Title of Invention	REFERENCE OSCILLATOR WITH AUTOMATIC COMPENSATION FOR AGING AND TEMPERATURE .
Abstract	A receiver 10, uses an adaptive algorithm 48 to tune a low-cost crystal oscillator 24 according to a temperature compensation profile 50, so as to produce a precision master reference frequency 26 despite temperature 49, initial tolerance, and aging effects. An automatic frequency control system 20 also tunes the crystal oscillator 24. The adaptive algorithm 48 adjusts the temperature compensation profile 50 for the crystal oscillator 24 according to the adjustments made by the automatic frequency control 20 should a received signal's 13 quality factor exceed the quality factor 55 associated with the temperature compensation profile 50.

Title of Invention

REFERENCE OSCILLATOR WITH AUTOMATIC COMPENSATION FOR AGING AND TEMPERATURE .

Abstract

A receiver 10, uses an adaptive algorithm 48 to tune a low-cost crystal oscillator 24 according to a temperature compensation profile 50, so as to produce a precision master reference frequency 26 despite temperature 49, initial tolerance, and aging effects. An automatic frequency control system 20 also tunes the crystal oscillator 24. The adaptive algorithm 48 adjusts the temperature compensation profile 50 for the crystal oscillator 24 according to the adjustments made by the automatic frequency control 20 should a received signal's 13 quality factor exceed the quality factor 55 associated with the temperature compensation profile 50.

Full Text	REFERENCE OSCILLATOR WITH AUTOMATIC COMOENSATION FOR AGING AND TEMPERATURE BACKGROUND Field of the Invention Tha Present invention relates to electronic circuitry, and more specifically to electronic circuitry tuning a reference oscillator. Description of Related Art Because the usable electromagnetic spectrum is a limited resource, governmental agencies regulate its use and exploitation. For example, in the U.S. the Federal Communication Commission (FCC) divides the usable electromagnetic spectrum into frequency ranges, or bands. Each band may be assigned to a, specific function, or may be reserved for future use. Since only a few bands are allocated to a specific function, such as portable communication, it is important that the band be efficiently utilized. Thus, the frequency bands assigned to wireless communications should accommodate many wireless users. To allow each frequency band to accommodate so many users, a wireless communication device typically uses a reference oscillator to accurately find the carrier frequency used by its particular base station. in a particular use, the wireless device is a mobile handset configured to communicate with one or more base stations. To find the carrier frequency, reference oscillators operating in a wireless communications device typically generate a highly accurate frequency reference. Because low-cost oscillators are. prone to substantial initial error, large individual variations, and degraded performance over time, they have not been used to generate such an accurate frequency reference. Accordingly, known wireless communication devices use more expensive high-precision reference oscillators, and associated precision circuitry, to obtain the necessary accuracy and repeatability. The precision circuitry in the wireless communication device or mobile handset generally includes a voltage-controlled temperature-compensated crystal oscillator (VC-TCXO). The VC-TCXO provides a reference frequency, which is used by the handset to , find, or lock onto the carrier frequency. Depending upon the handset's current temperature, the VC-TCXO adjusts its tuning to maintain a constant reference frequency output. Such VC-TCXO's, however, are relatively expensive, adding significantly to the cost of a handset. Moreover, despite a VC-TCXO's expense, the reference frequency output eventually drifcs with age. In addition, VC-TCXOs typically have an indeterminate initial accuracy resulting from manufacturing tolerances that complicates a handset's design. Further, conventional temperature compensation circuitry for a VC-TCXO has degraded accuracy at temperature extremes. Despite this degraded accuracy, users expect a handset to provide reliable communication that mimics or betters that of a landline phone. To meet expectations, the handset must consistently establish communication in an acceptable time period and do so over a receiver's wide temperature range. For example, a handset is expected to operate in sub-Arctic conditions, and also operate after being subject to the torturous heat of a car dashboard in summer. As a result, it is desirable that a handset VC-TCXO should typically maintain a frequency accuracy of better than about +/- 2 parts per million (ppm) over the temperature range -30° C to +85° C. As generally described earlier, a VC-TCXO is not stable over time. VC-TCXO modules may drift at approximately 1 ppm per year, yet are often expected to operate for many years. The typical AFC (Automatic Frequency Control) loop used to tune the VC-TCXO modules can cope with offsets of no greater than about +/- 4 ppn with respect to its desired reference frequency. Thus, in just a few years, such VC-TCXO modules will have drifted to the point that they may not be able to assist in acquiring or capturing a base station's carrier. Systems have been developed to account for the aging of a VC-TCXO. For example, U.S. Pat. No. 6,064,270 discloses a handset having a VC-TCXO that, should the base station's beacon carrier escape capture, will perform a random search for the carrier by adjusting the VC-TCXO's reference frequency +/- 4 ppm (or some other suitable amount). If the carrier is acquired, the offset is remembered for future use. Although this system may provide acceptable age compensation when coupled with an expensive VC-TCXO, it is not practically implemented with less expensive oscillators having more substantial drift and corresponding large offsets. A system having such large offsets that searched blindly for the carrier frequency would produce delays intolerable to the user. In addition, the search accounts only for the gross offsets encountered during aging and does not address temperature effects, necessitating the continued use of an expensive VC-TCXO. Accordingly, there is a need in the art for an improved receiver capable of providing a reference frequency that is robust to temperature and aging effects without the use of an expensive VC-TCXO. SUMMARY It is desirable to accurately generate a reference frequency signal using a low-cost oscillator. It is therefore an object of the invention to enable low-cost oscillators to accurately generate reference signals despite initial tolerance, temperature and aging effects. It is a further object of the invention to enable such oscillators to adaptively adjust to temperature and aging effects by recalibrating accordingly. To overcome the disadvantage in the prior art and meet the objectives of this invention, a computational device implements an adaptive algorithm. The adaptive algorithm adjusts stored frequency correction information to compensate a reference oscillator. During this compensation, the reference oscillator adjusts its master reference signal responsive to the stored frequency correction information. Because this algorithm adaptively accounts for initial frequency error, aqing, and temperature effects, a low-cost VCXO may be used as the reference oscillator. In accordance v/ith one aspect of the invention, the frequency correction information comprises a plurality of predetermined frequency correction factors spanning a temperature range. The computational device may receive current temperature information, which enables the computational device to derive a current frequency correction factor from Che frequency correction factors. The reference oscillator tunes or adjusts the frequency of its reference signal according to the current frequency correction factor. Should the reference signal frequency be offset from a desired frequency, the computational device may adjust one or more of the stored frequency correction factors according to the frequency offset. In such a manner, the stored correction factors are adjusted as needed tc provide superior frequency accuracy. Quality factors may be associated v/ith the frequency correction factors. In this embodiment, when a current freauency correction factor is calculated, a currant quality factor is also determined from the quality factors. A received signal's quality factor is compared to the current quality factor before the temperature correction factors arc; updated. In such a manner, an older, but higher quality temperature compensation factor would not be replaced by a newer, but lower quality temperature compensation factor. Advantageously, the present invention enables a communications device to use a low-cost oscillator as a reference oscillator, avoiding the cost of the VC-TCXO. Because the invention provides for adaptive adjustment of the frequency correction factors, the low cost oscillator also systematically compensates for oscillator aging. Further aspects and features of the invention are set forth in the following description together with the accompanying drawings. ACCOMPANYING BRIEF DESCRIPTION OF THE ACOMPANYING DRAWINGS FIG. 1 illustrates a receiver having an adaPtive VC-TCXO correction loop according to one embodiment of the invention. FIG. 2 illustrates the temperature correction profile for a VCXO. FIG. 3 illustrates a portion of the profile of Figure 2 being calibrated according to one embodiment of the invention. FIG. 4 is a flowchart of a method for providing a reference signal in accordance wich one embodiment of the invention. FIG. 5 is a flowchart of a method for calibrating a communication device in accordance with one embodiment of the invention. Use of the same reference symbols in different figures indicates similar or identical items. DETAILED DESCRIPTION Referring now to Figure 1, a block diagram is shown of a receiver 10 using a low-cost reference oscillator 24 . Despite being low cost, the reference oscillator 24 generates a precision reference frequency-signal 26 of sufficient accuracy for wireless communication. Since the receiver 10 does not reauire an expensive VC-TCXO, the receiver 10 is desirable for use in a wireless communications device such as a mobile handset. In the illustrated example, the receiver 10 is a super heterodyne receiver 10 that adaptively accounts for initial frequency error, temperature effects and aging drift of the low precision oscillator 24. .More specifically, the receiver 10 incorporates a logic block 4 3 to remember, predict, and adapt frequency correction information. The frequency correction information may be stored in a memory 44 in analog or digital form. In digital form, the frequency correction information may be sampled to create a plurality of temperature correction factors. These correction factors may be responsive to temperature fluctuations, initial error conditions, and aging effects, for example. Accordingly, the low cost oscillator 24 provides a sufficiently repeatable and accurate frequency signal for use in the local oscillator block 15. Before proceeding to discuss the logic block 43, the receiver 10 will be generally described. The receiver 10 includes an antenna 12 that receives a transmitted signal such as a base station carrier signal. The antenna signal 13 is down converted with a local oscillator (LO) signal 11 in a mixer 14 to form an intermediate frequency (IF) signal 34. The generation of the LO signal 11 by the local oscillator block 15 will be described further herein. An IF gain and selection block 16 amplifies and filters the I? signal 34 produced by the mixer 14. A freauencv discriminator 13 frequency demodulates the amplified and filtered IF signal to produce a demodulated signal 19. The demodulated signal 19 is low-pass filtered by an automatic frequency control (AFC) loop filter 20 to form an AFC signal 21. The AFC signal 21 couples to the reference oscillator 24 to tune the frequency of the reference signal 26. The reference signal couples to a phase-locked loop (PLL) 28 that is part of the local oscillator block 15. A loop filter 30 then couples the PLL output to a voltage-controlled oscillator (VCO) 32 that produces the LO signal 11. Those of ordinary skill in the art will appreciate that the receiver 10 of Figure 1 is an idealized version for use in a wireless handset . In an actual handset receiver, multiple frequency translations may require additional mixing stages. Furthermore, AFC loop filter 2 0 may include I and Q channels (not illustrated) . The I and Q signals so produced may be. digitized and processed before being filtered by the AFC loon filter 20. The receiver's logic block 4 3 will new be described. The logic block 43 includes a computation engine 46, which implements an adaptive algorithm 4 3 that self-learns and accounts for initial error conditions and temperature and aging effects of the reference oscillator, enabling the receiver 10 to capture signals without requiring the use of precision TC-VCXOs. The computation engine 4 6 provides a current correction factor 40 according to the algorithm 48. .Those of ordinary skill in the art will appreciate that algorithm 4 8, unlike the remaining elements in Figure 1, is not structure but is shown symbollically to indicate its relationship to the receiver 10. The correction factor 40 may compensate for initial tolerance error, aging effects, and current temperature, for example. The computation engine 4 6 may be implemented using any suitable logic device that may be programmed to produce this factor 40, including a microprocessor (not illustrated) or a state machine (not illustrated). Logic block 4 2 also preferably includes a memory device 44. It will be appreciated that the memory device 44 mav be alternatively integrated with the computational device 46. The current temperature correction factor 4 0 is combined with the AFC signal 21 in summer 23 to produce a reference oscillator correction signal 22. This reference oscillator correction signal 22 couples to the tO oscillator block 15 to adjust the LO signal 11. In this manner, the receiver 10 has both "coarse" and "fine" control of its master reference signal 26 and, ultimately, the LO signal. The coarse control is provided by the logic block 4 3 through the current temperature correction factor 40. The coarse control is desirably operated so that the fine control provided by the AFC loop can capture the transmitted signal, typically requiring the coarse control to be within about +/- 4 ppm of the required frequency. In this fashion, the coarse control provided by the logic block 43 effectively enables the low cost VCXO to provide a level of performance conventionally provided by the more expensive precision TC-VCXO. However, unlike receivers using conventional TC-VCXOs, the receiver 10 continually recalibrates the low cost VCXO as mandated by changes in temperature and aging. The adaptive algorithm 4B uses a predetermined frequency offset profile specific to whatever VCXO is used as the reference oscillator 24. This profile provides the frequency conpensation necessary to adjust, the reference oscillator 24 over its operating temperature range. Thus, the frequency offset profile may also be denoted as the temperature compensation profile. Referring to FIG. 2, an illustrative profile 50 for a low-cost oscillator is shown. Several techniques may be employed to derive the profile 50 to account for the component-to-component variances of low-cost VCXOs. In one embodiment, a given VCXO would be individually tested over the expected receiver's temperature range to calibrate the profile 50 exactly to the VCXO's particular behavior. Such individual testing could, however, increase manufacturing costs and reduce the savings introduced by using a low-cost VCXO in place of a precision TC-VCXO. Thus, in another embodiment, rather than test a given VCXO over the entire expected operating range, each VCXO would be initially calibrated only at a single ambient temperature. Prior to this calibration, the expected profile may be determined from manufacturer data, typically representing the averaging of many individual VCXOs. The expected profile is, as the name implies, only expected and may be departed from considerably by a given individual VCXO. For example, 10 pprn variance:? can be observed from the expected profile predicted by a given VCXO manufacturer. For a given VCXO, the manufacturer will generally publish a data sheet indicating an expected frequency offset profile. averaging the test results of several VCXOs from a aiven batch over the expected receiver's temperature range (typically -30° to +120' C) . It will be appreciated that, given the large component-to-component tolerances expected for a low-cost VCXO, the actual temperature compensation profile may vary widely from individual VCXO to VCXO, even within the same manufacturer batch. Thus, it is desirable that the profile 50 be individually calibrated for a given VCXO because of the wide tolerance. Given that the receiver will usually be at ambient temperature and this temperature is normal for a manufacturing plant, the individual calibration may be performed at ambient temperature. It will be appreciated that another temperature may be selected for calibration purposes. Prior to calibration, data indicative of the profile 50 may be placed in memory by storing samples 52 of the profile in a non-volatile memory 44. Each sample 52 corresponds to a discrete temperature and its associated frequency offset/temperature compensation value. Accordingly, the samples 52 represent initial temperature correction factors. The temperature correction factors 52 or the offset profile 50 may be taken at regular temperature intervals as shown. Alternatively, the profile 50 may be sampled more in regions where the slope is changing and less in regions of relatively little slope change. A thermistor (not illustrated in Figure 1) or another suitable temperature-sensing device provides the receiver's temperature 49. The logic block 43 may correlate a subset of the temperature correction factors 52 with the receiver's temperature 49 to provide a current correction factor signal 40. In operation, if the receiver's temperature matches the temperature corresponding to or.e of the temperature 'correction factors 52, the "correlation" would comprise merely using the frequency offset of the matched factor 52 as the current correction factor signal 40. In such a case, the subset of samples used for correlation has just one member. More typically, however, the receiver's temperature will not correspond to any of the temperatures used for the temperature correction factors 52. One simple way to correlate the receiver's temperature in this case is to assume a linear slope between adjacent samples immediately framing the receiver's temperature. For example, suppose the receiver's temperature is 90' C and the two closest temperature correction factors give a value of +10 ppm at 100° C and +5 ppm at 80° C. Assuming a linear slope between these two temperature corrections factors would give the current correction factor as 7.5 ppm corresponding to the receiver's temperature of 90° C. In such a case the subset of samples used for correlation has two members. Alternatively, three or more samples may be used to predict a quadratic or higher order slope for the profile 50 at the receiver's temperature. Given this expected profile 50, an individual VCXO may be calibrated at ambient temperature v.'ith a powerful transmitted signal to ensure that the receiver will capture the transmitted signal with its AFC loop. As discussed earlier, a conventional AFC loop can only handle approximately about +/- 4ppm in frequency offset at typical operating signal-to-noise ratios. However, these conventional AFC loops can accommodate larger offsets such as -/- 10 ppm should the signal-to-noise ratio be unusually large. Such a situation occurs when a mobile user is close to the transmitting base station or, in this case, receiving a beacon frequency in a laboratory or manufacturing setting. Thus, even if a VCXO being calibrated is off 1Oppm from its expected profile at ambient temperature, the AFC loop will be able to capture the signal given a high enough signal-to-noise ratio - an occurrence easily established in a controlled setting. To begin the calibration, the logic block 43 correlates the receiver's temperature 49 with a subset of the temperature correction factors 52 in the manner just described to produce a current temperature correction factor 40. The reference oscillator 24 tunes accordingly and the AFC 100p filter 20, if necessary, produces an AFC signal 21 to capture the transmitted signal. For example, if the reference signal's frequency is too high, the AFC signal 21 tunes . the reference oscillator 24 to lower the frequency of the reference signal 26. Conversely, if the reference signal' frequency is too low, the AFC signal 21 tunes the reference oscillator 24 to increases the frequency of the reference signal 26. In either case, the reference signal 26 has a frequency offset from that of a desired frequency as determined by the base station's carrier frequency. The resulting AFC signal 21 corresponding to this frequency offset may be used to recalibrate the expected profile to produce an adjusted profile. In other words, the AFC signal 21 correlates with a certain ppm frequency offset. This frequency offset, whether positive or negative, may be applied to calibrate the subset of the temperature correction factors within the expected profile. For example, if the AFC signal 21 used to capture the manufacturing signal during calibration at ambient temperature correlates to a +10 ppm frequency offset, the subset of samples used for correlation may be calibrated by adding 10 ppm to each sample in the subset. Alternatively, the subset may be calibrated linearly according to their contribution to the initial correlation as seen in Figure 3. Here, the receiver's temperature is 63° C thus requiring an 80% contribution from the temperature correction factor at 60* and a 20% contribution from the temperature correction factor at 75°, assuming a linear correlation. If the resulting AFC signal 21 corresponds to a +10 ppm shift, then the temperature correction factor at 60° would be shifted up 0 ppm and the temperature correction factor at 75" would be shifted up 2 ppm to calibrate the subset. In addition, should this initial calibration result in a sufficiently large shift, such as the 8 ppm discussed with respect to Figure 3, all of the remaining temperature correction factors may be shifted accordingly, for example, by shifting 1 ppm. As discussed, the value of the AFC signal 21 correlates to a frequency offset. The computation engine 46 receives the AFC signal 21 and may correlate this signal to the corresponding frequency offset using a lookup table. Alternatively, the current temperature correction factor 40 may be adjusted by the computation engine 46 until the AFC signal 21 is minimized or "zeroed out." At this point, the current temperature correction factor 4 0 accounts for and eliminates the frequency offset. Assuming that ambient temperature is used for this initial calibration, the profile 50 may have a temperature correction factor 52 at. ambient temperature to eliminate any requirement to correlate with more than one temperature correction factor 52 to derive the current temperature correction factor 40 (the subset of factors used would have just one member) . To distinguish the factors 52 that are calibrated accurately from those that are not, each factor 52 may be assigned a quality factor 5 5 having an arbitrary scale such as from 1 to 100. The quality factor 55, as the name implies, relates to the quality of the received signal used to calibrate the receiver as determined, for example, by the signal-to-noise ratio. Alternatively, the quality factor may relate to an automatic gain control signal as determined by the received signal, the carrier-to-noise ratio, or another suitable factor relating to the quality of the received signal. For the profile shown in Figure 2, the temperature correction factor for 30° is associated with a quality factor of 95, the temperature correction factor for 45° is associated with a quality factor of 20, and the temperature correction factor for 60" is associated with a quality factor of 95. Before the initial calibration, all the temperature- correction factors 52 would be associated with a low quality factor, e.g., a value of 50. After the initial calibration, the subset of temperature correction factors 52 that are calibrated as discussed with respect to Figure 3 are assigned a high quality factor such as a value of 100 because the transmitted signal in this laboratory setting is such that a very high signal - to-noise ratio is assured v/ithin the receiver. Note that in a situation such as illustrated in Figure 3, the quality factors associated with the temoerature correction factors 52 within the subset may be updated proportionally to their contribution to the current temperature correction factor 40. Alternatively, should the subset consist of just one member corresponding to the calibration temperature, just this factor would receive the high quality factor. Because the initial calibration typically occurs at ambient temperature, the receiver's temperature will tend to change gradually from ambient if the receiver is moved, for example, from indoors to an outdoors location during use by a consumer. As the receiver heats or cools, the receiver's temperature will correlate with temperature correction factors 52 that were not initially calibrated. Thus, the receiver "self-learns" or adaptively adjusts these factors 52, using the transmitted signal as a frequency reference. As in the initial calibration, the receiver's temperature is correlated with a subset of the temperature correction factors 52 to produce a current temperature correction factor 40. The transmitted signal is then captured by the receiver's AFC loop. Should the AFC signal 21 indicate a frequency offset is present, the subset is calibrated in the manner just described with respect to Figure 2. In a wireless receiver, the reference oscillator thus becomes locked to the base station's transmitted frequency. Note, however, that because this calibration is occurring outside the initial calibration conditions on a real-world transmitted signal, there is a danger that the receiver could calibrate improperly on a noisy transmitted signal. Thus, the computation engine 46 may respond to a valid signal indication 51. In a mobile handset, the computation engine 4 6 may use the identification of the pilot signal or some other suitable traffic indication to serve as the valid signal indication 51. Having assured itself that the received signal is valid, the computation engine may then calculate the-quality factor of the received signal. For example, the logic block 4 3 may have a lookup table that correlates the current carrier-to-noise ratio of the received signal to a quality factor on the scale of 1 Co 100. This received quality factor is compared to the quality factor as given by the- subset of temperature correction factors 52 used to derive the current temperature correction factor 40. If the received quality factor is greater, the subset is calibrated as discussed with respect to the initial calibration. However, because the received quality factor will not necessarily equal 100 as in the initial calibration, the amount by which the subset would be updated according to the frequency offset may be reduced proportional to the received quality factor. For example, should the frequency offset correlate to a 10 ppm update of the subset and the received quality factor is 90, then just 90% of the 10 ppm may be applied to the subset. In this fashion, the stored temperature correction factors 52 become progressively more accurate as the receiver adapts to shifts in receiver'- temperature. It will be appreciated that other methods may be used to proportionally apply the frequency offset As the reference oscillator 24 ages, its expected profile 50 may change as well. To compensate for this aging effect, the quality factors associated with the temperature correction factors 52 may be systematically decreased over time. For example, should the quality factors correspond to a scale of 1 to 100, the quality factors could be reduced by 5 every three months. In this fashion, the temoerature correction factors 52 will be more readily recalibrated as time passes so as to account for aging effects. The rate at which the duality factors are reduced to account for this aging depends upon the characteristics of the individual reference oscillator 24 within a given receiver. Note that once calibrated, even low-cost VCXOs exhibit little hysteresis such that repeated use of the receiver at the same temperature with the same current temperature correction factor 40 will result in signal capture with little additional offset needed from the AFC correction signal 21. Thus, the adjustment of the quality factors to account for aging may be very gradual. Because the receiver may experience extremes in temperature without prior calibration, there is the possibility that the "fine" control provided by the AFC loop will be unable to capture the transmitted signal, despite the "coarse" control provided by the logic block 4 3 via the current temperature correction factor 40. In such a situation, the logic block 43 may be configured to search for Che transmitted signal by incrementing or decrementing the current temperature correction factor 40. The required increment will depend upon the amount of fine control provided by the conventional AFC loop. For example, if this AFC loop can accommodate frequency errors of +/- 4 ppm in the master reference signal 26, the increment or decrement should be not greater than 4 ppm. Rather than search blindly, the logic block 43 stores the cumulative record of positive or negative calibrations of the temperature correction factors 52. If the receiver's temperature correlates with a certain temperature correction factor that has been recalibrated with positive increments in ppm several times, it would be logical to then assume that the required increment in the current temperature correction factor 40 should be positive rather than negative. In addition, the logic block 43 can gauge the scale of the increment by this history. For example, should the recent recalibrations have all been significant (such as greater than 5 ppm), the usual increment in the current temperature correction factor could be increased from, say, 4 ppm to 5 ppm. The logic block 43 may store other parameters relating to the history of the calibrations for diagnostic ourposes. These parameters include the time of each calibration and the number of times each temperature correction factor 52 has been updated. Referring now to FIG. 4, a method 100 of calibrating the stored frequency correction information is shown. The method 100 is preferably implemented en a communications device, for example, a mobile wireless handset. It will be appreciated that method 100 may be used on a wide variety of communications equipment. Method 100 initially stores a frequency offset or temperature compensation profile in a memory in the communications device, as shown in block 102. As described earlier, the frequency offset profile may be in the form of discrete temperature compensation factors or be in analog form. Alternatively, the profile may be approximated by a polynomial function and the necessary polynomial coefficients stored in memory. Optionally, a quality factor may be associated with the correction information as previously described. As shown in block 112.. the quality factor may also be stored in the communications device. A current temperature signal is provided in block 103. It will be appreciated the current temperature may be detected by any of several well known devices, such as a thermister, and the temperature signal may take one of several known forms such as a voltage or current signal, Responsive to the current temperature, a current correction signal corresponding to the current temperature is retrieved from the stored frequency correction information in block 104. Should the frequency correction information be stored as a plurality of temperature correction factors, the current temperature is correlated with a subset of the factors as described previously The correction signal tunes a reference oscillator in block 106. Ideally, the tuning performed in block 106 tunes the oscillator to precisely the desired frequency. More often, however, a frequency error in the tuning of the reference oscillator may be detected in block 103. Such an error may be detected and measured in an AFC loop, for example. The determined frequency error is used in block 110 to generate undated frequency correction information for the current temperature. In one implementation, the method 100 also determines the quality of the signal received according to the tuning of the reference oscillator as shown in block 114. For example, this received quality factor may relate to the strength or signal to noise ratio of the received signal. It will be appreciated that other factors may be used to determine the received quality factor. In block 116, the received quality factor determined for the updated correction information may be compared to the quality factor associated with the stored frequency correction information used to derive the current correction signal. Responsive to this comparison, a decision can be made whether to proceed to store the updated correction information in block 12 0. For example, if the stored frequency correction information used to derive the current correction signal has an associated quality factor of 70, and the received signal has only an associated quality factor of SO, then the updated frequency correction information will not replace the better, although older, stored frequency correction information. On the other hand, when the received quality factor is better than the quality factor associated with the stored correction information used to derive the current correction signal, then the updated frequency correction information is stored in block 120. Also, the received quality factor is also associated with the undated correction information and stored, as shown in block 122. It will also be appreciated'that other factors may be used to determine when the updated correction information will be stored. Such fetctors may be used to reduce the risk that high quality frequency correction information will be replaced by false or low quality frequency correction information. For example, block 118 shows that a validation signal may be determine whether the updated correction information is stored. This validation signal indicates whether a valid signal is being received. The validation signal may be generated, for example, when a predetermined signal is positively received and identified. In one specific example, the validation signal is generated when a pilot signal is confirmed. It will be appreciate:: that the generation cf the validation signal may be responsive to other factors. To account for aging effects, block 124 shows that the stored quality factors may be slowly degraded over time. In such a manner, even frequency correction information which had the highest quality factor may be subject to adjustment after a period of time. Referring now to FIG. 5, a method 130 for initial calibration of a communications device configured to implement the method 100 of Figure 4 is shown. In block 132, the initial frequency correction information for a reference oscillator is determined. As discussed Dreviously, this determination may be based on manufacturer provided information. In block 134, this initial frequency correction information is stored in memory. Because of the large individual tolerances encountered with low-cost reference oscillators, this initial frequency correction information is only approximate and does not necessarily correspond to the behavior of a given individual reference oscillator. Thus, this initial frequency correction information should be calibrated according to the individual behavior of the reference oscillator being tested. To do so, a strong calibration signal is generated and the reference oscillator tuned according to the initial frequency correction information in block 136. As used herein, this "strong" calibration signal denotes a signal powerful enough to enable a conventional AFC loop to capture this calibration signal even if the reference. oscillator has a large frequency offset;, e.g., +/- 10 pom. To eliminate this frequency offset, undated or calibrated frequency correction information is generated in block 140. Finally, in block 142 this undated frequency correction information is stored. A quality factor may be associated with this stored information in block 144. To better appreciate the contribution of the receiver 10, the operation of a conventional receiver will be briefly discussed. It will, however, be understood that conventional receiver operation is well known. In conventional prior art receivers, an AFC signal typically is directly coupled to an expensive high performance reference (or crystal) oscillator to provide an LO signal. In response to the AFC signal, the high cost reference oscillator adjusts the frequency of a master reference signal. The master reference signal typically coucles to a phase-locked loop (PLL) that is part of the local oscillator control block. A loop filter then generally couples the PLL to a voltage-controlled oscillator (VCO) that produces the LO signal. The LO signal mixes with the received signal from the antenna to produce an IF signal. This IF signal will be demodulated and passed through an AFC low pass filter, and feed back into the reference oscillator. Thereby a conventional AFC "loop" is formed. Conventional AFC low pass filters are designed go that the AFC control provided is fast-acting and can correct +/- 4 ppm frequency inaccuracies in the master reference signal. Should the high performance reference oscillator produce a master reference signal precisely at the required frequency, the AFC loop would never need to adjust the reference oscillator. However, even precision reference oscillators will tend to be slightly inaccurate and require adjustment by the AFC loop. As noted in the background section, conventional AFC loops could correct for about a +/- 4 ppm deviation by the reference oscillator yet still acquire the carrier signal at typical operating "signal-to-noise ratios. Conventional control of the reference oscillator by using a precision TC-VCXO, however, could not adaptively adjust for temperature and aging effects experience by even the expensive, high performance reference oscillator. Advantageously, the receiver 10 enables a low cost reference oscillator to accurately and reliably provide a stable reference frequency by "locking on" to an accurate transmitted signal such as a bass station's carrier frequency. Further, the receiver 10 adaptively adjusts to improve the reference oscillator's performance over time. It will be appreciated that the present invention need nor, use a manufactured low-cost VCXO module. For example, one could use discrete crystal and oscillator stages that are combined to form a low-cost VCXO solution. Moreover, a conventional high-cost VC-TCXO may still be used within the present invention and benefit from the adaptive aging and temperature compensation features. The receiver described herein is suitable for all v/ireless protocols such as TDMA and CDMA. Moreover, although described with respect to a super heterodyne receiver, the present invention is widely applicable to ether types of receivers that must tune a reference oscillator to capture a transmitted signal. For example, the present invention is applicable to direct conversion receivers. Thus, although the invention has been described with reference to particular embodiments, the description is only an example of the invention's application and should not be taken as a limitation. Consequently, various adaptations and combinations of features of the embodiments disclosed are within the scope of the invention as encompassed by the following claims. WE CLAIM: 1. A method of providing a reference frequency signal, comprising : storing (102) temperature correction information for a reference oscillator ; associating and storing (112) a stored quality factor in association with the stored temperature correction information ; receiving (103) current temperature data ; generating (104) a current correction signal by correlating the current temperature data with the stored temperature correction information ; tuning (106) the reference oscillator using the current correction signal, the reference oscillator outputting the reference frequency signal; determining (108) a frequency offset in the reference frequency signal; generating (110) updated temperature correction information based on the frequency offset; receiving (114) a signal and associating a received signal quality factor for the received signal; comparing (116) the received signal quality factor with the stored quality factor ; and modifying the stored temperature correction information responsive to the comparing step. 2. The method as claimed in claim 1, which involves storing (120) the updated temperature correction information. 3. The method as claimed in claim 2, wherein the updated temperature correction information is stored only if the received signal quality factor exceeds the stored quality factor. 4. The method as claimed in claim 3, wherein ihe stored quality factor is reduced proportionally to the passage of time. 5. The method as claimed in claim 1, wherein the received signal quality factor is determined by the carrier-to-noise ratio of the received signal. 6. The method as claimed in claim 2, which involves determining (118) whether the received signal is a valid signal; and wherein the updated temperature correction information is only stored if the received signal is a valid signal. 7. The method as claimed in claim 6, wherein the received signal is determined to be a valid signal if it is a pilot signal of a base station. 8. The method as claimed in claim 1, which involves : receiving (136) a strong signal; determining (140) an initial frequency offset in the reference frequency signal based on the strong signal; and calibrating (142) initial temperature correction information based on the initial frequency offset. 9. A circuit for generating a reference frequency signal in a wireless communication device comprising: a reference oscillator (24) ; a memory (44) configured to store temperature correction information for the reference oscillator, the memory (44) further configured to store a stored quality factor associated with the stored temperature correction information ; logic circuitry (43) coupled to the reference oscillator and configured to receive current temperature data (49) and generate a current correction signal (40) by correlating the current temperature data with the stored temperature correction information, wherein the reference oscillator (24) is tuned using the current correction signal (4) and outputs the reference frequency signal ; and a computational engine (46) having an adaptive algorithm (48) executed by the logic circuitry (43), the adaptive algorithm configured to determine a frequency offset in the reference frequency signal and generate updated temperature correction information based on the frequency offset; wherein the adaptive algorithm is further configured to receive a signal and associating a received signal quality factor for the received signal, compare the received signal quality factor with the stored quality factor (116), and modify the stored temperature correction information responsive to the comparison. 10. The circuit as claimed in claim 9, wherein the memory is configured to store the updated temperature correction information. 11. The circuit as claimed in claim 10, wherein the updated temperature correction information is stored only if the received signal quality factor exceeds the stored quality factor. A receiver 10, uses an adaptive algorithm 48 to tune a low-cost crystal oscillator 24 according to a temperature compensation profile 50, so as to produce a precision master reference frequency 26 despite temperature 49, initial tolerance, and aging effects. An automatic frequency control system 20 also tunes the crystal oscillator 24. The adaptive algorithm 48 adjusts the temperature compensation profile 50 for the crystal oscillator 24 according to the adjustments made by the automatic frequency control 20 should a received signal's 13 quality factor exceed the quality factor 55 associated with the temperature compensation profile 50.

Full Text

REFERENCE OSCILLATOR WITH AUTOMATIC COMOENSATION FOR AGING AND TEMPERATURE
BACKGROUND
Field of the Invention
Tha Present invention relates to electronic circuitry, and more specifically to electronic circuitry tuning a reference oscillator.
Description of Related Art
Because the usable electromagnetic spectrum is a limited resource, governmental agencies regulate its use and exploitation. For example, in the U.S. the Federal Communication Commission (FCC) divides the usable electromagnetic spectrum into frequency ranges, or bands. Each band may be assigned to a, specific function, or may be reserved for future use. Since only a few bands are allocated to a specific function, such as portable communication, it is important that the band be efficiently utilized.
Thus, the frequency bands assigned to wireless communications should accommodate many wireless users. To allow each frequency band to accommodate so many users, a wireless communication device typically uses a reference oscillator to accurately find the carrier
frequency used by its particular base station. in a particular use, the wireless device is a mobile handset configured to communicate with one or more base stations.
To find the carrier frequency, reference oscillators operating in a wireless communications device typically generate a highly accurate frequency reference. Because low-cost oscillators are. prone to substantial initial error, large individual variations, and degraded performance over time, they have not been used to generate such an accurate frequency reference. Accordingly, known wireless communication devices use more expensive high-precision reference oscillators, and associated precision circuitry, to obtain the necessary accuracy and repeatability.
The precision circuitry in the wireless communication device or mobile handset generally includes a voltage-controlled temperature-compensated crystal oscillator (VC-TCXO). The VC-TCXO provides a reference frequency, which is used by the handset to , find, or lock onto the carrier frequency. Depending upon the handset's current temperature, the VC-TCXO adjusts its tuning to maintain a constant reference frequency output. Such VC-TCXO's, however, are
relatively expensive, adding significantly to the cost of a handset. Moreover, despite a VC-TCXO's expense, the reference frequency output eventually drifcs with age. In addition, VC-TCXOs typically have an indeterminate initial accuracy resulting from manufacturing tolerances that complicates a handset's design.
Further, conventional temperature compensation circuitry for a VC-TCXO has degraded accuracy at temperature extremes. Despite this degraded accuracy, users expect a handset to provide reliable communication that mimics or betters that of a landline phone. To meet expectations, the handset must consistently establish communication in an acceptable time period and do so over a receiver's wide temperature range. For example, a handset is expected to operate in sub-Arctic conditions, and also operate after being subject to the torturous heat of a car dashboard in summer. As a result, it is desirable that a handset VC-TCXO should typically maintain a frequency
accuracy of better than about +/- 2 parts per million
(ppm) over the temperature range -30° C to +85° C.
As generally described earlier, a VC-TCXO is not stable over time. VC-TCXO modules may drift at
approximately 1 ppm per year, yet are often expected to operate for many years. The typical AFC (Automatic Frequency Control) loop used to tune the VC-TCXO modules can cope with offsets of no greater than about +/- 4 ppn with respect to its desired reference frequency. Thus, in just a few years, such VC-TCXO modules will have drifted to the point that they may not be able to assist in acquiring or capturing a base station's carrier.
Systems have been developed to account for the aging of a VC-TCXO. For example, U.S. Pat. No. 6,064,270 discloses a handset having a VC-TCXO that, should the base station's beacon carrier escape capture, will perform a random search for the carrier by adjusting the VC-TCXO's reference frequency +/- 4 ppm (or some other suitable amount). If the carrier is acquired, the offset is remembered for future use. Although this system may provide acceptable age compensation when coupled with an expensive VC-TCXO, it is not practically implemented with less expensive oscillators having more substantial drift and corresponding large offsets. A system having such large offsets that searched blindly for the carrier frequency would produce delays intolerable to the user. In
addition, the search accounts only for the gross offsets encountered during aging and does not address temperature effects, necessitating the continued use of an expensive VC-TCXO.
Accordingly, there is a need in the art for an improved receiver capable of providing a reference frequency that is robust to temperature and aging effects without the use of an expensive VC-TCXO.
SUMMARY
It is desirable to accurately generate a reference frequency signal using a low-cost oscillator. It is therefore an object of the invention to enable low-cost oscillators to accurately generate reference signals despite initial tolerance, temperature and aging effects. It is a further object of the invention to enable such oscillators to adaptively adjust to temperature and aging effects by recalibrating accordingly.
To overcome the disadvantage in the prior art and meet the objectives of this invention, a computational device implements an adaptive algorithm. The adaptive algorithm adjusts stored frequency correction information to compensate a reference oscillator.
During this compensation, the reference oscillator adjusts its master reference signal responsive to the stored frequency correction information. Because this algorithm adaptively accounts for initial frequency error, aqing, and temperature effects, a low-cost VCXO may be used as the reference oscillator.
In accordance v/ith one aspect of the invention, the frequency correction information comprises a plurality of predetermined frequency correction factors spanning a temperature range. The computational device may receive current temperature information, which enables the computational device to derive a current frequency correction factor from Che frequency correction factors. The reference oscillator tunes or adjusts the frequency of its reference signal according to the current frequency correction factor. Should the reference signal frequency be offset from a desired frequency, the computational device may adjust one or more of the stored frequency correction factors according to the frequency offset. In such a manner, the stored correction factors are adjusted as needed tc provide superior frequency accuracy.
Quality factors may be associated v/ith the frequency correction factors. In this embodiment, when
a current freauency correction factor is calculated, a currant quality factor is also determined from the quality factors. A received signal's quality factor is compared to the current quality factor before the temperature correction factors arc; updated. In such a manner, an older, but higher quality temperature compensation factor would not be replaced by a newer, but lower quality temperature compensation factor.
Advantageously, the present invention enables a communications device to use a low-cost oscillator as a reference oscillator, avoiding the cost of the VC-TCXO. Because the invention provides for adaptive adjustment of the frequency correction factors, the low cost oscillator also systematically compensates for oscillator aging.
Further aspects and features of the invention are set forth in the following description together with the accompanying drawings.
ACCOMPANYING
BRIEF DESCRIPTION OF THE ACOMPANYING DRAWINGS
FIG. 1 illustrates a receiver having an adaPtive VC-TCXO correction loop according to one embodiment of the invention.
FIG. 2 illustrates the temperature correction profile for a VCXO.
FIG. 3 illustrates a portion of the profile of Figure 2 being calibrated according to one embodiment of the invention.
FIG. 4 is a flowchart of a method for providing a reference signal in accordance wich one embodiment of the invention.
FIG. 5 is a flowchart of a method for calibrating a communication device in accordance with one embodiment of the invention.
Use of the same reference symbols in different figures indicates similar or identical items.
DETAILED DESCRIPTION
Referring now to Figure 1, a block diagram is shown of a receiver 10 using a low-cost reference oscillator 24 . Despite being low cost, the reference oscillator 24 generates a precision reference frequency-signal 26 of sufficient accuracy for wireless communication. Since the receiver 10 does not reauire an expensive VC-TCXO, the receiver 10 is desirable for use in a wireless communications device such as a mobile handset.
In the illustrated example, the receiver 10 is a super heterodyne receiver 10 that adaptively accounts for initial frequency error, temperature effects and aging drift of the low precision oscillator 24. .More specifically, the receiver 10 incorporates a logic block 4 3 to remember, predict, and adapt frequency correction information. The frequency correction information may be stored in a memory 44 in analog or digital form. In digital form, the frequency correction information may be sampled to create a plurality of temperature correction factors. These correction factors may be responsive to temperature fluctuations, initial error conditions, and aging effects, for example. Accordingly, the low cost oscillator 24 provides a sufficiently repeatable and accurate frequency signal for use in the local oscillator block 15.
Before proceeding to discuss the logic block 43, the receiver 10 will be generally described. The receiver 10 includes an antenna 12 that receives a transmitted signal such as a base station carrier signal. The antenna signal 13 is down converted with a local oscillator (LO) signal 11 in a mixer 14 to form an intermediate frequency (IF) signal 34. The
generation of the LO signal 11 by the local oscillator block 15 will be described further herein.
An IF gain and selection block 16 amplifies and filters the I? signal 34 produced by the mixer 14. A freauencv discriminator 13 frequency demodulates the amplified and filtered IF signal to produce a demodulated signal 19. The demodulated signal 19 is low-pass filtered by an automatic frequency control (AFC) loop filter 20 to form an AFC signal 21. The AFC signal 21 couples to the reference oscillator 24 to tune the frequency of the reference signal 26. The reference signal couples to a phase-locked loop (PLL) 28 that is part of the local oscillator block 15. A loop filter 30 then couples the PLL output to a voltage-controlled oscillator (VCO) 32 that produces the LO signal 11.
Those of ordinary skill in the art will appreciate that the receiver 10 of Figure 1 is an idealized version for use in a wireless handset . In an actual handset receiver, multiple frequency translations may require additional mixing stages. Furthermore, AFC loop filter 2 0 may include I and Q channels (not illustrated) . The I and Q signals so produced may be. digitized and processed before being filtered by the
AFC loon filter 20. The receiver's logic block 4 3 will new be described.
The logic block 43 includes a computation engine 46, which implements an adaptive algorithm 4 3 that self-learns and accounts for initial error conditions and temperature and aging effects of the reference oscillator, enabling the receiver 10 to capture signals without requiring the use of precision TC-VCXOs. The computation engine 4 6 provides a current correction factor 40 according to the algorithm 48. .Those of ordinary skill in the art will appreciate that algorithm 4 8, unlike the remaining elements in Figure 1, is not structure but is shown symbollically to indicate its relationship to the receiver 10. The correction factor 40 may compensate for initial tolerance error, aging effects, and current temperature, for example. The computation engine 4 6 may be implemented using any suitable logic device that may be programmed to produce this factor 40, including a microprocessor (not illustrated) or a state machine (not illustrated). Logic block 4 2 also preferably includes a memory device 44. It will be appreciated that the memory device 44 mav be alternatively integrated with the computational device 46.
The current temperature correction factor 4 0 is combined with the AFC signal 21 in summer 23 to produce a reference oscillator correction signal 22. This reference oscillator correction signal 22 couples to the tO oscillator block 15 to adjust the LO signal 11. In this manner, the receiver 10 has both "coarse" and "fine" control of its master reference signal 26 and, ultimately, the LO signal. The coarse control is provided by the logic block 4 3 through the current temperature correction factor 40. The coarse control is desirably operated so that the fine control provided by the AFC loop can capture the transmitted signal, typically requiring the coarse control to be within about +/- 4 ppm of the required frequency. In this fashion, the coarse control provided by the logic block 43 effectively enables the low cost VCXO to provide a level of performance conventionally provided by the more expensive precision TC-VCXO. However, unlike receivers using conventional TC-VCXOs, the receiver 10 continually recalibrates the low cost VCXO as mandated by changes in temperature and aging.
The adaptive algorithm 4B uses a predetermined frequency offset profile specific to whatever VCXO is used as the reference oscillator 24. This profile
provides the frequency conpensation necessary to adjust, the reference oscillator 24 over its operating temperature range. Thus, the frequency offset profile may also be denoted as the temperature compensation profile.
Referring to FIG. 2, an illustrative profile 50 for a low-cost oscillator is shown. Several techniques may be employed to derive the profile 50 to account for the component-to-component variances of low-cost VCXOs. In one embodiment, a given VCXO would be individually tested over the expected receiver's temperature range to calibrate the profile 50 exactly to the VCXO's particular behavior. Such individual testing could, however, increase manufacturing costs and reduce the savings introduced by using a low-cost VCXO in place of a precision TC-VCXO. Thus, in another embodiment, rather than test a given VCXO over the entire expected operating range, each VCXO would be initially calibrated only at a single ambient temperature. Prior to this calibration, the expected profile may be determined from manufacturer data, typically representing the averaging of many individual VCXOs. The expected profile is, as the name implies, only expected and may be departed from considerably by a
given individual VCXO. For example, 10 pprn variance:? can be observed from the expected profile predicted by a given VCXO manufacturer. For a given VCXO, the manufacturer will generally publish a data sheet indicating an expected frequency offset profile.
averaging the test results of several VCXOs from a aiven batch over the expected receiver's temperature range (typically -30° to +120' C) .
It will be appreciated that, given the large component-to-component tolerances expected for a low-cost VCXO, the actual temperature compensation profile may vary widely from individual VCXO to VCXO, even within the same manufacturer batch. Thus, it is desirable that the profile 50 be individually calibrated for a given VCXO because of the wide tolerance. Given that the receiver will usually be at ambient temperature and this temperature is normal for a manufacturing plant, the individual calibration may be performed at ambient temperature. It will be appreciated that another temperature may be selected for calibration purposes.
Prior to calibration, data indicative of the profile 50 may be placed in memory by storing samples
52 of the profile in a non-volatile memory 44. Each sample 52 corresponds to a discrete temperature and its associated frequency offset/temperature compensation value. Accordingly, the samples 52 represent initial temperature correction factors.
The temperature correction factors 52 or the offset profile 50 may be taken at regular temperature intervals as shown. Alternatively, the profile 50 may be sampled more in regions where the slope is changing and less in regions of relatively little slope change. A thermistor (not illustrated in Figure 1) or another suitable temperature-sensing device provides the receiver's temperature 49. The logic block 43 may correlate a subset of the temperature correction factors 52 with the receiver's temperature 49 to provide a current correction factor signal 40. In operation, if the receiver's temperature matches the temperature corresponding to or.e of the temperature 'correction factors 52, the "correlation" would comprise merely using the frequency offset of the matched factor 52 as the current correction factor signal 40. In such a case, the subset of samples used for correlation has just one member.
More typically, however, the receiver's temperature will not correspond to any of the temperatures used for the temperature correction factors 52. One simple way to correlate the receiver's temperature in this case is to assume a linear slope between adjacent samples immediately framing the receiver's temperature. For example, suppose the receiver's temperature is 90' C and the two closest temperature correction factors give a value of +10 ppm at 100° C and +5 ppm at 80° C. Assuming a linear slope between these two temperature corrections factors would give the current correction factor as 7.5 ppm corresponding to the receiver's temperature of 90° C. In such a case the subset of samples used for correlation has two members. Alternatively, three or more samples may be used to predict a quadratic or higher order slope for the profile 50 at the receiver's temperature.
Given this expected profile 50, an individual VCXO may be calibrated at ambient temperature v.'ith a powerful transmitted signal to ensure that the receiver will capture the transmitted signal with its AFC loop. As discussed earlier, a conventional AFC loop can only handle approximately about +/- 4ppm in frequency offset
at typical operating signal-to-noise ratios. However, these conventional AFC loops can accommodate larger offsets such as -/- 10 ppm should the signal-to-noise ratio be unusually large. Such a situation occurs when a mobile user is close to the transmitting base station or, in this case, receiving a beacon frequency in a laboratory or manufacturing setting. Thus, even if a VCXO being calibrated is off 1Oppm from its expected profile at ambient temperature, the AFC loop will be able to capture the signal given a high enough signal-to-noise ratio - an occurrence easily established in a controlled setting.
To begin the calibration, the logic block 43 correlates the receiver's temperature 49 with a subset of the temperature correction factors 52 in the manner just described to produce a current temperature correction factor 40. The reference oscillator 24 tunes accordingly and the AFC 100p filter 20, if necessary, produces an AFC signal 21 to capture the transmitted signal. For example, if the reference signal's frequency is too high, the AFC signal 21 tunes . the reference oscillator 24 to lower the frequency of the reference signal 26. Conversely, if the reference signal' frequency is too low, the AFC signal 21 tunes
the reference oscillator 24 to increases the frequency of the reference signal 26. In either case, the reference signal 26 has a frequency offset from that of a desired frequency as determined by the base station's carrier frequency. The resulting AFC signal 21 corresponding to this frequency offset may be used to recalibrate the expected profile to produce an adjusted profile. In other words, the AFC signal 21 correlates with a certain ppm frequency offset.
This frequency offset, whether positive or negative, may be applied to calibrate the subset of the temperature correction factors within the expected profile. For example, if the AFC signal 21 used to capture the manufacturing signal during calibration at ambient temperature correlates to a +10 ppm frequency offset, the subset of samples used for correlation may be calibrated by adding 10 ppm to each sample in the subset. Alternatively, the subset may be calibrated linearly according to their contribution to the initial correlation as seen in Figure 3. Here, the receiver's temperature is 63° C thus requiring an 80% contribution from the temperature correction factor at 60* and a 20% contribution from the temperature correction factor at 75°, assuming a linear correlation. If the resulting
AFC signal 21 corresponds to a +10 ppm shift, then the temperature correction factor at 60° would be shifted up 0 ppm and the temperature correction factor at 75" would be shifted up 2 ppm to calibrate the subset.
In addition, should this initial calibration result in a sufficiently large shift, such as the 8 ppm discussed with respect to Figure 3, all of the remaining temperature correction factors may be shifted accordingly, for example, by shifting 1 ppm. As discussed, the value of the AFC signal 21 correlates to a frequency offset. The computation engine 46 receives the AFC signal 21 and may correlate this signal to the corresponding frequency offset using a lookup table.
Alternatively, the current temperature correction factor 40 may be adjusted by the computation engine 46 until the AFC signal 21 is minimized or "zeroed out." At this point, the current temperature correction factor 4 0 accounts for and eliminates the frequency offset. Assuming that ambient temperature is used for this initial calibration, the profile 50 may have a temperature correction factor 52 at. ambient temperature to eliminate any requirement to correlate with more than one temperature correction factor 52 to derive the
current temperature correction factor 40 (the subset of factors used would have just one member) .
To distinguish the factors 52 that are calibrated accurately from those that are not, each factor 52 may be assigned a quality factor 5 5 having an arbitrary scale such as from 1 to 100. The quality factor 55, as the name implies, relates to the quality of the received signal used to calibrate the receiver as determined, for example, by the signal-to-noise ratio. Alternatively, the quality factor may relate to an automatic gain control signal as determined by the received signal, the carrier-to-noise ratio, or another suitable factor relating to the quality of the received signal. For the profile shown in Figure 2, the temperature correction factor for 30° is associated with a quality factor of 95, the temperature correction factor for 45° is associated with a quality factor of 20, and the temperature correction factor for 60" is associated with a quality factor of 95.
Before the initial calibration, all the temperature- correction factors 52 would be associated with a low quality factor, e.g., a value of 50. After the initial calibration, the subset of temperature correction factors 52 that are calibrated as discussed
with respect to Figure 3 are assigned a high quality factor such as a value of 100 because the transmitted signal in this laboratory setting is such that a very high signal - to-noise ratio is assured v/ithin the receiver. Note that in a situation such as illustrated in Figure 3, the quality factors associated with the temoerature correction factors 52 within the subset may be updated proportionally to their contribution to the current temperature correction factor 40. Alternatively, should the subset consist of just one member corresponding to the calibration temperature, just this factor would receive the high quality factor.
Because the initial calibration typically occurs at ambient temperature, the receiver's temperature will tend to change gradually from ambient if the receiver is moved, for example, from indoors to an outdoors location during use by a consumer. As the receiver heats or cools, the receiver's temperature will correlate with temperature correction factors 52 that were not initially calibrated. Thus, the receiver "self-learns" or adaptively adjusts these factors 52, using the transmitted signal as a frequency reference.
As in the initial calibration, the receiver's temperature is correlated with a subset of the
temperature correction factors 52 to produce a current temperature correction factor 40. The transmitted signal is then captured by the receiver's AFC loop. Should the AFC signal 21 indicate a frequency offset is present, the subset is calibrated in the manner just described with respect to Figure 2. In a wireless receiver, the reference oscillator thus becomes locked to the base station's transmitted frequency. Note, however, that because this calibration is occurring outside the initial calibration conditions on a real-world transmitted signal, there is a danger that the receiver could calibrate improperly on a noisy transmitted signal. Thus, the computation engine 46 may respond to a valid signal indication 51. In a mobile handset, the computation engine 4 6 may use the identification of the pilot signal or some other suitable traffic indication to serve as the valid signal indication 51.
Having assured itself that the received signal is valid, the computation engine may then calculate the-quality factor of the received signal. For example, the logic block 4 3 may have a lookup table that correlates the current carrier-to-noise ratio of the received signal to a quality factor on the scale of 1
Co 100. This received quality factor is compared to the quality factor as given by the- subset of temperature correction factors 52 used to derive the current temperature correction factor 40. If the received quality factor is greater, the subset is calibrated as discussed with respect to the initial calibration. However, because the received quality factor will not necessarily equal 100 as in the initial calibration, the amount by which the subset would be updated according to the frequency offset may be reduced proportional to the received quality factor. For example, should the frequency offset correlate to a 10 ppm update of the subset and the received quality factor is 90, then just 90% of the 10 ppm may be applied to the subset. In this fashion, the stored temperature correction factors 52 become progressively more accurate as the receiver adapts to shifts in receiver'- temperature. It will be appreciated that other methods may be used to proportionally apply the frequency offset
As the reference oscillator 24 ages, its expected profile 50 may change as well. To compensate for this aging effect, the quality factors associated with the temperature correction factors 52 may be systematically
decreased over time. For example, should the quality factors correspond to a scale of 1 to 100, the quality factors could be reduced by 5 every three months. In this fashion, the temoerature correction factors 52 will be more readily recalibrated as time passes so as to account for aging effects. The rate at which the duality factors are reduced to account for this aging depends upon the characteristics of the individual reference oscillator 24 within a given receiver.
Note that once calibrated, even low-cost VCXOs exhibit little hysteresis such that repeated use of the receiver at the same temperature with the same current temperature correction factor 40 will result in signal capture with little additional offset needed from the AFC correction signal 21. Thus, the adjustment of the quality factors to account for aging may be very gradual.
Because the receiver may experience extremes in temperature without prior calibration, there is the possibility that the "fine" control provided by the AFC loop will be unable to capture the transmitted signal, despite the "coarse" control provided by the logic block 4 3 via the current temperature correction factor 40. In such a situation, the logic block 43 may be
configured to search for Che transmitted signal by incrementing or decrementing the current temperature correction factor 40. The required increment will depend upon the amount of fine control provided by the conventional AFC loop. For example, if this AFC loop can accommodate frequency errors of +/- 4 ppm in the master reference signal 26, the increment or decrement should be not greater than 4 ppm. Rather than search blindly, the logic block 43 stores the cumulative record of positive or negative calibrations of the temperature correction factors 52. If the receiver's temperature correlates with a certain temperature correction factor that has been recalibrated with positive increments in ppm several times, it would be logical to then assume that the required increment in the current temperature correction factor 40 should be positive rather than negative.
In addition, the logic block 43 can gauge the scale of the increment by this history. For example, should the recent recalibrations have all been significant (such as greater than 5 ppm), the usual increment in the current temperature correction factor could be increased from, say, 4 ppm to 5 ppm. The logic block 43 may store other parameters relating to
the history of the calibrations for diagnostic ourposes. These parameters include the time of each calibration and the number of times each temperature correction factor 52 has been updated.
Referring now to FIG. 4, a method 100 of
calibrating the stored frequency correction information is shown. The method 100 is preferably implemented en a communications device, for example, a mobile wireless handset. It will be appreciated that method 100 may be used on a wide variety of communications equipment.
Method 100 initially stores a frequency offset or temperature compensation profile in a memory in the communications device, as shown in block 102. As described earlier, the frequency offset profile may be in the form of discrete temperature compensation factors or be in analog form. Alternatively, the profile may be approximated by a polynomial function and the necessary polynomial coefficients stored in memory. Optionally, a quality factor may be associated with the correction information as previously described. As shown in block 112.. the quality factor may also be stored in the communications device.
A current temperature signal is provided in block 103. It will be appreciated the current temperature
may be detected by any of several well known devices, such as a thermister, and the temperature signal may take one of several known forms such as a voltage or current signal,
Responsive to the current temperature, a current correction signal corresponding to the current temperature is retrieved from the stored frequency correction information in block 104. Should the frequency correction information be stored as a plurality of temperature correction factors, the current temperature is correlated with a subset of the factors as described previously
The correction signal tunes a reference oscillator in block 106. Ideally, the tuning performed in block 106 tunes the oscillator to precisely the desired frequency. More often, however, a frequency error in the tuning of the reference oscillator may be detected in block 103. Such an error may be detected and measured in an AFC loop, for example.
The determined frequency error is used in block 110 to generate undated frequency correction information for the current temperature. In one implementation, the method 100 also determines the quality of the signal received according to the tuning
of the reference oscillator as shown in block 114. For example, this received quality factor may relate to the strength or signal to noise ratio of the received signal. It will be appreciated that other factors may be used to determine the received quality factor.
In block 116, the received quality factor determined for the updated correction information may be compared to the quality factor associated with the stored frequency correction information used to derive the current correction signal. Responsive to this comparison, a decision can be made whether to proceed to store the updated correction information in block 12 0. For example, if the stored frequency correction information used to derive the current correction signal has an associated quality factor of 70, and the received signal has only an associated quality factor of SO, then the updated frequency correction information will not replace the better, although older, stored frequency correction information. On the other hand, when the received quality factor is better than the quality factor associated with the stored correction information used to derive the current correction signal, then the updated frequency correction information is stored in block 120. Also,
the received quality factor is also associated with the undated correction information and stored, as shown in block 122.
It will also be appreciated'that other factors may be used to determine when the updated correction information will be stored. Such fetctors may be used to reduce the risk that high quality frequency correction information will be replaced by false or low quality frequency correction information. For example, block 118 shows that a validation signal may be determine whether the updated correction information is stored. This validation signal indicates whether a valid signal is being received.
The validation signal may be generated, for example, when a predetermined signal is positively received and identified. In one specific example, the validation signal is generated when a pilot signal is confirmed. It will be appreciate:: that the generation cf the validation signal may be responsive to other factors.
To account for aging effects, block 124 shows that the stored quality factors may be slowly degraded over time. In such a manner, even frequency correction
information which had the highest quality factor may be subject to adjustment after a period of time.
Referring now to FIG. 5, a method 130 for initial calibration of a communications device configured to implement the method 100 of Figure 4 is shown. In block 132, the initial frequency correction information for a reference oscillator is determined. As discussed Dreviously, this determination may be based on manufacturer provided information. In block 134, this initial frequency correction information is stored in memory. Because of the large individual tolerances encountered with low-cost reference oscillators, this initial frequency correction information is only approximate and does not necessarily correspond to the behavior of a given individual reference oscillator. Thus, this initial frequency correction information should be calibrated according to the individual behavior of the reference oscillator being tested. To do so, a strong calibration signal is generated and the reference oscillator tuned according to the initial frequency correction information in block 136. As used herein, this "strong" calibration signal denotes a signal powerful enough to enable a conventional AFC loop to capture this calibration signal even if the
reference. oscillator has a large frequency offset;, e.g., +/- 10 pom. To eliminate this frequency offset, undated or calibrated frequency correction information is generated in block 140. Finally, in block 142 this undated frequency correction information is stored. A quality factor may be associated with this stored information in block 144.
To better appreciate the contribution of the receiver 10, the operation of a conventional receiver will be briefly discussed. It will, however, be understood that conventional receiver operation is well known. In conventional prior art receivers, an AFC signal typically is directly coupled to an expensive high performance reference (or crystal) oscillator to provide an LO signal. In response to the AFC signal, the high cost reference oscillator adjusts the frequency of a master reference signal.
The master reference signal typically coucles to a phase-locked loop (PLL) that is part of the local oscillator control block. A loop filter then generally couples the PLL to a voltage-controlled oscillator (VCO) that produces the LO signal. The LO signal mixes with the received signal from the antenna to produce an IF signal. This IF signal will be demodulated and
passed through an AFC low pass filter, and feed back into the reference oscillator. Thereby a conventional AFC "loop" is formed.
Conventional AFC low pass filters are designed go that the AFC control provided is fast-acting and can correct +/- 4 ppm frequency inaccuracies in the master reference signal. Should the high performance reference oscillator produce a master reference signal precisely at the required frequency, the AFC loop would never need to adjust the reference oscillator. However, even precision reference oscillators will tend to be slightly inaccurate and require adjustment by the AFC loop. As noted in the background section, conventional AFC loops could correct for about a +/- 4 ppm deviation by the reference oscillator yet still acquire the carrier signal at typical operating "signal-to-noise ratios.
Conventional control of the reference oscillator by using a precision TC-VCXO, however, could not adaptively adjust for temperature and aging effects experience by even the expensive, high performance reference oscillator. Advantageously, the receiver 10 enables a low cost reference oscillator to accurately and reliably provide a stable reference frequency by
"locking on" to an accurate transmitted signal such as a bass station's carrier frequency. Further, the receiver 10 adaptively adjusts to improve the reference oscillator's performance over time. It will be appreciated that the present invention need nor, use a manufactured low-cost VCXO module. For example, one could use discrete crystal and oscillator stages that are combined to form a low-cost VCXO solution. Moreover, a conventional high-cost VC-TCXO may still be used within the present invention and benefit from the adaptive aging and temperature compensation features.
The receiver described herein is suitable for all v/ireless protocols such as TDMA and CDMA. Moreover, although described with respect to a super heterodyne receiver, the present invention is widely applicable to ether types of receivers that must tune a reference oscillator to capture a transmitted signal. For example, the present invention is applicable to direct conversion receivers.
Thus, although the invention has been described with reference to particular embodiments, the description is only an example of the invention's application and should not be taken as a limitation. Consequently, various adaptations and combinations of
features of the embodiments disclosed are within the scope of the invention as encompassed by the following claims.
WE CLAIM:
1. A method of providing a reference frequency signal, comprising :
storing (102) temperature correction information for a reference oscillator ;
associating and storing (112) a stored quality factor in association with the stored
temperature correction information ; receiving (103) current temperature data ; generating (104) a current correction signal by correlating the current temperature data
with the stored temperature correction information ; tuning (106) the reference oscillator using the current correction signal, the reference
oscillator outputting the reference frequency signal; determining (108) a frequency offset in the reference frequency signal; generating (110) updated temperature correction information based on the frequency
offset; receiving (114) a signal and associating a received signal quality factor for the received
signal;
comparing (116) the received signal quality factor with the stored quality factor ; and modifying the stored temperature correction information responsive to the comparing
step.
2. The method as claimed in claim 1, which involves storing (120) the updated temperature
correction information.
3. The method as claimed in claim 2, wherein the updated temperature correction information is
stored only if the received signal quality factor exceeds the stored quality factor.
4. The method as claimed in claim 3, wherein ihe stored quality factor is reduced proportionally to
the passage of time.
5. The method as claimed in claim 1, wherein the received signal quality factor is determined by
the carrier-to-noise ratio of the received signal.
6. The method as claimed in claim 2, which involves determining (118) whether the received
signal is a valid signal; and
wherein the updated temperature correction information is only stored if the received signal is a valid signal.
7. The method as claimed in claim 6, wherein the received signal is determined to be a valid signal
if it is a pilot signal of a base station.
8. The method as claimed in claim 1, which involves :
receiving (136) a strong signal;
determining (140) an initial frequency offset in the reference frequency signal based on
the strong signal; and calibrating (142) initial temperature correction information based on the initial
frequency offset.
9. A circuit for generating a reference frequency signal in a wireless communication device
comprising:
a reference oscillator (24) ;
a memory (44) configured to store temperature correction information for the reference oscillator, the memory (44) further configured to store a stored quality factor associated with the stored temperature correction information ;
logic circuitry (43) coupled to the reference oscillator and configured to receive current temperature data (49) and generate a current correction signal (40) by correlating the current temperature data with the stored temperature correction information, wherein the reference oscillator (24) is tuned using the current correction signal (4) and outputs the reference frequency signal ; and
a computational engine (46) having an adaptive algorithm (48) executed by the logic circuitry (43), the adaptive algorithm configured to determine a frequency offset in the reference frequency signal and generate updated temperature correction information based on the frequency offset;
wherein the adaptive algorithm is further configured to receive a signal and associating a received signal quality factor for the received signal, compare the received signal quality factor with the stored quality factor (116), and modify the stored temperature correction information responsive to the comparison.
10. The circuit as claimed in claim 9, wherein the memory is configured to store the updated
temperature correction information.
11. The circuit as claimed in claim 10, wherein the updated temperature correction information is
stored only if the received signal quality factor exceeds the stored quality factor.

A receiver 10, uses an adaptive algorithm 48 to tune a low-cost crystal oscillator 24 according to a temperature compensation profile 50, so as to produce a precision master reference frequency 26 despite temperature 49, initial tolerance, and aging effects. An automatic frequency control system 20 also tunes the crystal oscillator 24. The adaptive algorithm 48 adjusts the temperature compensation profile 50 for the crystal oscillator 24 according to the adjustments made by the automatic frequency control 20 should a received signal's 13 quality factor exceed the quality factor 55 associated with the temperature compensation profile 50.

Documents:

1467-KOLNP-2003-CORRESPONDENCE.pdf

1467-KOLNP-2003-FORM 27.pdf

1467-kolnp-2003-granted-abstract.pdf

1467-kolnp-2003-granted-assignment.pdf

1467-kolnp-2003-granted-claims.pdf

1467-kolnp-2003-granted-correspondence.pdf

1467-kolnp-2003-granted-description (complete).pdf

1467-kolnp-2003-granted-drawings.pdf

1467-kolnp-2003-granted-examination report.pdf

1467-kolnp-2003-granted-form 1.pdf

1467-kolnp-2003-granted-form 18.pdf

1467-kolnp-2003-granted-form 2.pdf

1467-kolnp-2003-granted-form 3.pdf

1467-kolnp-2003-granted-form 5.pdf

1467-kolnp-2003-granted-pa.pdf

1467-kolnp-2003-granted-reply to examination report.pdf

1467-kolnp-2003-granted-specification.pdf

1467-kolnp-2003-granted-translated copy of priority document.pdf

« Previous Patent

Next Patent »

Patent Number

227664

Indian Patent Application Number

1467/KOLNP/2003

PG Journal Number

03/2009

Publication Date

16-Jan-2009

Grant Date

14-Jan-2009

Date of Filing

11-Nov-2003

Name of Patentee

KYOCERA WIRELESS CORPORATION

Applicant Address

10300, CAMPUS POINT DRIVE, SAN DIEGO, CA

Inventors:

#	Inventor's Name	Inventor's Address
1	FORRESTER TIMOTHY	11082 MULGRAVE ROAD, SAN DIEGO, CA 92131

PCT International Classification Number

H03L/1/00

PCT International Application Number

PCT/IB2002/01646

PCT International Filing date

2002-05-14

PCT Conventions:

#	PCT Application Number	Date of Convention	Priority Country
1	09/859,853	2001-05-16	U.S.A.