FLOW VELOCITY MEASURING DEVICE WORKING ACCORDING TO THE SING-

AROUND-PRINCIPLE RELATED APPLICATION

This application claims priority and benefit from Swedish patent application No. 0702127-2, filed September 24, 2007, the entire teachings of which are incorporated herein by reference. TECHNICAL FIELD

The present invention relates to measuring the flow velocity of a streaming fluid such as for determining the flow of the streaming fluid, i.e. the amount of fluid per time unit that passes a channel. BACKGROUND

When measuring the flow velocity of a fluid, sound signals can be emitted in the fluid and the run time of the signals for a fixed distance be determined, since the propagation speed of the signals in the fluid is influenced by the flow velocity of the fluid, provided that the signals are emitted in a suitable direction. Such a method is disclosed in the published International patent application WO 02/40945, the method using the "sing-around" principle. According to this principle run times of a large number of sound signals which are emitted close upon each other and travel the same distance are measured. When a sound signal is detected at the end of the distance, the signal is reflected back to the beginning of the distance and from there back to the end of the distance. This signal will then be mixed with or superposed on a later emitted sound signal, so that detection of the arrival of the later emitted signal will be uncertain and impaired with an error that in many cases is not negligible. To decrease this source of error, the emitted sound signals can according to the mentioned patent application be pole switched in accordance with a specific pattern in order to reduce the effect of a reflected signal in the case where the "sing-around" principle is used. In U.S. patent 5,796,009, which corresponds to the Swedish patent having publication No.

503 614, measuring flow velocity in a fluid is disclosed using the "sing-around" method, hi the conventional way, the detection of the reception of a sound pulse triggers the emission of a subsequent sound pulse. The emission occurs with a delay, which has such a duration or length, that a complete separation in time between the subsequent pulse and the reflection of the first mentioned pulse at the receiver is accomplished. However, it may be difficult to arrange such a large delay with a sufficient accuracy. SUMMARY

It is an object of the present invention to provide a flow velocity meter having a reliable simple design and a good accuracy.

The basic process described in the mentioned International patent application can be further improved. Since the reflected signal varies in amplitude in relation to the direct signal and partly in phase depending on the temperature of the fluid, the fluid velocity of which is being measured, the zero crossing - observe that a zero crossing is in the typical case used to trigger the emission of the next pulse in the sing-around loop and to determine the run time whereas of course other positions or levels in a received signal can alternatively be used - of the received mixed signal will vary depending on the reflected signal, which has been mixed with the useful signal. To improve the measuring principle, a suitably chosen delay that can be variable, can be introduced by the control electronic circuits used, for example by the signal generator which sends activation pulses to the emitters/detectors used in the measuring process. Of course, the delay should be chosen so that the zero crossing in the mixed signal is close to the zero crossing of the signal, the run time of which is to be just then measured.

Generally, when determining the flow velocity and/or the flow of a fluid, which is flowing through a pipe or a channel, a sequence of sound signals is emitted between two sound emitting elements/sound detectors in order to propagate in the fluid and then, the sound signals are detected when they arrive to the respective sound emitting element/sound detectors. The emission of a sound signal from one of the two sound emitting elements/sound detectors is started at the arrival of an electrical activation signal to the sound emitting element/sound detector. In this way the sound signal is emitted as an oscillation having a definite frequency, which substantially corresponds to the eigenfrequency of the sound emitting element/sound detector, in particular if the electrical activation signal has a suitable shape. The electrical activation signal is generated, except the activation signal for the first sound signal in the sequence, in response to the fact that an immediately preceding emitted sound signal has been detected by the other sound emitting element/sound detector. The total length of time of the propagation of the sound signals between the two sound emitting element/sound detectors is determined for every sequence of sound signals and from these determined total lengths of time a measure of the flow velocity and/or the flow can be calculated.

The electrical activation signals of the sound signals can be supplied to the respective sound emitting element/sound detector with suitably chosen small delays, which are smaller than the period corresponding to said definite frequency of the sound signals. In this way the effect of errors, existing in the detection of a sound signal, on the calculation of flow velocity or the flow, can be eliminated or at least be reduced, even significantly reduced. Such small delays can be introduced that have lengths which are relatively well determined, hi addition, the choice of those sound signals, for the activation signal of which a delay shall be introduced, can be done in a

particular way, so that the effect of a possibly inadequate adjustment of the lengths of the delays does not to any substantial extent influence the finally calculated value of the flow velocity and/or the flow, still provided that the used lengths of the delays are relatively exactly known.

For example, the one or more sequences of sound signals can be emitted in one of the following ways. hi a first method a subsequence of a number of sound signals can be emitted without any intentionally introduced or set delays and thereafter a subsequence including the same number of sound signals, all of which are activated with a small constant delay. For example, the constant delay can correspond to an eighth of the period of the oscillation frequency of the sound signals. This pair of subsequences can be emitted along or against the flow direction and if necessary the same pair of subsequences can also be emitted in the opposite direction, i.e. against or along the flow direction, respectively, so that totally four subsequences, each including the same number of sound signals, are emitted. Then the flow velocity and/or the flow can be calculated. The sequential order in relation to one another between these four subsequences can be varied to possibly further reduce errors. Emitting a sound signal along the direction of the flow means that the directional vector of the sound signal has a positive component in the flow direction of the fluid and emitting a sound signal against the direction of the flow means that the directional vector of the sound signal has a negative component in the flow direction of the fluid.

As an alternative to this method a procedure comprising pole switching may be used. Then, a subsequence of a number of sound signals are emitted using pole switching according to a particular pattern in the way described in the above cited published International patent application WO 02/40945 and also a subsequence having the same number of sound signals and the same pattern of pole switching and in addition, with a constant delay of the activation of all sound signals. As above, this pair of subsequences may be emitted along or against the direction of the flow and if necessary, the same pair of subsequences can also be emitted in the opposite direction, so that totally four subsequences are emitted, and thereupon the flow velocity and/or the flow can be calculated. The sequential order in relation to one another between the four subsequences can also be varied to possibly further reduce errors.

In another method a sequence of sound signals is emitted, which are activated with delays varying within the sequence according to a particular pattern in order to reduce the detection errors.

In still another method all sound signals in the sequence can be activated with a constant delay, that compensates for the difference in phase between a directly detected sound signal and a reflected sound signal interfering with the detection of the direct sound signal.

Additional objects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objects and advantages of the invention may be realized and obtained by means of the methods, processes, instrumentalities and combinations particularly pointed out in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS

While the novel features of the invention are set forth with particularly in the appended claims, a complete understanding of the invention, both as to organization and content, and of the above and other features thereof may be gained from and the invention will be better appreciated from a consideration of the following detailed description of non-limiting embodiments presented hereinbelow with reference to the accompanying drawings, in which:

- Fig. 1 generally shows the structure of the sound emitter/sound detector part of a flow velocity meter,

- Fig. 2 is a diagram showing the shape of activation pulses, - Fig. 3 a generally shows the principle of the sing-around process,

- Fig. 3b generally shows the same principle but including pole switching,

- Fig. 4a shows times for emission and detection,

- Fig. 4b shows the same times but taken along a time axis,

- Fig. 5 a generally shows the structure of control electronic circuits of a flow velocity meter, - Fig. 5b shows the structure of a delay circuit,

- Fig. 6 is a diagram showing received signals including a direct signal and a reflected signal,

- Fig. 7 is a diagram of a received signal obtained as a superposition of a reflected signal on a direct signal,

- Fig. 8 is a diagram similar to Fig. 7 for a direct signal delayed by a quarter of the pulse length, - Fig. 9 is vector diagram showing direct and reflected signals, and

- Figs. 10a - 10b are vector diagrams showing a resultant for different cases of delay and pole switching.

DETAILED DESCRIPTION

In Fig. 1 a cylindrical pipe conduit 1 is shown, in which a fluid is flowing with a flow velocity v in the direction of the arrow 3. A first sound emitter and sound receiver 5, for shortness sake here called the first emitter/detector, is designed as including a flat surface having a normal located in an angle φ to the flow direction of the fluid. The flat surface of the emitter/detector forms the bottom of short, cylindrical tubular piece 9, which starts from or connects to the pipe conduit 1 and forms the same angle φ with the pipe conduit. The flat surface of the first

emitter/detector 5 has its normal arranged with a positive component in the direction of the flow, so that the flat surface of the emitter/detector is affected by a negative dynamic pressure from the fluid. A second similar sound emitter and sound receiver 6, here called the second emitter/detector, is designed and placed in the same way with its flat surface at the bottom of a second short tubular piece 9, having the same design and the same cylinder axis as the first tubular piece 8 . The flat surfaces of the emitters/detectors 5, 6 are opposite and parallel to another. The second emitter/detector 6 is affected by a positive dynamic pressure or stream thrust, as the normal of its flat surface has a component against the direction of the flow. Both of the short tubular pieces 8, 9 can be seen as parts of a cylindrical tube, that intersects the pipe conduit in an angle φ. The flat surfaces of the emitters/detectors 5, 6 may, for a suitable electrical activation, be made to emit sound waves and to detect sound waves arriving at the surfaces. Then, the sound waves emitted from one of the emitters/detectors can be detected by the opposite emitter/detector. The sound waves will have different run times for the distance between the emitters/detectors depending on the flow velocity v of the fluid in the pipe conduit 1. Each emitter/detector 5, 6 has a positive terminal and a negative terminal. When supplying a suitable electrical activating voltage to the positive terminal of one the emitters/detectors, periodically repeated sound pulses with a certain polarity from the emitter/detector are obtained, whereas when supplying the same voltage to the negative terminal of the emitter/detector periodically repeated pulses of the opposite polarity are emitted from the emitter/detector. The pulses are emitted from the above mentioned flat surface of the emitter/detector in a direction towards the flat surface of the opposite emitter/detector. Correspondingly, it is true that when the emitters/detectors work as detectors and they detect a received signal with a certain polarity, electrical signals will be output on their positive and negative terminals, that are the inverted counterparts of each other. Conventionally, the emitter/detectors are the piezoelectrical type, so that a variation of the electrical voltage at the emitters/detectors results in a change of the size of the emitter/detector, which sets the emitters/detectors in an elastic vibration movement that in turn results in the emission of sound pulses of decreasing magnitudes. The vibration movement occurs with the resonance frequency / _res , also called eigenfrequency, which is determined by the geometrical dimensions of the emitter/detector. Therefore, the emitter/detector is best activated by electrical pulses that are applied exactly with the resonance frequency. Furthermore, the emitters/detectors 5, 6 can be activated by a double electrical activation pulse, as shown in Fig. 2. This figure can also be seen as showing the dimensional change of the emitter/detector in the direction of the normal, disregarding elastic effects.

The sing-around process for an accurate determination of the run time of the signal between the emitters/detectors 5, 6 is illustrated in Fig. 3a. From for example the first emitter/detector 6 a sound signal pi of the above given type is at a time t = 0 emitted with a positive polarity. It is received by the second emitter/detector 6, and the time of arrival is determined by the detection of a specific position/specific level in the received signal, for example the above mentioned zero crossing, and the arrival is signalled to the first emitter/detector, emitting, with a certain small delay due to processing times in control electronic circuits and drive circuits, a new, second identical sound signal p ₂ . The first signal pi is reflected against the second emitter/detector 6. The second sound signal p ₂ reaches the second emitter/detector at a time, that is detected by it, and with a certain small delay a third similar sound signal p ₃ is thereafter emitted from the first emitter/detector. About the same time but somewhat earlier on account of the delayed emission of the second sound signal, the first reflection pi' of the first signal arrives to the first emitter/detector 5 and is there reflected further towards the second emitter/detector 6. Thus, the third sound signal p ₃ and the second reflection P ₁ " of the first sound signal arrive approximately at the same time to the second emitter/detector, the reflection arriving somewhat earlier due to the small delays in the emission of the second sound signal and the third sound signal. The reflection interferes with the detection of the set position of the third signal and hence the determination of the run time of the third signal from the first emitter/detector 5 to the second emitter/detector 6 will be incorrect.

This process including emission of signals is repeated and is allowed to continue for a large number of emitted sound signals. The run time of the travel of every sound signal between the emitters/detectors 5, 6 is determined but it may as mentioned above be impaired by errors. To reduce these errors some sound signals may be emitted with a negative polarity. Thus, according to the principle picture in Fig. 3b, for example the fourth and fifth emitted sound signals Xi ₄ , n ₅ have a negative polarity. It is possible to find a particular pattern of emitting sound signals with positive and negative polarities, see the above cited International patent application, which reduces the mentioned errors in the determination of the run times.

Between the time when a sound signal is detected and the time when the next sound signal is emitted there always exists, according to the discussion above, some small delay, that depends on the emitters/detectors used and the drive circuits used therefor and the necessary control electronic circuits. Using the control electronic circuits this delay can be set to a chosen value larger than or equal to a minimum value. By choosing such a delay in a suitable way for different kinds of sequences of emitted sound signals also the above mentioned errors in the detection can be reduced.

For the determination of run times of pulses and signals in the system the following times

can be defined, see Fig. 4a: tfiuid_aiong = the time for the sound through the fluid along the flow, i.e. from the surface of the emitter/detector 5 to the surface of the emitter/detector 6, tfiuid_against = the time for the sound through the fluid against the flow, i.e. from the surface of the emitter/detector 6 to the surface of the emitter/detector 5, temitter/detector_i_2 = the time for the travel of the signal through the two emitters/detectors, i.e. strictly the sum of the time period between the time when an electrical activation signal reaches one of the emitters/detectors 5, 6 and the time when the emitter/detector starts to emit its two pulses and the time period between the time when an ultrasound signal reaches the other of the emitters/detectors and the time when it starts to output an electrical detection signal, and teiectromcs = the total time of the travel of the signal through drive circuits and control electronic circuits, which time can be made variable by a suitable design of the control electronic circuits.

The different times are also shown in the diagram of Fig. 4b, where α, β, γ are positive quantities, which fulfill the conditions α < l, β < l, γ < l and (α + β) < 1. The total time that passes between the time when the control electronic circuits supply a command to emit a sound signal along the flow in the pipe conduit, i.e. from the first emitter/detector 5, and the time when the control electronic circuits have registered the reception of a detection signal from the second emitter/detector 6 and are capable of supplying a new command to emit a sound signal, is

1 loop_along ^~ t _eml tter/detector_l_2 ' tfhud_along t _e lectronics

The corresponding time for a sound signal against the flow is:

1 loop_against ⁼ t _e mitter/detector_l_2 + tfluid_against + teiectromcs

For the second reflected signal pi" of the first emitted sound signal P ₁ it takes a time ~ tfiuid_aiong + tfl _uld _ _a g _a i _nst before it is mixed with the useful signal, i.e. the direct, not reflected signal p ₃ , counted from the time when the signal has passed the emitter/detector, see Fig. 4a. Of course, a useful signal is always mixed with a reflection of the sound signal, which has been emitted two loops earlier (reflected between the emitters/detectors 5, 6). Also the reflection of the sound signal, which has been emitted four loops earlier, can be mixed with a useful signal and can to some extent interfere with the detection of this signal. The word loop is here used as a synonym for cycle and designates the course of events, when a command to emit a sound signal is supplied

up to the detection of this sound signal.

The time period from the time when a single signal is supplied from the control electronic circuits that a sound signal should be emitted along the flow until a signal has been received that the twice reflected signal has been detected and the control electronic circuits thereafter are ready to supply a new activation signal, provided that no more sound signals are emitted during the intermediate time, is then:

temitter/detector_l_2'2 + tfi _ul( j_along + tfl _ul(1 _ _a g _alns t + tfl _ulc j_along ^~ * ^~ t _em itter/detector_l_2' 2 + t _elec t _rO nics

For a sequence of signals the time period from the time when a first signal is supplied from the control electronic circuits that a sound signal should be emitted along the fluid until a signal has been received that the third sound signal after the sound signal has been detected, without interference of any reflected signal and the control electronic circuits are ready to supply a new signal to the emitter/detector is

3'(t _e mitter/detector_l_2' 2 + tfi _ul< j_ _alon g + t _em ,tter/detector_l_2' 2 + teieςtronics)

The difference, i.e. the time when the third direct sound signal arrives after the twice reflected first signal in the sequence of sound signals, is then

2 ^" t _eml tter/detector_l_2 + 2 't _e lectronics ^ ^" tflmd along " tfluid_agamst

This means that if the total time t _e i _ectromcs for the travel of the signal through the drive circuits and the processing in the control electronic circuits is increased from its minimum value t _e iectronics_imn, which is given by the times of transmission, reaction and processing in the included components, is increased by a delay of x seconds, an increase of 2x seconds of the above mentioned time difference is obtained .

In the calculations above it has been assumed that the time teiectr _o m _c s is constant. Instead, if this time is changed between loops following each other, i.e. if the delay x is changed, so that for each loop the combined times of processing in the control electronic circuits are t _e iectromcs_i, teiectromcs_2, t _e iectromcs_3, ••• for cycles Nos. 1, 2, 3, ... and the delays are X _J > 0, j = 1, 2, 3, ..., i.e. teiectronics _j = t _e iectro _mC s_mm + X _j , j = 1, 2, 3, ..., where teiectromcs_mm is the shortest possible time for the travel of the signal through electronic circuits and drive circuits, the time difference for the case above between direct and reflected signal, which is due to the processing in the control electronic

circuits, is either

telectromcs 1 ^"■" telectronics 2

or

telectromcs_2 ^" ^ ^" telectromcs_3

depending on the fact whether the change of the processing time in the control electronic circuits occurs in the processing before the command to emit a sound signal is supplied, or after a signal that a sound signal has been detected has been received. It is the latter case that is illustrated in Fig. 4b. hi the control electronic circuits the activation signal can be generated with a delay that does not have to be synchronized with or against some other clock. The delay can be relatively small but with a suitably chosen magnitude it can give a considerably increased accuracy in the detection of the desired zero crossing. The delay can for example be achieved by using delays in logic electronic circuits, such as by using a suitably chosen number of gate delays. Thus a number of inverters connected in series after one another can be used that in turn increment a counter so that the counter is increased by one for each gate delay cycle, until the number of steps for the desired total delay has been reached. As an alternative, the counter can of course be counted down from a value, which indicates the total desired delay, until the counter reaches zero or switches. Such a delay having a magnitude of (a fixed number of gate-delays)*the number of cycles counted as steps on a counter can if needed be calibrated against a clock in the control electronic circuits, such as a clock controlled by a crystal in a microprocessor, at uniform time intervals, e.g. each hour. In the control electronic circuits a control unit such as a microprocessor can write the length of the delay, before the signal pulse is sent. The same delay, x according to Fig. 2, that has been input from the beginning can be used for all loops in a sequence of sound signals but the delay can also be changed between each loop in a sing-around-sequence depending on the principle chosen. If the time delay is changed between the loops a new value for the delay has to be written between emitting and receiving an ultrasound signal, hi particular, this method can be used in the case where no pole switching or no change of the polarities of the pulses according to the mentioned International patent application is performed.

In Fig. 2 the shape of a double activating pulse is shown that can be emitted from a signal generator or drive circuit associated with an emitter/detector to the respective emitter/detector for

activating emission of a sound signal. The given times are: x = the delay before emission of a pulse that is introduced after each detection of a zero crossing before the next activating pulse is generated. t _p = the length of each of the two primary pulses of activation, which is determined in correspondence to the eigenfrequency of the ultrasound emitters/detectors. t _m = the length of the intermediate time between the two primary pulses. tf = the pulse period, i.e. the time period between the start of the two primary pulses. This can be chosen, in order to get a resulting sound signal that is as well defined as possible, with regard to the eigenfrequency of the emitter/detector, so that t _f = \lf _res , i.e. the pulse period is equal to the period time of the eigenfrequency. This is assumed to be valid in the sequel of this description.

If the pulses are emitted with a duty cycle of 50 %, which also can be required to get a resulting sound signal that is as well-defined as possible, the length t _p of the pulses, i.e. the pulse time, and the intermediate time t _m between them have the same size and are then equal to half the pulse period l/(2/ _πs ). Thus, in the latter case the pulse length or pulse time t _p for e.g. an eigenfrequency of 1 MHz is equal to 0.5 μs.

The double activation pulse according to Fig. 2 is generated at the start of each loop. By introducing a delay x for the activating electrical signal, i.e. a delay of the time, when it starts, the reflected sound signal can be displaced in relation to the direct sound signal, hi Fig. 6 the thinner line shows the reflected sound signal and the thicker line the direct sound signal at the receiver without any introduced delay. The direct sound signal and the reflected sound signal are mixed with one another and give the sound signal shown in Fig. 7, which is the actually detected sound signal. The grey line shows the next emitted activation signal, i.e. the next activating double pulse, that is supplied from the drive circuits to the ultrasound emitter/detector.

It appears from these diagrams that by introducing a small chosen delay in the activation of the direct sound signal, the direct sound signal can be displaced, so that it is in phase with the interfering reflected sound signal or at least so that the phase difference between the direct sound signal and the interfering reflected sound signal is substantially reduced. Then the zero crossings of the two signals will be close to one another. Such a displacement can in that case always be obtained using a delay, which is shorter than the period time l/f _τes of the oscillation of the sound signals and which for a sequence of sound signals emitted with a constant delay is shorter than half the period time l/(2f _κs ).

However, it may be difficult to set the delay accurately enough. Then it is possible to use delays which have for example permanently set, small values, for example for each of sequences of sound pulses emitted after one another, and still obtaining an increased accuracy in the

calculation or the flow velocity and/or the flow, which is finally done by evaluating the total detected run times of the sound pulses in each sequence. Then, the values of the delays can generally be fractions of half the period time, i.e. x = 1/n ^• 1/(2/^), 2/n ^• \/(2f _τes ), ..., (n-1) /n ^• l/(2/res), where n is a positive integer, typically equal to 2, 4, 8, ... Such methods will now be described.

For the delay x = t _f /8 for the activating electrical signal in each loop in a sequence of sound signals the direct sound signal will, as described above, be displaced by the length of time tf/4 from the interfering reflected sound signal, see Fig. 8.

Thus, the interfering reflected sound signal arrives in this case a time tf/4 earlier compared to the direct sound signal. The read loop time will thereby increase by ~ t _f /8 due to the delay. This delay of the loop time can be compensated for in the software that is used to finally provide, from the measured times, a value of the flow velocity. The prolonged loop time depends on the delay from the chosen detected zero crossing until the next pulse is emitted, see Fig. 8.

The same delay x for the activating electrical signal can be used in measuring along or against the flow of the fluid.

By varying the magnitude of the delay in this way, from one sequence of sound signals to the next sequence of sound signals, with a small amount in a suitable way, for example with an amount, which is smaller than half the pulse period, i.e. so that 0 < x < t _f /2, the effect of the interfering sound reflection on the measurement of the flow velocity can to a large extent be sorted out in the final calculation of the flow velocity and perhaps also to a still higher degree in a calculation of the flow of the fluid, as mentioned above.

Together with the method of pole switching that is described in the above mentioned International patent application, a still better result can be achieved, as the direct sound signal and the reflected sound signal are not pure sine curves but have varying amplitudes. When a switch of pole is performed the absolute amplitude of the reflected sound signal varies less for positive and negative polarities around the detection of the zero crossing. As to the phase of the sound signals, a switching of pole corresponds to a phase displacement of half a period or 180°. Such a displacement can here be assumed to correspond to a delay in the electronic circuits corresponding to half of half the pulse period, i.e. a delay of x = t _f /4. Then, it then turns out that it is possible to use in the measurement sequences of sound signals, which are emitted with different polarities and/or with delays in suitable patterns, for which the magnitudes of the delays have an amount which is smaller than half of half the pulse period, i.e. so that 0 < x < t _f /4. As above, delays with for example small constantly set amounts can be used, for example for each of sequences of sound pulses emitted after one another. Then instead, the length of the delays may be fractions of half of

half the period time, i.e. x = 1/n ^• l/(4f _τes ), 2/n ^• l/(4f _τes ), ..., (n-1) /n ^• l/(4f _τes ), where n as above is a positive integer, which typically can assume any of the values 2. 4, 8, ... Hence, the effect of the interfering sound reflection on the measurement of the flow velocity can be better eliminated in the final calculation of the flow velocity and/or the flow, which will be described in more detail below.

By using the vector diagram in Fig. 9, the advantage of using both pole switching and small delays according to the discussion above can be approximately described. The sound signals are assumed to be pure sine signals and can then be represented as vectors, the lengths of which correspond to the amplitude of the sound signal and the angular positions of which correspond to the phase of the sound signal, hi Fig. 9 is:

D = the vector of the direct sound signal and reference, amplitude = D, angle = 0°

R ₁ = the reflected, interfering sound signal when its original direct sound signal has the same polarity as the considered direct signal, amplitude = R ₂ , angle = 45° (this case corresponds to telectronics min Lf/ 1 u) R ₂ = the reflected, interfering sound signal when its original direct sound signal has a polarity opposite that of the actual direct sound signal, amplitude = R ₂ , angle = 225° R ₃ = the reflected, interfering sound signal when its original direct sound signal has the same polarity as the considered direct signal and the direct sound signal is emitted with a displacement = t _p /4 in relation to the original direct sound signal of the interfering, reflected sound signal, amplitude = R ₃ , angle = 135°

R ₄ = the reflected, interfering sound signal when its original direct sound signal has a polarity opposite that of the actual direct sound signal and the direct sound signal is emitted with a delay = t _p /4 in relation to the original direct sound signal of the reflected, interfering sound signal, amplitude = R ₄ , angle = 315°. The resultant is shown in the four different cases in Figs. 10a - 1Od. The difference in amplitude is not of any interest in these figures but the change of the direction of the resultant, which is equal to the time error for the zero crossing. If these four variants are added together, a much smaller error of detection of the zero crossing is obtained.

R ₁ => positive error R ₂ -> negative error R ₃ => positive error R ₄ => negative error

R ₁ and R ₂ are generated in half the loops if pole switching is used in the manner described in the cited International patent application.

According to the International patent application the vectors D and Ri are added in half the number of loops for sing-around and in the rest of the loops the vectors D and R ₂ are added together. The detection of zero is displaced in the positive direction due to the sound reflection in one of the cases and almost as much in the negative direction in the other case. If the amplitudes of RJ and R ₂ are small enough, the error in difference of time between measurements along and against the flow at different temperatures will be negligible.

If the demands of accuracy are increased or in the case where the emitters/detectors are not as good and give sound reflections having higher amplitudes, typically > 0.2 * the direct signal, a measurement can also be done, in which the reflection is displaced 90° in relation to R ₁ and R ₂ half the times. This is done by delaying the emitted signal a time corresponding to tf/8 in the signal generator, which is used to generate the activation pulses, this giving a displacement of tf/4 between the sound signal and corresponding to a displacement of 90°. This does not have to be exact to produce a distinct effect and can reduce, by a factor > 10 times, the systematic error in determining the flow velocity and/or the flow due to the reflection .

This is achieved, executing the measurement first according to the mentioned International patent application without any delay, i.e. with x = 0, along or against the flow direction of the fluid using the required number of loops and then executing a new measurement for the same number of loops with a constant delay x = t _f /8 for each loop. The difference between the measurements is that a delay has been added to the loop, before the supplying of activation pulses is executed.

Then, in a fourth of the loops the signals D and R ₁ are added to each other, in a second fourth of the loops D and R ₂ are added to each other, in further another fourth of loops D and R ₃ are added to each other and in the remaining fourth of the loops D and R ₄ are added together. To get a maximum accuracy in the measuring using delays, delays x can be used having for example those different lengths from x = 0 to x < t _f /4, which the resolution in the used delay circuit, see below, makes possible in the measurements performed according to the process described in the International patent application, using the signal generator described herein.

Different variants in which a delay of the activation pulses emitted from the signal generator are used together with the prior art pole switching method, may include:

1. Measurements are executed according to the International patent application using the signal generator described herein, introducing a delay after a measurement has been done along and against the flow until the next measurement along and against the flow will be performed with a new delay. The length of time x of the delay is varied with different values between x = 0 and x

< tf/4 with the resolution allowed by the used delay circuits. The different delay values are used the same number of times. The systematic error in the determination of flow velocity and/or the flow will then be ~ 0 after all delays have been used. According to the description above one can for example have a resolution as to the delay time set by the electronics, which is equal to tf/8, which makes only two delays having the value x = 0 and the value x = t _f /8 possible for the first case above.

If the resolution instead is t/32, delays having lengths from x = 0 to 7*t _f /32 can be used, also for the first case above. Then, the systematic error can be further reduced by performing, at each measurement occasion, two measurements along the flow direction, i.e. measurements of the total run time for two sequences including the same number of sound signals emitted along the flow direction are performed, and two measurements against the flow direction, i.e. measurements of the total run time for two sequences including the same number of sound signals emitted along the flow direction are performed, so that the measurements in each flow direction are done using two different delays X ₁ and X ₂ . Then, the order of the measurements can be varied according to the following scheme:

- measurement against the flow with delay X ₁ ,

- measurement along the flow with delay X ₁ ,

- measurement along the flow with delay x ₂ , and

- measurement against the flow with delay x ₂ . Using such a procedure, in which the order is shifted for the two delays X ₁ and X ₂ for measuring along or against the flow direction, the effect of changes of flow and changes of temperature can be reduced. The pairwise change of delays can generally be chosen, so that measurements are done, in which the difference between the delays in the pair is t _f /8. In the case where the resolution is t _f /32 the delays chosen for the pairs then are (X ₁ , X ₂ ) = (0, 4*t _f /32), (t/32, 5*tf/32), (2*tf/32, 6*t/32), (3*t _f /32, 7*V32).

2. Measurements are executed without pole switching at emitters and receivers but using the signal generator as described herein, the delays changing between sing-around loops. If this is done in a particular sequence, the different phase differences are obtained the same number of times, see the example described below. A resolution according to point 1 can be used with delays having lengths from x = 0 to x < t _f /2. If one wants to achieve similar results as described above with the different resultants, the delays for a resolution of t _f /8 should be changed in the loop according to: x = 0, x = t _f /8, x = t _f /4, x = 3*t _f /8. This will not produce results as good as measurements according to point 1, but in the case where it is not possible to use pole switching at the emitter and receiver, this principle can be used. The amplitude of a reflected signal at a zero

crossing is not the same for the delays x = 0 and x = t/4 according to point 1 above, in which pole switching is used. That would give approximately the same effect as shown in Figs. 10a - 10b for the different resultants. (If the errors are added together they will be equal to ~ 0)

Examples of sequences if the resolution of the delay circuit is tf/8 are given in Table 1 below. The phase difference equals the sum of the two earlier used delays.

Table 1 x = 0 the first emission of pulse in the sing-around loop x = 0 the second X = O the third x = t _f /8 the fourth x = 0 the fifth x = 2*t _f /8 the sixth x = 0 the seventh x = 3*t _f /8 the eighth x = 0 the ninth x = 0 the tenth

Table 2. Result of the above described so quence

In Fig. 5a the structure of electrical/electronic components of the flow velocity meter is generally shown, which are included in a control electronic device 11. It supplies activation signals to the sound emitters/sound detectors 5, 6 and receives detected signals from them. In the figure only the case is shown where emitter/detector 5 works as a generator of sound pulses, while emitter/detector 6 works as a detector. To be capable of managing the reverse case, the control electronic circuits can be provided with suitable multiplexers and gates and/or a doubling of some structural elements can be made, compare the above mentioned International patent application.

The signal processing in the control electronic device 11 can be controlled by a monitoring unit 13, for example a microprocessor, that can receive commands from an operator via input devices, not shown, like control buttons, a keyboard, etc. If desired, it can be provided with or connected to calculating circuits, not shown, for determining for example times for detection windows, suitable delays and measured real flow velocity and/or measured real flow. Such circuits for calculating the real flow velocity and the flow correspond, if they are provided, to the above mentioned software. The monitoring unit should also be connected to some output device or display device for indicating or showing the result of the flow velocity measurements. In a simple design for example only the total number of measured clock pulses can be shown on a simple display, not shown, whereupon an operator knowing the number of sing-around cycles can calculate a value of the flow velocity.

The control electronic device 11 further includes a signal generator/drive circuit 15, which after the reception of an enable-signal supplies the two activation pulses to the emitter/detector 5 for emission of the sound pulse. The activation pulses are also fed to a counter 17 for detection window for starting it and to a counter device 19 for determination of the total run time. The latter includes special logic circuits, not shown, such as two separate counters, so that one of these counters, at the reception of successive activation pulses in a run time determination in one direction, counts these pulses and compares the number of counted pulses to the total number of sing-around-cycles, which the measurement shall comprise. When the total number has been reached, the second included counter is set to an enable-state, so that it will stop at the next reception of a signal from the zero crossing detector 21. Furthermore, a disable-signal is delivered to the signal generator/drive circuit 15, so that no more activation pulses will be supplied. The second counter is counting, from its start at the reception of the first activation pulse until the reception of a detection pulse, clock pulses from a clock pulse generator, not shown, which outputs pulses of a relatively high frequency. When such a last detection pulse has been received, the second counter is accordingly stopped and the number of counted clock pulses is displayed or transferred to some other device or unit to be shown or further processed.

The zero crossing detector 21 receives the signal detected by emitter/detector 6 but is only in an active state, after having received an enable-pulse from the counter 17 for the detection windows. After having received such an enable-pulse, it detects the next zero-crossing and then outputs a signal, which again sets itself in a non-active state of detection. The same signal is transferred to the counter 19 for run time determination, which as already been mentioned, only reacts to the signal, when the predetermined number of activation pulses for a sing-around-process have been received. The same signal is also transferred to a delay circuit 23, which will be described in more detail below and which after a after set delay time delivers a signal to the signal generator/drive circuit 15 to again emit the double activation pulse. The various included components may need different parameters and they can be stored in a suitable way, for example in memory cells in a memory 25. Such memory cells or registers may contain the total number of measured clock pulses after the latest sing-around-process, a start value for the counter 17 of the number of clock pulses until an enable-signal shall be supplied to the zero crossing detector 21, the total number of cycles, i.e. the total number of activation pulses, for a sing-around-process, the number of delay cycles, etc.

The delay circuit 23 may have a structure as shown in Fig. 5b. Before the next signal from the zero crossing detector 21 is emitted, the total number of steps or cycles, which shall be included in the next delay are transferred from the corresponding memory cell or register (hold circuit) to a counter 31. The transfer may for example be initiated by the activation signal from the signal generator/drive circuit 15. When the signal then arrives from the zero crossing detector, the delay is for example started on the increasing flank and the counter is decremented for each such received flank on its counting input. The same signal is also further fed in parallel to one or more simple logic gates 33 connected in series with one another and the increasing flank is transformed into a new pulse. This gate or these gates causes/cause a predetermined delay. The new pulse is fed back to an OR gate 35, through which the detection signal also passes and is therefrom fed both to the counter 31 and to the logic gate or gates to give a new delay cycle. When the counter has counted down to 0, the delay is finished and the signal to the signal generator/drive circuit 15 is delivered to emit a new activation pulse.

Instead of delay cycles accomplished by feedback using one or more logical gates 33, the counter 31 may of course count clock pulses from a clock generator, not shown, the start of the counting being provided by the signal from the zero crossing detector 21.

While specific embodiments of the invention have been illustrated and described herein, it is realized that numerous other embodiments may be envisaged and that numerous additional advantages, modifications and changes will readily occur to those skilled in the art without

departing from the spirit and scope of the invention. Therefore, the invention in its broader aspects is not limited to the specific details, representative devices and illustrated examples shown and described herein. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents. It is therefore to be understood that the appended claims are intended to cover all such modifications and changes as fall within a true spirit and scope of the invention. Numerous other embodiments may be envisaged without departing from the spirit and scope of the invention.