JPH08226946A - Jitter spectrum analyzer - Google Patents

Abstract

PURPOSE: To aim at operation approximation in the case where a jitter amount (fn) is small compared with a time interval T and convert into the jitter amount (fn) of regular time intervals by a comparatively simple operation means without finding a curve 58 for performing curve fitness. CONSTITUTION: An average time interval value T is calculated from data arrays of irregular time intervals by an average time interval operation part 56, and time series data of jitter amounts tn, τn of each time are computed. A regular time interval data operation part 12 or the like for calculating the jitter value of the n-th position is provided from three data of the jitter value τn of the n-th position, the jitter value τn-1 just before it and the jitter value τn+1 just after it by using the time series data. The data array of the jitter values (fn) for the regular time interval T is output by successively calculating it for the time series data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a jitter spectrum analyzing apparatus which receives measurement data having a jitter component contained in a signal under measurement, that is, non-uniform time interval data, and which can be operated by an FFT processing section. The present invention relates to a conversion device for converting a constant time interval data string.

[0002]

2. Description of the Related Art FFT processing is performed in order to analyze a jitter spectrum of an input signal. However, as input data of this FFT processing, FFT processing cannot be performed unless it is a data string having a jitter amount at fixed time intervals. Therefore, it is necessary to convert the data sequence measured as a data sequence at non-equal time intervals due to jitter into a data sequence format that can be FFT processed. For this purpose, for example, by using the spline interpolation method, a curve fitting (curve fitting)
It is necessary to perform a pre-processing conversion to obtain a curve to be fitted), read the value on the curve at an averaged fixed time interval position from this curve, and obtain an isochronous data string of the jitter amount.

As an example of the prior art, there is an example of conversion into a data string of equal time intervals by a spline interpolation method. This will be described with reference to FIGS. 5, 6, and 7. As shown in FIG. 5, the equal time interval data conversion unit 50 of the present apparatus has a time interval measurement unit 52 and a measurement data memory 54.
And an average time interval calculation unit 56 and an equal time interval data calculation unit 58. This output data is supplied to the FFT calculation unit 70 to obtain the spectrum of the jitter component.

The time interval measuring unit 52, as shown in the waveform data example of FIG. 6, digital data 100dat obtained by counting the time interval values T1 to Tn of the rising edges of the two pulses of the signal under measurement 100 with high resolution. Are continuously stored in the measurement data memory 54 in the order of n points. Since this data string contains a jitter component, it is a data string at non-equal time intervals.

The average time interval calculation unit 56 calculates the average time interval value T by adding and averaging the obtained n points of data, and T = Σ (Tn) / N with n = 1 to N. Ask in. This is supplied to the equal time interval data calculation unit 58.

The equal time interval data calculation unit 58 obtains the jitter amount data string (fn) at the reference time point (nT) obtained by integrating the average time interval value (T). That is, the measurement time point corresponding to the reference time point (see FIG. 3) (tn = ΣTi, where i = 1 to n), the difference, that is, the jitter amount (τn = nT−
tn) is calculated. Thus, the data string of the jitter amount at the measurement time point is obtained. Now, plotting this time series, (tn, τn), with the horizontal axis representing time and the vertical axis representing the amount of jitter, it becomes as shown in FIG. 7 (marked with a cross). Next, by the spline complement method, a curve 58curv that fits the curve is obtained by passing through the time series coordinate points (marked with X) of this jitter amount, and the value on the curve at the reference time (marked with ○) is calculated. , The data amount of the jitter amount at equal time intervals (f
get n). By supplying the data string of the jitter amount (fn) to the FFT calculation unit 70, the spectrum of the jitter component can be obtained by the FFT calculation.

[0007]

As described above, in order to obtain the jitter amount (fn) at each reference time point position, it is necessary to find the curve-fitting curve 58curv. The calculation for obtaining this curve has a drawback that it takes time.

Therefore, the problem to be solved by the present invention is to find the curve 58curv that fits the curve, paying attention to the fact that the calculation can be approximated when the jitter amount (fn) is sufficiently smaller than the time interval T. Instead, the purpose is to convert the jitter amount (fn) at equal time intervals by a relatively simple calculation means.

[0009]

Thus, the time interval measuring section 52 measures the time interval sequence of the signal under measurement 100, and the spectrum of the jitter component from the data sequence T1 to TN at the non-equal time intervals due to this jitter is measured. For the purpose of analysis, a conversion device is realized to convert a data string of non-uniform time intervals due to jitter into a data string (fn) of jitter components at equal time intervals that can be FFT processed.

As a concrete means of the equal time interval data calculating unit 12, fn = {Tτn / τn-τn-1} {1 + 1/2
The jitter value fn for each equal time interval is obtained by the calculation means of {{τn + 1 + τn-1-2τn} / (τn−τn-1) ² }}. That is,
Jitter value τn at non-equal time intervals and immediately preceding jitter value τn-1
Then, the jitter value (fn) for the current position (nT) is calculated from the three data of the jitter value τn-1 immediately after. This is sequentially calculated and an equal time interval data string of the jitter amount (fn) is output.

[0011]

The jitter component is the time interval value T of the signal under measurement 100.
If it is smaller, the approximate conversion formula (13) can be used to easily calculate the data converted into the jitter sequence at the current position at equal time intervals. Moreover, it is only necessary to use the three data of the jitter value τn at the current position, the jitter value τn-1 immediately before, and the jitter value τn-1 immediately after.

[0012]

The embodiment of the present invention is an example in which the jitter value (fn) is calculated by the approximate calculation means. About this,
It will be described with reference to FIGS. 2, 3, and 4. In the example,
The jitter component will be described as a function f (t) of t. As shown in FIG. 1, the isochronous data conversion unit 10 of this apparatus has a time interval measurement unit 52 and a measurement data memory 54.
And an average time interval calculation unit 56 and an equal time interval data calculation unit 12. This is a part changed by the equal time interval data calculation unit 12 from the conventional one.

In general, the pulse train shown in FIG. 3 is expressed as x (t) = Σp (t−nT) at values of n = −∞ to + ∞. Here, T is the repeating average pulse period, and t is t time. Since the actual signal under measurement 100 includes the jitter component f (t), x (t) =
It is expressed as Σp (t−nT + f (t)) (3). In the jitter analysis, this jitter component f (t) is a measurement target. In the time interval measuring unit 52, the time interval values T1 to Tn of the signal under measurement 100 are measured and stored in the measurement data memory 54 as in the conventional case.

In the equation (3), since the rising time of the pulse is the time when the equation in p () becomes zero, t-
nT + f (t) = 0 ... (4) is solved for t. This can be obtained according to the value of n, so tn
I will write. The relationship between f (t) and tn can be shown by FIG. That is, tn is y = nT-t,
It is the x coordinate of the intersection of y = f (t). If f (t) ≡0, then tn = nT, and at equal time intervals, f
Since (t) exists, tn deviates from nT. On the other hand, according to FIG. 3, the rising time (tn) of the pulse is obtained from the measured time interval data (Tn) by tn = ΣTi (here, i = 1 to n). Next, using T obtained by the average time interval calculation unit 56, the jitter amount τn = nT−
tn is calculated. Thus, time series data (tn, τ
n) is obtained. From equation (4), it is clear that this (t
The time series data of (n, τn) is equal to (tn, f (tn)). That is, it is time-series data obtained by sampling f (t) at each time tn.

Next, the equation (4) is generalized to t-τ + f
Consider the equation (t) = 0 (5). It is assumed that this is solved for t and the solution t = g (τ) is obtained. Since tn is a solution when t = nT in the equation (5),
tn = g (nT). Therefore, time series (nT, τn)
Is (nT, τn) = (nT, nT-tn) = (nT, n
T-g (nT)). Now, h (t) ≡t−g (t)
, The time series (nT, τn) is a time series in which h (t) is sampled at every time nT. That is, τn
Can be regarded as h (t) isochronous sampling data. Here, if the functional relationship between f (t) and h (t) is clarified, it is understood that τn can be used to obtain isochronous sampling data of f (t).

The relationship between f (t) and h (t) is obtained.
The equation (5) can be easily rewritten as an equation τ = t + f (t) ... (6) solved for τ. On the other hand, the equation solved for this is t = g (τ) = τ−h (τ) ... (7). By substituting the equation (6) into the equation (7), t = t + f (t) -h (t + f (t)) ∴f (t) = h (t + f (t)) ... (8) is obtained. , F (t) and h (t) have been clarified.

Now, let | f (t) | << T. This condition is sufficiently satisfied in ordinary jitter measurement.
When the Taylor expansion up to the second-order of f (t) is performed on the right side of the equation (8), h (t + f (t)) = h (t) + h '(t) f (t) + (1/2) h "(T) f (t) ² ... (9) where h '(t) and h" (t) are the primary and 2 of h (t).
Is the derivative of Substituting equation (9) into equation (8), f (t) = {h (t) + (1/2) h "(t) f (t) ² } / {1-h '(t)} ... (10) is obtained.Equation (10) is not a closed expression for f (t), so the right side is further substituted for f (t) on the right side, and 2 of h (t) is obtained. When it is taken up to the next, it becomes like the expression (11): f (t) = {h (t) / {1-h '(t)}} + (1/2) [{h "(t) / {1 -H '(t)}] [h (t) / {1-h' (t)}] ² ... (11)

From equation (11), from τn, fn≡f (nT)
Then, the formula to be obtained is derived. Now, assuming that the spectrum of f (t) has only frequency components sufficiently smaller than 1 / T, h (nT) = τn h ′ (nT) ≈ (1 / T) (τn−τn-1) It is assumed that h ″ (nT) ≈ (1 / T ² ) (τn + 1 + τn−1−2τn) ... (12) holds. Substituting these equations (12) into the equation (11), fn = { T [tau] n / ([tau] n- [tau] n-1)} {1+ (1/2) {{([tau] n + 1 + [tau] n-1-2 [tau] n)} / ([tau] n- [tau] n-1) ² }} ... (13) is obtained. From the sequence of τn using this equation (13),
By calculating fn, it is possible to calculate the approximate value series of the equal time interval data of f (t) which is the jitter component.

That is, from the equation (13), as shown in FIG. 2, three data of the jitter value τn at the present non-uniform time interval, the immediately preceding jitter value τn-1 and the immediately following jitter value τn-1 are obtained. Therefore, it is shown that the jitter value for the equal time interval T of the current position can be easily converted and calculated. From this, it is possible to convert to an approximate value series data format that is capable of FFT processing without obtaining the curve 58curv that is curve-fitted as in the conventional case. Jitter value (fn) for this equal time interval
The calculation means is provided in the equal time interval data calculation unit 12.

[0020]

Since the present invention is configured as described above, it has the following effects. If the absolute value | f (t) | of the jitter component is sufficiently smaller than the time interval value T (= T) of the signal under measurement 100, then | f (t) |
(T) | << T is satisfied. From this, it is possible to easily convert and calculate the jitter sequence data of the current position at equal time intervals by using the approximate calculation formula (13). Moreover, it is only necessary to use the three data of the jitter value τn at the current position, the jitter value τn-1 immediately before, and the jitter value τn-1 immediately after. As a result, there is an effect that the data can be converted into a data string format that can be FFT processed without obtaining the curve 58curv that is curve-fitted as in the conventional case.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an isochronous interval data conversion unit for jitter spectrum analysis according to the present invention.

FIG. 2 is a diagram illustrating conversion of a non-uniform time interval data string into an equal time interval data string according to the present invention.

FIG. 3 is a diagram illustrating a time interval T of the signal under measurement 100 according to the present invention.

FIG. 4 is a diagram illustrating a relationship between f (t) and tn according to the present invention.

FIG. 5 is a configuration diagram of a conventional isochronous interval data conversion unit for jitter spectrum analysis.

6 is a diagram illustrating waveform data and measured values of a signal under measurement 100. FIG.

FIG. 7 is a diagram for explaining conversion from a conventional curve fit curve to a jitter amount of an equal time interval data string.

[Explanation of symbols]

 10, 50 Equal time interval data conversion unit 12 Equal time interval data calculation unit 52 Time interval measurement unit 54 Measurement data memory 56 Average time interval operation unit 58 Equal time interval data calculation unit 58 curv curve 70 FFT operation unit 100 Measured signal 100dat Digital Data τn, τ1 to τn Non-equal time interval jitter sequence T1 to Tn Non-uniform time interval data sequence fn, f2 to fn-1 Equal time interval jitter sequence T Average time interval

Claims (2)

[Claims] 1. A time interval measuring section (52) measures a time interval sequence of a signal under measurement (100), and a jitter component having an equal time interval from a data sequence (T1 to TN) of non-uniform time intervals due to this jitter. In the device for converting into the data sequence (fn) of the above, the average time interval calculation unit (56) calculates the average interval value T from the data sequence of unequal time intervals, and from this, the time series of the jitter amount (τn) at each time is calculated. The data is calculated, and using this time series data, the jitter value at the n-th position is determined from the three data of the jitter value τn at the n-th position, the jitter value τn-1 immediately before, and the jitter value τn-1 immediately after. A jitter spectrum analyzer characterized in that it is provided with an equal time interval data calculation means for calculating (fn) and is equipped with the above. 2. The equal time interval data calculating unit means: fn = {Tτn / (τn−τn-1)} {1+ (1/2)
{{(Τn + 1 + τn-1-2τn)} / (τn−τn-1) ² }}
The jitter spectrum analyzer according to claim 1, wherein the jitter spectrum analyzer is calculated by the following equation.