DE102011004830B4 - Phase method for measuring the propagation velocity of sound waves with dynamic measurement window - Google Patents

Abstract

A method for measuring the propagation velocity of sound waves, comprising the steps of: feeding at least two sound waves having at least two different frequencies (f1 to fn) into a measurement path, the sound waves being fed in succession; Receiving the transmitted sound waves after passing through the measuring section; Determining the phase angles (φ1 to φn) of the received sound waves with respect to a respectively associated reference signal of the same frequency (f1 to fn); Predetermining an approximation of the propagation velocity of the sound waves in the measurement path from the measured phase angles (φ1 to φn); Defining the propagation velocity in the measurement path using one of the measured phase angles (φ1 to φn) as a difference from a reference point with known propagation velocity and known phase angle (φb) within a measurement window, using the approximated predetermined propagation velocity to position the measurement window.

Description

The invention relates to a method for measuring the propagation velocity or the transit time of sound waves, in particular for determining the speed of sound in liquid and gaseous media. For this reason, the following emphasis is placed on the determination of the speed of sound, because running time and speed can be converted into each other with known sound path length. The application preferably takes place in media with fluctuating sound attenuation and a sound dispersion which is not negligible in the frequency range used (frequency dependence of the speed of sound).

Known measuring methods for determining the speed of sound are based on a propagation time measurement of either pulse-shaped or continuous sound signals. In this case, the sound path length must be known in the rule and the duration of the sound signals is measured either absolute or as a difference compared to a reference period.

The impulse methods can be subdivided into impulse-runtime and sing-around methods after the absence or presence of feedback. With them, the sound propagation time is measured absolutely. They can only be used if the group and phase velocity in the medium to be measured are identical, ie no dispersion occurs.

The principle of pulse transit time measurement is to directly measure the transit time of a sonic pulse from transmitter to receiver. For this purpose, the received pulse must always be detected at the same point in time. The simplest possible realization of such a detection is the edge triggering. In her, the reception time is determined by exceeding a fixed trigger voltage in the received pulse. This results in a fluctuating sound attenuation and thus also a fluctuating received signal amplitude to some significant measurement errors or even complete failure of the entire measuring system. By increasing the fundamental frequency of the sound pulses, this error can be reduced due to the then steeper voltage increase but not eliminated, since the sound attenuation increases with the square of the sound frequency and thus it can not be increased arbitrarily. Therefore, a larger number of more complex detection methods has been developed in which the amplitude of the received signal has no or only a very small influence on the detection time and thus on the accuracy of the transit time measurement.

The simplest extension is a retriggering for the next signal zero crossing. That is, after exceeding a predetermined trigger threshold, the time of the next zero crossing of the received signal is used as a reception time. This is almost independent of the amplitude of the signal. For larger amplitude fluctuations and a slow increase in the pulse amplitudes over several periods of the carrier oscillation, the risk of a total failure of the edge triggering, and thus of the entire measurement, is very high here as well. The reason for this is that the same period of the carrier oscillation can no longer be reliably detected under these conditions, which then also applies to the determined signal zero crossing in the case of a subsequent signal zero crossing triggering. Therefore, z. In EP 0 324 731 A1 instead of a trigger threshold, the position of the maximum (center of gravity) of the envelope of the received pulse is first determined. The reference point for the subsequent zero-crossing triggering is then the position of the next following peak value of the carrier oscillation of the pulse. However, since the temporal position of the envelope maximum is largely independent of its height, depending on the accuracy requirement imposed on the transit time measurement, the downstream determination of a signal zero crossing can also be dispensed with and the position of the envelope maximum can be used directly in some applications. Such a procedure is described in EP 1 293 801 A2 described there, where dispensed with the formation of an envelope and instead the position of the global maximum of the carrier oscillation of the pulse is determined.

Instead of the zero-crossing detection after the determination of a reference time, a phase angle measurement between the carrier oscillation of the received pulse and a reference signal can also take place. In this combination, a higher tolerance to received signal interference, such. B. the signal fluctuations at a low signal-to-noise ratio achieved. So working runtime procedures are in DE 10 2005 004 331 A1 . EP 0 201 989 A2 and DE 4 407 369 A1 described.

It should be noted that for each absolute propagation time measurement, the measured transit time is always greater than the actual transit time of the sound signals through the measurement path in the medium to be measured. The difference between these two times is called the dead time of the measuring system. She is independent of the speed of sound in the test section and has, inter alia, after Sorge, G .: "Fascination Ultrasound", Publisher BG Teubner, 1st edition, Stuttgart 2002, page 65, Henning, B .; Zuber, A .: "Ultrasonic measuring instrument and system for process measurement in liquids", Dissertation, Technical University Magdeburg (1991), pages 14 and 16 and Burda, W .: "Ultrasonic resonator for process-technical determination of the properties of static and dynamic fluids" , Dissertation, Publisher Shaker (1993), page 9, both mechanical and electrical causes. The mechanical causes essentially include acoustic delays which occur during the transition of the sound signals from the sound transducer into the medium in the measuring path and vice versa. The electrical causes are signal delays in the electronic parts of the measuring system. If the measuring system used consists of high-quality components, then the dead time can generally be regarded as a measuring system-specific constant independent of the conditions, which is determined in a single calibration of the measuring system and taken into account as a correction factor in all subsequent measurements. Such a correction is z. In DE 2 110 570 A described. Above all, measuring systems for use in educational institutions, such. However, for example, secondary schools often consist partly or even entirely of low-cost components. To US Pat. No. 6,738,312 B1 the dead time can then generally no longer be considered as a measuring system-specific constant. Rather, it is assumed here that they have a significant dependence on the measurement conditions, such. As the temperature shows. This leads to significant errors in the speed of sound detection.

In US Pat. No. 6,738,312 B1 Therefore, a pulse transit time method is presented that operates in reflection mode and with separate transmit and receive transducers, wherein the receive transducer is located in the sound path between the transmit transducer and the reflector. As a result, part of the sound pulse emitted by the transmitting transducer runs directly to the receiving transducer and another part first to the reflector and then back to the receiving transducer. To detect these two received pulses, a threshold triggering with subsequent determination of the next peak value of the carrier oscillation of the pulses is used. If the distance between the receiving transducer and the reflector is assumed to be the measuring path length, the transit time of the sound pulses is the quotient of twice the receiving transducer-reflector distance and the time difference between the arrival of the two receiving pulses at the receiving transducer. Since the dead time of the measuring system delays the arrival of both pulses in the same way, it has no influence on the determined sound velocity value.

The sing-around method, also referred to as the pulse-orbit-repetition rate method, was developed by E. Hiedemann ("Wave Propagation in Solid Bodies, Ultrasound," Solid State Physics, Part 1, (Fiat Review of German Science, Volume 8) edited by G. Joos, Wiesbaden 1947, pp. 154-178). Here, the generation of the transmission pulse by means of a trigger signal controlled pulse generator. After the sonic pulse thus generated has passed through the measuring path, it is detected, which causes the generation of a new trigger signal and, as a result, also transmission pulse. This feedback loop continues to run as long as no sound pulse is lost due to interference or can not be detected by excessive damping of the sound. The resulting pulse repetition frequency can be easily determined with a frequency meter (Toda, K., Rabe, A .: "A Method for Measuring the Change of Ultrasound Velocity in a Liquid, Journal of Applied Physics", Volume 53 (1982), Pp. 194-196; Ficken, GW, Hiedemann, EA: "Simple form of the sing-around method for the determination of sound velocities", Journal of Acoustic Society of America, Vol. 28, No. 5 (1956), p 921-923). With known measuring length from her the duration of the sound pulses can be calculated, with pulse repetition frequency and running time are inversely proportional to each other. This method, in its conventional form, is not suitable for measurements in media with a very low sound attenuation, since otherwise repeated back and forth reflected echo pulses would interfere with subsequently transmitted sound pulses. This almost always leads to an apparent shift of the received pulses and as a result to a falsification of the self-adjusting pulse repetition frequency.

This restriction is with the in US 4,926,395 A introduced extended variant of the sing-around method eliminated. The trigger-controlled transmission and the received signal detection as well as the resulting feedback loop also remain unchanged here. As transmission signals, however, instead of short, only a few oscillations of sound pulses, longer bursts of sound are used, which are generated from a continuous, sinusoidal reference signal and thus have the same frequency and phase position as this. After transmitting and traversing the measuring section, the phase shift of the receiving burst obtained is determined relative to the transmission burst or the reference signal, and the frequency of the reference signal is changed until the transmitting and receiving bursts oscillate in phase. If this is achieved, the length of the measuring path is an integer multiple of the wavelength of the sound signals and the counted periods of the reference signal from the transmission time to the detection of the received signal divided by the set reference signal frequency gives the Sound transit time through the measuring section. With a known measuring line length, the speed of sound can then be calculated from this. Since the set sound frequencies and the respectively resulting phase angle can be determined or measured very accurately and small temporal fluctuations of the received signal detection have no effect on the determined transit time of the sound signals, the accuracy of the sound velocity measurement is significantly higher than in conventional sing-around method , Furthermore, when the transmit and receive bursts oscillate in phase, all the echo bursts resulting in low attenuation media are also in phase with them. As a result, only constructive interference can occur, which only slightly increase the amplitude of the received signal bursts, which in principle does not lead to a falsification of the determined transit time or sound velocity values.

With solid bodies, sound waves can also be detected with the aid of a laser interferometer if they arrive at their surface after passing through the interior of the solid and set it in motion. Such a method, which was developed for geophysical investigations of seabeds, is published in US 2008/0007715 A1 disclosed. Here, a laser interferometer and a plurality of detectors arranged in matrix form on a solid surface (the seabed) are used. The laser interferometer sends a convergent object beam sequentially toward each individual detector, thus measuring the changes in the distances of all the detectors from the laser interferometer in a very rapid sequence. As a result, the movement of the detectors or the solid surface and thus also the sound waves arriving there can be measured.

The measuring methods working with continuous sound signals can be divided into those with broadband and those with narrowband, ie sinusoidal, sound signals. The wideband methods use either amplitude-modulated sound signals as in DE 4437 205 A1 or frequency modulated sound signals as in DE 3 412 089 A1 , The transmission signal is impressed as a code, which is then detected in the received signal. This allows, as with the pulse transit time method, a direct measurement of the sound wave propagation time. An increase in the measurement accuracy can also be achieved here by an additional phase angle measurement between the carrier oscillation of the received signal and a reference signal. The narrowband methods can in turn be divided into optical methods, resonance or interferometric methods and phase methods. They have some decisive advantages over the pulse measuring methods, but also disadvantages. The latter includes a much higher level of electronic and partly equipment-related effort, which is necessary above all for automated measurement. Due to the development of digital measurement technology in conjunction with its progressive miniaturization and cheapening, this disadvantage no longer has any great significance compared to pulse methods. Their main advantages are the high resistance to received signal interference, since they can be easily filtered out, the high transmission strength and reception sensitivity, since the transducers can be operated at or near their resonance frequency and, due to the narrow band, the insensitivity to the occurrence of sound dispersion in the zu measuring medium.

The principle of the optical method is based on the diffraction phenomena of light in a sound-through liquid or gaseous medium, which must be optically transparent. Since the refractive index of such a medium is density-dependent, it varies spatially and temporally with the sound pressure of a propagating sound wave, with the local changes in the refractive index being proportional to the instantaneous values of the sound pressure. It follows that the distance of the maxima and minima of the refractive index is a sound wave length, corresponding to the position of the compression and dilatation areas. The sound field thus produces in the sound-transmitted medium an optical phase grating with the sound wave length as a grating constant, which can be determined by the methods known from optics from the position of the diffraction images. The speed of sound then results from the multiplication of the lattice constants with the sound frequency.

The most accurate measurements of the sound frequency are possible with resonance methods. Their long-known operating principle is based on the generation of standing sound waves in the gaseous or liquid medium to be examined. For this purpose, two planar transducers or a planar transducer and a planar reflector with a known distance are generally opposite in plane. From the frequency of standing waves and the transducer or transducer-reflector distance, so the sound path length, then the speed of sound or -laufzeit can be determined. Depending on whether the sound path length or the sound frequency is varied to produce the standing waves, measuring systems operating according to this method are referred to as either acoustic interferometers or resonance cells.

The development of the former goes to Pierce, GW: "Piezoeiectric Crystal Oscillators Applied to the Precision Measurement of the Velocity of Sound in Air and CO ₂ at High Frequencies", Proceedings of the American Academy of Arts and Sciences, Volume 60 (1925), No. 5, pp. 271-302. In this variant, the resonance method is often referred to as interferometric method. With them, the change in the sound path length usually with the aid of a micrometer screw. Since mechanically moving parts are needed here, acoustic interferometers are rather unsuitable for automated measurements. Resonance cells, on the other hand, can do without moving parts, which is why automatic measurement systems have already been developed on their basis. As an example, here is the Burda, W. in his aforementioned dissertation "ultrasonic resonator for process-technical determination of the properties of static and dynamic fluids", published by Shaker (1993), and Burda, W. and Becker, E.-J. in "Ultrasonic Resonance Meter for Fluid Measurement" Tech. Mess., R. Oldenbourg Verlag, Issue 10 (1993), pp. 376-382. The decisive disadvantage of all Resorianzverfahren is that because of the required multiple reflection of the sound waves can be measured only in media with a very low sound attenuation. To be examined liquids must z. B. be absolutely free of bubbles.

Even with phase measurement methods, which also use sinusoidal, continuous sound signals, extremely accurate transit time measurements are possible. With you, the phase angle of the received signal is measured against a reference signal, which is usually identical to the transmission signal or coincides with it in frequency and phase. In this case, in contrast to the resonance method, multiple reflections and thus standing waves are undesirable because they disturb the phase angle measurement or even completely prevent it. They can therefore also be used for transit time measurement in media with strong sound attenuation. As is known, phase angles can only be measured modulo 2π, that is to say the value range [0; 2π] are limited, a phase measurement in principle provides ambiguous statements, which is why additional considerations or measurements are required to obtain unique propagation time values. Also with you there are those with and those without feedback.

In the former, either the sound frequency ( DE 1 156257 B ) or the sound path length changed until the measured phase angle assumes a predetermined value. In DE 1 219 700 B is z. B. a generator circuit proposed, which performs the frequency adjustment automatically. From the sound frequency, the sound path length and the phase angle can then be determined the sound velocity change from a reference point. However, since the phase angle measured values, as already mentioned, are limited to the value range [0; 2π] are limited, also sonic velocity changes can be measured only up to a certain limit. In this case, the higher the sound frequency and the greater the sound path length, the lower the maximum measurable change in speed. Furthermore, it must be noted that measurements with variable sound frequency can only be performed in media without sound dispersion in the frequency range used.

Phase measurement methods without feedback work in the simplest case with a fixed sound frequency. Schaafs and Kalweit, a so-working device in "The phase comparison interferometer for measuring minute changes in sound velocity up to 1 mm / s", Acustica, Volume 10 (1960), pp 385-393 presented. It should be noted that the term "phase-comparison interferometer" as used herein in the sense of the definition given here, and also in the literature for interferometer generally common definition, is not applicable, since this is a phase measurement method. To avoid interference signals in the phase angle measurement by structure-borne noise, crosstalk signals (parts of the electrical excitation signals that go directly to the receiving transducer) and multiple reflections in measurements in media with low sound attenuation, as in 3 Frequently, sinusoidal sound signals are used in the form of longer signal bursts, with the arrival and decay phase of a receiving burst not being used for the phase angle measurement. Even such sound signals can be assumed to be continuous here since their keying, apart from the desired noise suppression, does not affect the phase measurement. Such a, with sound bursts working phase measurement method is in DE 2950 862 C2 described. Due to the limited value range of the phase angle measured values, only speed differences with respect to a known reference point with known propagation time v _b and known phase angle φ _b can be determined within predefined limits, the measurement window, instead of absolute velocity values. The size of such a measurement window is again inversely proportional to the wavelength of the sound signals used and in speed measurements also inversely proportional to the sound path length.

positioniert. Dabei ist φ_o der Phasenwinkel und v_o die Schallgeschwindigkeit an der oberen Messfensterbegrenzung, s die Schallweglänge, f die Schallfrequenz und Δt der Laufzeitunterschied zu dem Bezugspunkt mit v_b und φ_b. Um φ_o auf den Wertebereich 0 ≤ φ_o < 2π zu begrenzen, muss die Gleichung 1 wie folgt erweitert werden:

Before starting a measurement it must be ensured that all velocity values of the measurement series to be recorded lie within the measurement window. Since the window position is invariable during the entire recording of the measurement series, it can also be called a static measurement window. For this purpose, starting from the reference point to the upper or lower window limit, ie the maximum or minimally measurable speed, associated phase angle φ _o or φ _u calculated. When selecting the first variant, the measurement window is also displayed

positioned. Where φ _{o is} the phase angle and v _{o is} the sound velocity at the upper measurement window boundary, s is the sound path length, f is the sound frequency and Δt is the transit time difference to the reference point with v _b and φ _b . In order to limit φ _o to the value range 0 ≤ φ _o <2π, equation 1 must be extended as follows:

ermittelt werden. Hierbei ist frac die Restfunktion, die durch frac(x) = x – [x] definiert ist. Sie gibt also die Differenz zwischen x und [x] an, wobei [x] die Gaußklammer ist, die auch als Abrundungs- oder Ganzzahlfunktion bezeichnet wird. Der Wert von frac ist immer positiv.Subsequently, in the sound-through medium, the speed of sound v by the measurement of the phase angle φ with

be determined. Here, frac is the residual function defined by frac (x) = x - [x]. It thus gives the difference between x and [x], where [x] is the Gaussian bracket, which is also called a rounding or integer function. The value of frac is always positive.

Since the sizes sound path and frequency can not be reduced arbitrarily and numerous applications, such. As measurements in bubble-containing liquid media, even require the highest possible frequency of sound, which is often too small resulting from the sound frequency to be used resulting measurement window in its static form. One way to solve this problem while preserving the static window property is to use "virtual" sound at a much lower frequency. They are based on the principle of virtual beating. This means that under identical measuring conditions, two sound signals with different, but not too far apart, frequencies are either spatially as in Uhlig, K .: "A method for measuring the speed of sound by analog conversion of the transit time of a sound signal into a frequency in an AC voltage , Dissertation, University of Kiel (1971) or in time as in DE 19841 154 C2 and DE 100 36 567 A1 sent separately and measured the respective phase angle. In the case of spatial separation, therefore, the sound waves in both sample spaces must propagate at the same speed, while with a temporal separation both measurements have to take place so quickly in succession that the speed of sound can be considered constant throughout the measuring cycle. Compliance with the first condition requires in comparison to the second in today's measurement a much higher cost, which is why the second variant of the method should be given preference in the rule. Subsequently, the phase angle of the virtual beat signal, which, however, has never existed in real time, is determined from the two phase angles thus measured, and the sound velocity is calculated therefrom. The calculation of the speed value again takes place as a difference from a known reference value. The two sound signals are thus effectively superimposed virtually in the computer. Because the frequency of the virtual beat signal is much lower compared to the real sound signals, a suitable choice of the frequency pair used can be used to generate a static measurement window which is sufficiently large for every conceivable application. Since both sound frequencies are always relatively close to each other, it is also possible to measure in sound-dispersion media. Furthermore, for this reason, low-attenuation sound transducers with a narrow resonance curve, which allow a high transmission strength and reception sensitivity can be used.

The main drawback with the use of virtual sound signals is the reduction of the accuracy of the determined speed measurement values compared to the simple phase measurement. With constant measurement accuracy of the phase angle measurements, this decreases with increasing size of the measuring window, which therefore never needed to be greater than necessary for the respective application. In addition, in the worst case, the measurement errors of the two phase angle measurements add up. The enlargement of the signal bandwidth by the use of two sound frequencies, on the other hand, does not have a significant disadvantage due to their small size, even when measured in media with sound dispersion.

Furthermore, it is also possible in the phase measurement method to determine the transit time or speed of sound waves absolute. For this purpose, the ambiguity of the phase angle measurement is eliminated by distance or frequency variation. In the distance variation, the phase angle of the received signal is measured at a constant sound frequency at at least two different sound path lengths. Since mechanically moving parts are again required here, this variant is rather unsuitable for automation of the measurement. In the frequency variation, however, no moving parts are needed, see Kaarsberg, E.A .: "Elastic-waves velocity measurements in rocks and other materials by phase-delay methods", Geophysics, Volume 40 (1975), No. 6, pp. 955-960. Here, with constant sound path length, the phase angle of two or more received signals is measured at at least two different sound frequencies. In both variants, it must be ensured that the phase angle difference between two adjacent measuring points is always less than 2π. Furthermore, the phase angle measurements must be carried out sufficiently quickly in succession, so that it can be assumed during the measurement of all phase angles of stationary measurement conditions, ie a constant speed of sound in the sound-through medium. If the sound path lengths or frequencies are subsequently plotted over the measured phase angles, the speed of sound results from the increase in the straight line passing through them.

The more measuring points are recorded and the broader the frequency spectrum used, the more accurate is the determined speed value. However, the more measuring time is required, which severely limits the number of possible measuring points per speed value with constantly changing media properties. A sufficiently high accuracy of the determined speed values can generally only be achieved with a larger number of measuring points and the widest possible spectrum of the used sound frequencies. In the cited publication by Kaarsberg, E.A. B. recorded up to 42 measuring points to determine a speed value, which results in a very high accuracy, but only in static measurement conditions is possible. In addition, more vapor-deposited transducers must be used by the high bandwidth for phase measurement techniques, resulting in lower transmit power and receive sensitivity. This method is therefore unsuitable for measurements in media with a strong sound attenuation and above all sound dispersion as well as at a rapidly changing switching speed.

It was therefore the object to develop a measuring method with which the sound propagation time or speed can be determined quickly and with high accuracy even with dynamically changing measuring conditions, high sound attenuation and dispersion in the medium to be examined.

Accordingly, a method is proposed for measuring the propagation velocity of sound waves, comprising in summary the following steps: determining n sound frequencies f ₁ to f _n with n ≥ 2; Feeding all sound waves with the frequencies f ₁ to f _n in a measuring section, wherein the sound waves are fed successively in time; Receiving the transmitted sound waves after passing through the measuring section, wherein the transmission of the sound waves with the frequencies f ₂ to f _{n takes place} only after the receipt of each previously transmitted sound wave; Determining the phase angle φ ₁ to φ _{n of} the received sound waves with respect to a respectively associated reference signal of the same frequency f ₁ to f _n ; Predetermining an approximation of the propagation velocity of the sound waves in the measurement path from the slope of the line passing through all measured phase angle measurement points (f ₁ , φ ₁ ) to (f _n , φ _n ); and re-determining the exact propagation velocity in the measurement path using one of the measured phase angles φ ₁ to φ _n as the difference from a reference point with known propagation velocity and known phase angle φ _b within a previously positioned measurement window having a lower and upper velocity limit v _u and v _o wherein the approximated predetermined propagation velocity is used to position the measurement window.

The measurement thus takes place without a static measurement window, but by means of a dynamic measurement window repositioned for each transit time measurement.

Since very little, in the simplest case, only two sound frequencies f ₁ and f ₂ , each with one phase angle measurement per frequency (f ₁ , φ ₁ ) and (f ₂ , φ ₂ ), are used to predetermine the propagation velocity of the sound waves, the velocity value thus obtained is used very inaccurate, which is why it is only a rough predetermination of the speed of sound. After the measuring window has been positioned using this value, finally, according to equation 3, one of the measured phase angles becomes the exact one Speed of sound is determined as the difference from a reference point with known sound velocity and known phase angle.

To increase the accuracy in the approximate prediction of the propagation velocity of the sound waves, the number of sound frequencies used in this case can be increased. The slope of the straight line passing through it is then preferably determined by linear regression.

An increase in accuracy in the approximate prediction of the propagation velocity of the sound waves can also be achieved by transmitting a plurality of sound waves of the same frequency and the subsequent averaging of the measured phase angles.

The frequency and, as a rule, also the phase position of the transmitter and associated reference signal match.

Measure the used sound frequencies so that the occurring phase angle difference between two neighboring frequencies is always smaller than 2π.

The measurement uncertainty of the approximately predetermined transit time must be smaller than the measurement window, so that there is a measurement window position in which it is ensured that the exact yet unknown runtime value is within the measurement window. This condition is most certainly satisfied if the sound wave with the lowest of all frequencies used is used to determine the exact duration, since the measurement window is the largest here.

The measurement window should generally be laid over the approximately predetermined transit time so that it is centered in the phase angle measurement with respect to the magnitude of the measurement error.

Determining the distance between two adjacent sound frequencies can be done by the first frequency is set and the following frequency is selected so that taking into account the phase angle measurement accuracy, the maximum possible phase difference is always less than 2π. The maximum possible phase difference occurs at the smallest possible sound velocity in the measurement path.

Due to the dynamic repositioning of the measuring window at each measuring point, the condition prevailing with a static measuring window that all sound velocity values of a series of measurements to be recorded must be within the measurement window fixed (static) before the start of the measurement.

Rather, as already mentioned, it must only be ensured that the measurement uncertainty of the respectively roughly predetermined sound velocity value is smaller than the dynamic measurement window, ie that the exact, as yet unknown, sound velocity value is located inside the window. As a result, the dynamic measurement window can be much smaller than it could be a static measurement window in most applications. The generation of low-frequency, virtual sound signals for enlarging the measurement window as in DE 198 41 154 C2 and in the method presented by Uhlig, K. this is no longer necessary, which is why the simple phase method for accurate sound velocity determination is sufficient. In addition, the sound velocity values determined have a greater accuracy in comparison to a phase measurement with virtual sound signals, since in the first place only the uncertainty of a phase angle measurement is received and, secondly, the measurement window is generally considerably smaller. Another advantage over phase methods with virtual sound signals is the much lower bandwidth of the sound signals used in the old rule, which increases the accuracy of measurements in media with sound dispersion and increases when using only weakly damped transducers averaged over all the sound frequencies used transmit power and reception sensitivity. This, in turn, is advantageous for measurements in media with a strong sound attenuation. So z. B. in measurements in liquid media with the in DE 198 41 154 C2 described method typically uses signal bandwidths between 100 kHz and 500 kHz, while here in a comparable application generally only signal bandwidths of a few 10 kHz for the rough prediction of the speed of sound are needed, as will emerge from the example below. In the final accurate determination of the speed of sound, the signal bandwidth used is even close to 0 Hz. As long as the sound velocity value to be determined is within the measurement window, a measurement distortion by sound dispersion can thus be completely ruled out here.

In the following an embodiment of the invention will be explained in more detail with reference to the drawings. Show it:

1 Time course of two reference signals with the frequencies f ₁ and f ₂ and the associated sampled received signals (bursts) each having the same frequency and determining the phase angle of the received signals relative to the respective reference signal;

2 Graphical representation of the approximate prediction of the speed of sound when using two sound frequencies;

3 Accuracy of the approximated predetermined speed of sound for a given phase angle measurement accuracy and the resulting spread of these sound velocity values (v _{max -v min} )

The invention relates to a phase method with dynamic measurement window for determining the propagation speed or transit time of sound waves. For this purpose, in a first method step, two or more continuous or keyed, sinusoidal sound signals with at least two different frequencies f ₁ to f _n are transmitted successively and received again after passing through the measurement path, wherein in each case the phase angle φ of the received signal measured against an associated reference signal of the same frequency becomes. The time interval between the individual sound signals should be so small that the speed of sound in the sounded-through medium can be regarded as constant throughout the measuring process. Subsequently, in a second method step, the transit time of the sound signals is predetermined approximated from the phase angles thus obtained. This is followed by a third process step, in which the exact transit time difference with respect to a reference point with known sound velocity and known phase angle is determined from one of the measured phase angles, wherein the previously determined with low accuracy sound velocity value is used for positioning the measurement window required for this purpose.

For most applications, when using suitable measuring technology, two phase angle measurements carried out at different sound frequencies are sufficient to roughly predetermine the speed of sound or the duration of the sound waves. Only in the case of strong disturbances of sound propagation or due to a very small measurement window with correspondingly high sound frequencies can it be necessary to record more than two measurement points to increase the accuracy in the approximated prediction of the speed of sound. Therefore, in the following considerations, the simplest case, that is to say two phase angle measurements with differing sound frequencies per speed value to be determined, is assumed. Here, as already mentioned in the previous section, the frequency difference between the two sound signals is generally much lower than in the DE 198 41 154 C2 sound signals used to generate a virtual beat signal.

As already mentioned, the two phase angle measurements must be carried out so quickly in succession that the switching speed can be regarded as constant throughout the measuring process. Preferably, sampled, sinusoidal sound signals in the form of longer bursts are used for noise suppression. The two reference and received signal pairs with the frequencies f ₁ and f ₂ and the phase angles φ ₁ and φ ₂ between the reference and the respective associated received signal are in 1 shown. The measurable phase angles are based on the value range [0; 2π] limited.

Thus, a sound wave with the first frequency f, is sent through the measuring path with the medium to be examined and received again after passing through the measuring path. Subsequent to time, another sound wave with the second frequency f _{2 is} likewise sent through the measuring path and received again. The determination of the two phase angles φ ₁ and φ _{2 of} the two received signals relative to a reference signal of the same frequency takes place either directly after the signal reception or to minimize the measurement time only after the completion of all to be carried out to determine a speed of sound transmission and reception processes.

Then, the sound frequencies are plotted against the measured phase angles, as in 2 is shown, the speed of sound results from the rise a of the straight line passing through it. If more than two measurement points have been recorded at more than two different sound frequencies, the increase in the straight line passing through them is generally determined with the aid of linear regression. The speed of sound v can now with constant sound path s with v = 2π · s / a (4) be calculated. Are, as in 2 shown, only two measurement points have been recorded, so the sound velocity with the simplified shape v = 2π · Δf / Δφ · s (5) where Δf is the frequency difference and Δφ is the phase angle difference between the two measurement points.

Before starting a measurement, the two sound frequencies f ₁ and f _{2 to be used} and thus also the frequency difference Δf between them and the sound frequency to be used for the final determination of the propagation velocity with the simple phase method must be determined.

In order to ensure reliable determination of the propagation velocity, the magnitude of the random errors in the phase angle measurement φ _F , ie the scattering width ± φ _{F of} the two phase angle measurements φ ₁ and φ _{2, must be} known in order to obtain the maximum and minimum possible phase angle difference in the required velocity measurement range Δφ _max = φ ₂ - φ ₁ + 2φ _F = Δφ + 2φ _F (6) Δφ _min = φ 2 - φ ₁ - 2φ _F = Δφ - 2φ _F (7) to investigate. Together with Equation 5, this results in the spread of the roughly predetermined sound velocity value as follows:

für die Geschwindigkeitsmessung gibt, bei der Δφ_max = 2π ist. Um eine fehlerfreie Messung sicher zu stellen, darf die kleinste mögliche Schallgeschwindigkeit in dem zu messenden Medium den Wert v_k nicht unterschreiten. Sie sollte bei einer Messung immer deutlich größer als v_k sein, um hier eine gewisse „Sicherheitsreserve” zu besitzen.Here, V _min and V _{max are} the minimum or maximum possible sound velocity at a measuring point. This is clarified in 3 where v _{g is} the measured speed of sound. Furthermore, it can be seen from equations 8 that, with a constant frequency difference Δf and sound path length s, the phase angle Δφ increases between two measuring points. Therefore, it follows from the condition Δφ _max <2π for the maximum permissible phase angle that there is always a lower limit

for the velocity measurement in which Δφ _max = 2π. In order to ensure error-free measurement, the smallest possible speed of sound in the medium to be measured must not fall below the value v _k . It should always be significantly larger than v _{k in} order to have a certain "safety margin" here.

mit der oberen Fehlergrenze der Phasenwinkelmessungen φ_F,max = T·π·Δf / 2 (12) die bei einer Messung nicht erreicht werden darf. Es muss hier somit immer die Bedingung φ_F < φ_F,max erfüllt sein. Das dynamische Messfenster ist also mit v_u = 2π· Δf / Δφ + T·π·Δf·s; v_o = 2π· Δf / Δφ – T·πΔf·s (13) zu positionieren.In the simple phase method with which the exact sound velocity is determined, the relationship applies to the lower and upper measuring window limits, ie the smallest and largest measurable speed of sound v _u and v _o v _u = s / t; v _o = s / t - T, (10) where T is the period of the sound waves and t is their transit time at v _u . It is best to position this measurement window centered on the random errors φ _F in the phase angle measurements. For this purpose, it is assumed that the magnitude of the measurement uncertainty of the grab predetermined sound velocity value is equal to the size of the measurement window according to equation 10, so that at a measurement uncertainty-centered window position v _min = v _u and v _max = v _o is. This results from equations 8 and 10

with the upper error limit of the phase angle measurements φ _{F, max} = T · π · Δf / 2 (12) which can not be reached during a measurement. Thus, the condition φ _F <φ _{F, max} always has to be fulfilled here. The dynamic measurement window is so with v _u = 2π · Δf / Δφ + T · π · Δf · s; v _o = 2π · Δf / Δφ - T · πΔf · s (13) to position.

As an example, a complete measuring process with a dynamic measuring window is described below:
The sound path length should be s = 2 · 10 ^-6 m and the measurement error in the phase angle measurement φ _F = 1.5 °. Furthermore, two sound frequencies with f ₁ = 2 × 10 ⁶ Hz (2 MHz) and f ₂ = 2.047 × 10 ⁶ Hz (2.047 MHz) should be used. To re-determine the exact speed of sound, the sound signal with the lower frequency f, is used. Then, according to the equation 12, φ _{F, max} = 2.115 ° and the condition φ _F <φ _{F, max are} thus satisfied. The lower limit for the sound velocity measurement is a value of v _k ≈ 948 m · s ^-1 .

Before starting the actual measurement, the ultrasonic measuring system used must be calibrated once. For this purpose, the phase angle at the sound frequency used for later determination of the speed of sound is repeatedly measured in a medium with a precisely known speed of sound. The number of phase measurements must be so high that the measurement error of the phase angle averaged over all measurements is negligibly small. In this calibration, a Berugsphasenwinkel of φ _b = 68.40 ° was measured at a speed of sound of v = 1400 m · s ^-1 and the sound frequency f ₁ .

Now the speed of sound in the medium to be examined can be determined. A phase measurement at the frequency f _{1 gave} a phase angle φ ₁ = 102.69 ° and at the frequency f ₂ a phase angle φ ₂ = 328.29 °. The positioning of the dynamic measurement window with equations 11 gives a lower limit of v _u = 1472.39 m · s ^-1 and an upper limit of v _o = 1528.66 m · s ^-1 . It follows from equations 8 that the speed of sound is in the range of v _min = 1480.31 m · s-1 to v _max = 1520.20 m · s ^-1 and thus secured within the set measurement window. After the successful positioning of the dynamic measurement window results according to the equation 2 at the upper window boundary, a phase angle of φ _o = 282.74 °. Now with the equation 3, the determination of the speed of sound, which gives a value of v = 1500 m · s ^-1 .

After some time, a phase measurement is carried out again in the medium to be examined, with a phase angle φ = 336.37 ° at the frequency f ₁ and a phase angle φ ₂ = 199.00 ° at the frequency f ₂ . The measurement window can now be re-positioned with the equations 11 to the limits v _u = 1491.66 m · s ^-1 and v _o = 1549.44 m · s ^-1 . Here too, with v _min = 1499.79 m · s ^-1 and v _max = 1540.76 m · s ^{-1, it is} again ensured that the speed of sound is within the set measurement window. As a phase angle at the upper window boundary, a value of φ _o = 156.38 ° then results according to Equation 2. Now, with the equation 3, the sound velocity can be redetermined, which now has a value of v = 1520 m · s ^-1 results.

If a minimum sound velocity of 1000 m · s ^{-1 is} assumed, then the lower measurement limit for the velocity of sound v _k ≈ 948 m · s ^{-1 is} sufficiently far away to ensure a reliable determination even in the case of an unexpectedly slightly higher measurement error in the phase measurement to ensure the speed of sound. A sound velocity of 1000 m · s ^-1 is a very low value for liquid media, which is only undercut in a few cases, so that the selected sound frequencies depending on the application only very rarely or never have to be changed. Since the positioning of the measuring window also takes place automatically, this method is considerably more comfortable to use than any previously existing phase measuring method for the relative determination of the speed of sound. Similar examples with an identical statement and values typical of gaseous or solid media could also be found.

Claims (12)

A method for measuring the propagation velocity of sound waves, comprising the steps of: feeding at least two sound waves having at least two different frequencies (f ₁ to f _n ) into a measurement path, the sound waves being fed in succession; Receiving the transmitted sound waves after passing through the measuring section; Determining the phase angles (φ ₁ to φ _n ) of the received sound waves with respect to a respectively associated reference signal of the same frequency (f ₁ to f _n ); Predicting an approximation of the propagation velocity of the sound waves in the measurement path from the measured phase angles (φ ₁ to φ _n ); Determining the propagation velocity in the measurement path using one of the measured phase angles (φ ₁ to φ _n ) as a difference from a reference point with known propagation velocity and known phase angle (φ _b ) within a measurement window, wherein the approximated predetermined propagation velocity is used to position the measurement window. A method according to claim 1, characterized in that for accuracy increase in the approximate prediction of the propagation velocity of the sound waves more than two sound signals with more than two sound frequencies (f ₁ to f _n ) and thus more than two measured phase angles (φ ₁ to φ _n ) be used. A method according to claim 2, characterized in that when using more than two sound waves at more than two different sound frequencies, the predetermination of the propagation velocity of the sound waves with the aid of the linear regression of the measured phase angle (φ ₁ to φ _n ) as a function of the respective frequency (f ₁ to f _n ) takes place. A method according to claim 1, characterized in that when using more than two sound waves with at least two different sound frequencies and with an at least twice occurring sound frequency all measured phase angles of sound waves of the same frequency are averaged. Method according to one of claims 1 to 4, characterized in that each transmit signal coincides with the respective associated reference signal in frequency and phase. Method according to one of the preceding claims, characterized in that in the predetermination of the propagation velocity of the sound waves, the measured phase difference between two sound signals with adjacent sound frequencies must always be less than 2π. Method according to one of the preceding claims, characterized in that the measuring window is positioned centered relative to the predetermined propagation speed of the sound waves and taking into account the phase angle measuring accuracy. Method according to one of the preceding claims, characterized in that the scattering width of the predetermined approximated propagation velocity caused by the random error in the phase angle measurement (φ _F ) is smaller than the measurement window for the determination of the propagation velocity. Method according to one of the preceding claims, characterized in that continuous sinusoidal sound signals or sampled quasi-continuous sound signals are used. Method according to one of the preceding claims, characterized in that the sound waves are emitted temporally in such a succession that the duration of the sound waves within the measuring section during the transmission of all sound signals can be regarded as constant. Method according to one of the preceding claims, characterized in that the sound wave with the lowest frequency is preferably used for the determination of the propagation velocity. Method according to one of the preceding claims, characterized in that the determination of the sound frequencies (f ₁ to f _n ) takes place in such a way that the first sound frequency f _{1 is} predetermined and the other sound frequencies (f ₂ to f _n ) are determined from the condition, that taking into account the phase angle measurement accuracy at all propagation velocities to be determined, the maximum measured phase difference between two sound signals with adjacent frequencies must always be less than 2π.

Images