The invention relates to an arrangement for generating amplitude- modulated, ultrasonic transmiβsion pulses. An arrangement of this type is used preferably for generating transmission pulses with a small frequency spectrum and, in particular, with a low side- lobe level in order to achieve better detection of doppler- shifted echo signals.

Such an arrangement is known in an embodiment comprising a linear amplifier, connected to a high-power, fixed-frequency oscillator and driven with a control voltage of a suitably selected amplitude behaviour to obtain amplitude modulation of the oscillator output voltage. This arrangement has the disadvantage that it becomes complicated and costly if it is required to generate high power, ultrasonic transmiβsion pulses in accordance with a predetermined amplitude behaviour with sufficient accuracy.

It is an object of the present invention, therefore, to provide an arrangement as set forth in the opening paragraph, whereby the above disadvantage is substantially obviated.

According to the invention, the arrangement for generating amplitude-modulated, ultrasonic transmission pulses comprises a combination of: a. a switching circuit controlled with a plurality of switching pulses per pulse interval, which switching circuit includes an inductive load circuit for generating the transmission pulses; b. a unit for generating timing signals determining the switching pulse frequency; c. a memory, of which the memory locations are filled with the informat on, required per transmission pulse interval, about the pulse width of the switching pulses to be generated, which information is derived from the desired amplitude pattern of the transmission pulses to be generated;

OMPI

d. a switching-pulβe generator for producing the switching pulses, uβing the timing βignalβ on the one hand and the pulβe width information in the memory on the other hand; e. an addreββ generator for generating addresses on the supply of the timing signals to read out the memory.

The invention will now be deβoribed with reference to the accompanying five figures of which:

Pig. 1A iβ a timing diagram of a limited number βf pulsed signals of varying width to obtain an amplitude-modulated ultra¬ sonic transmission pulβe, while

Fig. 1B is a fragment of the timing diagram of Pig. 1A;

Fig. 2 is a timing diagram of an ultrasonic transmission pulse, whose amplitude is governed by the pulsed signals shown in Fig. 1 ;

Fig. 3 is a block diagram of an arrangement for- generating ampli¬ tude-modulated ultrasonic transmission pulses according to the invention;

Figs. 4A-G are a number of timing diagrams for illustrating the active state of various units forming part of the switching-pulse generator to obtain switching pulses for the switching circuit; and

Fig. 5 is a block diagram of a combination of arrangements for generating amplitude-modulated ultrasonic transmission pulses according to the indention to obtain a transmission pulse having a defined wavefront.

The timing diagram of Pig. 1A shows a sequence of pulsed signals generated at a given frequency f , while each pair of pulses has the same pulse width with an alternating polarity. Apart from such a pulse width behaviour, the pulse width increases in time to a given value and subsequently decreases. When a sequence of signals of this sort is applied to an inductive load circuit of suitable dimensions, it will produce as signal component the first harmonic of this sequence, the characteristics of which harmonic being illustrated in Fig. 2. This figure shows an amplitude-modulated signal, of which the frequency f

the transmitting frequency, is given by th9 relationship: f - The amplitude of this signal depβndβ on the pulβe width of the pulsed signal in relation therewith and can be calculated by the Fourier method of analysis. For a desired amplitude behaviour of the abovementioned signal component it is thus possible to * calculate the appurtenant pulse widths of the sequence of pulsed signals to be generated.

An arrangement which stores information on calculated values of pulse width and generates a sequence of pulsed signals to obtain an amplitude-modulated, ultrasonic transmission pulse will now be described with reference to Figs. 3 and 4A-G.

Pig. 3 Illustrates an address generator 1, a permanent memory 2, a switching-pulse generator 3 _» a switching circuit 4, and a unit 5 for generating timing signals with a wide application of digital techniques.

The address generator 1 produces addresses at frequency f . These addresses are used to asβign information in memory 2 concerning the pulse width of the switching pulses to be generated in switching pulse generator 3« Since in the case in question high- power switching pulses must be generated in switching cirouit 4 and, in view of space and economy, an accumulator of minimum size must be used, the supply voltage delivered by the accumulator will drop considerably during the generation prooesβ. This will occur to a greater extent as the width of the switching pulses to be generated is greater and, hence, the power from the accumula¬ tor is drawn over a longer period. If now an amplitude-modulated ultrasonic transmission pulse with a symmetrical waveform is iesired and the series of pulse-width values to be produoed in the first and second halves of the generation cycle are mirror images of each other, the decrease in tha supply voltage, especial¬ ly during the second half of the generation cycle of the switching pulse will cause an asymmetric amplitude behaviour of the trans¬ mission pulse. To prevent that the amplitude behaviour of the symmetrically generated transmission pulβe be affeoted by

in the βupply voltage, the pulse width values required in the second half of the transmission pulse generation cycle must be somewhat greater than thoβe arranged in a mirror-image relation¬ ship in the first half of the cycle. It is therefore of great advantage to provide the arrangement for generating amplitude- modulated, ultrasonic transmission pulseβ with a pulse-width error generator 6, which iβ connected to the supply voltage T of the switching circuit 4 and which delivers the information to be gathered about the variation of this voltage to the permanent memory 2. Thiβ information combined with that of address generat 1 gives an insight into the pulse width Information to be selected to obtain a transmission pulse with a symmetrical wave¬ form from switching circuit 4. A feasible embodiment of pulse- width error generator 6 comprises a voltage sensor 7 connected to the βupply voltage T, an A/D converter θ connected to sensor 7 and a. register 9 connected to A/D converter 9. Eaoh time interval in which two consecutive switching pulses are produced at a frequency f , the information obtained with A/D converter 8 is written into register 9 for subsequent-βtorage into memory 2.

Further, with the arrangement of Fig. 3 a choice can be made from several (m) transmitting frequencies f , using the frequency selection information T applied to addresβ generator 1. For this purpose unit is designed to generate timing signals pertaining to frequencies f and f . Memory 2 thereto comprises a main memory 10 and an assembly of submemorieβ 11A-M corresponding with the number of transmitting frequencies f . Addresβ generator 1 selects the desired submemory in accordance with the nature of the frequency selection information T. On the βupply of addresses from address generator 1, the submemory selected delivers the standard pulse width information pertaining to these addresses to the main memory 10. The supply of both the standard pulse widt information and the aforementioned voltage amplitude information to main memory 10 results in the delivery of oorractei pulse widt information concerning the transmission pulse from memory 10 to the switching pulse generator 3»

Since the transmission pulβe iβ generated in accordance with the corrected digital pulβe width information and with the aid of counting circuitβ incorporated in βwitching pulβe generator 3 _» the period T pertaining to transmitting frequency f is divided into a suitable number of time increments T (see Fig. 1B); the s clock pulses Φ 8 (to be generated at frequenoy f8) corresponding with the time increments are supplied by unit 5« The corrected pulse width information from memory 2 is therefore relative to time increments T _β , in which the βwitching pulse will commence, and to time incrementβ T _B , in which this pulse will terminate.

Using a sequence of suitably selected switching pulses generated following this principle, switching circuit 4 is activated with each switching pulse to derive the transmission signal of Pig. 2 from the inductive load 12 in the switching circuit. For this purpose the switching circuit 4 in Fig. 3 is provided with two parallel circuits 13 and 14 connected to the supply voltage T, each of which circuits consisting of a pair of series-connected power transistors 15, 16 and 17, 18 respectively. Each of these power transistors iβ driven by a baβe control element 19 _> 20, 21 and 22 respectively, receiving the required control signals from the switching pulse generator - Further, the center parts of circuits 13 and 14 are bridged by inductive load 12, which comprises a filter circuit 23 with an ultrasonic transducer at the secondary side. Two transistors 15 and 18 or 17 and 16, located in different circuits 13 and 14, and active at different voltage levels, must in turn be brought into the conducting state by means of switching pulses, whereas the other two transistors 17, 6 and 15, 18 must be cut off. In this way a voltage as shown in Fig. 1A is set up across the inductive load 12 of switching circuit 4«

Such switching of the power transistors 15- 18 however requires that a power transistor 16 or 18 operating at the lower voltage level is cut off at least during the period in which power transis- tor 15 or 17 operating at the higher voltage level and in the same parallel circuit, is conducting to prevent a short circuit

occurring in switching circuit 4. However, it iβ difficult to realise a switching program in which a transistor 1 or 17 operating at the higher voltage level is switched to the conductin state and the adjacent transistor 16 or 18 operating at the lower voltage level to the cut-off βtate at the same time, because the switching of transistor 16 or 18 to the out-off βtate iβ slower than the switching of transistor 16 or 18 to the conducting βtate. Hence, to βwitch a power transistor 15 or 17 operating at the higher voltage level to the conducting state, transistor 16 or 17 operating at the lower voltage level should already be in the cut-off state. Further, when power transistor 15 or 17 operating at the higher voltage level is cut off, an induction voltage will be built up acroββ the primary winding. If no measures are taken, this voltage will cause damage to transistor 16 or 18 connected to tranβiβtor 15 or 17 to be cut off. To prevent a negative induction voltage across the cut-off transistor 16 or 18, two measureβ have to be taken: the first measure consists in maintaini the cut-off state of transistor 16 or 18, operating at the lower voltage level, for some time after the adjacent transistor 15 or 17 operating at the higher voltage level has been cut off; the second measure consists in inserting a diode, normally in the cut¬ off state, parallel to transistor 16 or 18, operating at the lower voltage level and connected to transistor 15 or 17 to be cut off. With the use of pulsed switching signals, having a pulβe width that approaches half of the period, such a diode will also bridge transistor 15 or 17 operating at the higher voltage level.

The switching of two interconnected transistors 15 and 16 or 17 and 18 therefore implies that the switching period in which transistor 16 or 18, operating at the lower voltage level, must be in the cut-off state, commences earlier but finishes later than the switching period in which the adjaoent transistor 15 or 17 must be in the conducting state. To obtain suitable switching signals for this purpose the memory 2 also comprises a reference pulse circuit 25 connected to the main memory 10, while the switching pulse generator 3 is provided with a switoh 26 connected to circuit 25 and a first and a second pulse shaping network 27

OMP

/λr-m. ^W I ^I P ^P

and 28 connected to switch 26. As already stated, the main memory

10 supplies the digital standard pulβe width information. With this information the reference pulse circuit 25 generates a

"reference" pulse I , situated symmetrical within the pertaining time interval determined by two timing signals T . A feasible embodiment of such a reference pulse circuit 25 is obtained with a subtracting unit 29, which stores the half value of the standard pulse width information at the start of the above time interval, and with an adding unit 30, which stores the full value of the standard pulse width information. The period of operation of adding unit 30 follows that of subtracting unit 29. Both units

29 and 30 are driven by Increment pulses Φs from the timing unit 5«

Switch 26 controlled with the timing signals Tw is used to activate the first and the second pulse shaping networks 27 and 28 alternately, such that in one time interval the control signals to be generated are received and processed by power transistors 15 and 16 and in the following time interval by the other two power transistors 17 and 18. Since the two networks 27 and 28 are similar, only the first pulse shaping network 27 will be described, making refe>rence to Figs. 4A-G.

Pig. 4A illustrates a timing diagram of a reference pulse I with a pulse width p specified in the standard pulse width information, as generated by the reference pulse generator 15 within the time intervals determined by the timing signals T . and •£ -. The reference pulse I is used to obtain a control pulse for the power transistor 15, operating at the higher voltage level, and a control pulse for the adjacent power transistor 16, operating at the lower voltage level, where the latter control pulse is initiated a certain period τ earlier, and may continue for a corresponding period τ longer than the former control pulse. To obtain such control pulses, the first network 27 comprises a first starting element 31 _> second starting element 32, a first counter 33 _» a second counter 34, a register 35 and a logic circuit 36.

On receiving the leading edge of reference pulse I ( see Fig. 4A) ,

OMPI WIPO

the firβt starting element 31 will deliver a starting βignal to the firβt counter 33 _» in which a given digital number repreβenti a certain period τ from register 35 is stored (βee Fig. 4B). This leading edge iβ alβo determinative for the start of the generation of the control βignal for the other transistor 16

(βee Fig. 4D). AS soon as counter 33 bas decreased to zero, this counter stops and delivers a βwitching βignal to the logic circuit 3 (βee Fig. 4E). Thiβ moment iβ determinative for the leading edge of the control pulse of Fig. 4E, which activates transistor 15. On receiving the trailing edge of the same refere pulse (see Fig. 4A), the βecond βtarting element 3 will deliver a βtart βignal to the βeoond counter 34 (see Fig. 4C) _ι which als stored the number representing the period τ from register 35» At the instant counter 34 has decreased to zero, this counter stops and delivers a βwitching βignal to the first starting element 3 (βee Fig. 4B). Starting element 3 reactivates counte in which again the number representing the time interval 1 from register 35 is stored. The instant at which the first counter 33 activated for the second time is also the instant at which the control signal for transiβtor 15, operating at the higher voltag level, iβ to be terminated (βee Fig. 4E). The moment when the first counter 33 bas again decreased to zero is determinative fo terminating the generation of the control signal for the other transistor 16 (see Fig. 4D)« To obtain the active state of power transistor 15 according to Fig. 4E, the logic circuit 3 compris two separate outputs connected to the base control element 19, which consists of an inductive transmission circuit (not shown i the figure) with a double primary winding, energised separately by the signals generated at the separate outputs of the logic circuit 36. On the leading edge of the signal generated at one output the power transistor 15 will start conducting through one of the primary circuits, and on the leading edge of the signal generated at the other output this transistor will be cut off through the other primary circuit. The arrangement described abo enables to generate an ultrasonic transmiβsion signal having an accurately defined amplitude modulation pattern.

OM WI

Such an arrangement is ideally suited to generate an ultrasonic transmiβsion pulse with a negligibly low side-lobe level, using pre-established standard pulβe width information, which is later adaptable with the pulse-width error generator 6: this creates better possibilities to determine a doppler frequency in an echo signal from an object moving in water. For an example of a trans¬ mission pulse of such characteristics, reference is made to a pulse with an amplitude behaviour characterised by the socalled Hamming function: f(t) - A I + B sin ² (ct + d)] where A, B, c and d are βuitably selected constants.

Furthermore, the above described method of generation of ultrasonic transmission pulses offers a suitable possibility to generate such pulses with a directed plane wavefront in accordance with the principle of interference. This will be explained with reference to Fig. 5.

The arrangement shown in this figure comprises a number of power amplifiers 37A-N, to which a corresponding number of transducers 38A-N are connected separately; these transducers are arranged in a circle. Each of the power amplifiers 37A-N consists of a switching pulse generator 3 and a switching circuit 4 _» as described with reference to Fig. 3-.

This arrangement further comprises a memory 2 of which the operation has already been described with reference to Fig. 3-

To transmit a plane wavefront in a desired direction, it is not necessary to activate all transducers; it is sufficient to use a limited number of transducers which are oriented substantially in the direction of propagation of the plane wave. This will be illustrated below in an example in which the number of transducers is limited to six. To generate a plane wave at an angle β with respect to a fixed reference line R (see Fig. 5), the six trans¬ ducers 38B-G are oriented in a direction which corresponds nearest

--g fREX OMPI

with the direction of propagation of the plane wavefront. Tranβducβrβ 38B-G have an identical program for the generation of βwitching pulses, aβ dββoribβd with reference to Figβ. 1A-B, 2 and 4A-G, but theβe six βwitching programs are not completed in phase. Only the switching programs for transduoerβ 38B and 38G are carried out simultaneously, like the βwitching programs for transduoβrβ 38C and 58F and thoβe for transducers 38D and 38E. On the other hand, the switching program for transducers 380 and 38F are delayed with respect to that for transducers 38B and 38G, while the switching program for 38D and 38E is again delayed with respect to that for transducers 38C and 38F. The delay times between the various programs should be selected and carried out accurately, as this will be to the benefit of the quality of the plane wave. The required delays can be obtained with the aid of digital delay lines 39a-f. These delay lines contain the word concerning the reference pulse for the period T (see Fig. 1B). Therefore, a first number of memory cells of the delay line will be filled for example with the binary number 0, then with the binary number 1 for a period corresponding with the pulse width p, and again with the binary number 0 for the remaining period T . Since delay lines 39a-f are identical to each other, only delay line 39a will be described.

Delay line 39a comprises a shift register 40, of which the word length is determined by the number of time increments T 5 during a period T . Shift register 40 contains the reference pulse in binary form. Delay line 39a further comprises a write pulse genera tor 41 i a register 42 and a read pulse generator 43- The write pulβe generator 41 generates a write pulse, which ensures that, on the application to shift register 40, the binary reference puls required for a following period T iβ transferred from memory 2 into the shift register 40. This write pulse is also applied to register 42, which in response delivers to the read pulse genera¬ tor 43 a digital number representing the required delay. After count-down of this digital number in generator 43 _» this generator delivers a read pulse to the shift register 40 to shift the binary information on the pulse shape out of shift register 40.

To get the information of the six delay lines 39a-f into the appropriate six power amplifiers, 37B-G in the case in question, a selection circuit 44 is incorporated between the six delay lines 39a-f and the power amplifiers 37A-H. With the given angle β the selection circuit 44 connects the βix power amplifiers 37B-G, separately to delay lines 39a-f to transmit the plane wavefront at an angle β with respect to the reference line H.

OMPI /., WIPO ,Λ.