BACKGROUND
In U.S. Pat. No. 8,072,281 (“Hanafi”), entitled “Method, system and apparatus for accurate and stable LC-based reference oscillators,” issued Dec. 6, 2011 and incorporated herein by reference, the Temperature Null (TNULL) phenomenon has been analyzed and illustrated. On operating an LC tank at its TNULL phase, the oscillation frequency exhibits minimal frequency variation versus temperature. Related U.S. Pat. No. 8,884,718 entitled “Method and apparatus to control the LC tank temperature null characteristic in a highly stable LC oscillator,” issued Nov. 11, 2014, is also incorporated hereby by reference.
The oscillator which can force the LC tank to oscillate at its TNULL phase utilizing the TNULL phenomenon is said to be a “TNULL oscillator”. It is also called a “Self-Compensated Oscillator” in the sense that it exhibits minimal frequency variations across temperature without the need for external compensation circuitry. It is denoted as “SCO” for simplicity.
FIG. 1 presents the generic SCO. The oscillator consists of:
- 1. An LC tank circuit 101 to define the oscillating frequency.
- 2. A gm-cell 103 to compensate the losses of the tank circuit to start and sustain oscillations.
- 3. A phase block (Φ) 105 to adjust the phase of the tank impedance.
Herein, the phase Φ is programmed such that it is as close as possible to the inverted tank TNULL phase “—ΦGNULL”; thus the phase of the tank impedance becomes ΦGNULL. This is achieved by equating the frequency at the temperature range extremes To−ΔT and To+ΔT. In this case, the frequency deviates within this temperature range by the oscillator inherent behavior and no mechanism is applied to control the oscillator behavior.
The profile of the frequency variation versus temperature at the TNULL phase is denoted by the “Temperature Null Characteristic” or the “TNULL Characteristic”.
In Hanafi, a first order model for the tank variation versus temperature was analyzed and the theoretical expectation for the TNULL characteristic was introduced as shown in FIG. 2. The first order model of the tank versus temperature included the temperature variations of the inductor DC (direct current) losses only.
Practically, there are other factors that affect the TNULL characteristic, such as and not limited to:
- 1. The temperature varying harmonics induced by the active circuitry.
- 2. The temperature varying parasitic capacitances imposed by the routing interconnects and the active circuitry.
- 3. The temperature varying non-ideal effects in the inductor of the tank such as the skin depth effect and the proximity effect.
- 4. The temperature variation of the capacitance of the tank.
Due to such factors, the practical TNULL characteristic deviates from the theoretical expectations of the first order model. The shape of the practical TNULL characteristic varies according to the weight of each factor and the combination of the different factors. FIG. 3 compares three examples for possible practical TNULL characteristics to the theoretical TNULL characteristic as expected based on the first order model. The TNULL characteristic is the shape of the frequency deviation Δf(T) versus temperature, where Δf(T) is the frequency deviation referred to the oscillation frequency at the extremes of the temperature range To−ΔT and To+ΔT when operating at TNULL. Note that the frequency at the temperature range minimum To−ΔT is equal to the frequency at the temperature range maximum To+ΔT on operating at TNULL. This equality is an intrinsic feature of the TNULL characteristic since the TNULL phase is obtained by equating the phase-frequency plots at both To+ΔT and To−ΔT. Hence, Δf(T) is given as:
The excursion in the TNULL characteristic from the nominal frequency in the temperature range of To−ΔT to To+ΔT has several disadvantages such as increasing the overall frequency deviation versus temperature, complicating the trimming and calibration process and violating the initial accuracy specification which is the value of the frequency deviation at To that is usually the room temperature.
SUMMARY
Techniques are described that enable controlling the TNULL characteristic by controlling the magnitude and direction of the frequency deviation versus temperature, and thus, compensating the frequency deviation.
It is worth noting that compensating an SCO operating at TNULL is more convenient than compensating a conventional LC oscillator operating at a phase away from ΦGNULL, for example as in U.S. Pat. Nos. 7,332,975 and 8,134,414. This stems from the fact that the magnitude of the frequency deviation of the SCO is much smaller than that of the conventional LC oscillators. Thus, the SCO offers a better initial point to apply frequency compensation which yields the following advantages in the compensation system:
- 1. Operating at TNULL requires a smaller dynamic range for the compensation circuits because the frequency excursions that should be compensated in the case of the SCO are appreciably smaller than those in the case of the conventional LC oscillator.
- 2. Lower frequency deviation at TNULL implies that a lower temperature-to-frequency gain is required at the compensation loop, resulting into lower noise translation from the temperature sensor and the compensation circuitry to the oscillator output phase noise.
- 3. The oscillator may become less sensitive to process corners when trimmed to operate at TNULL.
BRIEF DESCRIPTION OF THE DRAWING FIGURES
The present invention may be further understood from the following detailed description in conjunction with the appended drawing figures. In the drawing:
FIG. 1 is a block diagram of a self-Compensated Oscillator (SCO).
FIG. 2 is a graph illustrating a theoretical TNULL characteristic based on the first order temperature model.
FIG. 3 is a graph illustrating three possible practical TNULL characteristics versus the theoretical TNULL characteristic derived from the first order model.
FIG. 4 is a block diagram of a SCO with the TNULL characteristic compensation applied through varying the LC tank phase versus temperature (Phase Compensation).
FIG. 5 is a graph illustrating SCO frequency deviation before and after applying the compensation of the TNULL characteristic.
FIG. 6 is a graph illustrating compensating the frequency deviation of the SCO inside and outside the TNULL range.
FIG. 7A is a block diagram of a SCO with TNULL characteristic compensation applied through varying the LC tank impedance (Ztank) versus temperature (Impedance Compensation).
FIG. 7B is a block diagram of a SCO with TNULL characteristic compensation applied through varying the input impedance of the following buffer (Load Compensation).
FIG. 8 is a block diagram of a SCO with TNULL characteristic compensation using phase compensation, impedance compensation and load compensation simultaneously.
FIG. 9A is a block diagram of an example of a compensation block that can be utilized to compensate the SCO TNULL characteristic.
FIG. 9B is a block diagram of an example of a compensation block that can be utilized to compensate the SCO TNULL characteristic.
FIG. 9C is a block diagram of an example of a compensation block that can be utilized to compensate the SCO TNULL characteristic.
FIG. 9D is a block diagram of an example of a compensation block that can be utilized to compensate the SCO TNULL characteristic.
FIG. 10 is a block diagram of a an IQ SCO with phase compensation applied on the IQ SCO.
FIG. 11 is a phasor diagram of the IQ SCO of FIG. 10.
FIG. 12 is a block diagram of a an IQ SCO with impedance compensation applied on the IQ SCO.
FIG. 13 is a diagram of digitally controlled capacitor units for compensating Ztank.
FIG. 14 is a diagram of an analog varactor for compensating Ztank.
DETAILED DESCRIPTION
Referring now to FIG. 4, there is shown the SCO with the proposed Phase Compensation applied. In FIG. 4, blocks 401, 403 and 405 correspond to blocks 101, 103 and 105 of FIG. 1. As explained in Hanafi, the (Φ) control is adjusted such that the oscillator operates at ΦGNULL. Afterwards, the compensation block 407 generates a temperature-dependent control signal S(T). This control signal is then used to control the (Φ) block in order to generate a phase (Φ) between the voltage and current which follows a specific profile across temperature. The SCO output frequency depends on the value of Φ according to a specific sensitivity function; thus, the SCO exhibits a temperature-dependent frequency shift corresponding to the control signal S(T). The control signal profile across temperature is adjusted such that the generated frequency shift substantially cancels the TNULL characteristic (the inherent behavior of the oscillator deviation at TNULL) as shown in FIG. 5.
The control signal profile can also be adjusted to compensate the oscillator inherent frequency deviation outside the TNULL operation range as well, as shown in FIG. 6. As illustrated in FIG. 6, the SCO is operating at the TNULL of the temperature range of To−ΔT to To+ΔT and the oscillator deviates significantly outside this range. The control signal in this case is used to compensate the frequency deviation outside the TNULL range as well.
FIG. 7A presents a variation of the proposed compensation mechanism. In FIG. 7A, blocks 701a, 703a and 705a correspond to blocks 101, 103 and 105 of FIG. 1. In this case, a compensation block 707a is used to induce the compensating frequency shift by varying the value of Ztank i.e. S(T) controls the tank impedance of the SCO. As proposed earlier, the induced frequency shift can compensate the SCO inherent frequency deviation inside and outside the TNULL range.
Moreover, FIG. 7B shows a further compensation mechanism. In FIG. 7B, blocks 701b, 703b and 705b correspond to blocks 101, 103 and 105 of FIG. 1. This time a compensation block 707b provides a compensation signal S(T) that controls the input impedance of the active circuit which interfaces the SCO. Normally, an oscillator is followed by an active buffering circuit such as output buffer 709b which delivers the oscillator signal from the oscillator to the required recipients while protecting the oscillator from any possible undesired loading. The input impedance (Zin) of such a buffer is considered as a part of the SCO tank impedance; thus, controlling the buffer input impedance (Zin) across temperature induces a controllable frequency shift across temperature. This compensation mechanism is denoted as “Load Compensation”. Load compensation can compensate the SCO inherent frequency deviation inside and outside the TNULL range.
Finally, the SCO can be compensated using a mix of all these techniques phase compensation, impedance compensation and load compensation as shown in FIG. 8. In FIG. 8, blocks 801, 803 and 805 correspond to blocks 101, 103 and 105 of FIG. 1. Compensations blocks 807a, 807b and 807c provide phase, impedance and load compensation, respectively.
Generally, the control signal (S(T)) generated by the compensation block can take several forms depending on the SCO architecture. For example and not for limitation, the control signal can be an analog signal, digital signal or a mix of both analog and digital signals. Furthermore, the control signal can be a voltage signal, a current signal or a mix of both current and voltage signals.
FIG. 9A to FIG. 9D shows different examples for generating the control signal (S(T)). In FIG. 9A, the temperature sensor 901 detects the die temperature and generates an analog signal (A(T)) that is substantially proportional to temperature. Afterwards, a control circuit, the profile generator block 903a, utilizes A(T) to generate S(T) with the specific temperature-dependent profile required to compensate the SCO TNULL characteristic. In FIG. 9A, the whole compensation process is done in the analog domain.
FIG. 9B illustrates a different concept. In FIG. 9B, blocks 901 and 903a correspond to blocks 901 and 903a of FIG. 9A. Herein, the SCO frequency is controlled using a digital signal; it is a Digitally-Controlled SCO (DCSCO). Thus, the control signal (S(T)) is transferred into the digital domain using an Analog-to-Digital Converter (ADC) 905 and then used to compensate the DCSCO.
In FIG. 9C, the output of the temperature sensor 901 is transferred into the digital domain by an ADC 907 and then the compensation profile is generated digitally (block 903d). The digital control signal is then used to compensate the DCSCO. FIG. 9D illustrates a concept similar to FIG. 9C except that the output of the digital compensation block is transferred back to the analog domain using a Digital-to-Analog Converter (DAC) 911 and then used to compensate the SCO.
Furthermore, the topologies explained in FIG. 9A to FIG. 9D are for the sake of example and not for limitation. For instance, the SCO can be compensated using a combination of these topologies depending on the SCO architecture.
Compensation Examples
The following description presents some techniques for the proposed TNULL characteristic compensation. The presented techniques are demonstrated just for example and not for limitation.
Example I
FIG. 10 shows the LC oscillator in a quadrature configuration. The quadrature configuration consists of two identical oscillator cores, the I-core and the Q-core, coupled together with the transconductance cells “gmc” (1011a and 1011b). The I-core includes a tank 1001I and an amplifier 1003I. The Q-core includes a tank 1001Q and an amplifier 1003Q. As explained in Hanafi, the IQ oscillator can be configured to work as an SCO. The phase between the voltage and the current in the tank circuits is given by:
Where gmc is the coupling transconductance and gm is the oscillator core transconductance. The initial phase is adjusted to force the oscillator to operate at the TNULL Phase (−ΦGNULL). The compensation block 1007 then generates a temperature-dependent profile that modulates the gmc/gm values and thus modulating the V-I phase. The control signal can modulate either gm or gmc and can be of analog nature, digital nature or a mix between analog and digital.
FIG. 11 shows the phasor diagram for the IQ oscillator. The oscillator is initially adjusted to operate as an SCO by adjusting V-I angle to ΦGNULL. Afterwards, the compensation block modulates the ΦGNULL by ΔΦ(T) using the control signal S(T). The modulated ΔΦ(T) should induce a frequency shift that cancels the inherent frequency deviation of the SCO.
Example II
FIG. 12 illustrates the proposed compensation for a quadrature oscillator core through the LC tank impedance. In FIG. 12, block 1201I, 1201Q, 1203I, 1203Q, 1211a and 1211b correspond to blocks 1001I, 1001Q, 1003I, 1003Q, 1011a and 1011b of FIG. 10. As explained in Example I, the gmc/gm ratio is chosen to adjust the V-I phase to operate at the Null Phase TNULL. Afterwards, the compensation block modulates the tank impedance Ztank to compensate the oscillator inherent frequency deviation. The control signal can be in analog and/or digital form, and can modulate any part of the tank impedance as explained above.
FIG. 13 shows an example of compensating Ztank 1300, including a tank circuit 1310. In this example, the capacitive part of Ztank is modified using capacitor units 1301-1, 1301-2, . . . , 1301-n that are digitally switched on or off to compensate the frequency deviation of the SCO. Each capacitor unit includes a capacitor and a switch (e.g., C1 and S1 in the case of capacitor unit 1301-1).
FIG. 14 shows another example for compensating the capacitive part of Ztank. In FIG. 14, elements 1400 and 1410 correspond to elements 1300 and 1310 in FIG. 13. In this case, an analog varactor 1403 is connected in parallel to the tank circuit and its control voltage S(T) is supplied by the compensation block. A hybrid solution can utilize both the digitally-controlled capacitor units and the analog-controlled varactor as well.
It will be appreciated by those skilled in the art that the present invention may be embodied in other specific forms without departing from the spirit or essential character thereof. The foregoing description is therefore intended in all respects to be illustrative and not restricted. The scope of the invention is indicated by the appended claims, not the foregoing description, and all changes which come within the meaning and range of equivalents thereof are intended to be embraced therein.