Abstrict A gain-controlled amplifier (24) in a sonic flow meter accepts
an output signal from each transducer of a plurality of pairs of
transducers (1a, 1b; 2a, 2b; 3a, 3b; and 4a, 4b) in sequence. A
ramp generator (62) generates a time-varying-gain output, which
increases between the transmission of a pulse by one transducer
and the reception thereof by another. This time-varying-gain signal
is one component of the amplifier's gain-control signal, which includes
as another component an AGC value fetched from a memory (50) and
rendered into analog form by a digital-to-analog converter (42).
The value in the memory (50) is the difference between the time-varying-gain
value and the value that other circuitry (30 36) has determined
to be necessary by monitoring the amplifier output produced during
a previous operation of the same transducer pair. A third component
of the gain-control signal is provided by further circuitry (40
72 76 102), which acts to reduce the gain-controlled value if
it determines that noise is present. It concludes that noise is
present if the receiving transducer produces signal magnitudes that
exceed a predetermined threshold before sound is expected from a
transmitting transducer.
Claims What is claimed as new and desired to be secured by Letters Patent
of the United States is:
1. A flow meter comprising:
(A) a plurality of sonic transducers, at least a plurality of which
are operable as transmitting transducers and at least a plurality
of which are operable as receiving transducers, transducers, each
of the transmitting transducers being operable to transmit sonic
pulses along a different sonic path associated therewith through
a fluid to an associated receiving transducer, each receiving transducer
responding to the reception of a sonic pulse to generate a transducer
output representative thereof;
(B) a driver for alternately driving each transmitting transducer
individually to transmit sonic pulses along the sonic path associated
therewith, each sonic pulse transmission defining the beginning
of a new cycle of operation;
(C) an amplifier, having an output port, a gain-control port adapted
to receive a gain-control signal, and an input port adapted to receive
an amplifier input, for amplifying the amplifier input with a gain
determined by the gain-control signal to generate an amplifier output;
(D) means for selectively coupling the amplifier input port to
receive the transducer output of the receiving transducer associated
with the sonic path associated with the transmitting transducer
currently being driven by the driver;
(E) an AGC circuit responsive to the amplified signal to generate
therefrom an AGC signal representing an AGC value associated with
the path taken by the sonic pulse from which the amplified signal
resulted;
(F) memory means responsive to the AGC signal to store the AGC
value represented thereby separately from each AGC value associated
with another path;
(G) means for (i) retrieving from the memory means, for use during
a given cycle of operation in which the driver drives a given transmitting
transducer, the AGC value represented by the AGC signal during a
previous cycle in which the driver drove the given transmitting
transducer and (ii) generating a retrieved-value signal from the
retrieved AGC value;
(H) a TVG circuit for generating a TVG signal that changes monotonically
during a time interval between the time at which the driver sends
a pulse and the time at which a receiving transducer receives it;
(I) gain-control means responsive to the retrieved-value signal
for generating a gain-control signal dependent thereon and applying
the gain-control signal to the gain-control port of the amplifier
during the given cycle of operation, the gain-control means also
being responsive to the TVG signal so that the gain-control signal
depends on both the TVG signal and retrieved-value signal; and
(J) measurement means responsive to the amplifier output for measuring
the time intervals between transmission of sonic pulses and reception
thereof, for comparing the time intervals for different paths, for
computing therefrom the flow rate of the fluid flowing through the
paths, and for generating an indication of the flow rate.
2. A flow meter as defined in claim 1 wherein the amplitude of
the gain-control signal generated by the gain-control means is proportional
to the sum of values proportional to the amplitudes of the retrieved-value
signal, the TVG signal, and a constant.
3. A sonic flow meter comprising:
(A) a plurality of sonic transducers, at least a plurality of which
are operable as transmitting transducers and at least a plurality
of which are operable as receiving transducers, each of the transmitting
transducers being operable to transmit sonic pulses along a sonic
path associated therewith through a fluid to an associated receiving
transducer, each receiving transducer responding to the reception
of a sonic pulse to generate a transducer output representative
thereof.
(B) a driver for driving a transmitting transducer to transmit
sonic pulses of a predetermined frequency along the sonic path associated
therewith;
(C) an amplifier, connected to receive a transducer output, for
amplifying the transducer output to produce an amplified signal
therefrom;
(D) a first zero-cross detector for monitoring the amplified signal
and generating a first crossing signal when the amplified signal
crosses from a first polarity to an opposite polarity;
(E) a second zero-cross detector for monitoring the amplified signal
and generating a second crossing signal when the amplified signal
crosses back from the opposite polarity to the first polarity;
(F) a window generator responsive to the first and second crossing
signals for establishing a time window that brackets a point in
time delayed by one-half cycle of the predetermined frequency from
a zero crossing detected by the first zero-cross detector and generating
a receive signal in response to the second crossing signal only
if the second crossing signal occurs during the time window; and
(G) processing means for measuring the time between the sound transmission
and the receive signal for a plurality of paths, computing therefrom
the flow rate of the medium, and producing an indication thereof.
4. A flow meter comprising:
(A) a plurality of sonic transducers, at least a plurality of which
are operable as transmitting transducers and at least a plurality
of which are operable as receiving transducers, each of the transmitting
transducers being operable to transmit sonic pulses along a different
sonic path associated therewith through a fluid to an associated
receiving transducer, each receiving transducer responding to the
reception of a sonic pulse to generate a transducer output representative
thereof;
(B) a driver for alternately driving each transmitting transducer
individually to transmit sonic pulses along the sonic path associated
therewith, each sonic pulse transmission defining the beginning
of a new cycle of operation;
(C) an amplifier, having an output port, a gain-control port adapted
to receive a gain-control signal, and an input port adapted to receive
an amplifier input, for amplifying the amplifier input with a gain
determined by the gain-control signal to generate an amplifier output;
(D) means for selectively coupling the amplifier input port to
receive the transducer output of the receiving transducer associated
with the sonic path associated with the transmitting transducer
currently being driven by the driver;
(E) an AGC circuit responsive to the amplified signal to generate
therefrom an AGC signal representing an AGC value associated with
the path taken by the sonic pulse from which the amplified signal
resulted;
(F) memory means responsive to the AGC signal to store the AGC
value represented thereby separately from each AGC value associated
with another path;
(G) means for (i) retrieving from the memory means, for use during
a given cycle of operation in which the driver drives a given transmitting
transducer, the AGC value represented by the AGC signal during a
previous cycle in which the driver drove the given transmitting
transducer and (ii) generating a retrieved-value signal from the
retrieved AGC value;
(H) gain-control means responsive to the retrieved-value signal
for generating a gain-control signal dependent thereon and applying
the gain-control signal to the gain-control port of the amplifier
during the given cycle of operation;
(I) measurement means responsive to the amplifier output for measuring
the time intervals between transmission of sonic pulses and reception
thereof, for comparing the time intervals for different paths, for
computing therefrom the flow rate of the fluid flowing through the
paths, and for generating an indication of the flow rate;
(J) a noise detector, responsive to the amplifier output, for determining
whether the amplifier output, during a predetermined time interval
before a receiving transducer is expected to receive a pulse from
a given path, exceeds a predetermined threshold and for generating
a noise signal if it does; and
(K) means responsive to the noise signal to store in the memory,
in place of the current value associated with the given path, a
value that represents a lower gain.
Description FIELD OF THE INVENTION
This invention relates generally to the field of sonic flow meters
and, more particularly, to a receiver for a sonic flow meter which
can monitor the flow rates in multiple flow paths, each with widely
different and varying characteristics.
BACKGROUND OF THE INVENTION
A sonic flow meter is an apparatus for measuring the flow rate
of liquids through a channel. Typically, the meter includes at least
two transducers, one transducer positioned upstream from the other
transducer. The flow meter determines the flow rate of the fluid
by transmitting pulses of sonic energy between the two transducers
and by measuring the transit time of pulses through the fluid. When
the fluid in the channel is stationary, the transit times of the
pulses are the same, regardless of which direction the pulses travel
through the fluid. However, when the fluid is moving, the motion
of the fluid decreases the transit time of the pulses transmitted
in the downstream direction and it increases the transit time of
pulses transmitted in the upstream direction. The increase or decrease
in the transit time is proportional to the flow rate of the fluid.
Therefore, by measuring the difference between the upstream and
downstream transit times, the flow meter can determine the flow
rate of the fluid.
The flow rate V is related to the difference in transit time (T.sub.u
-T.sub.d) in the following way:
where D is the distance between the transducers and x is the angle
that the sound-propagation path between the transducers forms with
the fluid-flow direction.
A common way of measuring transit time is to count the number of
cycles of a clock signal between the time at which the sonic pulse
is transmitted and the time at which it arrives at the receiving
transducer. Of course, the counts for the upstream and downstream
measurements must start and stop at corresponding points in the
transmitted and received signal pulses if the resulting flow-rate
determination is to be accurate. In addition, the detector on the
receiver end must not mistake noise for the received signal.
A threshold detector, which senses the received signal's crossing
of a preselected threshold, is a conventional apparatus for satisfying
these requirements. Generally, the threshold is set high enough
that the ambient noise does not trigger the detector, but it is
set low enough to detect the received signal. From threshold-crossing
time and the signal amplitude, one can calculate the times at which
the signal crosses the zero axis, and the actual arrival time of
the signal can thus be determined.
When a threshold detector is used, however, the strength of the
received signal affects the determination of arrival time. That
is, a weak signal may appear to arrive later than a strong signal
even though the actual transit times for the signals are identical.
There are two reasons for this, both having to do with the shape
of the received signal. Typically, the received signal is a series
of oscillations that first increase and then decrease in amplitude.
When the signal begins with a positive excursion followed by a negative
excursion, the detector is generally set to determine the point
in time at which the first negative excursion reaches a threshold
level. If the signal is too weak, however, the receiver may miss
the first negative excursion and detect a negative excursion of
a subsequent cycle of the signal. The resulting measurement of transit
time is then seriously in error.
Even if the receiver does not miss the first negative excursion,
there is another reason why the weaker signal may appear to arrive
later than the stronger signal. The slope of the strong signal at
a zero crossing is greater than the slope of the weaker signal at
its corresponding zero crossing. Thus, even though the two signals
arrive at the same time as determined by their zero crossings, the
weaker signal reaches the threshold level somewhat later than the
stronger signal and thus appears to have a longer transit time.
Of course, the error attributable to this factor is significantly
smaller than the error attributable to missing the first negative
excursion completely. Nevertheless, this error can be significant
in some circumstances, such as those in which the distance between
the transmitting and receiving transducers is relatively short.
To avoid the errors just described, flow meters generally employ
gain-controlled amplifiers to produce a received signal of known
amplitude for processing by the threshold detector to generate the
flow-rate readings. The degree of amplification is selected to guarantee
that the first negative excursion will be detected. And by carefully
setting the amplitude of the amplified signal, thereby fixing the
slope of the signal at its zero crossings, the actual transit times
can be determined with a high degree of accuracy.
Changes in the environment in which the transducers are placed,
however, may require recalibrating the flow meter to maintain accurate
and reliable operation. For example, repositioning the transducers
by moving them farther apart reduces the strength of the received
signal. Unless the meter is recalibrated to account for the weaker
signal, the resulting flow measurements may be erroneous. As a rule,
a flow meter is calibrated for one set of transducers, and any recalibration
must be done manually. Recalibration is therefore both time-consuming
and inconvenient, and it requires a skilled technician using specialized
test apparatus. Thus, conventional flow-meter circuitry cannot easily
be used to monitor a number of other transducers in situations in
which each set of transducers may be positioned in a different region
of the channel or in a different channel.
Changes in the ambient temperature also tend to affect the accuracy
of the meter readings, although to a much smaller degree than the
changes mentioned above. The circuitry commonly used to provide
threshold-level detection tends to be sensitive to temperature changes,
so such changes affect the time at which the detector senses the
arrival of the received signal Techniques to compensate for such
effects are available, but they add unwanted expense to the flow
meter.
Another shortcoming of existing flow meters that operate as described
above is their inability to operate reliably in the presence of
noise. One type of noise is a change in the composition of the stream.
If unwanted material, such as bubbles injected by a pump or debris
floating down a river, is introduced into a stream being monitored,
the strength of the received signal may be attenuated or become
erratic. The change in the signal is likely to result in incorrect
flow-rate readings and, if large enough, may lead to a complete
loss of monitoring ability.
A second type of noise is the ambient acoustical signals that are
often present in the channel. In an open channel such as a river,
this noise may be generated by a passing motor boat or a turbine
generator in a nearby power plant. In a closed channel such as a
pipe in a chemical plant, it may be generated by a valve opening
or closing or by a pump. Regardless of the source, acoustical noise
of sufficient magnitude can completely obstruct the operation of
the flow meter by providing a false signal that the meter treats
as the received signal
In summary, conventional sonic flow meters have numerous shortcomings
that seriously limit their usefulness. The limitations are most
acutely felt in hydro-plant applications, in which many flow rates
must be monitored, the environment is constantly changing, and noise
is common.
SUMMARY OF THE INVENTION
The invention is an improved flow-meter receiver circuit that is
capable of monitoring different sets of transducers, each of which
may be located in a different fluid conduit having potentially time-varying
acoustical characteristics. The circuit automatically establishes
and holds calibration for each set of transducers. It provides relatively
high accuracy and noise immunity but low temperature sensitivity.
In addition, the circuit generates a signal that provides information
relating to compositional changes in each of the streams being monitored.
The flow meter includes a transmitter, a receiver, and a plurality
of transducer pairs. The transmitter generates sonic pulses and
sends them through a fluid channel between the two transducers of
a selected transducer pair. Each pair of transducers defines a different
sonic path. The receiver monitors the receiving transducer for received
pulses, and the times of reception of the pulses are then used to
determine the pulses' transit times, i.e., the times required for
them to traverse the channel. To achieve satisfactory accuracy in
the transit-time measurements, a variable-gain amplifier in the
receiver first amplifies the received pulses to produce an amplified
signal that has a predetermined amplitude. Determinations of the
transit times are then based upon the amplified signal.
A TVG (time-varying gain) circuit and an AGC (automatic gain control)
circuit together generate a signal that controls the gain of the
amplifier so that it produces an amplified signal having the desired
amplitude. The TVG circuit generates a first signal, which ramps
up in amplitude during a, period between production of the sonic
pulse and its expected reception. Thus, after each pulse is transmitted,
the TVG circuit signal starts at an initial value and increases
monotonically up to a final value that is determined by the separation
of the two transducers. The AGC circuit generates a second signal
that, when added to the final value of the first signal, causes
the amplitude of the amplified signal to move closer to a reference
signal that represents the desired amplitude of the amplified signal.
In accordance with one aspect of the invention, the AGC circuit
includes a memory for storing an AGC value corresponding to each
path. When the receiver receives a sonic pulse, the AGC circuit
sets the gain of the receiver amplifier in accordance with a previously
stored AGC value. The AGC circuit then compares the amplified signal
with the reference signal, calculates a revised AGC value that will
reduce the difference between the amplified signal and the reference
signal, and stores the revised AGC value in a memory for use during
the next cycle of that path. This operation is repeated for successive
pulses until the amplified signal reaches the desired amplitude
and the AGC values remain stable from pulse to pulse. In this way,
the receiver automatically establishes the proper calibration of
the receiver and generates a measure, i.e., the stored AGC value,
of the acoustical characteristics of the fluid stream.
In accordance with another aspect of the invention, the receiver
includes a noise detector and an attenuation circuit that control
the AGC circuit. The noise detector senses the presence in the amplified
signal of any component that exceeds a preselected threshold level
and occurs before the expected time of arrival of the transmitted
pulse at the receiving transducer. Such components constitute noise,
and if the detector senses noise during a transmit-receive cycle,
the attenuation circuit causes the AGC circuit to reduce the previously
stored AGC value by a predetermined amount. Such reductions are
repeated until the noise is no longer detected. When the noise is
no longer detected, the AGC circuit functions in the manner previously
described to raise the amplitude of the amplified signal back up
to the desired level. Thus, the circuit desensitizes the receiver
to detected noise by reducing the gain of the amplifier until the
noise no longer exceeds the preselected threshold level. If the
amplitude of the received sonic pulse is greater than the amplitude
of the noise, the receiver continues to be capable of generating
transit-time measurements in the presence of the noise.
According to another aspect to the invention, the receiver also
includes what we call a "filter" circuit to improve the
reliability and accuracy of the transit-time measurements. The filter
comprises a first detector for detecting a first zero crossing of
the amplified signal, a second detector for detecting the next zero
crossing of the amplified signal, and a window-generator circuit.
When the first detector detects the first zero crossing, the window-generator
circuit enables a gate for a preselected period of time after the
first zero crossing is detected. Only detection signals from the
second detector which occur while the gate is enabled may be used
to determine the transit time for the sonic pulse. Detection signals
coming from the second detector during any other period of time
are rejected. Since the zero crossings of the received signal provide
a more accurate indication of the arrival time of the signal than
do the non-zero-threshold crossings, the filter circuit enhances
the performance of the receiver.
BRIEF DESCRIPTION OF THE DRAWINGS
This invention is pointed out with particularity in the appended
claims. The above and further advantages of the invention may be
better understood by referring to the following description in conjunction
with the accompanying drawings, in which:
FIG. 1 is a block diagram of a sonic flow meter which embodies
the invention;
FIG. 2 is a block diagram of a portion of the receiver shown in
FIG. 1;
FIG. 3 is a representative example of an amplified signal;
FIG. 4 is a block diagram of the control circuitry within the receiver
shown in FIG. 1;
FIG. 5 is an alternative embodiment of the control circuitry shown
in FIG. 4; and
FIG. 6 is a timing diagram of the signals generated in the embodiment
illustrated in FIG. 5.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
FIG. 1 is a block diagram of a sonic flow meter including four
transducer pairs 1a, 1b; 2a, 2b; 3a, 3b; and 4a, 4b. The flow meter
uses at least one transducer pair in each conduit for which it is
to take flow-rate data, and it may use more than one pair in a given
conduit so as to take measurements in more than one path through
the conduit. A transducer pair is placed in a fluid stream within
the conduit of interest such that one transducer is upstream from
the other transducer. The flow meter then transmits sonic pulses
between the two transducers, first in one direction, for example,
an upstream direction, and then in the other direction, namely,
the downstream direction, to measure an upstream transit time and
a downstream transit time of sound through the moving fluid. The
difference between the upstream and downstream transit times is
used to calculate the flow rate for the fluid stream.
Each transducer of a pair may act as a transmitting transducer
or a receiving transducer depending upon whether the meter is measuring
the upstream or the downstream transit-time at that moment. A switch
7 selects the particular transducer pair which is to provide the
transit-time measurements, and within the pair it selects the transmitting
transducer and the receiving transducer.
The flow meter includes a transmitter 5 which produces signals
for the transmitting transducer 1a, and a receiver 6 which monitors
signals from the receiving transducer 1b. The transmitter 5 drives
the transmitting transducer 1a, causing it to produce a short pulse
of sonic energy that propagates through the fluid toward the receiving
transducer 1b. In the embodiment described herein, the frequency
of the sonic pulse is between 100 kHz and 1 MHz. The receiver 6
monitors the receiving transducer 1b by way of line 8 and it signals
the time of arrival of the transmitted sonic pulse.
The flow meter also includes a counter 10 which measures the transit
times of the transmitted sonic pulses. Transmission of the sonic
pulse resets a counter 10 and causes it to begin counting. When
the receiver 6 detects the arrival of a transmitted pulse, it stops
the counter 10 thereby producing a measure of the transmitted pulse's
transit time. The transit-time information from the counter 10 is
then passed to a processor 14. The receiver also sends the processor
information, to be described below, from which the processor determines
whether to accept the counter output as a reliable indication of
pulse transit time.
Switch 7 then reverses the sending and receiving functions of transducers
1a and 1b, and the foregoing procedure is repeated. After the processor
14 has received both the upstream transit time and the downstream
transit time, it calculates the flow rate for the channel. The processor
typically uses the calculated flow rate for further processing,
and it may also send it to other systems such as a display system
16 which displays it to the flow-meter operator.
A path-selector circuit 18 selects a signal path for a particular
transit-time measurement and causes the switch 7 to make the proper
connections. That is, the selector circuit selects the transducer
pair, e.g., transducers 1a and 1b, that will generate the desired
transit-time information. Within the selected pair, it determines
which transducer is to transmit and which one is to receive.
FIG. 2 is a block diagram of a portion of the receiver circuit
6. As illustrated, the receiver 6 includes an input matching circuit
20 followed by a voltage-controlled variable-gain amplifier 24.
The matching circuit 20 serves at least two purposes. First, it
limits the maximum amplitude of the signal which is permitted to
reach the amplifier 24. This prevents a strong, unwanted signal
such as noise or the drive signal of the transmitter 10 from overloading
the amplifier 24 and thereby temporarily disabling it. Second, it
matches the impedance of the transducer 1b and the amplifier 24
to maximize the signal power that the amplifier 24 receives. The
amplifier 24 amplifies the signal from the matching circuit 20 to
produce an amplified signal on line 22. The receiver 6 uses the
amplified signal 22 to detect the arrival of the transmitted pulse
and to stop the counter 10 shown in FIG. 1.
The amplified signal on line 22 generally has a waveform such as
that illustrated in FIG. 3. The signal 22 consists of a series of
oscillations that first increase in amplitude and then decrease
in amplitude. The period T of each cycle in the signal is, of course,
the reciprocal of the carrier frequency of the sonic pulse. The
transmitter 5 drives the transmitting transducer 1a so as to generate
an amplified signal 22 which, as a rule, has a first peak 22a of
a predictable polarity. For purposes of the following description,
the polarity is positive. In other words, each amplified signal
has a positive first peak 22a followed by a negative second peak,
which is referred to hereinafter as the first negative peak 22b.
In the first embodiment of the invention described herein, the time
at which the pulse is considered to arrive--i.e., the time at which
the counter 10 stops--is that of the occurrence of this first negative
pulse 22b, and the amplifier gain is controlled so as to keep the
amplitude of this peak at a reference level.
Specifically, a gain-control signal on line 26 controls the gain
of the amplifier 24 and thus the magnitude of the oscillations in
the amplified signal on line 22. After receiving each transmitted
pulse, the receiver 6 adjusts the gain-control signal on line 26
for the next received signal until the oscillations of the amplified
signal on line 22 have a desired amplitude. The receiver 6 accomplishes
this by first determining the difference between the amplitude of
the first negative-going peak 22b and a reference level 28 (FIG.
3), which represents the desired amplitude of the first negative-going
peak 22b, and then adjusting the gain of the amplifier 24 to reduce
the difference which will be observed for the next received pulse
on the selected path. As shown in FIG. 2 the receiver circuitry
that performs this function includes an error-voltage generator
30.
The error-voltage generator 30 determines the amplitude of the
first negative-going peak 22b by employing conventional sample-and-hold
techniques. As will be described in more detail below, a sample
signal on line 32 causes the error-voltage generator 30 to begin
sampling the amplified signal 22 as soon as the receiver 6 detects
that the received pulse has started. After approximately one quarter
of the period of the first oscillation, i.e., T/4 has elapsed,
the sample signal 32 causes the generator 30 to hold the amplitude
of the amplified signal 22 existing at that point in time. The generator
30 then compares the sampled level to the reference level 28 (FIG.
3) and generates an error signal 34 proportional to the difference
between the two levels. If the sampled level is greater than the
reference level 28 the error signal 34 is negative, causing the
gain of the amplifier 24 to be decreased for the next received pulse.
If the sampled level is less than the reference level 28 on the
other hand, the error signal 34 is positive, causing the gain of
the amplifier 24 to be increased for the next received pulse.
When the sampled voltage is zero, the error-voltage generator 30
produces an error signal on line 34 having maximum error value.
After each non-zero sample, the error signal 34 of the generator
30 slowly decays toward the maximum error value. The time constant
of the decay is substantially longer than the period of one oscillation
of the sonic pulse but is small enough that the error signal 34
will be approximately equal to the maximum value by the time the
next pulse is transmitted.
A damping circuit 36 whose purpose will be explained presently,
attenuates the error signal 34 before sending it to a first analog
sum circuit 38. The first analog sum circuit 38 has three input
ports 38a, 38b, and 38c. Input port 38a receives the attenuated
error signal from the damping circuit 36 and the other two input
ports 38b and 38c receive output signals from a noise hold and scaler
circuit 40 and a D/A converter 42 respectively, both of which will
be described shortly. The analog sum circuit 38 adds its inputs
to produce a first analog signal on line 39 which is then converted
to a digital AGC value on line 44 by an A/D converter 46. In this
embodiment, the digital AGC value is an eight-bit-wide signal. The
digital AGC value from line 44 propagates by way of lines 48 to
a RAM 50 which, under control of a memory-cycle timing circuit
52 triggered by a signal on line 54 stores that value for use in
a later cycle, in which the memory-cycle timing circuit causes it
to be fetched and to be converted to analog form by the D/A converter
42.
The D/A converter 42 converts the digital AGC value on line 44
back to a second analog signal on line 56. The second analog signal
is fed back to input port 38c of the analog sum circuit 38 where
it is added to the modified error signal on line 37 to generate
the next revised digital AGC value 44 for the selected path. The
second analog signal is also sent to a second analog sum circuit
58 which produces the gain-control signal on line 26.
The second analog sum circuit 58 has three input ports 58a, 58b,
and 58c. Input port 58a receives the second analog signal 56 input
port 58b receives a fixed gain signal 60 which is externally set
to a constant value, and input port 58c receives a TVG output signal
62a from a TVG ramp generator 62. The gain control signal 26 may
be a weighted sum of these three signals.
The TVG generator 62 produces a TVG output signal on line 62a that
ramps up from an initial value to a final value after each transmission
of a sonic pulse. The generator 62 has a start input port 64 and
a stop input port 66. The start input port 64 receives a transmit
signal on line 68 which marks the time at which the transmitter
5 transmits the sonic pulse. The stop input port 66 receives a range-gate
signal on line 70 which establishes the period during which the
TVG output signal 62a ramps up. The range-gate signal on line 70
is initially low, it goes high prior to the transmit signal on line
68 marking the beginning of range gate, and then it returns to
low just before the transmitted pulse reaches the receiver transducer
1b, marking the end of the range gate. In other words, the duration
of range gate is adjusted to correspond to the transit time or,
equivalently, to the distance between the transmitting transducer
1a and the receiving transducer 1b.
Before the TVG ramp generator 62 receives the transmit signal 68
the TVG output signal 62a is at the initial value. Upon receiving
the transmit signal 68 the TVG output signal 62a begins ramping
up until the generator 62 receives the signal marking the end of
range gate. At the end of range gate, the ramping stops, and the
generator 62 holds a final value determined by the duration of the
range gate. The TVG output signal on line 62a is returned to the
initial value before the next transmit signal 68 and the process
is repeated for the next transmit-receive cycle.
The gain-control portion of the circuit operates as follows. Assume
that the stored AGC value for the selected path is less than required
to achieve the desired amplitude of the amplified signal 22. Also
assume that no noise is present, so the output of the noise hold
and scaler circuit 40 is zero. (The noise hold and scaler circuit
40 will be described later.) After a sonic pulse is transmitted
on a selected path, the TVG output signal on line 62a ramps up to
and holds its final value, which is determined by the length of
the range gate. In the meantime, the memory-cycle circuit 52 causes
the RAM 50 to produce on line 48 the digital AGC value stored at
a location identified by a path code on line 71 corresponding to
the selected path. The D/A converter 42 converts the digital AGC
value on line 48 to a corresponding analog value, which propagates
over line 56 to the second analog sum circuit 58. The second analog
sum circuit 58 adds the final value of the TVG output signal on
line 62a, the fixed gain signal on line 60 and the second analog
signal on line 56 to produce the gain-control signal 26 thereby
setting the gain of the amplifier.
It should be understood that the duration of the range gate increases
in proportion to the separation of the two transducers 1a and 1b,
so the final amplitude of the TVG output signal 62a also increases
in proportion to the separation of the two transducers. In this
way, the receiver compensates for the increased signal attenuation
associated with moving the transducers farther apart and thus reduces
the dynamic range required of the circuit's AGC function.
In response to an appropriately timed sample signal on line 32
the error-voltage generator 30 samples the amplified signal on line
22 and produces an error signal on line 34 corresponding to the
difference between the reference level and the amplitude of the
first negative-going peak of the amplified signal on line 22. Since
by assumption the initial AGC value was less than required, the
error voltage generator 30 will produce a positive error signal
on line 34 causing the stored AGC value to be increased. After
the damping circuit 36 attenuates the error signal on line 34 by
a preestablished amount, the first analog sum circuit 38 adds the
modified error signal on line 37 and the second analog signal on
line 56 which corresponds to the stored AGC value, to generate
a first analog signal on line 39.
The memory-cycle timing circuit 52 then causes the RAM 50 to release
the bus 48 and concurrently causes the A/D converter 46 to convert
the first analog signal 39 to the revised digital AGC value and
place it on the bus 48. The RAM 50 then writes this revised digital
AGC value into the memory location which had been occupied by the
previously stored digital AGC value for the selected path.
In short, the receiver 6 examines the magnitude of the amplified
signal 22 corresponding to a transmitted pulse, compares it with
the desired magnitude for the amplified signal, and then revises
the stored AGC value in the RAM 50 so that the amplified signal
for the next transmitted pulse along the selected path moves closer
to the desired magnitude.
The attenuation factor N of the damping circuit 36 is typically
less than one. If N were equal to one, the stored AGC values would
reach the desired value after the first received pulse is examined.
The receiver 6 however, would be more sensitive to noise and prone
to oscillation. Selecting an attenuation factor which is less than
unity reduces the receiver's sensitivity to noise and its susceptibility
to oscillation, although it additionally makes it necessary for
the receiver to examine several received pulses on a selected path
in order to arrive at a final stable AGC value.
With the aid of FIG. 4 the portion of the receiver circuitry that
controls the error-voltage generator 30 the memory-cycle timing
circuit 52 and the counter 10 (shown in FIG. 1) will now be described.
As illustrated in FIG. 4 the receiver also includes a threshold
detector 72 which senses when the receiving transducer lb has received
a sonic pulse. The threshold detector 72 is a differential amplifier.
One of its inputs is the amplified signal on line 22. The other
input is a first threshold voltage 74. A received pulse is deemed
to have begun when the amplified signal first goes more negative
than the first threshold 74 (FIG. 3) and the output of the threshold
detector 72 has thus become high.
The threshold-detector output is one input of a three-input AND
gate 76 which receives the two other inputs. One is an inverted
ABLANK signal on line 78. The ABLANK signal is asserted when it
is expected that the receiving transducer will detect extraneous
signals. For example, when the sonic pulse is transmitted, it is
likely that the receiving transducer will pick up the transmission
signal directly from the transmitter circuitry. To prevent the receiver
circuit from responding to this extraneous signal, the ABLANK signal
is asserted during transmission to mask such signals. The Q output
80a of a flip-flop 80 is the remaining, third input of the AND gate.
The Q output at port 80a is normally high but goes low immediately
after the output of AND gate 76 goes high in response to a received
pulse.
The output of the three-input AND gate 76 is one input of another
AND gate 82. The range-gate signal on line 70 after being inverted
by an inverter 84 is the other input of AND gate 82. Thus, the
VALID output of gate 82 on line 86 goes high to indicate receipt
of the signal only if the amplified signal on line 22 is more negative
than the first threshold after the end of the range gate, i.e.,
after a time at which the pulse might begin to be expected.
The positive-going transition of the VALID signal 86 causes the
output of a one-shot, gate generator 88 to assume its unstable
high state for slightly less than T/4 after which it goes low.
This period is the time it takes for the signal to reach the first
negative peak after the threshold is detected, and the Q output
of the gate generator 88 is the sample signal on line 32 that causes
the error voltage generator 30 shown in FIG. 2 to sample the amplified
signal on line 22.
The up-transition of the sample signal on line 32 also resets flip-flop
80 with the following results. First, the Q output of the flip-flop
80 goes to low, thereby disabling the three-input AND gate 76 and
shielding the gate generator 88 from all subsequent signals coming
from the threshold detector 72. Secondly, the Q complement output
of the flip-flop 80 changes from low to high, thereby producing
a signal on line 81 that stops the counter 10. Prior to the next
transmission, the flip-flop 80 is reset by the TRANSMIT signal 68
so that the circuit is ready to react to the next received signal.
In summary, the circuitry that controls the gate generator 88 works
as follows. The threshold detector 72 responds to any signal appearing
at the output of the amplifier 24 which goes more negative than
the first threshold voltage 74. The three-input AND gate 76 prevents
the output of the threshold detector from reaching the gate generator
88 as long as the ABLANK signal is asserted, thereby shielding the
circuit from known extraneous signals. Similarly, AND gate 82 prevents
the output of the threshold detector 72 from reaching the gate generator
88 during the range gate, thereby shielding the circuit from noise
during that time interval. After the end of range gate, however,
the output of the threshold detector 72 is able to reach the gate
generator 88. Thus, when the detector 72 detects the arrival of
the sonic pulse and its output thereby changes from low to high,
it causes the gate generator 88 to produce the sample signal on
line 32. The rising edge of the sample signal 32 resets flip-flop
80 causing a low-to-high transition in its Q-complement output
so as to stop counter 10 at a count that represents the pulse transition
time. Simultaneously, the Q output assumes a high value so the three-input
AND gate 76 again shields the gate generator 88 from all signals
that come from the threshold detector 72 until the next TRANSMIT
signal on line 68 sets flip-flop 80.
In response to the falling edge of the sample signal on line 32
which marks the end of the sample period, a pulse generator 90 produces
a short-duration pulse, which an OR gate 92 forwards to the memory-cycle
timing circuitry 52 of FIG. 2 to cause the A/D converter 46 to convert
the revised AGC value to digital form and causes the RAM 50 to store
the resultant digital value into the memory location identified
by the path code 71 (FIG. 2).
The other input to OR gate 92 is the OVERFLOW signal, which circuitry
not shown in the drawings generates if a received signal has not
been detected within a predetermined time interval after the sonic
pulse was transmitted. The predetermined interval is selected to
be greater than the expected transit time for the pulse. If the
gain of the amplifier 24 is initially too low, it is possible that
none of the peaks of the amplified signal will be more negative
than the first threshold voltage 74 and the threshold detector
72 will not detect the received pulse. Thus, the threshold detector
72 will not trigger the generation of the sample signal 32. In that
case, the OR gate 92 forwards the OVERFLOW signal over line 54 to
the memory-cycle timing circuit 52 thereby causing a revised AGC
value to be written into the RAM 50. Since the error voltage generator
30 does not receive a sample signal 32 for the last transmitted
pulse, the error signal 34 will be approximately equal to the maximum
error value. This causes the revised AGC value to be greater than
the last stored AGC value, so the gain of the amplifier 24 for the
next transmitted sonic pulse increases.
After enough transmit-receive cycles have been performed on the
selected path, the gain of the amplifier 24 will increase to a level
high enough to generate an amplified signal 22 that the threshold
detector 72 can detect. It is likely, however, that the threshold
detector 72 will first detect one of the larger negative pulses
that occur after the first negative pulse. To assure that the receiver
synchronizes on the first negative pulse, the reference level 28
(FIG. 3) is set to be approximately 16db above the first threshold
voltage 74. Thus, if there is a negative pulse before the one that
the threshold detector 72 is sensing, its amplitude will rise above
the first threshold level as the peak of the detected pulse rises
towards the reference level. In this manner, the receiver finds
and triggers on the first negative-going peak of the amplified signal
22.
As was described in connection with FIG. 2 the receiver 6 operates
in such a manner that the first negative peak gravitates to the
reference level 18. A signal of that magnitude results in optimum
measurement accuracy. Although the amplified signal tends toward
that optimum magnitude, however, changes in conditions can cause
the magnitude to vary, possibly to the extent that measurement accuracy
is unacceptable. To reduce the effects of such changes, the circuitry
of FIG. 4 includes "SQM" (signal-quality measurement)
circuitry, including an SQM detector 104 an SQM flip-flop 108
and an SQM gate 112 which provides the processor 14 (FIG. 1) with
an indication of the likely accuracy of the transit-time measurement
that it receives from the counter 10.
The SQM detector 104 compares the amplified signal 22 with a second
threshold voltage 106. (See also FIG. 3.) The SQM detector 104 is
an operational amplifier which produces a low output when the amplified
signal 22 is more positive than the second threshold voltage and
a high output otherwise. The output of the SQM detector 104 is the
D input of a D-type flip-flop 108. At the expected time of occurrence
of the first negative peak 22b, a positive-going edge at the clock
terminal 108c of flip-flop 108 causes it to store the value then
at its D input port 108a and place this value on its output line
108d, holding it there until reception of a resetting TRANSMIT signal
on line 68. The output of the SQM flip-flop 108 thus indicates whether
the first negative peak was of sufficient magnitude to be reliable.
This output on line 108d is one input of an AND gate 112 whose
other input is the Q-complement output of flip-flop 80 and is thus
high from the reception of the pulse to the end of the cycle. The
INDICATOR output of AND gate 112 is thus an indication of whether
the currently received pulse is "good," as determined
by the magnitude of its first negative peak. If the pulses received
from a given transducer pair for both the forward and the reverse
measurements are good, the processor 14 makes a velocity determination
from those measurements. Otherwise, the processor "ignores"
the counter output, since that output is likely to be inaccurate.
As noted above in reference to FIG. 2 the receiver embodying the
invention also includes the noise hold and scaler circuit 40. When
noise is detected, the scaler circuit 40 imposes a downward adjustment
of the stored AGC value to reduce the receiver's sensitivity to
noise. The scaler circuit 40 monitors a NOISE signal line 96 for
a high state, which indicates that noise has been detected. When
it detects the high state, the scaler circuit 40 sends an INHIBIT
signal on line 98 to the error-voltage generator 30 and sends a
noise-attenuation signal on line 100 to the first analog sum circuit
38. The INHIBIT signal 98 prevents the error-voltage generator 3
from responding to the sample signal 32 and thus allows the error
signal on line 34 to continue its decay toward its maximum error
value. The polarity of the noise-attenuation signal on line 100
on the other hand, is opposite that of the maximum error value.
Its magnitude is preselected so that, when the analog-sum circuit
38 adds it to the maximum error value, the output on line 39 causes
a decrease in the AGC value stored in the RAM 50.
An AND gate 102 generates the NOISE signal t which noise hold and
scaler 40 reacts. This signal is high if the amplified signal on
line 22 exceeds the first threshold 74 during the range gate on
line 70--i.e., during the period before the received pulse is expected.
Specifically, AND gate 102 which is controlled by the range-gate
signal on line 70 monitors the output of the three-input AND gate
76. After a transducer has transmitted a pulse, and while the range
gate is still on, both the three-input AND gate 76 and AND gate
102 are enabled and thus allow any output of the threshold detector
72 to pass to the NOISE signal line 96. Consequently, if the threshold
detector 72 senses noise during the range-gate period, it causes
the NOISE line 96 to go to a high state, which indicates the presence
of noise and thereby causes the noise-hold-and-scaler circuit 40
to reduce the amplifier sensitivity in the manner just described.
FIG. 5 depicts an alternate embodiment of the present invention.
In addition to the circuit elements described in connection with
FIG. 4 and correspondingly numbered in FIG. 5 the circuit of FIG.
5 includes further elements, which increase measurement accuracy.
Unlike the arrangement of FIG. 4 in which the gate-generator one-shot
88 triggers flip-flop 80 and thus stops the counter 10 a fixed
time period after the amplified signal reaches the first threshold
74 that of FIG. 5 employs the additional elements to trigger flip-flop
80 and thus stop counter 10 at the first positive-going zero crossing
of the amplified signal. Unlike the times of occurrence of threshold-crossing
points, upon which the FIG. 4 arrangement bases its termination
of the transit-time measurement, the times of occurrence of zero-crossing
points do not depend upon the amplitude of the oscillations. Thus,
they provide a more reliable indication of the times of pulse arrival
at the receiving transducer.
In the FIG. 5 arrangement, a zero-crossing detector 116 provides
a high output where the amplified signal 22 which it receives at
its inverting terminal 116a, is below the ground level presented
to its non-inverting terminal 116b. Otherwise, the detector output
on line 116c is low. A hold-off-delay generator 118 monitors the
signal on output line 116c. The output of the generator 118 is normally
low. When it receives a low-to-high transition, however, the generator
118 produces on its output line 120 a short pulse delayed from the
transition by slightly less than half the period of the transmitted
frequency; i.e., it produces a pulse shortly before the next positive-going
zero crossing is expected. The delayed pulse passes to an AND gate
122 which is controlled by the VALID signal on line 86. If the
VALID signal is high and thereby indicates that the amplified signal
on line 22 has reached the first threshold voltage, AND gate 122
permits the short pulse to pass to the input port of a window-generator
circuit 124.
The window generator is a monostable multivibrator whose output
is normally low. When the window generator 124 receives the delayed
pulse from the delay generator 118 it raises the signal on its
output line 124a to a high state for a short preset interval and
then returns it to a low state. The duration of the preselected
interval is chosen so that the signal on output line 124a will be
high for an interval that brackets the expected time of occurrence
of the next positive-going zero crossing.
The signal on the output line 124a passes to an AND gate 126 whose
other input is the output of an inverter 128 that inverts the VALID
signal 86 to produce a signal that, as will be explained presently,
goes high upon the positive-going zero crossing of the amplified
signal. If that zero crossing occurs during the window defined by
the output of the window generator 124 therefore, AND gate 126
produces a low-to-high transition in response. This transition on
the output line 127 of AND gate 126 resets flip-flop 80 and the
subsequent high-to-low transition triggers the pulse generator 90.
In contrast to the flip-flop 80 and pulse generator 90 of FIG.
4 therefore, the corresponding elements in FIG. 5 receive the output
of gate 126 whose output represents actual detection of a zero
crossing, rather than the output of the gate generator 88 whose
output represents a prediction of a negative peak. An additional
difference between the two embodiments is that the output of the
three-input AND gate 76 of FIG. 5 is fed back to the non-inverting
input terminal of the threshold detector 72 through a resistor 130.
This has the effect of pulling the first threshold voltage 74 to
zero volt when the output of the three-input AND gate 76 is high.
As will now be explained, this is what causes the tracking edge
of the VALID signal to occur on a zero crossing so that the output
of gate 126 represents an actual zero crossing.
With the zero-cross detector 116 in the circuit, the receiver operates
as illustrated in the signal timing diagrams shown in FIG. 6. Before
the transmitted pulse arrives, the range-gate signal on line 70
will have ended, as FIG. 6b illustrates. Furthermore, the inverted
ABLANK signal on line 78 will no longer be low, and the TRANSMIT
signal on line 68 will have issued. As the first negative pulse
22b of the amplified signal on line 22 occurs, the zero-crossing
detector 116 senses at time T.sub.0 the signal transition from positive
to negative and raises its output on line 116c to a high state,
where it remains until time T.sub.4 when the signal crosses zero
in the positive direction. (See FIG. 6g.) The transition of the
output of the zero-crossing detector 116 to the high state starts
the delay generator 118 which responds by generating a pulse after
a preset time has elapsed, namely, at time T.sub.3 as FIG. 6h illustrates.
At a time T.sub.1 after T.sub.0 the amplified signal on line 22
crosses the first threshold voltage 74 thereby causing the output
of the threshold detector 72 to become high, as FIG. 6cillustrates.
This in turn causes a number of other changes that all take place
around time T.sub.1. First, the feedback through resistor 130 pulls
the first threshold voltage 74 to zero volt so that the threshold
detector 72 will be set up to detect the next zero crossing, as
FIG. 6a illustrates. Second, since the inverted ABLANK signal 78
is not being low at this time, since flip-flop 80 had been set by
the TRANSMIT signal 68 and since the range-gate signal 70 has ended,
the outputs of gates 76 and 82 both go high. That is, the VALID
signal on line 86 goes high, as FIG. 6d indicates, to trigger the
gate generator 88 and thereby start the sample signal on line 32
as FIG. 6 indicates. The VALID signal on line 86 remains in a high
state until the threshold detector 72 detects the next positive-going
zero crossing and returns its output, depicted by FIG. 6c, to a
low state.
The sample signal 32 is asserted for a predetermined duration,
namely, T.sub.2 -T.sub.1 as FIG. 6e indicates. During this period,
the error-voltage generator 30 samples the amplified signal and
then holds the value existing at time T.sub.2 which marks the end
of the predetermined period.
Since the VALID signal on line 86 remains high until T.sub.4 the
pulse from the delay generator 118 occurring at time T.sub.3 passes
through the AND gate 122 to trigger the window generator 124. As
FIG. 6i illustrates, the window generator 124 responds by generating
a high signal on its output line 124a for an interval that brackets
the expected time of occurrence T.sub.4 of the positive-going zero
crossing. When the threshold detector 72 detects a positive-going
zero crossing, the VALID signal on line 86 switches to a low state,
which, because of inverter 128 enables AND gate 126 whose output
on line 127 is depicted in FIG. 6j. If the positive-going zero crossing
occurs while the window-generator output is high, the signal on
the output line 127 of AND gate 126 also changes to a high state,
as FIG. 6jillustrates, and remains in that state until the output
line 124a of the window generator 124 returns to a low state. The
positive transition of the gate-126 output triggers flip-flop 80
which stops the counter 10 and the subsequent negative transition
of the gate-126 output triggers the pulse generator 90 which starts
a memory cycle during which a revised AGC value is stored in the
RAM.
If the first positive-sloped zero crossing of the amplified signal
does not occur when expected--that is, if it occurs outside the
window--then the output of AND gate 126 remains low. Thus, the counter
will not be stopped until the OVERFLOW signal is generated, and
the memory cycle will not occur. In that case, the detected signal
is treated as a bad signal and is essentially ignored.
The flow meter embodying the invention can monitor the flow rates
on more than one set of transducers without manual recalibration
of the receiver for each transducer pair. The AGC circuit in the
receiver automatically establishes the proper calibration for each
set of transducers by determining and then storing in the RAM the
separate AGC value necessary to yield an amplified signal of a desired
amplitude in each path. The time-varying-gain component of the gain-control
signal reduces the dynamic range required of the AGC circuit and
substantially increases the adaptability of the receiver to transducer
pairs having widely differing amounts of signal attenuation. Since
the calibration for each path is set automatically and then stored
as a unique AGC value for the path, the receiver can respond to
changes in the fluid composition without operator intervention.
Moreover, the stored AGC value provides a direct and useful indicator
of the change in the attenuation characteristics of the fluid stream.
Visual display of the stored AGC value indicates to the operator
whether other materials or contaminants have been introduced into
the fluid stream and may also provide a measure of the amount of
material added.
In battery-powered flow meters the stored AGC values offer a significant
advantage over other approaches. Because battery life is limited,
it is desirable to power down the receiver when flow-meter readings
are not being taken, but conventional AGC arrangements require significant
time and energy to re-establish the AGC level and make a new reading
when power is again applied to the receiver. At the cost of only
minimal power for the memory between readings, the present arrangement
maintains the stored AGC values so that they are available when
power is reapplied to the receiver. This eliminates the need to
reestablish the AGC value, and it thereby reduces the time and energy
necessary to make the next reading.
In addition, the noise-control function of the circuit makes it
possible to operate the receiver in environments that may be too
noisy for more-conventional flow meters. The receiver responds to
noise occurring during the range gate by systematically and gradually
reducing the gain of the amplifier until the noise is below the
threshold voltage. By reducing the gain in this manner, the receiver
is more likely to continue being able to detect received signals
in a noisy environment. Although it will not provide meaningful
AGC values during periods of noise, the circuit quickly reestablishes
a stable AGC value soon after the noise has ended.
It is apparent in light of the embodiment of the invention described
above that various alterations, modifications, and improvements
will readily occur to those skilled in the art. Such obvious alterations,
modifications, and improvements, though not expressly described
above, are nonetheless intended to be implied and are within the
spirit and scope of the invention. Accordingly, the foregoing discussion
is intended to be illustrative only, and not limiting; the invention
is limited and defined only by the following claims and equivalents
thereto. |