of FCLK. In other words, logic state one exists for only one cycle of FCLK. It is during this period that the logic level 1 of signal ST1 is applied to NAND gate 24. In this same period, the output DATA of comparator 18 is at a logic level 1 and DATA is at a logic level 1. This is what occurs when the negative-going edge of output DIN from counter-divider 16 occurs early with respect to the positive transition of TCLK. The DATA signal is applied to the other input of NAND gate 24 whose output ENABLE will be a logic level 0. ENABLE at the logic level 0 is applied to the P input of counter-divider 16. The application of logic level 0 to the P input of counter-divider 16 disables the counter-divider for one FCLK cycle nd effectively selects the division ratio for counter-divider 16 at (N-M).

In the present embodiment M=1, or one cycle of FCLK. The value of M is determined by the time duration of the pulse ST1 at logic level 1. M can be assigned other integer values by modifying central logic 30 so that ST1 at logic level 1 will exist for more cycles of FCLK. For example, if a value of 2 is assigned to M, then ST1 would exist for two cycles of FCLK.

Upon the next transition of FCLK, control logic enters into logic state two as represented by the onset of the pulse ST2 at time t.sub.3. ST1, at logic level 0, is applied at time t.sub.2 to an input of NAND gate 38. At the same time, ST1 is also applied back to the input of AND gate 40. This causes the value of pulse DI to be at logic level 0. Therefore, when the next transition of FCLK occurs, at time t.sub.3, control logic 30 enters state two as represented by the generation of pulse ST2. Control logic 30 remains in state two until the next transition of FCLK after TCLK goes to logic level 0 at which time control logic 30 returns to state zero.

Flip-flop 34 remains in logic state two so long as TCLK is at logic level 1. This avoids an early return to state one which would cause a premature clocking out of the output of counter-divider 16 and could result in multiple binary values being generated during a given cycle of TCLK. By holding the logic in state two until TCLK goes to logic level 0, there will be only one binary value produced at the output of comparator 18 for each cycle of TCLK.

The transition of states from zero through two takes place in control logic 30 in the following manner. Prior to time t.sub.1, control logic 30 is in state zero since TCLK is a logic level 0. At time t.sub.1, TCLK ST1 and ST2 are at logic level 1 and the AND gate passes pulse D1. Upon the next positive transition of FCLK, applied to a CK input of flip-flop 32 via conductor 44, ST1 goes to logic level 1. Concomitantly, ST1 goes to logic level 0 and terminates pulse D1.

TCLK is also applied at a logic level 1, via conductor 46, to an input of NAND gate 36. Prior to time t.sub.3, ST2 is at logic level 0, and this logic level 0 is applied via conductor 50 to the other input of NAND gate 36. This results in a logic level 1 appearing at the output of NAND gate 36. This output is applied to an input of NAND gate 38. At this time, ST1 is applied to the other input of gate 38 which is at a logic level 0. Therefore, the output of gate 38 is a logic level 1, resulting in a positive transition of pulse D2. Now, upon the next positive transition of FCLK, applied to the CK input of flip-flop 34 via conductor 48, ST2 appears as a logic level 1 at the Q input of flip-flop 34. ST2 or state two is latched until the next positive transition of FCLK immediately following the value of TCLK falling to a logic level 0 value as at time t.sub.4. At this time and in the absence of ST1 and ST2, control logic 30 has returned to state zero and it remains there until the next positive transition of TCLK at time t.sub.5.

The output ENABLE from control logic 30 appearing at the output of NAND gate 24 is normally at logic level 1 such that counter-divider 16 is enabled or allowed to increment on every positive transition of FCLK. The character of ENABLE whether logic level 0 or a logic level 1 is controlled by DATA each time ST1 goes to a logic level 1. The ENABLE output will change counter-divider 16 from a divide-by-(N-M) to a divide-by-N. In one embodiment of the present invention, N=5 and M=1 and therefore N-M=4. Counter-divider 16 is a two-bit binary counter. When the counter-divider's output DIN has its negative transition early with regard to TCLK a shown prior to time t.sub.1, output DATA of phase comparator 18 goes to a logic level 1, and ENABLE goes to a logic level 0 upon the next occurrence of logic level 1 ST1. Then, at time t.sub.2, counter-divider 16 does not increment on the next positive going transition of FCLK. Since counter-divider 16 normally functions as a divide-by-4 counter-divider, the loss of one increment that occurs when ENABLE goes to a logic level 0 for one cycle of FCLK causes the output DIN from counter-divider 16 to occur at 1/5th the rate of FCLK. Thus, ENABLE changes the counter from dividing FCLK by 4 to dividing FCLK by 5. Since an additional cycle of FCLK is needed before the next negative going transition of DIN, DIN has been delayed with respect to TCLK and brought in phase therewith.

On the other hand, when output DIN of counter-divider 16 has its negative going transition late with respect to TCLK, such as at time t.sub.5, output DATA of phase comparator 18 goes to a logic level 0, and ENABLE remains at a logic level 1 during the next occurrence of logic level 1 of ST1. This allows counter-divider 16 to increment on every positive going transition of FCLK. Counter-divider 16 divides FCLK by 4 and causes the next negative going transition of DIN to occur without delay. Thus, if FCLK is within the frequency limits (i.e., within 4 to 5 times the frequency of TCLK) output DIN of counter-divider 16 will be brought in phase with TCLK.

The process continues with either a divide-by-4 or a divide-by-5 as selected by control logic 30 and DATA in order that output DIN of counter-divider 16 remains in-phase with TCLK. The process gives rise to the generation of the single-bit digital function DATA. Each single digital bit is indicative of the direction of phase error. The single digital bits may either be single weighted digital functions or multi weighted digital functions. In the present embodiment, the digital bits are single weighed digital functions.

Variable frequency oscillator 14 (VFO) can be any device that has an output within a desired frequency range and whose frequency can be altered by a selected variable such as capacitance, voltage, temperature and resistance. One form of variable frequency oscillator is shown in FIG. 4 as a voltage controlled oscillator responsive to a digital input. The voltage controlled oscillator, otherwise known as a voltage-to-frequency converter, is available from several manufacturers.

In one embodiment of the invention, oscillator 14 has a center frequency of 1.125 megahertz and varies from between 1 megahertz to 1.25 megahertz. In that embodiment, clock 12 produced a 250 kilohertz square wave.

Various elements comprising the system of FIG. 4 are available from Texas Instruments Incorporated and are identified by the following codes: AND gate 40 is an SN 74HC11N; flip-flops 32 and 34 and comparator 18 are SN 74HC74N; NAND gates 36, 38 and 24 and SN 74HC00N; and counter-divider 16 is SN 74HC161N.

In the Patent, the output of the low pass filter 20 corresponding to loop filter 208 of FIG. 3 is connected to a utilization device. In the present modification, the output from the loop filter 20 is connected directly to an input of the VCO 14 by way of cable or conductor 21. In addition, a steering or modulating voltage V.sub.m is applied to an input of the loop filter 20. With these modifications the operation of the frequency synthesizer of FIG. 4 is the same as that described in conjunction with the other figures of this application. ##SPC1##

Now that the invention has been described, variations and modifications will become apparent to those skilled in the art. It is intended to that such variations and modifications be encompassed within the scope of the appended claims.