The invention relates to an ultrasonic flow meter adapted to perform measurements throughout a wide measuring range.

Such flow meters find application in heat meters used for recording the individual heat consumption in buildings connected to a block or city heat supply network. Fig. 1 shows an example of the sensor section of such a meter.

The sensor section consists of a measuring tube of a diameter D through which the liquid flow to be measured passes. Two ultrasonic transducers are mounted in the measuring tube in facing relationship at a distance _1 from each other.The two transducers transmit simulta¬ neously, and with the same phase, bursts of a sine wave measuring signal having a carrier frequency of, for example, 1.18 MHz. The difference between the times of arrival of the bursts at the two op¬ positely mounted transducers will be proportional to the velocity v of the liquid between the transducers.

With denoting the sound velocity in the liquid, jtl the time of propagation from transducer 1 to transducer 2 and _t2 the time of . propagation from transducer 2 to transducer 1, and with the liquid flowing into the direction from transducer 1 to transducer 2, then: tl = —: and t2 = c + v c - v

The difference dt in arrival time is then:

_d_ _# t. -= -t• _. - -tlι -= 1 - 1; ^β .. -l}(c+v?) - 7—l(;—c-?v) _= —22.1.v_r , ( 1 c-v c+v (c-v) . (c+v) or -or In a flow meter in accordance with the example described, the length 1 is 0.124 m and the diameter D of the measuring tube is

0.010 . The sectional area 0 is then 78.54 x 10~ ^{6 2} . The relation¬ ship between the volume flow velocity V in liters per hour and the linear flow velocity v in meters per second is then: v = V/283 Assuming the upper measuring limit of V, i.e., V ax, to be 1,000 liters per hour, then the maximum linear flow velocity vmax = 3.54 m/s. Assuming the lower measuring limit of V, i.e., Vmin, to be 101/h, then vmin = 0.0354 m/s. If the liquid to be measured is water having

a temperature of 25°C, then the sound velocity c ^» 1,500 m/s. As the liquid flow velocity v will always be very much smaller than the sound velocity , equation (1) may be simplified to: dt = 2.1.v/c ² (2) without an appreciable error being introduced.

In the present example, the maximum and minimum arrival time differences (dt ax and dtmin, respectively) will then be: dtmax = 0.39 /us dtmin = 0.0039/us (= 3.9 ns) Especially the lower measuring limit leads to, for example, extreme¬ ly brief time differences, while a measurement accuracy of 15% is re¬ quired, which corresponds to 0.6 ns.

The direct measurement of time differences on this order requires the use of extremely fast electronic circuitry and components. It is an object of the invention to provide an alternative method of measuring the time differences in which use is made of standard elec¬ tronic components. The starting point is that the measurements are performed by means of high frequency sine wave measuring signals hav¬ ing a carrier frequency of, for example, 1.18 MHz. As the transducers are used for transmission as well as reception, the measuring signal has to be transmitted in finite bursts, for example with a length of 64 cycles, after which the transducers can be used for reception.

Fig. 2 shows a block diagram of an embodiment of the electronic circuit arrangement in accordance with the invention. Fig. 3 shows the time sequence diagram pertaining to this circuit arrangement.

The crystal oscillator 1 coacts with the divider-by-four 2 to gen¬ erate the frequency of the measuring signal (1.18 MHz). The transducers in the measuring tube 17 are driven via the transmit gate 13 and the transducer circuits 15 and 16. The voltage across the transducers are also applied to two sample gates 18 and 20. At the point of time when a burst transmitted by one transducer reaches the other transducer, the two sample gates are concurrently opened 32 times for brief periods by 32 sampling pulses. The hold circuits 19 and 21 are charged by the 32 samples. The sampling pulses are produced in the sampling pulse gate 24. A 1.18 MHz block signal is applied to sampling pulse gate 24 via

divider-by-four 12 and needle pulses (pulse width of approximately 100 ns) are formed from 32 1.18 MHz cycles under the control of the samplin control circuit 23. These needle pulses cause sample gates 18 and 20 to open for brief periods. Divider-by-four 12 receives a 4.72 MHz signal from the voltage- controlled crystal oscillator 11. Via the phase locking circuit con¬ sisting of the mixer 8, the phase comparator 9 and the loop filter 10, and with the 288 Hz block signal derived via divider circuits 2-7 from crystal oscillator 1 as a reference, voltage-controlled crystal os- cillator 11 is shifted in frequency by 288 Hz relative to the frequency of crystal oscillator 1.

The sampling pulses then have (during the burst of sampling pulses) a frequency that is 72 Hz higher than the carrier frequency of the measuring signal. The result is that within a period of 1/72 second, a full cycle of the 1.18 MHz measuring signal is scanned by the sample pulses.

The divider circuits 2-6 and the AND- gate 14 are operative to control the (transmit) bursts so that each time a burst with a length of 64 cycles (54/us) is transmitted concurrently by the two transducer In Fig. 3, line A represents the control signal for transmit gate 1 while line B represents the transmit signal itself.

The bursts arrive 83 /us later at the oppositely mounted trans¬ ducers. Line C in Fig. 3 represents the signal received.

The beginning of the received burst is distorted due to transients caused by resonance of the transducers. Sampling will therefore be initiated only after the beginning of the burst, i.e., 112 cycles (95/us) from the beginning of the transmit burst. Subsequently, during a period of 32 cycles a sample is taken of each cycle of the measuring signal received. Line D shows the signal produced by sampling control circuit 23. Line E represents the sampling pulse burst.

Fig. 4 shows a detail of Fig. 3, i.e., the interval during which the measuring signal received is being sampled. Line C represents the signal received, while line E' represents the successive sampling pulses. The 32 samples charge hold circuits 18 and 20, respectively.

Line F represents the output voltage of the hold circuit. During the

32 samplings within a burst of the measuring signal, the hold circuit gradually adopts the value of the instantaneous voltage of the measuring signal received at the sampling point.

It can be calculated that during the sampling of burst, the sampling pulse shifts only through 6.6 ns relative to the cycle of the measuring signal (the total duration of one cycle is 847 ns), while the duration of the sampling pulse is approximately 100 ns. It may therefore be said that, practically, the 32 successive sampling pulses sample a single point of the measuring signal. The next transmit burst is transmitted 434/us (512 cycles of the measuring frequency) after the previous transmit burst, and the same sampling procedure is executed. However, the sampling pulses have now been shifted through 1/32 cycle (26.5 ns) relative to the cycle of the measuring signal. Consequently, in 32 successive bursts an entire cycle of the 1.18 MHz measuring signal is scanned.

The hold circuit, which during the reception of a burst each- time acquires the voltage of the measuring signal at the instant of sampling, thus produces an AC voltage having a frequency of 72 Hz and a shape identical to that of the measuring signal (normally a sine wave shape). A low-pass filter smoothes out the step-like ripple on the output voltage of the hold circuit.

Line C" in Fig. 5 shows each time a cycle of the received measuring signal in each burst. Line E" shows each time a sampling pulse in each burst. Fig. 5 shows that the instant of sampling shifts relative to the sine wave cycles of the received measuring signal. For the sake of clarity, in Fig. 5 it is assumed that the entire cycle is scanned in 8 samplings instead of in 32. Line F' represents the output voltage of the hold circuit, while line G represents the waveform obtained by smoothing out by means of the low-pass filter. The sampling is executed concurrently in both reception paths connected to the two transducers. However, when there is a phase difference between the two measuring signals, the same phase difference will exist between the two 72 Hz output signals.

Consequently, a transformation takes place from two pulse-shaped received measuring signals having a carrier frequency of 1.18 MHz to two continuous signals having a frequency of 72 Hz, while the phase difference between the two 72 Hz signals is identical to that between

the two original 1.18 MHz signals. Consequently, now the phase differen between two continuous, sine wave derivative measuring signals having a frequency of 72 Hz has to be measured, which can be performed with far greater accuracy than the measurement of a similar phase difference between two 1.18 MHz measuring signals. In other words, the minimum time difference of 3.9 ns to be measured has been increased by a factor 16384 to 64/us. This time and phase difference can be measured success¬ fully by means of conventional time and phase measuring circuitry. It is observed that the times, frequencies and numbers of cycles mentioned in the above are merely examples. Other combinations are possible and are considered to fall under the scope of the present invention.

*****