Acoustic Flowmeter With Envelope Midpoint Tracking

This invention relates to an acoustic flowmeter and more particularly to such a flowmeter having improved accuracy and reliability.

Various systems have been developed for measuring the velocity of a fluid flow by utilizing the effect the fluid velocity will have on the transit time of acoustic pulses transmitted through the fluid medium along the axis of flow. One such system uses a dual "sing-around" technique in which one series of pulses is transmitted in one direction, e.g. upstream, and another series of pulses is transmitted in the opposite direction, e.g. downstream. The time when each pulse is received determines when the next pulse will be transmitted in the same direction. Accordingly, the repetition rate of the pulses in each direction is predominantly determined by the transit time between spaced transducers in the fluid medium. The transit time, in turn, is proportional to the flow velocity of the fluid medium. Thus, the pulse repetition rate provides an indication of fluid velocity.

Other systems have attempted to measure directly the elapsed time between transmission and reception of acoustic pulses in the form of repetitive triggered bursts of high frequency acoustic energy having a characteristic amplitude envelope. Because the beginning and end of the bursts is low in amplitude, one approach is to detect the peak amplitude of the envelope signal, and to measure time from peak to peak. However, because of the signal to noise ratio of broadband transducers at the relatively low acoustic frequencies employed, e.g. 500 kHz and below, precise peak detection has proven to be difficult. It would be desirable to have a system of measuring the time between transmission and reception of the signal envelope that is more accurate.

Accordingly, it is an object of this invention to provide an acoustic flowmeter that includes accurate, reliable, and inexpensive acoustic pulse transit time measuring system which does not rely on direct peak detection.

These and other objects of the invention are achieved with two spaced transducers by periodically transmitting a synthesized burst of ultrasound having a predetermined amplitude envelope and employing a feedback loop to track the midpoint of the received envelope. A variable time based generator produces a square wave signal, the leading (rising) edges of which repetitively trigger the waveform generator. The received signal is rectified and fed to an integrator via a polarity inverting circuit controlled by the square wave signal. The frequency of the square wave signal is controlled by the output level of the integrator. If the point at which the square wave signal changes from high to low (half cycle transition point) comes precisely at the midpoint of the received rectified envelope, the envelope will be cleaved into two halves. Thus, during the first half cycle of the square wave signal, the integrator output will change by a given amount in one direction and during the second half cycle of the square wave signal, the integrator output will fall by the same amount and return to the same output level, leaving the square wave signal at the same frequency. On the other hand, if the half cycle transition point occurs before or after the midpoint of the received envelope, the bipolar integrator inputs will be unbalanced and the integrator output will be left at a different level tending to lower or raise the frequency of the square wave signal, respectively. Thus, the output of the integrator produces an error signal which is "nulled" by shifting the half cycle transition point toward the midpoint of the preceding received envelope.

In the preferred embodiment, the half cycle transition point signal enables a zero crossing detector which issues an output signal upon the next zero crossing of the unrectified received signal. The zero crossing detector output thus tends to track the arrival of the midpoint of the bursts. The time interval between the transmission of the burst and the output of the zero crossing detector is preferably measured and averaged over numerous repetitions in one direction relative to the fluid flow and then repeated for the opposite direction.

Typically a number of measurements in the upstream direction are used to determine an average transit time T.sub.up in that direction. The same number of measurements is then made in the opposite direction to determine an average transit time T.sub.down for downstream propagation. In order to avoid inclusion of measurement values distorted by signal processing noise or the like, an analysis is carried out to determine the observed standard deviation for each series of measurements. If this observed deviation exceeds a predetermined limit, e.g. 1% to 2%, representing expected variation due to normal turbulence that value of T.sub.up and T.sub.down is disregarded. From the T.sub.up and T.sub.down, the flow can be computed from the following equation: ##EQU1## where: P=path length of the transmitted burst from transmitter to receiver;

F=a Flow Profile Factor; and

.theta.=the angle between direction of the transmitted ultrasound beam and a normal to the direction of flow of the stream.

The expression for observed standard deviation is ##EQU2## where T.sub.i =each individual measurement of T and

n=total number of measurements in the series.

A more precise calculation could be carried out by determining a corresponding flow velocity V.sub.n for each measurement. However, this requires both an upstream and downstream measurement for each determination in the series.

Other objects, features and advantages of the invention will be apparent from the following description of a preferred embodiment, including the drawings, in which:

FIG. 1 is a functional block diagram of an acoustic flowmeter system according to the invention; and

FIGS. 2 and 3 are timing diagrams illustrating the relationship between typical signals in the system of FIG. 1.

As shown in FIG. 1, an acoustic flowmeter uses a pair of acoustic transducers 10, 12 located at upstream and downstream positions in a conduit (not shown) where flow is to be measured. A waveform generator 14 when triggered generates a synthesized high frequency burst with a characteristic amplitude envelope having a predetermined frequency, amplitude and shape to fit the application, transducer and receiver. The form of the signal envelope 16 is shown in the drawing to the right of the waveform generator 14. The signal typically rises rapidly in peak-to-peak amplitude to a maximum at the center or midpoint and then rapidly tapers off. The ultrasound frequency within the burst is typically on the order of 500 kHz with an overall duration on the order of 10-100 microseconds. The signal 16 is converted to a powerful driving signal in transmitter 18 and passed via transmit/receive switch 20 or transmit/receive switches 20 and 22 to transducer 10 or 12, respectively, depending on whether upstream or downstream transmission is intended.

The transducer 10 or 12 converts the transmitter signal 16 output to a directional ultrasound burst which moves through the flowing medium to the other one of the transducers 10, 12. The sound energy is converted to an electrical signal which is amplified and filtered in receiver 24 to produce the received signal output 25.

At a rectifying circuit 26 the received signal 25 is rectified to produce the rectified signal 28, shown in FIG. 1. The signal 28 is composed entirely of positive components on a zero base line. The rectified signal 28 passes to a switch 30 which reverses the polarity of the signal at a time determined by the square wave output 32 of a voltage controlled oscillator circuit 34. The rectified envelope is passed via the polarity switch to analog integrator 38.

The integrator output forms the control voltage input to the voltage controlled oscillator circuit 34 which triggers the waveform generator 14. Circuit 34 also synchronously produces the square wave 32 which operates the polarity switch 30 and enables or arms a zero crossing detector 46. Detector 46 produces an output pulse upon the next zero crossing of the unrectified received signal following the half cycle transition point 44 of square wave 32 as shown in FIG. 1.

A set-reset flipflop 48 is set by a timing pulse (C) from the oscillator circuit 34 corresponding, for example, to the midpoint of the transmitted signal envelope and reset by the output of the zero crossing detector 46. The Q output of the flipflop is gated by the clock output of free running oscillator 54 in a pair of AND circuits 50, 52, corresponding to upstream or downstream operation of the flowmeter. The output of each AND circuit forms the clock input to a respective digital counter 56, 58. The output from the counters 56, 58, moreover, may be fed to a microcomputer or other device (not shown) where the transit times may be further processed.

A cycle counter circuit 60 counts trigger signals (A) from the voltage controlled oscillator 34. The counter 60 determines upstream or downstream operation of the flowmeter by interchanging the transmit/receive switches 20, 22 and counters 56 and 58, as shown. As illustrated in FIG. 3, the cycle counter circuit 60 is arranged to allow a given number of repetitions of signal envelopes 16 to obtain frequency lock, or balance, before enabling the appropriate counter 56, 58 through AND circuit 50, 52. Once measurement begins, for example, upstream, the cycle counter 60 continues upstream operation until a preset number of repetitions are averaged. If the counter 56 accumulates the total count from a fixed number of repetitions, the count will be proportional to the average. The cycle counter circuit 60 then switches to downstream operation and repeats the cycle as shown in FIG. 3.

In operation, the waveform generator 14 is triggered by pulse A which coincides with the rising (leading) edge 62 of the square wave 32 (signal B of FIG. 2). The ultrasound burst 16 is emitted from the upstream transducer 10. Gauged from the trigger pulse signal A, the midpoint of the envelope 16 always occurs at time T.sub.0 coinciding with the set pulse (signal C) from the VCO circuit 34 to flipflop 48. Thus, the Q signal goes high. Meanwhile, the first half cycle of square wave 32 is underway while the burst 16 is travelling through the fluid medium. The received envelope 25 is undergoing rectification as shown in line E when, prematurely, the half cycle transition point 44 reverses the polarity of the input in line F (waveform 36' in line F) before reaching the midpoint of the received envelope. Thus, the resulting integration is unbalanced in the negative direction and the output of the integrator 38 ends up at a lower level after reception of this first burst. In addition, however, the half cycle transition signal enabled the zero crossing detector which produced an output which reset the flipflop 48. Consequently, if counter 56 were enabled, a number of pulses (too few) would be counted by counter 56 in an attempt to measure the interval between the midpoints of the transmitted and received envelopes.

In the second cycle shown in FIG. 2, the error signal output of the integrator 38 on line G has slightly lowered the base frequency of the voltage controlled oscillator circuit 34 thus proportionally lengthening the overall period of the square wave 32 by an amount delta T sufficient to delay the half cycle transition point 44' until approximately the midpoint of the received envelope 25. In this example, the rectified signal 28 is cleaved in two as shown in lines F and G resulting in a balanced integration cycle which leaves the voltage control signal at the same level.

The half cycle transition point is now synchronized with the center of the received envelope. The system having locked onto the midpoint in this example, the correct number of clock pulses are counted by counter 56. Unless the transit time changes, the VCO frequency will remain approximately constant over the ensuing cycles of operation such that the average count will be proportional to the transit time in the forward direction. Since the burst generator trigger pulse A is synchronized to the square wave 32, the pulse repetition period of the bursts 16 is by design twice the average transit time in a given direction.

Using the zero crossing detector circuit 46 instead of the half cycle transition point 44 as the detector output ensures that the system will only detect received signals. Moreover, the zero crossing detector relaxes the requirement of signal detection to the point where the tracking circuit only has to point to the mid cycle in the signal envelope. The zero crossing detector then pinpoints an exact point in the midcycle, namely the zero crossing which always occurs, at least theoretically, at the same time within the envelope itself. Thus the zero detector serves to "fine tune" the measurement reference point.

If desired, the upstream and downstream readings can be subtracted from each other in which case the interval T.sub.0 to the midpoint of the transmitted signal can be ignored since it will cancel out in the difference. In this case, the midpoint of the synthesized waveform does not have to be known precisely in advance. In fact, the specific shape of the envelope and amplitude do not have to be known precisely in advance. All that is required for accurate transit time measurements is that the shape or contour of the envelope stay relatively constant over a sufficient number of cycles. Slow amplitude drift or even gradual changes in the contour of the envelope should not make any significant difference.

The system described above uses a very reliable measurement point in the signal envelope 25, namely the zero crossing point of mid cycle of the signal waveform instead of the relatively uncertain and variable peak amplitude. The bandwidth requirements of the system are thus reduced, increasing the signal to noise ratio of the detected signal. Bandwidth requirements of the transducer are also less of a problem with the dynamic tracking system of the invention. Reliability is greatly increased by use of the amplitude-independent adaptive control system of the invention.

Instead of single ended integrator 38, a differential integrator can be used, in which case the rectified signal input is switched to the inverting input of the differential integrator at the half cycle transition point. Alternatively, the integrator 38 can be replaced by other amplifier and low pass filter arrangements known to the art of closed loop feedback design. Any of the analog signal processing steps herein may be converted to digital or vice versa, if desired, for a particular application. In addition, for example, the same counter can be employed for both directions, the output bit of the cycle counter being used to indicate where to store the resulting count.

Other variations of components of the system besides those given in the discussion above may be made without departing from the spirit of the invention. The scope of the invention is set forth in the following claims .