US3440631A - Method and system for clearing repetitive voltage waveforms of statistical interferences - Google Patents

Description

A ril 22, 1969 A. BODMER 3,440,631

METHOD AND SYSTEM FOR CLEARING REPETITIVE VOLTAGE WAVEF ORMS OF STATISTICAL INTERFERENCES V Filed Sept. 29. 1965 Sheet of 5 I3 '16 [use 25 ompurlr1 haw R ss? @2 1 I s h i Z I f M T \L i 1' 5 I L F I 4+8): I i shot 7 I 28 Ni ii a v ms Q Shape onesho jmw 3/ Fig. 2

400m From tape 1 fl H LSmsoneshbt L reset reset I E] ..///Z .L;, W Mm, V 5355125011 fiZAY-fflml Triqqer 135/773 83m: I h

Delay W One shot H H I] [I to Mp2 H H H H H 4 run Oscillator 1 1 mm 1 n n H n n Inventor: ARTHUR BODMER A TTORNE Y April '22, 1969 A. BODMER 3 5 METHOD AND SYSTEM FOR CLEARING REPETITIVB VOLTAGE WAVEFORMS OF STATISTICAL INTERFERENCES Filed Sept. 29, 1965 Sheet ,3 of 5 b a? 01 .58 171.5 g; Q 5% Fig. 3

lnven tor ARTHUR BODMER A T TORNE Y From tap

April 22, 1969 A. BODMER 3,440,631

METHOD AND SYSTEM FOR CLEARING REPETI'I IVE VOLTAGE WAVEFORMS OF STATISTICAL INTERFERENCES Filed Sept. 29, 1965 Sheet 3 are Fig.4b Fig.4c

y- 9 Fig. 4d

Fig. 42 Fig. 4h

4 Fig. 4/

Inventor:

ARTHUR BODMER #72224 A TTORNE Y April 22, 1969 A. BODMER 3,440,631

METHOD AND SYSTEM FOR CLEARING REPETITIVE VOLTAGE WAVEFORMS OF STATISTICAL INTERFERENCES Filed Sept. 29, 1965 Sheet 4 of 5 current Rh 5 5 currem [6 ['5 513$ [:1 87m reset i 48 43 Stm't l & 44

9! Sto Start j g Side Fr oTn band SPech-m meter 45 M Timer i i 250sec I 47 Start Stop Tape deck I] Fig. 5

Inven torf ARTHUR BODMER AZ Q.

A TTORNEY 1 Filed Sept. 29, 1965 April 22, 1969 A. BODMER 3,440,631

METHOD AND SYSTEM FUR CLEARING HEPETITIVE VOLTAGE WAVEFQRMS 0F STATISTICAL INTERFERENCES Sheet 5 o! 5 Inventor.- ARTHUR BODMER A TTORNE Y United States Patent 3,440,631 METHOD AND SYSTEM FOR CLEARING REPETI- TIVE VOLTAGE WAVEFORMS 0F STATISTICAL INTERFERENCES Arthur Bodmer, Windisch, Switzerland, assignor to Varian International A-G, Zug, Switzerland Filed Sept. 29, 1965, Ser. No. 491,190 Claims priority, application Germany, Sept. 30, 1964,

Int. (:1. G11b5/02, /44 U.S. c1. s40 174.1 13 Claims The invention relates to a method to clear repetitive voltage waveforms, which show the dependence of some quantity, represented as a voltage, of another quantity, of statistical interferences (noise).

During the transmission of information of any kind statistical interference comes in, which in analogy to its appearance during the transmission of speech in information theory is known as noise. This noise cannot be attenuated below a certain limit, which invariably is defined by the setup and the working conditions.

During the transmission of information it is therefore necessary to make the ratio of the quantity representing the information to the noise level as high as possible. Normally this is done by plotting the voltage waveform representing the information against time. This voltage is amplified, so that the voltage representing the information is sufiiciently high above the noise level.

In a great variety of practical cases, mainly in modern measurement techniques, such as e.g., the measurement of NMR-spectra or the measurement of voltages from living cells, the information obtained, that is the voltages produced by the probe, is already masked by such great an amount of noise that there is practically no observable difference between the actual signal obtained and pure noise. Subsequent amplification would naturally amplify the noise by the same amount, so amplification would not improve the situation.

Because of the statistical nature of noise, the signal-tonoise-ratio (S/N-ratio) can be improved by several irepetitions of the signal. The signals, or the amplitudes of the voltage waveforms, add during repetitive occurrence, whereas the part of the sum due to nolse decreases. By repciitive generation the S/N-ratio increases with the square root of the number of repetitions, so that by the proper amount of repetitions a sufficiently noise-free signal can be produced Using this principle, telegraphic code pulses have been transmitted repetitively and the pulses Were added photographically. Noise produced only little blackening, while pulses produced strong blackening of the photosensitive emulsion, so that recognition of the pulses was obtained. Due to the nonlinear characteristic and because results obtained with photosensitive emulsions are only badly reproducible, this method is only suitable for signals containing information in the form of binary pulses.

It is known to clear continuous voltage waveforms by averaging short sections of the voltage waveform and to transmit these short sections (amplitudes) repeatedly. These amplitudes are stored on an endless steelwire in such a manner, that average values of the same short sections of the voltage waveforms during successive transmissions (corresponding following amplitudes) are superposed on those already stored. Because of the nonlinear characteristic of the steelwire used, this method also has considerable disadvantages, so that it is not suitable for measurement purposes, even if the steelwire would be replaced by modern magnetic tape.

A method has been developed particularly for the above mentioned measurement purposes, for clearing repetitive voltage waveforms, which show the dependence of one quantity, represented as a voltage, of another quantity, of statistical interferences (noise), in which method the average values of short sections of the voltage waveforms (amplitudes) are stored during one run and the average values of the same short sections during suc cessive runs (corresponding following amplitudes) are added to those already stored, and the sums of the amplitudes arrived at after several successive runs are fed to a recording instrument, and in which method the amplitudes are converted into digits, or pulses representing digits, respectively, which digits or pulses, respectively, are stored added to those already stored. The above described method can satisfy the requirements normally asked for, however for converting the amplitudes into digits, storing and addition as well as reconversion and feeding into the registration units a very complex technical setup is necessary, so that this method may beused only in those special cases, where a costly technical setup is admissible.

Therefore, it is the principal object of the invention to provide a method, which requires only a comparatively small apparatus, so that the noise-attenuation achieved may also be utilized in those cases, where only moderate costs are acceptable.

Normally, the conversion of amplitudes into pulse-pairs is done by comparing the amplitude in question with a linear ramp voltage, which starts with the first or starting pulse of the two pulses of a pulse-pair. If this linear ramp voltage starts at a point corresponding to an amplitude of zero, it is not possible to consider negative amplitudes; negative amplitudes, however, quite often occur with weak signals masked by a high noise level.

Accordingly it is another object of the invention to correctly record and add, respectively, such amplitudes having different prefixes,

It is another object of the invention to prevent distortion or interference of the readout of stored pulse-pairs by the addition of the corresponding following pulse-pairs or interference of new recordings by already stored pulsepairs, respectively.

It is still another object of the invention to assure, even with interference present, distinguishing between the leading and the trailing pulse of each pulse-pair.

It is a further object of the invention to provide an apparatus for carrying out the method of the invention, which apparatus is distinguished over the apparatus for carrying out the above mentioned prior method in a much simpler arrangement.

Still a further object of the invention is the determination of the time intervals for sampling an amplitude, because it must be assured that on each run the same amplitude is sampled.

According to the principal feature of the invention the digital method mentioned above is taken as a base and in aiteration of this method, the amplitudes are'converted into pulse-pairs having amplitude-proportional spacing, as known per se, and the pulse-pairs are recorded, and, in the case of corresponding following amplitudes the spacing of the recorded pulse-pairs is increased by the spacing of the new pulse-pairs, respectively.

According to another feature of the invention the pulses at each run are delayed by a constant spacing with regard to the preceding record, and the amplitudes and/or the linear ramp voltage are biased so that the linear ramp voltage reaches zero amplitude value after a period corresponding to said constant spacing.

According to a further feature of the invention, pulsepairs at each run are recorded into another one of a plurality of channels of a recording carrier.

According to another feature of the invention the first pulse of a pulse-pair representing an amplitude or a sum of amplitudes, respectively, is a double pulse of constant spacing.

A further feature of the invention is the same as the preceding feature and the double pulse is utilized to start and control the amplitude sampling sequence in such a way, that the first of the two pulses of the double pulse triggers a monostable multivibrator (One-shot), the time constant of which is longer than the spacing of the double pulse, the second pulse of the double pulse, together with the pulse from the One-shot sets a bistable multivibrator (Flip-Flop) into a predetermined state, or keeps it in that state, respectively, while the trailing edge of the pulse coming from the One-shot sets the Flip-Flop into the opposite stable state. The double pulse thus prepares the circuit to receive the second pulse of a pulse-pair, marking the end of an amplitude, so that upon occurrence of the second pulse the conversion of the amplitudes into a pulse-pair may start, the spacing of which is then added to that of the corresponding pulse-pair previously recorded. In the case described before, that the conversion takes place by comparing the amplitudes with a linear ramp voltage, the above mentioned Flip-Flop upon being reset into the predetermined state starts the linear ramp voltage.

According to the apparatus aspect of the invention, an apparatus for clearing repetitive voltage waveforms of statistical interferences is provided with an amplitude source as well as a registration unit, like the apparatus for carrying out the prior method.

Instead of using an expensive digital computer as in the prior apparatus according to another feature of the invention there is just used a converter for converting amplitudes into pulse-pairs, a storage unit for the pulsepairs and a converter to rcconvert the pulse-pairs into amplitudes.

Conveniently, a converter for conversion of the amplitudes into pulse-pairs is a comparison circuit, which releases a trigger circuit when the amplitude is equal with a linear ramp voltage, which trigger circuit in turn delivers the second pulse or pulse-pair to the memory.

A suitable memory is an endless magnetic tape, which may have several tracks. In a practical embodiment of the invention a simple tape recorder for amateur purposes, after some minor modifications, has proved to be suitable.

According to a further feature of the invention, there is provided an oscillator to define at the initial run the start of each amplitude sampling interval, and supplies the corresponding first pulse of a pulse-pair to the memory. The pulses thus obtained are read out in each successive run and serve to mark the amplitude to be sampled.

It has been mentioned already that the pulses in each new run are displaced with a constant spacing relative to the preceding record, when the amplitudes are converted into pulse-pairs by comparison of the amplitude with a linear ramp voltage, and negative going amplitudes are to be expected. This displacement preferably is achieved by means of a delay circuit connected to the input of the memory. But, because the second pulse of the pulse-pair thus generated has been displaced with the constant spacing by means of the bias of the amplitudes and/or the comparison voltage, this second pulse must not be delayed again, thus, according to a further feature of the invention, said delay circuit is bypassed by the connection between the trigger circuit and the memory.

According to a further feature of the invention for generating the double pulse marking the start of a pulsepair a One-shot is connected between the memory and the delay circuit to generate a double pulse. Because the first pulse of a pulse-pair goes through the delay circuit, and the second pulse of a pulse-pair is bypassed around the delay circuit, it is automatically assured that the first pulse is a double pulse and the second pulse of the same pulse-pair is a single pulse, respectively.

According to one embodiment of the system of the invention the double pulse is utilized for marking the start of an amplitude sampling interval in that a One-shot is connected to the memory output, the time constant of which One-shot is longer than the spacing of the double pulse, one output of which One-shot is connected to both inputs of a Flip-Flop, and the other output of which Oneshot leads to one input of an AND-gate, the other input of which is connected to the memory output, while its output leads to the set-input of the Flip-Flop which in the set-state starts a linear saw-tooth generator, and that the saw-tooth generator is connected to one side of the comparison circuit.

A further understanding of the invention may be gained from the following specification and by reference to the accompanying drawing, in which show:

FIG. 1 a schematic block-diagram of a system according to the invention;

FIG. 2 various pulse trains successively appearing in the system;

FIG. 3 a circuit diagram of the essential electrical part of a system of the invention;

FIGS. 4a-z' a non-distorted input signal-waveform, a noise voltage waveform, a combined voltage waveform representing a combination of the input signal and the noise voltage waveform, and the information which may be obtained therefrom after several runs, respectively;

FIG. 5 a schematic block-diagram of the application of a system of the invention to a nuclear resonance spectrometer; and

FIG. 6 a nuclear resonance spectrum obtained directly from a nuclear magnetic resonance spectrometer and there above the same spectrum after runs through the system represented in FIG. 5.

Referring to FIG. 1 the system of the invention represented as a block-diagram comprises essentially a double track-tape recorder 11 serving as a memory and at the output of which there is connected a pulse former 12 in such a manner that it selectively receives pulses from track I or II; a One-shot 13 having a time constant of 2.5 milliseconds connected parallel to one input of an AND-gate 14 at the output of the pulse former and the output 15 of which is connected to the other input of the AND-gate 14; a Flip-Flop 16, one input 17 of which is connected to the second output of the One-shot l3 and the second input 18 of which is connected to the output of the AND-gate 14; a reference voltage generator 19 at the output of said Flip-Flop 16; a comparison circuit 20 receiving from one side the comparison voltage from the generator 19 and on the other side the signal voltage from input 22 after amplification in an amplifier 21; a trigger 23 connected to the comparison circuit 20 producing a pulse at its output 24 as soon as the comparison voltage from generator 19 and the amplified signal voltage from amplifier 21 are equal; a One-shot 25 having a variable time constant between 4 and 8 ms.; a One-shot 26 having a predetermined time constant of 1.5 ms., and a pulse former 27, being connected to both the output 24 of the trigger 23 and both outputs of the last One-shot 26 and delivers the pulses coming from both devices in the proper form to the tape-deck 11, always to that track which at present is not read out at the output. Further there is provided an oscillator 28 having a period of 400 ms., always coupled to the input of the One-shot 25 and to the output 29 of the Flip-Flop 16, and which in a suitable position of the switch 30 may be connected to the input of the last One-shot 26; the oscillator 28 may be started by opening a switch 31. The input of the output pulse former 27 may be connected with the penultimate One-shot 25 instead with the trigger 23 by means of a switch 32. Finally, in parallel to the comparison voltage generator 19 there is connected to the output of the Flip- Flop 16 an output unit 33, which actually is a converter for converting the pulse-pairs, read out from the tapedeck 11 and formed in the intermediate stages, into amplitudes, which, as indicated, may be fed to a registration unit.

For operation of the system according to FIG. 1 reference is made to FIG. 2. Supposing that both tracks of the endless tape on the tape-deck 11 are empty, to prepare the system for operation, control pulses are to be written into this tape. Principally, it is possible to record the first run already together with the control pulses, for a clearer understanding of the operation this possibility shall not be considered. For this purpose the switches 30, 31 and 32 are switched to another position than shown in FIG. 1, so that the oscillator 28 is started, the input of the last One-shot 26 is connected to the oscillator 28, and the input of the output pulse former 27 is connected to the output of the penultimate One-shot 25. The oscillator 28 generates one pulse each after 400 ms., as shown in FIG. 2, line 9, which pulse triggers the One-shot 25 and the One-shot 26. As both outputs of the One-shot 26 are connected to the pulse former 27, a double pulse having a spacing of 1.5 ms. immediately upon occurrence of one pulse from the oscillator 28 is fed to the running tapedeck 11 and is there recorded; 4-8 ms. later, according to the adjustment of One-shot 25, there comes a further pulse via the pulse former 27 to the tape-deck 11, as shown in line of FIG. 2.

When the entire tape is marked with equally about 400 ms. spaced pulse-pairs, the signal, that is the amplitudes to be recorded, may be stored. For this purpose the pulses stored in the manner just described are applied to the pulse former 12, by connecting it with the necessary output of the tape-deck 11; the input of the tape-deck 11 correspondingly is switched over, for example according to FIG. 1. Also the switches 30, 31 and 32 are switched into the position shown in FIG. 1. As soon as the first pulse of a double pulse comes from the tape-deck 11 (see FIG. 2, line 1) the One-shot 13 (FIG. 2, line 2) is triggered. Thus voltage is applied to one input of the AND- gate 14, and as soon as the second pulse of the double pulse comes from the tape, AND-gate 14 applies a pulse to the input of the Flip-Flop 16. By this pulse the Flip- Flop 16 in each case, regardless in which state it was before, switched into a predetermined state, which shall be termed the reset state in the following specification. In the embodiment shown the pulse of One-shot 13 terminates one millisecond later, and by this resetting of the One-shot 13 the Flip-Flop is triggered into its set-state (see FIG. 2, line 3).

When Flip-flop 16 goes into its set state a pulse from one output 29 goes to the One-shot 25 and triggers same (FIG. 2 line 6). For the purposes of the specification it shall be assumed that the One-shot 25 is adjusted to a time constant of 6 ms. The One-shot 25 thus returns to its original state after further 6 ms., and at this time triggers the last One-shot 26, that is 2.5 ms.+6 ms.=8.5 ms. after the first pulse has come from the tape-deck 11. The Oneshot 26 then generates a double pulse, as described already above, which is applied to the empty track of the tape-deck 11, in the embodiment shown for example track II.

In the first run of the amplitudes to be recorded the second pulse of the pulse-pair from tape-deck 11 appears 6 ms. after the first pulse, on successive runs this pulse normally appears later. For convenience the second pulse of the pulse-pair from the tape is shown later in FIG. 2. As soon as this pulse appears the One-shot 13 is triggered, and 2.5 ms. later it returns intoits normal position, and therewith triggers the Flip-Flop 16 also in its reset state. When changing into this state the Flip-Flop 16 triggers the comparison voltage generator 19, so that this applies a linear ramp voltage to the comparison circuit 20 (see FIG. 2, line 4). The comparison circuit 20 and the amplifier 21 are adapted with respect to each other that 8.5 ms. later the amplitude value zero is reached, and a positive going amplitude value correspondingly later, as shown in FIG. 2, line 4. Negative going amplitudes of course are reached earlier.

As soon as the amplitude or the signal from amplifier 21 and the comparison voltage are equal, the trigger 23 is triggered and without delay applies a pulse via switch 32 and pulse former 27 to the tape-deck 11, where this pulse is recorded in track II (see FIG. 2, line 8).

This is repeated so often, that sufficiently signal amplitudes have added to distinguish over the influences of the noise voltage which tend to cancel out. As soon as this is achieved the output unit 33 and the registration device connected thereto are started, and the recorded pulsepairs from tape-deck 11 are applied via pulse former 12, One-shot 13 and Flip-Flop 16 to the converter 33, where they are reconverted into amplitudes which are registered in the registration device.

Details of the system of FIG. 1 are shown in FIG. 3. As shown in FIG. 3 the input pulse former 12 comprises five transistors in the grounded emitter configuration, a polarizing diode in the input and a clipper-Zener-diode, the One-shot 13 comprises two transistors; the AND-gate 14 is a simple diode gate, the Flip-Flop 16 comprises two transistors, the collector of the second transistor of Oneshot 13 being connected to the base of each of said two transistors of the Flip-Flop 16 via a pair of diodes, and the output of the AND-gate 14 is only connected to the base of a right hand transistor; the collector of the right hand transistor of the Flip-Flop 16 is connected to the input of the comparison voltage generator 19 consisting of three transistors and one Zener-diode, for producing a linear ramp comparison voltage, which is amplified in the emitter follower of the comparison circuit and applied to the base of a transistor, while the signal voltage coming from the signal input 22 via the compensation amplifier 21 is applied to the base of an equal transistor lying parallel to said first transistor.

Wire 24 and switch 32 connect the output of the comparison circuit 20 to the one stage trigger 23, which is connected with the three stage output pulse former 27 via a diode. The input of the output former 27 by means of a further decoupling diode is connected with both collectors of both transistors of the One-shot 26 having 1.5 ms. time constant, a capacitor being provided in each connection, so that just the flange of a pulse produced from the One-shot 26 after triggering thereof comes to the pulse former 27. A further diode and a switch 30 connect the input of the One-shot 26 with the collector of the right hand transistor of the One-shot 25 having two transistors and having a variable resistor 34 for varying the time constant of the One-shot 25 between 4 and 8 ms. At the input of the right hand transistor of the One-shot 25, via a decoupling diode, the collector of the left hand transistor of the Flip-Flop 16 is connected; via a further decoupling diode at the same input in parallel connection there is connected the output of the oscillator 28, a blocking oscillator having a double base diode, the capacity of which normally is shorted by the switch 31. The time constant of the oscillator by means of the variable resistor shown may be adjusted to the desired value. The output of the oscillator 28 further may be directly connected to the input of the One-shot 26 by repositioning of the switch 30, the input of the One-shot 25 being still connected with the output of the oscillator 28, however.

The output unit 33, a converter for converting the width of the pulse from Flip-Flop 16 into amplitudes, is connected to the collectors of both transistors of the Flip Flop 16, which converter comprises four transistors and the output of which converter is the slide of a potentiometer, so that the amplitude and the amplitude zero may be adjusted. The common adjustment of amplitude zero value and amplitude height necessarily obtained by a common potentiometer may be avoided by utilizing other circuit means, if desired.

The operation of the circuit shown appears to be clear from the specification of the block diagram FIG. 1; a description of the function of all transistors, diodes, Zenerdiodes or the compensation amplifier 21 appears not to be necessary for a worker skilled in the art.

In FIGS. 4a-i there is shown the improvement of the S/N-ratio which may be obtained with the system according to FIGS. 1 and 3. FIG. 4a shows a train of square wave pulses representative of an input signal, that is a waveform, as it would be without noise. FIG. 4b shows a noise voltage, while FIG. 40 shows a combination after mixing the noise voltage of FIG. 4b with the signal voltage of FIG. 4a. In the combination voltage of FIG. 4c the S/N-ratio is 1.07, practically such a voltage is of no value for measurement purposes.

In FIGS. 4d-i there is shown which waveform is obtained when a combination voltage of FIG. 40 repetitively is given through the system according to FIGS. 1 and 3. It is to be noted that the waveform more and more closely resembles the original signal of FIG. 4a; figured out the S/N-ratio improves as follows:

Number of runs S/N-ratio When using an appropriate number of runs for all practical purposes sufficiently good S/N-ratios may be achieved.

FIG. shows the combination of a system according to FIGS. 1 and 3 with a nuclear magnetic resonance spectrometer. For convenience there are only shown those parts of the spectrometer which are directly connected to the apparatus of the invention, while the remaining devices are omitted.

A nuclear magnetic resonance spectrometer principally operates in that an excitation RF. frequency is applied to the sample under investigation. This sample is simultaneously immersed into a magnetic field, the intensity of which is varied from an initial value to another, generally higher value, by applying a sweep current. This sweep current is generated in a sweep voltage integrator 40, which receives in dependence of the position of two relays 41 and 42 a high or a low input voltage, respectively, and accordingly generates a quickly rising search current or the slowly rising, normal sweep current, for the sweep coils.

The sample signal from the spectrometer, that is the signal including noise, is applied to a trigger 43, which on the one side is connected to the start/stop-circuit 44 0f the spectrometer and on the other side with the starting relay of the tape-deck 11 of the system of the invention. This last mentioned connection includes the switch of a counter relay 45, which opens the connection after a previously set number of runs and disables the entire system.

The start/stop-circuit 44 is connected with the search voltage relay 41 and a band offset-circuit 46. The sweep voltage relay 42 is connected to the system of the invention as shown in FIG. 3, so that upon occurrence of the first pulse of a sampling cycle this pulse switches the relay 42 into a position in which the sweep voltage integrator 40 starts to apply the slowly rising sweep voltage to the sweep coils.

In parallel to the relay 42 a timer 47 is connected to the system of FIG. 3, which timer takes care that 250 seconds after the start of the sweep current rise a relay 48 is activated for reset of the sweep current to its initial value and simultaneously the tape-deck 11 is disabled. The actual value of the timing value, in the embodiment shown 250 seconds, of course depends on the time in which the sweep current sweeps the desired range; under consideration of the tape speed further the length of the endless tape on the tape-deck 11 is dimensioned accordingly.

When starting the measurement, that is after the pulsepairs independent of the measured values are recorded on the tapedeck 11, a starting switch 49 is closed to trigger the start/stop-circuit 44. This circuit activates relay 41, so that a high voltage is applied to the input of the sweep voltage integrator and the quickly rising search current is applied to the sweep coils. When a predetermined magnetic field strength is obtained, a reference signal, a very strong pulse, is generated by the spectrometer, as usual in normal spectrometer work, and this pulse is fed from the trigger 43 on the one hand to the start/stop-circuit 44 to disable the search voltage, and on the other hand to the start relay of the tape-deck 11 to activate the same. If the band is to be offset, at this time the circuit 46 is put into operation, so that the corresponding side band of the frequency band is formed when the sweep current reaches the sweep coil. As soon as the first pulse of the first pulse-pair from the tapedeck 11 now running comes to the system of FIG. 3, relay 42 cares that the slowly rising, normal sweep current is applied to the sweep coils, and simultaneously the timer 47 starts. After the predetermined time has elapsed, in the embodiment shown after 250 seconds, the reset relay 48 is activated, and simultaneously the stoprelay of tape-deck 11, so that the sweep current and therewith the magnetic field is set back to its initial value and the tape-deck 11 stops. Simultaneously the counter relay 45 receives a pulse which is subtracted from the set number of runs.

The starting switch 49 again is closed, if desired of course automatically, and the operation is repeated until the number of runs set into the counter relay 45 is reached. Then, as previously mentioned, the connection between the trigger and the starting relay of the tapedeck 11 is opened, so that the cycle cannot be restarted; eventually an automatic for actuation of the starting switch 49 also is disabled.

The pulse-pairs stored in the tape-deck 11 then are played back in the manner described via the system of FIG. 3 and delivered to a registration instrument, where corresponding amplitudes are traced.

An example of an actual measurement with a nuclear magnetic resonance spectrometer after 1 and after runs is shown in FIG. 6. The lower trace shows the measurement or signal voltage directly obtained from the spectrometer, and the upper trace shows the measurement result substantially cleared of noise, but without zero level correction. Obviously, it is impossible even approximately to locate the resonance peaks in the lower trace, which peaks are clearly shown in the upper trace, quite in contrast, the form of the lower trace may mislead to expect that on those portions, in which actually resonance peaks are present, the signal voltage is not substantially distinguished over zero, while resonance peaks are to be found in other portions.

Since many changes could be made in the above construction and many apparently widely different embodiments of this invention could be made without departing from the scope thereof, it is intended that all matter contained in the above description or shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

What is claimed is:

1. A method for clearing repetitive voltage waveforms, which show the dependence of a quantity represented as a voltage, of another quantity, of statistical interferences (noise), in which method the average values of short sections of the voltage waveform are converted into pulsepairs having amplitude-proportional spacing and the pulse-pairs are recorded during one run, and the average values of the same short sections of the voltage waveform during successive runs are also converted into new pulsepairs having amplitude-proportional spacing and the spacing of the recorded pulse-pairs is increased by the spacing of the new pulse-pairs and the final spacings arrived at after several successive runs are reconverted into voltage signals which are fed to a recording instrument.

2. Method according to claim 1, in which the conversion of said average values into pulse-pairs is done by comparing the value in question with a linear ramp voltage, which starts with the first or starting pulse of the two pulses of a pulse-pair, and in which the pulses at each run are delayed by a constant spacing with regard to preceding record, and the voltage waveform signal and the linear ramp voltage are biased with respect to each other so that the linear ramp voltage reaches the zero value of said voltage waveform after a period corresponding to said constant spacing.

3. Method according to claim 1, in which the pulsepairs of each run are recorded into another one of a plurality of channels of a recording carrier.

4. Method according to claim 1, in which the first pulse of one of said pulse-pairs is a double pulse comprising two pulses having constant spacing.

5. Method according to claim 4, in which the first pulse of said double pulse triggers a monostable multivibrator the time constant of which is longer than the spacing of the double pulse, the second pulse of the double pulse, together with the pulse from the One-shot sets a bistable multivibrator into a predetermined state while the trailing edge of the pulse from the One-shot resets the bistable multivibrator into the opposite stable state.

6. A method for clearing repetitive voltage waveforms, which show the dependence of a quantity represented as a voltage, of another quantity, of statistical interferences (noise) including the steps of triggering with a first pulse of a double pulse as the starting mark of a deliberately generated pulse-pair a monostable multivibrator the time constant of which is longer than the spacing of the double pulse, the second pulse of the double pulse, together with the pulse from the One-shot sets a bistable multivibrator into a predetermined state while the trailing edge of the pulse from the One-shot resets the bistable multivibrator into the opposite stable state which after the second occurrence of said trailing edge is again said predetermined state, starting a linear ramp voltage by said bistable multivibrator upon resetting it into said predetermined state, comparing said ramp voltage with the average values of short sections of the voltage waveform to generate a pulse-pair the spacing of which is proportional to the average value of the short section being compared, said ramp voltage and said average values being biased with respect to each other so that said linear ramp voltage reaches the zero value of said voltage waveform after a predetermined period to cater for negative going values, recording the pulse-pairs thus generated, in a successive run repeating the generation of pulse-pairs, the already recorded pulse-pairs being utilized as triggering signals, increasing the spacing of the already recorded pulse-pairs by the spacing of the newly generated pulse-pairs, recording pulse-pairs with the thus increased spacing, the already recorded pulse-pairs delayed by a spacing corresponding to said predetermined period, reconverting the pulse-pairs arrived at after several successive runs into voltage signals, and feeding said voltage signals to a recording instrument.

7. Apparatus for clearing repetitive voltage waveforms of statistical interferences (noise) and having an amplitude source, a first converter means for converting the amplitudes into pulse-pairs, a memory means having an input and an output for recording said pulse-pairs, a second converter means to reconvert the pulse-pairs into amplitudes, and a registration instrument to record said regained amplitudes.

8. Apparatus according to claim 7, in which said first converter means is a comparison circuit, the output of which is connected to a trigger circuit to release the same when the amplitude is equal with an output signal from a linear ramp voltage generator, which trigger circuit in 'turn delivers the second pulse of a pulse-pair to the input of said memory means.

9. Apparatus according to claim 8, in which said memory means is an endless magnetic tape having preferably a plurality of tracks.

10. Apparatus according to claim 8, further comprising an oscillator means to define, at the initial run, the start of each amplitude sampling interval and to supply the corresponding first pulse of a pulse-pair to said memory means.

11. Apparatus according to claim 8, in which a delay network is connected to the input of said memory means and the connection between said trigger circuit and said memory means bypasses said delaying network.

12. Apparatus according to claim 11, in which a monostable circuit means to generate a double pulse is connected between said memory means and said delaying network.

13. Apparatus according to claim 12, in which said monostable circuit means having an input and two outputs, with its input is connected to the output of said memory means, the pulse width of said monostable circuit means being longer than the spacing of said double pulse, the first output of said monostable circuit means being connected to both inputs of a bistable circuit means having a set input, a reset input, a set output, and a reset output and the second output of said monostable circuit means leading to a first input of an AND-gate means having two inputs and one output, the second input of which is connected to the output of said memory means while the output leads to the set-input of said bistable circuit means, the set-output of which is connected to a linear saw-tooth generator means, which in turn is connected to one side of said comparison circuit.

References Cited UNITED STATES PATENTS 3,354,446 11/1967 Auril et al. 340-1741 3,373,415 3/1968 Gabor 340174.1 3,380,042 4/1968 Townsend et al. 340-174.1

TERRELL W. FEARS, Primary Examiner.

VINCENT P. CANNEY, Assistant Examiner.

Claims (1)

1. A METHOD FOR CLEARING REPETITIVE VOLTAGE WAVEFORMS, WHICH SHOW THE DEPENDENCE OF A QUANTITY REPRESENTED AS A VOLTAGE, OF ANOTHER QUANTITY, OF STATISTICAL INTERFERENCES (NOISE), IN WHICH METHOD THE AVERAGE VALUES OF SHORT SECTIONS OF THE VOLTAGE WAVEFORM ARE CONVERTED INTO PULSEPAIRS HAVING AMPLITUDE-PROPORTIONAL SPACING AND THE PULSE-PAIRS ARE RECORDED DURING ONE URN, AND THE AVERAGE VALUES OF THE SAME SHORT SECTIONS OF THE VOLTAGE WAVEFORM DURING SUCCESSIVE RUNS ARE ALSO CONVERTED INTO NEW PULSEPAIRS HAVING AMPLITUDE-PROPORTIONAL SPACING AND THE SPACING OF THE RECORDED PULSE-PAIRS IS INCREASED BY THE SPACING OF THE NEW PULSE-PAIRS AND THE FINAL SPACINGS ARRIVED AT AFTER SEVERAL SUCCESSIVE RUNS ARE RECONVERTED INTO VOLTAGE SIGNALS WHICH ARE FED TO A RECORDING INSTRUMENT.

Images