# Compensation Technique for the Nonlinear Behavior of Digitally-Controlled Oscillator (DCO) Gain

Systems and methods are provided for hopping a digitally controlled oscillator (DCO) among a plurality of channels, wherein a gain of the DCO KDCO is a nonlinear function of frequency. A first normalized tuning word (NTW) corresponding to a first channel of the plurality of channels is generated. A first normalizing gain multiplier X is generated based on the nonlinear function of frequency, on an estimate of the nonlinear function of frequency, at a first frequency corresponding to the first channel. The first NTW is multiplied by the first X to obtain a first oscillator tuning word (OTW). The first OTW is input to the DCO to cause the DCO to hop to the first channel. A system for hopping among a plurality of channels at a plurality of respective frequencies comprises a phase-locked loop (PLL), a digitally controlled oscillator (DCO), a multiplexer, and an arithmetic module.

**Description**

**CROSS-REFERENCE TO RELATED APPLICATIONS**

This application claims priority to U.S. Provisional Patent Application No. 62/549,004, filed Aug. 23, 2017 and entitled “A Compensation Technique for the Nonlinear Behavior of Digitally-Controlled Oscillator (DCO) Gain,” the entire contents of which are incorporated by reference herein.

**FIELD**

The technology described in this disclosure relates generally to digitally controlled oscillators (DCOs) and more particularly to the use of DCOs in phase locked loops (PLLs).

**BACKGROUND**

All-digital PLLs (ADPLLs) are widely used in advanced complementary metal-oxide-semiconductor (CMOS) based semiconductor devices. There they exploit the naturally fine resolution of voltage-controlled oscillators (VCOs), e.g., digitally-controlled oscillators (DCOs), thus reducing area and power dissipation versus analog PLLs.

**BRIEF DESCRIPTION OF THE DRAWINGS**

Aspects of the present disclosure are best understood from the following detailed description when read with the accompanying figures. It is noted that, in accordance with the standard practice of the industry, various features are not drawn to scale. In fact, the dimensions of the various features may be arbitrarily increased or reduced for clarity of discussion.

_{DCO }(f) over a wide frequency span from DC to the upper range of the core DCO frequency that is necessary to cover all GSM bands.

_{DCO }(f) over the frequency span covering four GSM TX and RX bands.

_{DCO }(f) over the frequency span covering the DCS-1800 band.

_{DCO }compensation algorithm.

**DETAILED DESCRIPTION**

The following disclosure provides many different embodiments, or examples, for implementing different features of the provided subject matter. Specific examples of components and arrangements are described below to simplify the present disclosure. These are, of course, merely examples and are not intended to be limiting. In addition, the present disclosure may repeat reference numerals and/or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed.

Further, spatially relative terms, such as “beneath,” “below,” “lower,” “above,” “upper” and the like, may be used herein for ease of description to describe one element or feature's relationship to another element(s) or feature(s) as illustrated in the figures. The spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. The apparatus may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein may likewise be interpreted accordingly.

DCOs can be implemented in a variety of forms. For example, an inductor-capacitor (LC) based DCO, which also can be referred to as an LC-tank, can achieve improved phase noise (PN) at lower power consumption, with low frequency pushing, and immunity from process and temperature variations. However, there is a strong non-linear behavior of the LC-based DCO across the overall tuning band. Such non-linearity can result in a potential stability issue and can increase uncertainty in the loop's behavior. In some embodiments, systems and methods described herein relate to techniques for compensating for the nonlinear behavior of digitally-controlled oscillator (DCO) gain and related systems and devices. Further embodiments describe interpolation based compensation techniques for the cubic behavior of DCO gain, and related systems and devices, such as center linear interpolation based compensation techniques, systems, and devices. For example, certain embodiments describe clock generation techniques, LC-tank oscillators (VCOs/DCOs), nonlinear oscillator gain, linear interpolation techniques, frequency hopping, and/or fast locking, and suitable combinations thereof.

Certain embodiments herein describe generating stable and finely tuned frequencies with a DCO, such as for use in generating clock signals in digital circuitry (e.g., in computer processors), generating frequency hopped signals such as for use in cellular communications, and the like. Certain embodiments describe a method for locking a phase locked loop (PLL), such as an all-digital PLL (ADPLL), to a frequency generated by a DCO that exhibits nonlinear gain as a function of frequency and thus otherwise may be difficult to lock to stably with a PLL. Certain embodiments provide one or more of the following advantages: the nonlinear behavior of DCO gain can be predicted and thus used to lock to the frequency of the DCO with a PLL without the need for measuring the frequency of the DCO at runtime, and hence provides one or more of (1) reduction of phase error when locking to the frequency of the DCO, (2) more rapid locking to the frequency of the DCO, (3) finer tuning to the frequency of the DCO, and (4) more stable generation of clock signals or communication frequencies.

The details of the example methods, systems, and devices of the present disclosure are described in the attached disclosure and drawings. It should be noted that the present technology is not limited to silicon based DCOs, nor to DCOs with cubic gain behavior, but also is applicable to DCOs based on other materials and/or with other nonlinear behavior.

**100** can include a PLL, such as ADPLL **110**, and a VCO, such as DCO **120**, coupled to the PLL. The gain K_{DCO }of DCO **120** can have nonlinear behavior, and ADPLL **110** can be configured so as to compensate for such nonlinear behavior. For example, it has been determined that within at least some ranges of frequency f, the K_{DCO }of DCO **120** can vary as a cubic function of frequency f, and ADPLL **110** can be configured so as to generate a normalizing gain multiplier x that can be used to compensate for the nonlinear (e.g., cubic) behavior of the DCO gain K_{DCO }in a manner such as provided herein. For example, ADPLL **110** can include gain estimation module **112** configured to calculate a normalizing gain multiplier x based on the K_{DCO }and a reference frequency f_{R }generated by the ADPLL in a manner such as described herein, multiplexer (MUX) **114** configured to select between the normalizing gain multiplier and an initial gain multiplier x_{0 }based on a calculation “complete” signal, and arithmetic module **116** configured to generate an oscillator tuning word (OTW) based on x. The ADPLL **100** can include an output (e.g., an output of arithmetic module **116**) configured to provide the OTW to DCO **120**, which generates a frequency based thereon with high accuracy and phase stability.

_{0}, is used for the OTW calculation in the beginning. Then the change in frequency Δf relative to an original locking frequency f_{0}, and the change in DCO gain ΔK_{DCO }can or will be injected into the gain estimation module (calculator) **112** to generate a normalizing gain multiplier x (calculated), which is provided to MUX **114**. The MUX **114** selects between the initial value x_{0 }and the normalizing gain multiplier x based on a calculation “complete” signal. When the calculation is done, the new and correct normalization value x can or will replace the original x_{0 }value and can be multiplied by the NTW at arithmetic module **116** to obtain OTW which is input to the DCO **120**. The DCO **120** will output a correct frequency after the correction, e.g., the cubic rule correction. The functionality of the arithmetic module **116** and the DCO **120** can be grouped together as **122**.

In some embodiments, a relationship that ties the frequency step Δf^{T }of a DCO, such as DCO **120** illustrated in ^{T }of the DCO is governed by, or can be expressed as, equation (1):

Δ*f*^{T}(*f*)=−2π^{2}(*L·ΔC*^{T})*f*^{3 } (1)

in which L is the LC-tank inductance. Both L and ΔC^{T }are constants for a stable process, voltage, and temperature (PVT). In some embodiments, L and ΔC^{T }are the only unknowns and subject to the PVT variations on the right-hand side of the equation, so it can make sense to group them together as a product. The resonating frequency f is controlled by the total capacitance C of the LC-tank. f is known precisely (within a frequency control word (FCW) resolution—1.5 Hz/LSB) by virtue of the ADPLL loop operation.

Because K_{DCO}=|Δf^{T}|, equation (1) can be rewritten as equation (2):

*K*_{DCO}(*f*)=2π^{2}(*L·ΔC*^{T})*f*^{3 } (2)

Taking the derivative of K_{DCO }with respect to frequency f results in equation (3):

In some embodiments, the K_{DCO }estimate, {circumflex over (K)}_{DCO}, is generated by gain estimation module **112** in a manner such as described elsewhere herein, and used in the ADPLL frequency synthesizer in the denominator of the DCO normalizing gain multiplier of value x, e.g., provided to MUX **114** for use in generating x based on {circumflex over (K)}_{DCO }and the external reference frequency f_{R }(which reference frequency can be generated by ADPLL **110**) such as illustrated in

In some embodiments, an exemplary purpose of this is to conveniently decouple the phase and frequency information throughout the system from the PVT variations that normally affect the gain K_{DCO }of the DCO **120**. For example, the frequency information can be normalized to the value of external reference frequency f_{R }using {circumflex over (K)}_{DCO}. Such normalization alternatively can be performed within the DCO, e.g., within a normalized DCO (nDCO) **130** such as illustrated in _{R}. In some embodiments, f_{R}/{circumflex over (K)}_{DCO }is a measure of the DCO gain estimation accuracy and affects precision of the frequency modulation. In some embodiments, the functionality of the arithmetic module **116** and the DCO **120**, indicated by **122**, can be represented by the nDCO **130**, and more specifically by a DCO gain normalization module **132** and a DCO **134**.

The normalizing gain multiplier x based on the estimated gain {circumflex over (K)}_{DCO }can be expressed as x=f_{R}/{circumflex over (K)}_{DCO}, or alternatively a normalizing gain multiplier x based on the actual K_{DCO }can be expressed as x=f_{R}/K_{DCO}. In either embodiment, the generation of x can be performed on a per-packet basis with a just-in-time method using dedicated hardware or software modules, e.g., gain estimation module **112** and MUX **114** such as illustrated in _{DCO }over multiple packets.

To further reduce the design complexity of digital coding, the nonlinear (e.g., cubic) behavior of the DCO gain can be estimated using one or more linear functions, e.g., using linear interpolation. For example, in some embodiments taking the derivative of x=f_{R}/K_{DCO }(e.g., based on the actual DCO gain K_{DCO }rather than the estimated DCO gain {circumflex over (K)}_{DCO}) as a function of frequency results in equation (5):

which could be conveniently written as equation (6):

Equation (6) reveals that, in some embodiments, the DCO gain variation ΔK_{DCO }is roughly 3 times that of the frequency variation.

_{DCO }interpolation error of 2.6% is too high for the linear interpolation that covers the entire GSM spectrum and is not shown in the graph.

_{DCO }values within a single band.

_{DCO }compensation algorithm. The x=f_{R}/K_{DCO }estimates from packets at different frequencies can be frequency independent or normalized first by either the inverse cubic equation Eq. 4 or linear interpolation, as depicted in **210**, the frequency of a completed packet is determined, and at **220**, a value of x is determined. The normalized value of x can be represented by x_{o}[k], where k is a sample index, and it corresponds to a certain frequency f_{0}, that preferably lies in the middle of a GSM band. The normalized value x_{o }calculated at **230** based on linear interpolation is represented by equation (7):

The term y_{0}[k] is denormalized for the particular packet frequency f, such that y[k] could be used as the DCO gain multiplier. In this method the just-in-time calculated sample, as considered too noisy, is not immediately substituted for the normalizing gain multiplier, but rather it is input to the filtering algorithm. Then, the filtering is performed at **240** according to the equation of a “leaky integrator” in equation (8):

*y*_{0}*[k*]=(1−α)·*y*_{0}*[k*−1]+α·*x*_{0}*[k]* (8)

where α is a coefficient of a first-order infinite impulse response (IIR) filter and y_{0}[k] is the filtered normalized value. For practical implementation reasons, α=2^{−λ} where λ is an integer. This way, the multiplication by can be realized as a right bit shift operation.

At **250**, the new packet frequency f is determined. The denormalization equation for the filtered multiplier y, calculated at **260**, is shown in equation (9):

In equation (9), y corresponds to ΔK_{DCO}, f_{0 }corresponds to the frequency at which the gain variation is 100%, and y_{0 }corresponds to the value of ΔK_{DCO }at which the gain variation is 100%. At **270**, the calculated value of y is written and/or output as the DCO gain multiplier.

^{3}+bX^{2}+cX+d), that is, as a cubic function of frequency, with high correlation. In some embodiments, the DCO frequency is operated doubled at the Bluetooth band (e.g., operated at around 4.8 GHz and the frequency then halved to around 2.4 GHz, so as to use smaller components than may be needed to generate frequencies directly in the Bluetooth band). For example, the nonlinear DCO gain variation can be in the GHz band.

The reduced digital design can benefit from the reduction from a cubic polynomial to a linear equation. For example, **310**, the simulated gain of the DCO in diamonds (corresponding to the diamonds illustrated in **320** (corresponding to the curve illustrated in **320** can be represented by y=−6.5668x^{3}+98.228x^{2}−488.77x+810.12. In some embodiments, such linear interpolation will or can result in a relatively large predicted error due to the cubic nonlinear curve of DCO gain in the middle of the operation frequency. Thus, in some embodiments, a different linear equation formula(e) can be used, e.g., a center linear interpolation such as shown in the straight, solid lines **330**, **340** illustrated in **330**) and F2 (corresponding to line **340**) that each are the same as equation (9) with a different parameter y_{0 }before (f−f_{0})/f_{0}. For example, in an exemplary embodiment, in F1 the value of y_{0 }can be changed from 3 to 2.8 (or other suitable value) and/or in F2 the value of y_{0 }can be changed from 3 to 3.3 (or other suitable value). It should be appreciated that the particular values of y_{0 }can depend on the center point of the curve (corresponding to point **350** in _{0 }in F1 and F2 is not limited to being larger or smaller than the original value of y_{0}. Additionally, or alternatively, it should be appreciated that if the cubic polynomial is not in the nonlinear portion of the curve, a linear interpolation from the start to the end (such as line **310** illustrated in **350** of the solid, straight lines **330**, **340** is not limited to the center of the operation band, but instead could be adjusted by the system design and design complexity.

**510**), e.g., such as described herein with reference to equation (1). The inductance L(f) effect optionally can be added (operation **520**). For the restricted resolution, the inductance variation can be taken into account, however, the inductance L(f) is not limited to add in the flow (e.g., is not required), because it can be a relatively minor influence on operation frequency.

After deriving non-linear gain of the DCO (operation **530**), a linear interpolation could be used to reduce the frequency error (operation **540**), e.g., a center linear interpolation such as described herein with reference to equation (7) and

Because the gain of the DCO varies nonlinearly as a function of frequency, the value of {circumflex over (K)}_{DCO }used to compensate for such nonlinearity similarly can change as a function of frequency. Accordingly, in each different channel (frequency band) to be generated by the DCO, there may be a difference in the DCO gain for use in generating the linear formulas (F1, F2). In some embodiments, {circumflex over (K)}_{DCO }could be calculated by adopting the Δf into the equation. Alternatively, the different values of DCO gain, KDCO_{1 }. . . KDCO_{N }could be stored in a lookup table and selected based on the channel (frequency band) selection (operation **550**). For example, when the system receives a channel hopping (frequency changing/frequency locking) request, the DCO OTW calculator may use the lookup table to obtain the corresponding DCO gain to achieve the accurate output frequency and to generate an OTW based thereon (operation **560**) which is provided to the DCO (operation **570**) for use in generating a frequency by the DCO, for example, twice the desired frequency used by the application f_{ckv }as shown in

In certain embodiments herein, the predicted DCO gain can be close to the actual silicon behavior, which can significantly reduce the phase error in the ADPLL at the beginning of the phase locking process. So this technique could also help to improve the locking time and settling time.

**640** can be performed by PLL **600** (such as an APDLL). For example, at a first stage an arbitrary channel CH, here CH20 (Channel 20 at frequency f_{CH20}), is used as the center point, and a frequency control word (FCW) generated by PLL **600** at this channel frequency is input to first arithmetic module **610**, such as a first multiplier module. Additionally, a first compensation factor for the arbitrary channel is generated by PLL **600** and input to the first arithmetic module **610**. For example, the first compensation factor can be or include normalization of K_{DCO }(DCO gain) at the arbitrary channel CH, e.g., CH20. Note that K_{DCO,CH20 }can be known exactly, e.g., by calibrating the DCO. The PLL **600** can include modules for generating x(CH)=f_{R}/K_{DCO,CH }for CH20 (such as a gain estimation module and multiplexer) in a manner similar to that described herein with reference to **610** of PLL **600** can be configured so as to multiply FCW_{CH20 }by the normalizing gain multiplier x(CH20) to obtain a preliminary OTW, referred to in _{PRE}. As shown in **612** represents the DCO frequency without channel (cubic) compensation, line **614** represents the DCO frequency with channel compensation such as applied by arithmetic modules **610** and **620**, and line **616** represents the DCO frequency without binary error compensation such as described further below.

Arithmetic module **620** of PLL **600** illustrated in _{DCO }of the DCO at a given frequency (channel) can be expressed using equation (2) above, and the derivative of K_{DCO }at that frequency (channel) can be expressed using equation (3) above. In the example shown in _{DCO }for that channel can be expressed as:

and the normalizing gain multiplier x(CH) for that channel can be expressed as:

Arithmetic module **620** (e.g., multiplier module) can be configured to multiply OTW_{PRE }by x(CH) so as to generate output OTW_{HOP }for the particular hopped frequency at channel CH. As shown in **622**), as compared to a flat DCO gain without channel compensation (line **624**).

Arithmetic module **630** of PLL **600** illustrated in _{HOP }into binary code, depicted in graph **632** in **630** at an arbitrary channel, e.g., OTW_{CH }for CH20 or other suitable channel which can correspond to a normalized FCW, can be provided at an output of PLL **600** and then input to DCO **640**. After compensations such as performed by arithmetic modules **610**, **620**, and **630** of PLL **600**, the FCW input to DCO **640** will be substantially linear to the channel and the gain K_{DCO }will be substantially a constant over all channels. As shown in _{DCO }across the range of channels CH1 to CH40. Line **642** corresponds to DCO gain without channel compensation, line **646** corresponds to DCO gain without binary error compensation, and line **644** corresponds to the DCO gain after compensation.

In one exemplary implementation, the channels CH and CH20 are at double the Bluetooth band, and the output of DCO **640** is input to an arithmetic module **650** (e.g., a module configured to divide the DCO output by 2), so that PLL **600** and DCO can have relatively small modules as compared to the size of modules that may be needed to directly generate a DCO output within the Bluetooth band. The desired Bluetooth frequency is represented by f_{ckv}.

**1200**, e.g., the DCO circuit **120** or the DCO circuit **640**, for channel hopping in accordance with some embodiments. The DCO circuit **1200** of ^{2}. In one exemplary embodiment, the FM sw-cap bank can include 128 unit-weighted 16 kHz units so as to cover the ±250 kHz GFSK modulation range. Exemplary characteristics of components of the DCO hardware of

An OTW, e.g., the OTW in the DCO circuit **120** or the DCO circuit **640**, may comprise the inputs to the exemplary DCO circuit **1200** listed in the above table, including FM, Fine, COAR, PVT, HOP1, HOP2.

**1300** includes an operation of generating a normalized tuning word (operation **1310**). For example, ADPLL **110** illustrated in **600** illustrated in

Flow chart **1300** also includes an operation of generating a normalizing gain multiplier based on the reference frequency and an interpolation of the nonlinear function of frequency of the DCO gain (operation **1320**). For example, ADPLL **110** illustrated in **600** illustrated in _{DCO }and the reference frequency f_{R}, such as MUX **114** of ADPLL **110** illustrated in **620** of PLL **600** illustrated in

Flow chart **1300** also includes an operation of multiplying a normalized tuning word (NTW) by the normalizing gain multiplier x to obtain an oscillator tuning word (operation **1330**). For example, ADPLL **110** illustrated in **600** illustrated in **116** of ADPLL **110** illustrated in **620** of PLL **600** illustrated in

Flow chart **1300** also includes an operation of inputting the OTW to the DCO to cause the DCO to hop to a channel (operation **1340**). For example, ADPLL **110** illustrated in **600** illustrated in **120** illustrated in **640** illustrated in

Note that any suitable combination of modules provided herein suitably can be integrated with or coupled to the PLL or the DCO as appropriate. For example, any suitable combination of some or all of the gain estimation module, multiplexer, and arithmetic module mentioned with reference to

Under one exemplary aspect provided herein is derived a nonlinear DCO gain behavior as a cubic function of oscillation frequency. For example, in some embodiments the DCO gain will increase cubically as the oscillation frequency increases. The present disclosure, among other things, shows that this nonlinear behavior is predictable and independent of the process, voltage and temperature (PVT) variation. Based on derivation of this nonlinear (e.g., cubic) gain, for example based on a center linear interpolation, the free run DCO frequency could be substantially or exactly achieved without the need for measurement of the gain. Under another exemplary aspect provided herein, a linear approach to mimic the silicon behavior is not enough for the finest resolution application. For example, a interpolation from center of the curve could greatly improve the predicted error between silicon and simulation behavior. Under another exemplary aspect provided herein, the present approach could be implemented for any type of DCO in the fast locking applications. For example, it could help to minimize the phase error in the beginning of locking progress.

Accordingly, a derived cubic rule of the nonlinear behavior of DCO gain is provided. A linear interpolation could help to get an accurate DCO gain with less complexity of design. A center linear interpolation technique further can improve the DCO gain. A “just in time” DCO gain calibration flow can include use of a calculator output a final normalized DCO gain, a multiplexer (mux) configured to select the right normalized DCO gain, and a multiplier configured to change the NTW to OTW when the calculation is completed. Typically, “calibration” implies that a series of steps that have been performed prior to the operation of hardware. However, the term “just in time calibration” is used here to describe a calibration-type process that can be performed during the operation of the hardware and “just in time” for the outputs to be generated. A completed DCO gain calibration flow for channel hopping (frequency band selection) can include use of a derived cubic formula, a inductance effect to the frequency, a non-linear DCO gain curve, a center linear interpolation equation, a lookup table, an OTW calculator, and a DCO. It could be adopted in any frequency hopping, fast locking and frequency prediction application. The calibration could be combined with any other type of DCO gain/frequency calibration, for example linear calibration (binary errors). The DCO design could be any type of LC-tank oscillator, for example inductors and/or transformers. The capacitor bank is not limited to the example and the combination is not limited to binary controlled or unit-weighted. The DCO gain after compensation can in be the same overall frequency range. The DCO frequency curve can be linear to the oscillation frequency after compensation.

In one embodiment, a method is provided for controlling a digitally controlled oscillator (DCO). A first normalized tuning word (NTW) corresponding to a first channel of the plurality of channels is generated. A first normalizing gain multiplier X is generated based on the nonlinear function of frequency, on an estimate of the nonlinear function of frequency, at a first frequency corresponding to the first channel. The first NTW is multiplied by the first X to obtain a first oscillator tuning word (OTW). The first OTW is input to the DCO to cause the DCO to hop to the first channel.

In another embodiment, a system is provided for controlling a digitally controlled oscillator (DCO). The system comprises: a phase-locked loop (PLL) is configured to generate a plurality of normalized tuning words, each NTW corresponding to a respective channel of the plurality of channels; a DCO having a gain that is a nonlinear function of frequency; a multiplexer configured to generate a plurality of normalizing gain multipliers X, each X being based on a reference frequency f_{R }and the nonlinear function of frequency, or an estimate of the nonlinear function of frequency, at a respective frequency of a channel of the plurality of channels; and an arithmetic module configured to generate a plurality of oscillator tuning words (OTWs) based on a respective NTW and a respective X, wherein the DCO hops among the channels based on respective OTWs of the plurality of OTWS.

In yet another embodiment, a normalized digitally controlled oscillator (nDCO) comprises: a digitally controlled oscillator (DCO) having a gain that is a nonlinear function of frequency; a multiplexer configured to generate a plurality of normalizing gain multipliers X, each X based on a reference frequency f_{R }and the nonlinear function of frequency, or an estimate of the nonlinear function of frequency, at a respective frequency of a plurality of frequencies; and an arithmetic module configured to generate a respective oscillator tuning word (OTW) based on each X, the DCO generating a respective frequency based on each OTW.

The foregoing outlines features of several embodiments so that those skilled in the art may better understand the aspects of the present disclosure. Those skilled in the art should appreciate that they may readily use the present disclosure as a basis for designing or modifying other processes and structures for carrying out the same purposes and/or achieving the same advantages of the embodiments introduced herein. Those skilled in the art should also realize that such equivalent constructions do not depart from the spirit and scope of the present disclosure, and that they may make various changes, substitutions, and alterations herein without departing from the spirit and scope of the present disclosure.

This written description and the following claims may include terms, such as “on,” that are used for descriptive purposes only and are not to be construed as limiting. The embodiments of a system, PLL, and/or DCO described herein can be manufactured, used, or shipped in a number of configurations.

## Claims

1. A method for controlling a digitally controlled oscillator (DCO), the method comprising:

- generating a first normalized tuning word (NTW) corresponding to a first channel of the plurality of channels;

- generating a first normalizing gain multiplier X based on the nonlinear function of frequency or an estimate of the nonlinear function of frequency, at a first frequency corresponding to the first channel;

- multiplying the first NTW by the first normalizing gain multiplier X to obtain a first oscillator tuning word (OTW); and

- inputting the first OTW to the DCO to cause the DCO to hop to the first channel.

2. The method of claim 1, wherein the first normalizing gain multiplier X is also generated based on a reference frequency fR.

3. The method of claim 2, wherein the reference frequency fR is the frequency of a second channel of the plurality of channels.

4. The method of claim 3, wherein the second channel is a central channel of the plurality of channels.

5. The method of claim 2, further comprising:

- generating a second NTW corresponding to a first channel of the plurality of channels;

- generating a second normalizing gain multiplier X based on the nonlinear function of frequency at a second frequency corresponding to a second channel or based on an estimate of the nonlinear function of frequency at the second frequency;

- multiplying the second NTW by the second normalizing gain multiplier X to obtain a second OTW; and

- inputting the second OTW to the DCO to cause the DCO to hop to the second channel.

6. The method of claim 1, wherein at least part of the method is performed in an all-digital phase locked loop (ADPLL).

7. The method of claim 1, wherein the first normalizing gain multiplier X is based on a linear interpolation of the nonlinear function of frequency at the first frequency.

8. The method of claim 7, wherein the interpolation is based on a center point of the nonlinear function of frequency.

9. The method of claim 7, wherein the nonlinear function of frequency is a cubic function of frequency.

10. The method of claim 7, wherein the interpolation is based on: y = y 0 + dy 0 df ( f - f 0 ) = y 0 ( 1 - 3 f - f 0 f 0 ) in which y corresponds to a change in DCO gain variation, f0 corresponds to a frequency at which the DCO gain variation is 100%, and y0 corresponds to the value of the change in DCO gain variation at which the gain variation is 100%.

11. A system for controlling a digitally controlled oscillator, the system comprising:

- a phase-locked loop (PLL) configured to generate a plurality of normalized tuning words, each NTW corresponding to a respective channel of the plurality of channels;

- a digitally controlled oscillator (DCO) having a gain that is a nonlinear function of frequency;

- a multiplexer configured to generate a plurality of normalizing gain multipliers X, each normalizing gain multiplier X being based on a reference frequency fR and the nonlinear function of frequency, or an estimate of the nonlinear function of frequency, at a respective frequency of a channel of the plurality of channels; and

- an arithmetic module configured to generate a plurality of oscillator tuning words (OTWs) based on a respective NTW and a respective normalizing gain multiplier X,

- wherein the DCO hops among the channels based on respective OTWs of the plurality of OTWs.

12. The system of claim 11, wherein the PLL comprises an all-digital PLL.

13. The system of claim 11, wherein at least one of the multiplexer and the arithmetic module are provided as part of a normalized DCO comprising the DCO.

14. The system of claim 11, wherein each normalizing gain multiplier X is based on a linear interpolation of the nonlinear function of frequency.

15. The system of claim 14, wherein the interpolation is based on a center point of the nonlinear function of frequency.

16. The system of claim 14, wherein the nonlinear function of frequency is a cubic function of frequency.

17. The system of claim 14, wherein the interpolation is based on: y = y 0 + dy 0 df ( f - f 0 ) = y 0 ( 1 - 3 f - f 0 f 0 ) in which y corresponds to a change in DCO gain variation, f0 corresponds to a frequency at which the DCO gain variation is 100%, and y0 corresponds to the value of the change in DCO gain variation at which the gain variation is 100%.

18. A normalized digitally controlled oscillator (nDCO), comprising:

- a digitally controlled oscillator (DCO) having a gain that is a nonlinear function of frequency;

- a multiplexer configured to generate a plurality of normalizing gain multipliers, each normalizing gain multiplier based on a reference frequency fR and the nonlinear function of frequency, or an estimate of the nonlinear function of frequency, at a respective frequency of a plurality of frequencies; and

- an arithmetic module configured to generate a respective oscillator tuning word (OTW) based on each normalizing gain multiplier,

- the DCO generating a respective frequency based on each OTW.

19. The nDCO of claim 18, wherein each normalizing gain multiplier is based on a linear interpolation of the nonlinear function of frequency.

20. The nDCO of claim 19, wherein the linear interpolation is based on a center point of a frequency range.

**Patent History**

**Publication number**: 20190068201

**Type:**Application

**Filed**: Aug 21, 2018

**Publication Date**: Feb 28, 2019

**Patent Grant number**: 10659057

**Inventors**: Chao Chieh Li (Hsinchu City), Min-Shueh Yuan (Taipei), Robert Bogdan Staszewski (Dublin), Chia-Chun Liao (Taipei City)

**Application Number**: 16/106,163

**Classifications**

**International Classification**: H03L 7/08 (20060101); H03L 7/099 (20060101); H03L 7/093 (20060101);