JP2006030112A - Measurement method for device characteristic - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a measurement method for device characteristics to enhance measurement accuracy by reducing the effect of external noises. <P>SOLUTION: All of data obtained by n-times of measurement under the same measurement conditions, are not integrated at one time, but are divided into m and put to repetitive integration of rather short time period. The results thereof are statistically processed to calculate the mean and standard deviation σ on the integrated data. When dispersion is found equal to or more than a prescribed value, it is determined that external noises have come in to delete data widely deviating from the mean. Lacking data corresponding to the deleted data are supplemented by performing additional measurement. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a device characteristic measurement method, and more particularly to a technique that is effective when applied to measurement that is easily influenced by external noise.

  As a technique studied by the present inventor, for example, in a 1 / f noise measurement system of a MOS (Metal Oxide Semiconductor) device or a measurement apparatus (spectrum analyzer or the like) controlled by software, when measuring a weak signal, A technique is used in which accuracy is obtained by performing integration over a long period of time and obtaining an average thereof.

  By the way, as a result of examination of the measurement technique as described above by the present inventors, the following has been clarified.

  As described above, a normal digital measurement apparatus has a function of performing long-time integration in order to improve measurement accuracy.

  However, as the integration time becomes longer, short-term instantaneous noise due to ON / OFF of the contact point of the external power device enters during that time, and an error (peak) occurs in the measurement result. When the electromagnetic shield is insufficient and there is an on / off switch of a large current circuit near the measurement system, electromagnetic noise generated from this switch enters. These noises are at a level sufficient to degrade the S / N ratio for a short time. Especially in weak signal measurement, the measured value may reach a meaningless level. Since these noises often occur suddenly for a short time, the longer the integration time, the more likely they are affected. As a result, the effect is offset even if the integration time is set longer to improve the measurement accuracy.

  In addition, while performing measurement for a long time, the characteristics of the device under measurement change, and the shape of the measurement result (characteristic curve) changes.

  In addition, spectrum analyzers, etc., have different optimum measurement condition settings depending on the frequency band to be measured. However, in conventional systems, the conditions for sweeping were fixed, so it was difficult to obtain the optimum spectrum. .

  Therefore, an object of the present invention is to provide a device characteristic measurement method that can reduce the influence of external noise and improve the measurement accuracy.

  Another object of the present invention is to provide a device characteristic measuring method capable of recognizing the occurrence of device characteristic variation.

  Another object of the present invention is to provide a device characteristic measurement method capable of reducing measurement time while ensuring measurement accuracy.

  The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

Of the inventions disclosed in the present application, the outline of typical ones will be briefly described as follows.
(1) The device characteristic measurement method according to the present invention does not integrate all measurement data under the same measurement conditions at once, but repeatedly performs integration for a short period of time in several times, and statistically analyzes the results. If there is more than a predetermined variation, it is determined that external noise has entered, data that deviates significantly from the average is deleted, and the missing data for the deleted amount is supplemented by additional measurements. Is.
(2) The method for measuring device characteristics according to the present invention is not to sweep a measurement condition such as voltage in one direction or in a reciprocal manner, but to make rough (rough) measurements every few, This is repeated by shifting one point at a time to complete the entire sweep.
(3) In the device characteristic measurement method according to the present invention, the frequency distribution of the external noise is measured in advance, the frequency at which the external noise exists is registered in the database, and the measurement at the frequency at which the external noise exists is performed at the time of measurement. Do not do it.
(4) The method for measuring device characteristics according to the present invention is to optimally set the measurement conditions (resolution bandwidth (RBW), integration time (video filter), etc.) of the spectrum analyzer for each measurement frequency.

  Of the inventions disclosed in the present application, effects obtained by typical ones will be briefly described as follows.

  (1) It is possible to eliminate the influence of instantaneous external noise due to on / off of a large current circuit.

  (2) When sweep measurement is performed for a long time, even if device characteristics change during the measurement, it is shown that there is a clear change in the data, and the initial characteristics and the characteristics after change can be estimated. .

  (3) Since it is possible to eliminate the influence of external noise mainly due to artificial radio waves, spectrum measurement is possible even if noise resistance measures such as a shield box are not sufficient.

  (4) Spectrum measurement in a wide frequency range can be performed with almost uniform accuracy and in the shortest time.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.

(Embodiment 1)
FIG. 1 is a flowchart showing a device characteristic measurement method according to Embodiment 1 of the present invention.

  An example of a device characteristic measuring method according to the first embodiment will be described with reference to FIG. The device characteristic measurement method according to the first embodiment is realized, for example, as software in a 1 / f noise measurement system of a MOS device, and is automatically executed by a program according to the following procedure.

  In an ordinary digital measuring instrument, in order to improve measurement accuracy, measurement is performed a plurality of times (n times) under the same measurement conditions, the values are integrated, and the average is obtained.

  In the first embodiment, n measurement data under the same measurement conditions are not integrated at a time, but the total number of measurements is divided into a relatively short time for measurement, and the integration of the measurement data for n / m times is m. Change to be done in batches. Note that n and m are natural numbers, and n is a multiple of m.

  That is, if the total number of measurements is n times, it is divided into m, measured n / m times (step S110), the measurement data is integrated (step S111), and the integration is performed m times (step S112).

  After completion of m integrations, statistical processing is performed to obtain an average of m pieces of integration data and a standard deviation σ (step S113).

  If the standard deviation σ is less than or equal to the specified value (step S114), since the data variation is small, it is determined that no external noise has entered, and the average of m pieces of integrated data is taken as the measured value. (Step S115).

  If the standard deviation σ is greater than the specified value and the variation is large (step S114), it is determined that there is data that is significantly larger than the average due to external noise or the like, and the integration that is most deviated from the average. Data is deleted (step S116).

  Further, if there is integration data that is 3σ or more away from the average (step S117), this data is also deleted (step S118).

  Then, additional measurement / integration is performed as many times as the number of deleted integration data (step S119).

  After completion of the additional measurement / integration, the process returns to step S113 to verify the data again.

  Therefore, according to the device characteristic measuring method of the first embodiment, it is possible to measure a weak signal (1 / f noise or the like) in an environment where external noise is likely to occur.

(Embodiment 2)
In the device characteristic measuring method according to the second embodiment of the present invention, the characteristic variation during the measurement is clearly shown.

In the conventional measurement, when measuring at n points by changing measurement conditions, V (1), V (2), V (3), ..., V (n-2), V (n-1), V (N)
However, V (k) = V (k−1) + ΔV
In this way, the measurement conditions are swept in a certain direction (monotonic increase or monotonic decrease).

  In the second embodiment, the entire sweep region is measured every m times, and is changed to measurement that is repeated m times. Note that n and m are natural numbers, and n is a multiple of m.

That is, V (1), V (1 + m), V (1 + 2m), ..., V (nm), V (2), V (2 + m), V (2 + 2m), ..., V (n-m + 1), ... , V (m), V (m + 1), V (m + 2), ..., V (n)
Thus, every m coarse (rough) measurements are shifted one point at a time and repeated m times to complete the entire sweep.

  When such a measurement is performed, the following effects appear.

  FIG. 2 is a diagram illustrating device characteristic variation during measurement in the device characteristic measurement method according to the second embodiment. FIG. 2 is a graph of measurement results such as Vd-VG characteristics or noise level-frequency characteristics. In the case of the Vd-VG characteristic, Vg is measured by sweeping Vg. In the case of the noise level-frequency characteristic, Vg is measured by sweeping the frequency. In FIG. 2, 21 is a measurement result, 22 is a characteristic before characteristic change (initial characteristic), 23 is a characteristic after characteristic change, and 24 is data for m pieces.

  When there is a device characteristic variation during the measurement, as shown in FIG. 2, the measurement result 21 becomes a saw-toothed curve, which clearly shows that there was a characteristic variation / characteristic deterioration. Further, if the envelopes of the upper limit and the lower limit of the measurement result 21 are obtained, the rough characteristics before and after the characteristic change can be known.

  After the measurement is completed, the characteristics of the device itself change, and the initial characteristics 22 cannot be measured again. However, since sweeping is performed every m times, one sweep is completed in a relatively short time. The outline of the initial characteristic 22 can be estimated by looking at the envelope of the measurement data (V (1), V (1 + m), V (1 + 2m),..., V (nm) data). Furthermore, it is possible to estimate the initial characteristic 22 by correcting data fluctuations that may be caused by characteristic deterioration by data processing. Similarly, the outline of the characteristic 23 after the fluctuation can be estimated.

  The above measurement method can be applied to a 1 / f noise measurement system, a DC characteristic measurement program, and the like.

  Therefore, according to the device characteristic measurement method of the second embodiment, when there is a device characteristic variation during measurement, a saw-tooth variation is seen in the measurement result, and therefore there is a variation during measurement. Is clearly shown. In addition, since sweeping is performed every several points, the initial characteristic and the characteristic after characteristic variation can be predicted by looking at the envelope of the measurement result.

(Embodiment 3)
The method for measuring device characteristics according to the third embodiment of the present invention is to remove external noise whose generation frequency is fixed.

  Most of the noise that continuously enters from the outside is an artificial radio wave, and the frequency of the power source, the radio wave for communication, etc. is fixed accurately. In many cases, the frequency spread (sideband) is limited.

  For example, the frequency of external radio waves varies depending on the environment in which the measurement device is placed, such as power supply noise of 50 Hz in Kanto, 60 Hz in Kansai, etc., but basically does not change much. This is because radio stations and the like are fixed.

  FIG. 3 is a diagram showing background noise in the device characteristic measurement method according to the third embodiment of the present invention. In FIG. 3, the horizontal axis indicates the frequency, and the vertical axis indicates the noise level.

  As shown in FIG. 3, when background noise is measured, external noise peaks 31 appear regularly at a certain frequency.

  Therefore, before the spectrum measurement, if the frequency and width of the external noise peak 31 appearing in the background noise are measured in advance using a dummy antenna and registered as a database, the measurement result of which frequency is the external noise. You can know if you are affected by.

  In general, the S / N ratio of spectrum measurement, as expressed in dB (decibel), the difference between noise and signal level is very large, and even if the background noise intensity is known, the level should be corrected therefrom. It is difficult.

  Therefore, the measurement is performed while avoiding the frequency where the measurement frequency of the spectrum measurement coincides with the frequency of the external noise peak 31 measured and registered in the database in advance. That is, without measuring at the frequency where the external noise peak 31 exists, measure the signal level at two frequencies outside the frequency where the external noise peak 31 exists, and then interpolate the signal level therefrom. Thus, a signal level that is not affected by external noise can be obtained by interpolation. If the signal intensity is sufficiently larger than the back noise level, no interpolation is performed.

  This is shown in FIG. FIG. 4 is a diagram illustrating an interpolation method of measurement data by interpolation. The horizontal axis is the measurement frequency, and the vertical axis is the signal level as the measurement result.

  As shown in FIG. 4, since the signal level includes the influence 41 of the external noise, measurement is performed while avoiding the frequency affected by the influence 41 of the external noise. Then, interpolation data 43 is created by interpolation from two measurement points 42 that are not affected by external noise on both sides of the frequency.

  Also, instead of measuring the background noise in advance and registering it in the database, you can measure the background noise in real time on the free channel of the spectrum analyzer and use this background noise to determine the frequency that should be avoided. Good. This method can cope with changes in the frequency distribution of background noise.

  Therefore, according to the device characteristic measuring method of the third embodiment, since a frequency in which external noise such as radio waves exists is registered in advance and measurement is performed while avoiding the frequency band, no external noise peak appears.

(Embodiment 4)
The device characteristic measurement method according to the fourth embodiment of the present invention performs spectrum measurement in a wide frequency range with uniform accuracy.

  When measuring a continuous spectrum with a spectrum analyzer, measurement is generally performed with a fixed resolution bandwidth (RBW) and integration time (video filter). However, if conditions are set so as to shorten the measurement time in the high frequency region (RBW: large, integration time: small), the measurement in the low frequency region becomes unstable. Conversely, if the conditions are set so as to stabilize the measurement in the low frequency region (RBW: small, integration time: large), the measurement time becomes long. This is shown in FIG.

  FIG. 5 is a diagram showing the frequency characteristics of 1 / f noise in the fourth embodiment. FIG. 5A shows a measurement result when the measurement time is set short (RBW: large, integration time: small), and FIG. 5B shows a measurement result when priority is given to stability in the low frequency region (RBW: Small, integration time: large), the horizontal axis represents frequency (usually 1 to 100 kHz), and the vertical axis represents noise level (dB).

  In the measurement result of FIG. 5A, an abnormal peak 51 due to feedthrough occurs in the low frequency region, and the spectrum is unstable in the low frequency region. On the other hand, the spectrum is stable in the high frequency region.

  In the measurement result of FIG. 5B, the spectrum is stable in both the low frequency region and the high frequency region, but it takes a long time to measure in the high frequency region.

  That is, in order to stably measure the spectrum in the low frequency region, it is necessary to reduce the RBW. In the low frequency region, it is necessary to perform integration for a long time. However, when this condition is set, the measurement accuracy is improved more than necessary in the measurement in the high frequency region, but the measurement time is greatly increased. In order to shorten the measurement time in the high frequency region, it is preferable to set the RBW longer and the integration time shorter. However, under these conditions, the stability in the low frequency region is poor and accuracy cannot be obtained.

  Therefore, in order to solve the above contradictory problem, as shown in FIG. 6, by making the RBW and the integration time variable for each frequency to be measured and setting them to optimum values, sufficient accuracy can be obtained in all frequency bands. Data can be obtained in a short time.

  FIG. 6 is a diagram illustrating set values of measurement conditions for spectrum measurement. 6A shows the set value of RBW, FIG. 6B shows the set value of integration time, the horizontal axis shows frequency (usually 1 to 100 kHz), and the vertical axis shows RBW and integration time, respectively.

  As shown in FIG. 6A, the RBW is set narrower as the frequency becomes lower. Further, as shown in FIG. 6B, the integration time is set longer as the frequency becomes lower. The degree to which the RBW and the integration time are changed is set by the user according to the application.

  Therefore, according to the device characteristic measurement method of the fourth embodiment, the measurement condition is set for each measurement point during the data sweep, so that an optimum spectrum can be obtained.

  The measurement methods of the first to fourth embodiments can be realized by improving control software in a measurement system or a measurement apparatus (such as a spectrum analyzer). Therefore, even a conventional measuring apparatus can be realized by improving the control method from the outside. Moreover, if it is incorporated in the measuring device itself, the capability of the measuring device itself can be improved. In addition, in the case of weak signal measurement, it is possible to simplify a shield box or the like that is required to prevent intrusion of external noise.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say.

  For example, in the above-described embodiment, the case where the present invention is applied to 1 / f noise measurement has been described. However, the present invention is not limited to this and can be applied to other measurements.

  The invention disclosed in the present application can be applied to a measuring apparatus such as a spectrum analyzer.

It is a flowchart which shows the measuring method of the device characteristic by Embodiment 1 of this invention. It is a figure which shows the device characteristic fluctuation | variation during a measurement in the device characteristic measuring method by Embodiment 2 of this invention. It is a figure which shows a background noise in the measuring method of the device characteristic by Embodiment 3 of this invention. It is a figure which shows the interpolation method of the measurement data by interpolation in the device characteristic measurement method by Embodiment 3 of this invention. (A), (b) is a figure which shows the frequency characteristic of 1 / f noise in Embodiment 4 of this invention. (A), (b) is a figure which shows the setting value of the measurement conditions of spectrum measurement in Embodiment 4 of this invention.

Explanation of symbols

21 Measurement result 22 Initial characteristic 23 Characteristic after fluctuation 24 Data for m 31 External noise peak 41 Effect of external noise 42 Measurement point 43 Interpolated data 51 Abnormal peak due to field-through

Claims (5)

A device characteristic measurement method that measures a device multiple times under the same measurement conditions and integrates the measurement data to obtain an average.
Divide the total number of measurements, perform measurement and integration, perform statistical processing on the integration data, and if the statistical processing results in large variations, delete the integration data that caused the variation , A device characteristic measuring method, wherein additional measurement is performed as many times as the number of deleted times. A device characteristic measuring method according to claim 1,
In the statistical processing, an average and a standard deviation of the integration data are obtained, and when the standard deviation is larger than a specified value, the integration data that is most deviated from the average is deleted. How to measure characteristics. A device characteristic measurement method for measuring a device by sweeping measurement conditions,
A device characteristic measurement method comprising: performing rough measurement for every m of the entire region of the sweep, repeating the measurement every m times by repeating the measurement every m times, and completing the entire sweep. .   The frequency distribution of the external noise is measured in advance, the frequency at which the external noise exists is registered in a database, and the frequency characteristic of the device is measured while avoiding the frequency at which the external noise is registered in the database. A method for measuring device characteristics. A device characteristic measurement method for measuring a frequency characteristic of a device using a spectrum analyzer,
A device characteristic measurement method, wherein the measurement condition of the spectrum analyzer is optimally set for each measurement frequency, and the frequency characteristic of the device is measured by changing the measurement condition for each frequency according to the set measurement condition.

Images