MXPA99003648A - Apparatus and associated method for calibrating a device - Google Patents

Abstract

Apparatus (82), and an associated method (202), for calibrating a device (44) responsive to values of a reference signal (89). The reference signal may be subject to short-term disturbances. In one implementation, a cellular radio base station utilizes a Stratum-2 oscillator (84) to which to phase-lock a base station VCO (44). Compensation is made for the aging of the Stratum-2 oscillator, thereby to provide a regulation signal causing the VCO to exhibit acceptable short-term and long-term frequency stability characteristics.

Description

"DEVICE, AND ASSOCIATED METHOD TO CALIBRATE A DEVICE" The present invention relates generally to methods and calibration apparatus for calibrating a device, such as a VCO (voltage controlled oscillator) that responds to the values of a reference signal. More particularly, the present invention relates to the calibration of a device that responds to a reference signal that exhibits acceptable short-term frequency stability characteristics but is susceptible to long-term frequency disturbances. The values of the reference signal are selectively modified with a reference signal that exhibits acceptable long-term frequency stability characteristics, but which may be susceptible to short-term frequency disturbances. One embodiment of the present invention is capable of operating in a PLL (phase-locked loop) circuit of e.g., a cellular radio base station. A network-generated signal, such as a PCM clock signal or a GPS clock signal, is provided to the radio base station. This signal exhibits good long-term frequency stability characteristics, but is susceptible to short-term frequency instability. A local oscillator, such as an OVCXO (crystal oscillator) controlled by oven voltage), placed in the radio base station, generates a reference signal that exhibits good frequency stability characteristics a. short term but is susceptible to long-term frequency instability, due to the aging of the crystal oscillator. A VCO of the radio base station is blocked_to the reference signal generated by the OVCXO. The reference signal is modified at selected intervals by means of a signal generated by network. An oscillatory signal generated by the VCO is caused in this way to exhibit stability characteristics, of acceptable frequency in the short term and in the long term.

BACKGROUND OF THE INVENTION Many types of devices must be calibrated, at least at selected intervals, to ensure their proper functioning. A VCO (voltage controlled oscillator) coupled in phase locked relationship with a reference signal is exemplary of this device. When, eg, the VCO forms a portion of a PLL circuit (phase locked loop), the oscillation frequency of the oscillatory signals generated by the VCO is blocked to that of a reference signal to which it responds. Operably a VCO.

Many types of radio communication devices use VCOs coupled in PLL circuits. The oscillatory signals formed by the VCO are used to form traffic signals that are transmitted by the transmitting apparatus. And, as the oscillatory signals generated by the VCOs of the receiving apparatus, for example, are used in the reception of receiving signals. The acceptable frequency stabilities of the oscillatory signals generated by the VCOs are required for proper operation of the radio communication apparatus. A radio base station capable of operating in a cellular communication system is exemplary of the radio communication apparatus that uses a VCO coupled in a PLL circuit. The acceptable frequency stability of the oscillatory signals generated by the VCO is required so that the downlink signals generated by the base station of the radio are properly transmitted to a mobile terminal, without interfering with the other downlink signals transmitted simultaneously. Acceptable frequency stability levels are similarly required to allow the radio base station to properly receive the uplink signals transmitted by the mobile terminals to the radio base station. The operating specifications promulgated by different standard adjustment bodies indicated, inter alia, the frequency stability requirements within which the operation of the cellular communication apparatus must be satisfied. The performance specifications for GSM, PCS 1900, and DCS 1800 mobile cellular radio communication systems all disclose strict synchronization accuracies to ensure that operable radio base stations in these systems generate at least signals that exhibit levels of acceptable frequency stability. Compliance with the required synchronization accuracies signaled in an appropriate performance specification required to which the radio base station is secured using a Pl-regulated loop, phase-locked circuit (PLL). In this circuit, a VCO is locked to a reference signal of high frequency stability. For example, a PCM clock signal is sometimes used to form the reference signal applied to the radio base station. A PCM clock signal is an 8 kHz reference signal generated by an operator of the network in a medium well controlled environment. The specifications of ETSI G.823 and G.824 indicate, inter alia, permissible levels of fluctuation in a PCM clock signal. When the PCM clock signal exhibits characteristics that are at least as good as the signal requirements stated in the appropriate specifications or these specifications, the radio base station can be operated to meet the performance specification promulgated by the body. of appropriate regulation adjustment. The quality of the reference signal is measured in stratum levels. The Stratum level of a PCM clock signal specifies maximum permissible frequency deviation of the reference signal. The aforementioned specifications of ETSI G.823 and G.824 indicate frequency stability standards that correspond to a level of "Stratum-2". Some networks, however, do not guarantee the reference signal, such as the PCM clock signal, so that it always falls within the requirements indicated in one of the appropriate specifications of ETSI G.823 and G.824. A reference signal of the least assured stability level is provided more economically. In some networks, therefore, a reference signal is not provided to the radio base stations. always guaranteed as being from a level of Stratum-2. Instead of this it is guaranteed that a reference signal of the level of Stratum-3 will be provided. The reference signal ~ provided by the network to the radio base stations is not always ensured as being from a level of Stratum-2, but the reference signal is usually from a level of frequency stability of Stratum-2 but during intermittent periods. During these periods, the reference signal is of an inadequate frequency stability level, that is, Stratum-3. This reference signal, therefore, is of good long-term frequency stability characteristics but is potentially deficient in short-term frequency stability characteristics. To ensure that the frequency stability standards required for the operation of the radio base stations are satisfied in these networks, some radio base stations include reference oscillators that generate reference signals for the quality of Stratum-2. An OVCXO (oven-controlled crystal oscillator) is exemplary of a Stratum-2 oscillator. An OVCXO, as well as some other type of Stratum-2 oscillators, exhibits short-term frequency stability, but is susceptible to displacement of long-term frequency caused by the aging of the oscillator. Conventionally, these oscillators must be calibrated regularly. Calibration is typically carried out using a manual procedure. This method is expensive, particularly when large numbers of radio base stations of a radio communication system have to be calibrated regularly. One way by which good long-term frequency stability characteristics of a PCM clock signal provided by the network can be used to correct the aging of the Stratum-2 oscillator placed in the radio base station, would reduce the need to calibrate manually the Stratum-2 oscillator. More generally, a way by which a device is allowed to calibrate with a reference signal subject to short-term disturbances would, of course, be advantageous. It has been in view of this background information related to the apparatus and methods of calibration, that significant improvements of the present invention have arisen.

COMPENDIUM OF THE INVENTION The present invention, accordingly, advantageously provides an apparatus and an associated method for calibrating a device that responds to the values of a reference signal susceptible to short-term disturbances. In an implementation, a modality of the present invention is capable of operating in a phase locked loop circuit in which an oscillator to be calibrated is provided with an adjustment signal. The adjustment signal is of a quality to allow the oscillator to form an oscillatory signal that exhibits acceptable short-term and long-term frequency stability characteristics. A reference signal generated by a reference source that has acceptable short-term frequency stability characteristics, but susceptible to long-term frequency deviation, is altered, during selected intervals in response to the values of a signal generated by a reference source having acceptable long-term frequency stability characteristics but which may be susceptible to instability of short-term frequency. The adjustment signal is therefore formed and used to adjust the oscillation frequency of the oscillator. In one aspect of the present invention, a radio base station of a communication system The cell is coupled to receive a signal generated by the network such as a PCM clock signal. The PCM clock signal exhibits good long-term frequency stability characteristics but is susceptible to short-term frequency instability. A Stratum-2 oscillator, such as an OVCXO, forms a portion of the radio base station and generates a reference signal that exhibits good short-term frequency stability characteristics, but is susceptible to long-term frequency instability caused , eg, by aging the oscillator. The operation of the embodiment of the present invention allows the local oscillator to form a signal that exhibits long-term and short-term frequency stability characteristics of acceptable levels. In another aspect of the present invention, an output oscillatory signal formed by an oscillator of an oscillator circuit is controlled in such a way that the output oscillatory signal is of acceptable long-term and short-term frequency stability characteristics. A first feedback element is coupled to receive the output oscillatory signal and to receive a first reference signal. The first reference signal exhibits frequency stability characteristics a. short term at least as good as a first "level of stability selected. The first feedback element forms a first difference signal representative of the deviation of the output oscillatory signal with respect to the first reference signal. A second feedback element is coupled to receive the output oscillatory signal and to receive a second reference signal. The second reference signal exhibits long-term frequency stability characteristics at least as good as a second level of selected stability. The second feedback element forms a second difference signal representative of the deviation of the output oscillatory signal with respect to the second reference signal. A compensation value generator is coupled to receive the second difference signal and to form a compression value signal. The compensation value signal is formed in response to the values of the second difference signal, when the second reference signal exhibits short term frequency stability characteristics at least as good as a third selected stability level. A regulator is coupled to receive the first difference signal and at least to selectively receive the compensation value signal. The regulator modifies the first difference signal that responds to the signal of compensation value and forms a control signal to control the frequency of the oscillator. In another aspect of the present invention, a phase locked loop circuit blocks the frequency of an oscillator of an oscillator circuit. A plurality of reference sources form a plurality of reference signals. Each of the reference signals has a noise component that exhibits a spectral density that has spectral components at different frequencies than the noise components of another of the reference signals. A combiner at least selectively couples to receive the reference signals generated by the plurality of reference signal sources. The combiner selectively combines the reference signals to form a resultant signal therefrom. The resulting signal adjusts the frequency of the oscillation of the oscillator, to thereby cause the frequency blocking oscillator to oscillate at a selected oscillation frequency. In these and other aspects, therefore, a method and associated apparatus calibrates a device in response to the values of a reference signal in which the reference signal is subject to short-term disturbances. The values of the portions of the reference signal are validated by associating a value of Admissibility with the portions of the reference signal. The indications of the portions of the reference signal and the admissibility values associated therewith are provided to a Kalman observer. The values stored in the Kalman observer are selectively altered in response to the indications provided to the Kalman observer and the admissibility values associated therewith. A regulation signal is formed that responds to the values stored in the Kalman observer. And, the regulation signal is applied to the device.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 illustrates a functional block diagram of an embodiment of the present invention positioned to calibrate a device. Figure 2 illustrates a functional block diagram of a phase locked loop circuit in which an embodiment of the present invention is operable. Figure 3 illustrates a functional block diagram of a PLL in a cellular radio base station wherein an embodiment of the present invention is capable of operating.

Figure 4 illustrates a functional block diagram of another phase locked loop circuit wherein one embodiment of the present invention is operable. Figure 5 illustrates a functional block diagram of the compensation value generator of an embodiment of the present invention. Figure 6 illustrates a functional block diagram of a portion of a mode of the compensation value generator shown in Figure 5. Figure 7 illustrates a flow diagram of the method that lists the steps of the method, the method of a method of the detailed description of the present invention.

DETAILED DESCRIPTION Figure 1 illustrates the regulation signal generator 10 of one embodiment of the present invention. The generator of the regulation signal is capable of operating to generate signals for calibrating a device, here the device 12. The generator of the regulation signal is coupled through the line 13 to receive a reference signal generated by a source 14. of reference signal.

The generator 10 of the regulation signal generates a regulation signal on the line 18, which is used to calibrate the device 12. The regulation signal generated on the line 18 is of a value that selectively responds to the values of the signal of reference applied to the regulation signal generator 10 on the line 13. The reference signal generated by the reference signal source 14 is susceptible to short-term disturbances, and the regulation signal generator 10 is capable of functioning, in part to determine when the values of the reference signal are of -Values -which exhibit short-term disturbances. When the portions of the reference signal are of values indicating that short-term disturbances are introduced therein, the values of the portions are not used by the regulation signal generator in the formulation of the regulation signal in the line 18. The regulation signal generator 10 includes a validator 22. The validator 22 is coupled to receive the reference signal generated on the line 13. The validator 22 is capable of operating to validate if the values of the reference signal they have the possibility of being free from short-term disturbances. And, the validator 22 associates an admissibility value with the value of the reference signal.

The regulation signal generator 10 also includes a Kalman objector 24, which is shown here as being operably connected to the validator 22 via lines 26 and 28. The observer 24 Kalman is further operably connected to the validator 22 through a counter-loop path 32. In the illustrated embodiment, the observer Kalman models the source of the reference signal and provides a counter-loop confidence interval value through the path 32 to the validator 22. The values of the counter-loop confidence interval are used. by the validator 22 to determine the possibility that the reference signal applied to the validator 22 is free of short-term disturbances. In a modality, the validator 22 dampens the sequences of the values of the reference signal and determines an average value of the damped sequence. If the average value is of a value within the values of the counter-loop confidence interval, the validator 22 validates the buffered sequence and passes an indication of the average value via line 26 to the observer 24 Kalman. The validator 22 is also capable of operating to associate an admissibility value with the damped sequence. The admissibility value, 'in a modality it is determined in response to the statistical variety of the average value of the damped sequence. If the individual values of the damped sequence are generally close to the average value, the variety of the average value is quite low and the admissibility value associated with this sequence is quite high. The admissibility value associated with the damped sequence validated by the validator 22 is provided, via line 28, to the observer 24 Kalman. The observer 24 Kalman, as mentioned above, forms a model of the source 14 of the reference signal. And the regulation signal generated on line 18 forms an improved version of the reference signal generated by the source 14 of the reference signal. The values of the regulated signal generated by the Kalman observer are altered by selected quantities in response to the average values of the damped sequences validated by the validator 22 and applied to the Kalman observer through line 26, together with the associated admissibility value with the same and applied to the observer Kalman on the line 28. That is, the observer Kalman forms a sequence based on a model of the source 14 of the reference signal. And, the sequence formed by the model is adjusted in response to the values provided by the validator 22 on lines 26 and 28. In one embodiment, the amount by which the values formed on the Kalman observer are adjusted and depend on the average value provided on line 26, together with the admissibility value provided on line 28. If the admissibility value is high, for a given average value, the amount by which the Kalman observer alters the sequence formed in the Kalman observer is greater than when the admissibility value is of a lower value. In this manner, the regulation signal generated on line 18 used to calibrate device 12 is formed from values modeled by the Kalman observer, but adjusted by values formed by the validator, when appropriate. When the reference signal generated by the source 14 of the reference signal exhibits short-term disturbances, the portions of the signal associated therewith are not used to adjust the values formed by the Kalman observer. And, even when the portions of the signal are validated by the validator, if the value of admissibility associated with them is low, the amount by which the values formed by the Kalman observer are adjusted is reduced. Figure 2 illustrates a "phase locked loop circuit shown generally at 42, where it is able to operate a mode of the present invention. During operation of the phase locked loop circuit, the oscillation frequency of a VCO (voltage controlled oscillator) 44 is controlled so that oscillator 44 generates an oscillatory signal on line 46, of a desired frequency. Through operation of one embodiment of the present invention, the oscillatory signal is caused to be of acceptable long-term and short-term frequency stability characteristics. Even when the exemplary phase lock loop circuit 42 shown in FIG. 2 will be described with respect to a mode in which the circuit 42 forms a portion of a radio base station capable of operating in a cellular communication system, circuit 42 can alternatively form portions of the other apparatus. The operation of the circuit 42 can be described in a similar manner with respect thereto. The phase locked loop circuit 42 includes a first reference signal source 52. The source.52 generates a reference signal on line 53 that exhibits good short-term frequency stability characteristics, but is susceptible to frequency deviation over long periods of time. The source 52 of the reference signal for example can be formed of an OVCXO that generates a quality signal of Stratum-2 but that exhibits frequency deviation over longer periods, e.g., in excess of one year. The first reference signal generated in line 53 is provided to a first feedback element 54. The oscillatory signal generated by the voltage controlled oscillator 44 is also provided through the feedback line 56 to the first feedback element 54. The feedback element 54 forms a first difference signal on the line 57 of values representative of the differences between the values of the first reference signal generated by the source 52, and the oscillatory signal generated by the VCO 44. The first signal of The difference formed in line 57 is provided at an input to a regulator 60. The oscillatory signal generated by VCO 44 on line 46 is also provided, via feedback line 56, to a second feedback element 58. The second feedback element 58 is further coupled to receive a second reference signal generated by a second reference signal source 62 through line 63. The second reference signal source 62 exhibits good stability characteristics, long frequency term, but is susceptible to short-term frequency instability.

The second reference signal generated by the signal source 62 is exemplary of a PCM clock signal. As described above, in some cellular networks, the PCM clock signal is not always guaranteed as being of a Stratum-2 quality. During some periods, which will be referred to as "retention", the second reference signal is less than the quality level of Stratum-2. The second feedback element 58 forms a second difference signal on the line 64 representative of the phase differences between the second reference signal and the oscillatory signal generated by the VCO 44. The second difference signal formed by the second element 58 of feedback is provided to a compensating value generator 65. The compensation value generator 65 is capable of operating to form a compensation signal of the values that respond to the values of the second reference signal but not to the values representative of the portions of the second reference signal that exhibit deviation to short term. The compensation signal is provided to a second input to the regulator 60 via line 66. The regulator 60 is capable of operating to form a regulation signal on line 68 to adjust the VCO frequency 44. The regulation signal formed by the regulator 60 is of values that correspond to the values of the first reference signal generated by the first reference signal source 52 that are adjusted at selected intervals by values of the signal of the reference signal. compensation value generated by the compensating value generator 65. The long-term frequency deviation to which the first reference signal is susceptible is compensated by the compensation value signal formed in response to the second stable reference signal in long-term frequency. When this regulation signal is provided to VCO 44, VCO 44 is maintained to oscillate at the desired frequency. Since the long-term frequency deviation of the first reference signal is compensated by the compensation value signal, the recalibration of the source 52 of the first reference signal is not necessary. The oscillatory signal generated by VCO 44 on line 46 in this manner is caused to exhibit good long-term and short-term frequency stability characteristics. Figure 3 illustrates a phase locked loop circuit shown here generally at 82, encompassed in a cellular radio base station. Again here, the oscillation frequency of a VCO 44 is controlled in such a way that an oscillation signal generated in the line 46 is of a desired oscillation frequency. The oscillation frequency of the oscillatory signal generated in line 46 is caused, through the operation of an embodiment of the present invention, to exhibit acceptable short-term and long-term frequency stability characteristics. The oscillation frequency of the VCO 44 uses a short-term stable Stratum-2 oscillator 84 in which the reference signal generated therefrom is calibrated using a stable long-term PCM clock signal provided to the radio base station wherein the PLL 82 is obtained through the line 86. The reference signal generated by the oscillator 84 is applied as a first input of a summing device 88 through the line 89. The oscillatory signal generated in the line 46 is provided through the feedback line 56 to a second input of the summing device 88. The PCM clock signal generated on line 86 is provided to a first input of the summing device 92. And the feedback line 56 on which the oscillatory signal is generated is provided to a second input of the summing device 92. The summing devices 88 and 92 each generate phase difference signals. The phase difference signal generated by device 88 is generated on line 93 and is representative of the phase differences between the oscillatory signal generated by VCO 44, and the reference signal generated by oscillator 84, Y, the phase difference signal formed by device 92 is generated on line 94 and is representative of the phase differences between the signal generated by the VCO and the PCM clock signal. The phase difference signal generated by the summing element is provided to the regulator 62, and the phase difference signal generated by the summing element 92 on the line 94 is provided to a compensation value generator 65 formed here of a processing device in which a calibration algorithm is capable of being carried out. The compensation value generator 65 generates a compensation value signal on the line 96 that is provided to a regulator 62. During the operation of an embodiment of the present invention, the compensation value generator 65 determines when the clock signal of PCM deteriorates temporarily, eg due to_ the retention of any of the elements of the network in the synchronization hierarchy on which the PCM clock signal depends. During these events the calibration of the reference source 84 is delayed.

During specific time periods, the sum of the two sources is carried out in the regulator 62. The contribution of the PCM clock signal defines the frequency deviation of the oscillator 84, Stratum-2 since there is a more recent previous sum . To calculate the deviation, the PCM clock signal is measured. If the PCM clock signal appears as deviating, this deviation is actually that of the Stratum-2 oscillator in an opposite direction. The measured deviation characteristics of the PCM clock signal are therefore the same as the deviation of the oscillator 84, Stratum-2. The operation of the compensation value generator 65 determines when the PCM clock signal is temporarily deteriorated. During these periods, the compensation value signal is not used by the regulator 62 to calibrate the reference source 84. And, the phase adjustment signal that is provided by the regulator 62 to VCO 44 includes, the calibration components that respond to the compensation value signal when the PCM reference signal is of good frequency stability characteristics.

Figure 4 illustrates a circuit of. phase lock loop (PLL), generally shown- at 102, wherein a mode of the present invention is also capable of operating. Here, again, during the operation of the PLL circuit, the frequency oscillatory of a VCO 44 is controlled. An oscillatory signal is generated by VCO 44 on line 46, and a feedback loop 56 provides indications of the oscillatory signal in a feedback arrangement, and, the oscillatory frequency at which the VCO 44 oscillates depends on an adjustment signal generated by a controller 103 and applied to the VCO 44. PLL circuit 102 is coupled to receive a multiple number, here N, of reference source signals.The reference source signals N each exhibit noise, with spectral densities having spectral components placed at different frequency intervals. Figure 3, three reference source signals are applied on lines 104, 106 and 108 to source-specific filters 112, 114 and 116, respectively Filters 112, 114 and 116 filter reference source signals applied The filter 112 forms a filtered signal on the line 117 which is applied to a first input of a summing element 118. The filter 114 forms a filtered signal in the line 119 that is applied to a first input of a sum element 122. And, the filter 116 generates a filtered signal on the line 123 that is applied to an input of a summing element 124. Line 56 of feedback, in which the oscillation generated by the VCO is formed, is also applied to the inputs of the sum elements 118, 122 and 124. Each of the sum elements 118, 122 and 124 forms a phase difference signal in three separate lines Í25. Each of these phase difference signals, in turn, is applied to a summing element 126. The reference source signals are thus combined in forms that depend on their respective noise spectral densities to form a single reference signal on line 127 that has less noise spectral density than any of the other reference source signals alone Through an appropriate selection of the reference source signals, the adjustment signal formed by the regulator 103 causes the operation of the VCO 44 to generate an oscillation signal having a stable long-term signal of low fluctuation, with an accuracy of high frequency. The reference source signals generated in lines 104, 106 and 108"can be selectively formed from the oscillation signal generated by the Stratum-2 oscillator, such as oscillator 84 shown in line 3, a clock signal of PCM, a GPS signal (Global Positioning System) or a reference signal formed by another type of reference signal source. analogy can be drawn between the generalized multiple reference mode of the PLL circuit 102 shown in Figure 4, with the PLL circuit 82 shown in Figure 3. Therefore, the compensation value generator 65 shown in Figure 3 it can be considered as being a specific filter source for the PCM clock signal. Figure 5 illustrates a regulation, signal generator that forms a compensating value generator 65 of an embodiment of the present invention. During operation, the regulation signal generator is capable of operating to calibrate a device, such as the device 12 shown in Figure 1. In an exemplary implementation, as described with respect to the aforementioned Figures 2 to 4, the The regulation signal generator forms a compensating value generator 65 of a PLL circuit. Although the following description of the exemplary operation of the regulation signal generator will be described with respect to an implementation in which the regulation signal generator forms a compensating value generator 65 of a PLL circuit of a radio base station In the case of a cellular phone, the operation of the generator can be described in a similar manner with respect to the calibration of the devices that are not a VCO of a PLL circuit.

The compensation value generator 65 is functionally shown as being formed of a validator 152 and a Kalman observer 154. In one embodiment, the validator 152 and the Kalman observer 154 comprise algorithms capable of being carried out by a processing device. A phase difference signal is provided to validator 152 through line 156. The signal is used to form measures proportional to the deviation of the average frequency of the reference signal during each sample period. The validator 152 is further coupled through a counter-loop path 158 to receive a confidence interval formed by the Kalman observer. The validator 152 is capable of operating to validate the calculated measurements of the signal of phase difference applied to it based on the confidence interval provided by the Kalman observer through the counter-loop path 158. The values indicative of the portions of the calculated measurements of the phase difference signal determined by the validator as being valid, are passed through the validator 152 on the line 162 and provided to the observer 154 Kalman. The admissibility values associated with these values are passed through the line 164. The observer Kalman generates a compensation value on line 166. The observer 154 Kalman uses a model of the aging of a Stratum-2 oscillator. The model forms a calculation of the aging of the Stratum-2 oscillator. This calculation, and the normal deviation of the calculation, forms a confidence interval, the indications of which is provided through the counter-loop path 158 towards the validator. Validator 152 in this manner is capable of delaying calibration during periods in which the PCM clock link is on hold. When in the retention state, the calculated measurements of the phase difference signals provided to the validator 152 are of values that are beyond the confidence interval provided to the validator by the Kalman observer. The aging of the Stratum-2 oscillator can be modeled with a state-space model defined as follows: W I0 °] N. { ">, where X] _ is the error in hertz; X2 is the time derivative of the oscillator aging; Up i = 1,2 are uncorrelated with white Gaussian noise with the spectral density ^ 1 and y 2 'respectively. The model is therefore a combination of a step and a ramp, that is, a line with an offset. This model is called a "high Alian variation" model and is the short-term description of the aging of an oscillator. Long-term aging is also represented by a non-linear model, typically a better model, described by the equation:? = / "(1 * AJn (Bn-1)) where f (t) is the frequency deviation during time t. Since the model used by the Kalman observer 154 is used among the samples provided to the validator 152, the short-term description model of the oscillator is used appropriately. The analysis of long-term aging indicates, however, that aging slows down and is not an accelerative process. The coincidence between the spectral densities of white Gaussian noise and the Alian variety is as follows:?, = 2A0? = 8p2 / j_. where hg and h_2 are the Alian vapacioines to take samples of a few minor and major decades of the "flicker floor" in the Alian variation model. A Stratum-2 oscillator exhibits approximately Alian variations and spectral densities of: During retention periods, the PCM clock signal exhibits frequency stability characteristics that correspond to the quality levels of Stratum-2 to Stratum-3. A Stratum-3 oscillator exhibits the following Alian variations and spectral densities: A0 = 8 * 10-M ?, = l 6-10- "h_y *? Ov and? 2 = 32-10-51 The measured quantity is the aging of the oscillator, x ^ Since the characteristics of the noise are usually different for each synchronization network, the noise characteristics do not form portions of the model used by the Kalman observer.The model of the measurements is therefore the same as the model of the aging of the oscillator thereby motivating the device 65 in Figure 3 to be viewed as a source-specific filter for the PCM clock signal.

Assuming that the noise of the system is constant between the sampling cases, it allows the transformation of the oscillator model in a discrete time model represented as follows: 01J J + tl -where T is the sample period. The covariance matrix of the vector? J_ k multiplied with the matrix ahead of it in the aforementioned equation, is independent of the time k of the sample and is referred to as Q. The variation of the calculated measure of the deviation of frequency in the sample k is called R ^. The observer Kalman in the exemplary mode forms a non-stationary Kalman observer. This observer Kalman includes the following model of state space: VL eWP.Qí where "* e? R (°?) The non-stationary Kalman observer exhibits a state vector x ^ which is the calculated state vector of the observed system based on the measurements of 1 to k-l, which minimizes EEx ^ - xj) 2], where X] < it is the true state vector. The observer Kalman is able to carry out the following calculations during the iteration of the kth: wy ** where e¡_ is the innovation, the measured output minus the predicted output. Then the co-variation of the innovation Sj- is the sum of the co-variance of measurement and the co-variation of prediction in the following way: sypßty The Kalman gain K ^, is then calculated as follows:? Ypfl1s- The observer Kalman is advanced in a step of time and the correction based on the calculated innovation and the Kalman gain is done as follows: Then, the co-variation, P, for ^ + i is calculated as follows: FÍPk-PFtSXHPk) F Q For each new iteration, the Kalman observer must obtain a new measurement and j, and a variation R ^ of measurement noise. In the general case, all matrices and values may depend on the sample k, but here, these values are constant. If no measurement is available, Rj ^ adjusts to infinity. In this way, the co-variation of the innovation also fits the infinity and Kj ^ 0, X] c + 1 = Fxjc and Pjc + 1 = FP] CFT + Q. As priority knowledge, no additional information is required except for the values of the model F and H matrices, the co-variation matrix of the Q system noise, the initial state vector xo and its covariation matrix Pg. To ensure that the Kalman observer is not operated erroneously, the following limitations are introduced into the Kalman observer: 1) Aging per second, x2, can never exceed one aging value per second; 2) The maximum correction, K (y-ypr¿-) can never exceed the normal deviation for three samples (plus the number of samples adhered to it that jumped one due to retention); and 3) If a sequence of 'measurements is skipped, x2' is divided by two.

Robust observer instead of a quick observer. If a sequence of measurements is skipped, it is reasonable to assume that the value before the skipped sequence was stained and the derivative of aging could be poorly calculated. In this case, a prolonged retention could cause the calculation of X] _ to deviate away from many aging measurements in the wrong direction. To fully calculate the risk, x2 must be set to 0 in a hold. A commitment is to divide and.-as mentioned. During the operation of the Kalman observer in the operation of the iterative steps listed above, the normal deviation, s, of JCT is calculated. The value of the normal deviation is used to calculate the confidence interval used by the validators. using the equation: threshold = max (? s, srj) confidence interval =] _ threshold where a and kj are predefined constants. If the difference between xi and the true aging of the actual oscillator is very large, many samples are required before the threshold has grown to be large enough. Once a mediation has been accepted, the normal deviation and therefore the threshold becomes very small. The limitations in Updates in the observer do not allow a sufficiently large correction to be obtained and the Kalman Observer is back at its starting point, ie large error and small threshold. To prevent an endless loop after a selected unsatisfactory period of time, a fast synchronization mode of the Kalman oberver is started instead. That is, when the performance of the Kalman observer unsatisfactorily follows the aging of the oscillator for a selected period of time or. When the radio base station is first powered up, the quick synchronization mode starts. The use of the fast synchronization mode allows the Kalman observer to obtain a calculation x within the proximity of the true aging of the oscillator. Then, the normal functioning of the Kalman observer resumes- The rapid synchronization mode is based on the assumption that the Kalman observer itself can handle smaller retentions since these smaller retentions cause more variety and therefore, more influence. little. The threshold of the confidence interval can therefore be raised to become: threshold = max (2, s, s] _) confidence interval = X] _ threshold where sj_ and 2 are predefined constants. Also, when in this mode, the oscillator model is changed to that of a Stratum-3 oscillator to force a higher bandwidth. And, 5 no maximum correction per update is used. The fast synchronization mode ends after a fixed period of time assuming that the aging calculated during this time is within a predefined interval. As the compensation value generated on line 166, the calculated aging of the stratum 2 oscillator, X] _, is provided of course. In one embodiment, the alteration of the compensation value is carried out selectively in response to an indication of the quality of the signal received on the line 156. This quality indication is typically the fraction of the measurements discarded since the last update of the compensation value or the normal deviation of the aging s. In this way, the quality indication is used to decide if the aging is currently Calculated from the stratum 2 oscillator by the observer 154 Kalman is used or not as the compensation value _ provided on line 166 or is considered as being, not reliable. If it is considered to be unreliable, the compensation value provided on line 166 is not altered.

In one embodiment, the validator 152 is operable to perform rejection of portions of the signal that are provided thereto on line 156. A sequence of the signal provided thereto is stored in a buffer and the average value together with its value of _ normal deviation, it is calculated for measurements during a selected period of time. The average is compared to a confidence interval that is provided through the path 158 of the counter-loop. If the average value is outside the confidence interval, the value is referenced as an isolated result. If the average value falls within the confidence interval, the validator accepts the buffered sequence and the determined average value along with its normal deviation is provided to the observer Kalman. If the value is determined to be an isolated result, the value is discarded and the observer uses Kalman's model to update the calculated aging of the oscillator and its normal deviation. A new measurement period is then started. The rejection method and the isolated result can be implemented either by using a measurement buffer and then calculating an average and normal deviation or a "flying" algorithm. The average value determined by the validator 152 together with its normal deviation is calculated from the. measurements made during a selected period of time. The storage of the sequences in a buffer reduces the need to carry out calculations at each instant of measurement. The average ? and the normal deviation of the sequence, s, is calculated as follows: where N is the number of symbols that form a sequence stored in a buffer and xj ^ is the measurement number k. The variation, See (?), Can be calculated as follows: Var (? = Var ±? '? T-ix¿ = - / V;' SKí Var ^ The calculated value of the variation of the average is used as the variation of the measurement and the average is used as the measurement in the Kalman observer, with the condition that the average is within the confidence interval of: calculated afiejamlento +/- threshold In an alternative modality, an observer Kalman is not stationary based on a model that describes a constant, updated at each instant of measurement, is used to form the average and its variation of the measurements calculated during a selected period of time. If the calculated constant is within the confidence interval, the same along with its variation, the Kalman observer is provided as the measurement and the measurement variation. In another embodiment, the validator 152 validates the signals provided thereto through line 156, using a change detection method. In this method, the values of the signal received in the validator 152 in line 156 during the selected time intervals are used to calculate the measurements that are stored in a buffer. A change detector is then used to find the changes in the buffer that could be caused by a hold. The periods of unlikely measurements caused by retention or other malfunction are isolated and discarded. The average of the remaining sequence stored in the buffer together with its normal deviation is provided to the observer Kalman. Change detection, as can be seen below, can also be implemented using multiple buffers. The lengths of the buffers are selected so that deviations caused by retention can be detected.

Figure 6 functionally illustrates the validator 152 using a two-buffers method that uses two buffers, here the buffers 172 and 174, each capable of storing N shaped values of a signal provided to the validator 152 through the line 156 The signal provided through line 156 is used to form measurements proportional to the average frequency deviation of the reference signal during each sample period. The measurements are filtered by a prefilter 176 from where samples are taken by the sampler I78 and the components of the sample signal are stored in the buffer 172. The measurements are subjected to sample measurement by the sampler 182 and the sampled components they are stored in the buffer memory 174. A change detector 184 detects changes in the measured values of the sampled signal components stored in the buffer memory 172, as described above. A determination is made by the switch 184 to use or discard portions of the stored sequence. In response to the determination, an element, switch 186 is either closed or open to thereby provide selective indications of the values validated in the validator to the Kalman observer 154.

During operation of the validator 152 of the embodiment illustrated in Figure 6, the average of each of the measurements M is stored in the buffers 172 and 174. The measurements stored in the buffer memory 172 are pre-filtered by the pre-filter 176 for suppress fluctuations by allowing small low frequency deviations to be detected and measurements stored in buffer 174 are stored only thus preventing, that the measures have influence with respect to each other. The operation of the change detector 184 in the values stored in the buffer memory 172 detects the low frequency changes of the values stored therein. In response to the determination made by the change detector 184, the values stored in the buffer 174 are selectively discarded. Non-discarded parts are provided to the Kalman observer, that is, the average and its corresponding variety is provided to the observer Kalman. Figure 7 illustrates a method generally shown at 202 of one embodiment of the present invention. The method calibrates a device that responds to the values of a reference signal in which the reference signal is subjected to short-term disturbances.

First and as indicated by block 204, the values of the portions of the reference signal are validated to have a tendency to be exempt from short-term disturbances where an admissibility value is associated with the portions of the reference signal . Then, and as indicated by block 206, the indications of portions of the reference signal having admissibility values at least as large as a selected quantity, together with the admissibility value associated therewith, are provided to a buffer of a Kalman observer. Then, and as indicated by block 208, the stored values that are stored in the Kalman observer are selectively altered in response to the indications provided thereto. Then, and as indicated by block 214, a regulation signal is formed in response to the stored values that are stored in the Kalman observer. And as indicated by block 216, the regulation signal is applied to the device. In this way, through the operation of one embodiment of the present invention, a device can be calibrated using a reference source having short frequency stability characteristics. acceptable term, together with a reference source that has acceptable long-term frequency stability characteristics. The above descriptions are of preferred examples for implementing the invention, and the scope of the invention should not necessarily be limited by this description. The scope of the present invention is defined by the following claims.

Claims (27)

R E I V IN D I CA C I O N E S:

1. An apparatus for regulating the frequency of an oscillatory output signal formed by an oscillator of an oscillator circuit, the apparatus comprises: a first feedback element coupled to receive the output oscillation signal and to receive a first reference signal, the first The reference signal exhibits short-term frequency stability characteristics at least as good as a first level of selected frequency stability, the first feedback element for forming a first difference signal, representative of the deviation of the oscillation signal from output in relation to the first reference signal; a second feedback element coupled to receive the output oscillation signal and to receive a second reference signal, the second reference signal exhibits long-term frequency stability characteristics at least as good as a second level of frequency stability selected, the second feedback element for forming a second difference signal, the second difference signal representative of the deviation of the output oscillation signal relative to the second reference signal; a compensating value generator coupled to receive the second difference signal, the compensation value generator to form a compensation value signal, the compensation value signal formed in response to the values of the second difference signal when the second reference signal exhibits short-term frequency stability characteristics at least as good as a third level of selected frequency stability; and a regulator coupled to receive the first difference signal and, at least, to selectively receive the compensation value signal, the regulator for modifying the first difference signal responds to the compensation value signal and to form a control signal to control the frequency of the oscillator.

2. The apparatus of claim 1, wherein the first reference signal exhibits short-term frequency stability characteristics, at least as good as the Stratum 2 level.

3. The apparatus of claim 2, further comprising a Stratum 2 oscillator, the Stratum 2 oscillator to generate the first reference signal.

4. The apparatus of claim 1, wherein the second reference signal exhibits characteristics of long-term frequency stability, at least as good as the Stratum 2 level.

The apparatus of claim 4, wherein the oscillation circuit forms a portion of a radio base station of a radio communication system. , and wherein the second reference signal to which the second feedback element is coupled to receive, comprises a signal generated per network.

The apparatus of claim 5, wherein the radio communication system comprises a cellular communication system and wherein the second reference signal comprises a PCM clock signal.

The apparatus of claim 1, wherein the second reference signal comprises a signal generated by a network, and wherein the compensation value generator eliminates the fluctuation and "vagabond" components of the signal generated by the network.

The apparatus of claim 7, wherein the oscillation circuit forms a portion of a radio base station of a radio communication system, and wherein the signal generated by the network comprises a PCM clock signal.

The apparatus of claim 7, wherein the oscillation circuit forms a portion of a radio base station of a radio communication system, and wherein the signal generated by the network comprises a clock signal received in the GPS receiver capable of operating in a GPS network.

The apparatus of claim 8, wherein the compensation value generator includes a Kalman observer and measurement validator, the observer Kalman to calculate the deviation of the frequency of the first reference signal and the measurement validator to validate the values of a second difference signal.

The apparatus of claim 10, further comprising a Stratum 2 oscillator for generating the first reference signal wherein the first reference signal is susceptible to frequency deviation due to aging of the Stratum 2 oscillator, and wherein the The first reference source calculated by the Kalman observer comprises calculations of the frequency deviation due to setting of the Stratum 2 oscillator.

The apparatus of claim 10, wherein the measurement validator validates the measurements by performing detection calculations. of change in the second signal of difference.

The apparatus of claim 10, wherein the measuring validator validates the values of the second difference signal that are determined to deviate from a calculated value less than a selected amount.

The apparatus of claim 10, wherein the measurement validator comprises a processing apparatus having a measurement alidation algorithm that is carried out thereon, the measurement validation algorithm when carried out for validate the measurements received during a selected time interval and to provide admissibility signals associated with the measurements indicative thereof to the Kalman observer.

The apparatus of claim 14, further comprising an intermediate buffer measurement for storing values of the measurements therein; the values stored in it allow, when analyzed, the operation of the deviation analysis.

The apparatus of claim 15, wherein the buffer comprises a plurality of buffers, the values stored in the plurality of buffers to allow different levels of filtering.

The apparatus of claim 16, wherein the measurements are validated by the measurement validator if the measurements fall within a range of Counter-loop confidence that is provided to the measurement validator by the Kalman observer.

18. The apparatus of claim 17, wherein the measuring validator further determines whether the values indicative of the second difference signal have a chance of not being accurate, and selectively provides the values indicative of the second difference signal to the Kalman observer in answer to this

19. The apparatus of claim 10, wherein the Kalman observer has a bandwidth associated therewith and wherein the observer Kalman uses multiple models of the aging of the Stratum 2 oscillator to vary the bandwidth associated with the Kalman observer.

20. The apparatus of claim 10, wherein the observer Kalman has associated with it the correction levels and state variables, and wherein the successive calculations made by the observer Kalman are within the maximum correction levels and within the variables of maximum status.

21. The apparatus of claim 10, wherein the update of the compensation values is selectively made in response to the indications of the quality levels of the second reference signal.

22. The apparatus of claim 21, wherein the indications of the quality levels comprise indications of normal deviations from the calculation. of aging.

23. The apparatus of claim 1, wherein the compensation value generator comprises a processing apparatus having a calibration algorithm capable of being operated thereon, the calibration algorithm when carried out is to form values of the signal of the compensation value.

24. A method for regulating the frequency of an output oscillation signal formed by an oscillator of an oscillation circuit, the method comprising the steps of: forming a first difference signal representative of the deviation of the output oscillation signal with Referring to a first reference signal, the first reference signal exhibits short-term frequency stability characteristics at least as good as a first level of selected frequency stability; forming a second difference signal representative of the deviation of the output oscillation signal relative to a second reference signal, the second reference signal exhibits long-term frequency stability characteristics at least as good as the second at the selected frequency stability level; forming a compensation value signal in response to the values of the second difference signal when the second reference signal exhibits short-term frequency stability characteristics, at least as good as a third level of selected frequency stability; selectively modifying the first difference signal in response to the compensation value signal to form a control signal; and apply the control signal to adjust the frequency of oscillation of the oscillator.

25. A phase locked loop circuit for regulating the frequency of an oscillator of an oscillating circuit, the phase locked loop circuit comprises: a first reference source for forming a first reference signal, the first The reference signal has a noise component that exhibits a first spectral density; at least one second reference signal to form at least the second reference signal, the second reference signal has a noise component exhibiting at least a second spectral density, the The noise component of at least the second reference signal has spectral components at different frequencies than the noise component of the first reference signal; a combiner at least selectively coupled to receive the first reference signal and at least the second reference signal formed by the first reference source and at least the second reference source, respectively, the combiner to selectively combine the first signal reference and the second reference signal and to form a resultant signal thereof, the resulting signal being to control the oscillation frequency of the oscillator to thereby block the oscillator in phase to oscillate at a selected oscillation frequency.

26. The phase locked loop circuit of claim 25 wherein at least the second reference source comprises a plurality of reference-sources, each reference source of the plurality forms a reference signal having a reference component. noise that exhibits a spectral density that has spectral components at different frequency intervals than the noise of the other signals reference formed by the other reference sources.

27. A calibration method for calibrating a device that responds to the values of a reference signal, the reference signal is subject to short-term disturbances, the method comprises the steps of validating the values of the portions of the reference signal that have the possibility of being free of short-term disturbances; associating an admissibility value with at least the portions of the reference signal validated during the validation step; providing indications of the portions of the validated reference signal during a validate step together with the admissibility values associated therewith to a Kalman observer; selectively altering the stored values that are stored in the Kalman observer in response to the indications and the admissibility values associated with them that are provided during the step of providing; to form a regulation signal that responds to the stored values that are stored in the Kalman Observer; Y Apply the regulation signal to the device to calibrate the device with it.