CH615542A5 - Method and device for correcting the characteristic curve of an analog-digital converter - Google Patents

Description

The invention relates to a method for correcting the characteristic curve of an analog-digital converter, which responds to a physical measurement variable and thereby delivers a measurement pulse sequence with a corresponding frequency, the characteristic curve representing the measurement pulse frequency as a function of the measurement variable being essentially a straight line within the working range of the converter. which, due to the nature of the transducer, intersects the coordinate axis representing the measuring pulse frequency outside the coordinate zero point when extended beyond the working range of the transducer

2nd

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

Which method of the measurement pulse sequence is superimposed on a correction pulse sequence and the correction pulse frequency is dimensioned such that the resulting characteristic curve, which represents the pulse frequency of the pulse sequence resulting from the superimposition as a function of the measured variable, runs through the coordinate zero point.

In many measuring instruments there is an exact linear dependency between the output signal and the measured variable, the characteristic curve representing the output signal depending on the measured variable being a straight line. Due to friction, other physical influences, the nature of the measuring instrument, etc., however, the output signals are not directly proportional to the measured variable, i.e. H. the characteristic curve or its extension does not run through, but to the side of the point of origin of the coordinate system in which the characteristic curve is shown. This creates difficulties, particularly when the output signal is a sequence of pulses, which are summed and registered by proportionally operating counters.

From CH-PS 592 387 an apparatus and a method for correcting the characteristic of an analog-to-digital converter can be seen.

The purpose of the invention is to create a method and a device for correcting the characteristic curve of an analog-digital converter, in which the pulse sequence is directly proportional to the measured variable over the entire working range of the converter.

The method according to the invention results from the characterizing part of patent claim 1.

The device according to the invention results from patent claim 7.

Exemplary embodiments of the subject matter of the invention are explained in more detail with reference to the drawing. Show it:

1 shows an example of a measurement characteristic curve in the form of a straight line, which represents the frequency of the output pulses of the converter as a function of the value of the measurement variable,

2 shows a representation corresponding to FIG. 1 to explain the correction of the measurement characteristic curve according to the invention, FIG. 3 shows a diagram to explain the superimposition of measurement pulses and correction pulses according to the invention, FIG. 4 shows a simple block diagram of a correction circuit for carrying out the method according to FIG the invention,

5 and 6 are diagrams for explaining the superimposition of measurement pulses and correction pulses according to two alternative modes of operation of the circuit according to FIG. 4, and

7 shows a detailed block diagram of a further embodiment of a correction circuit according to the invention.

With reference to FIG. 1, the difficulties encountered in the measurement methods used hitherto, which have already been indicated in the introduction, are briefly explained. 1 shows the characteristic curve of a measuring instrument of the type on which the invention is based, in which the measurement variable converter delivers a pulse train, the frequency of the pulses being linearly proportional to the measured variable.

As an example, a flow meter for measuring the flow volume q (m3 / s) is selected, in which senumformer a rotor or a ball is provided, the speed of which is detected as frequency f Hz in a known manner. The straight line denoted by the reference number A in FIG. 1 is obtained as the characteristic curve. The straight line A begins at a lower limit value fmjn, which corresponds to the lowest flow volume qmjn at which the transducer responds and which can therefore still be recorded in practice. The lower limit of the work area is therefore determined by the value qmjn. The dashed line extension of the straight line A does not run through the coordinate zero point, but intersects due to various interference factors,

615542

especially hydraulic losses, the ordinate at point P below the coordinate zero point.

There are of course also cases in which, in the manner indicated by the straight line A ', the intersection P' of the characteristic curve with the ordinate lies above the coordinate zero point. Flowmeters operating according to the hydrodynamic oscillator principle, for example swirl chamber flowmeters, which have no moving parts, have a characteristic line that intersects the ordinate above the coordinate zero point. For the present invention, it is irrelevant whether the point of intersection lies above or below the coordinate zero point.

To reproduce the frequency f, an integrating or directly indicating device, for example an adder for summing up the pulses, is normally used, which is incremented by one unit for each pulse. The frequency f is now not directly proportional to the flow volume q, since the straight line A representing the characteristic curve does not run through the coordinate zero point. The adder or the counter, on the other hand, has a characteristic curve starting from zero, since its content is directly proportional to the pulses supplied. To compensate for this deviation in proportionality, the adder is normally set so that it has the characteristic curve B shown in dash-dotted lines in FIG. 1, the gradient a of which is dimensioned such that it intersects the characteristic curve A of the transducer approximately in the middle qmed of the desired working range ai. This adjustment can be achieved by giving the adder a proportionality factor K = tana. In the process, measurement errors appear to be increasing towards the limits of the work area. Since in many cases only small measurement errors can be tolerated, the working area has to be narrowed very much to avoid too large measurement errors, which of course causes great difficulties in practical use.

The invention now aims at a complete correction of the measurement signals supplied by the measurement transducer, so that the measurement signals are not only proportional to the measurement quantity, but rather directly proportional to it, without having to accept an impairment of the measurement accuracy. The purpose of this correction is therefore that the extension of the characteristic curve intersects the coordinate zero point. This correction is achieved in a simple manner by superimposing a certain, but adjustable, correction frequency on the transducer frequency, so that the characteristic curve is shifted upward and intersects the coordinate zero point in its extension. This is shown in more detail in FIG. 2, in which the uncorrected frequency of the transducer is represented by straight line fg (corresponding to straight line A in FIG. 1), while the corrected frequency is shown by straight line fg corr. The constant frequency added to the frequency of the transducer is designated fk and applies:

fg corr fg fk

Within the working range denoted by a in FIG. 2, which extends from the lower limit value fmin to a desired upper limit value fmax, the corrected frequency fg corr is therefore directly proportional to q. It can be seen from the above explanations that it does not matter whether the uncorrected characteristic curve intersects the ordinate below or above the coordinate zero point. If the characteristic curve intersects the coordinate zero point above the characteristic curve, the required correction frequency only has to be subtracted from the frequency of the transducer. The correction frequency is therefore superimposed on the frequency of the transducer with the corresponding sign.

3rd

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

615 542 devices.

In order to achieve the desired parallel shift of the characteristic of the measurement transducer, a pulse generator is provided according to the invention, which delivers pulses with the desired correction frequency and which is connected to the output of the measurement transducer. In order to achieve an error-free frequency superposition in order to achieve the desired parallel shift of the characteristic of the transducer, certain conditions must be met, the three most important conditions being explained below. First, no correction pulses may be sent if the transducer is not delivering pulses, i.e. H. Correction pulses may only be sent when the transducer has delivered a pulse. Second, pulses delivered by the transducer must not coincide with correction pulses since the adder would then only register a single pulse. It may be necessary to separate completely or partially coinciding pulses so that the adder can record both pulses. Thirdly, correction pulses may no longer be sent when the measuring transducer stops or, if the frequency of the measuring transducer falls below the lower limit frequency fm, n of the working range, the number of correction pulses then superimposed on each measuring pulse must be limited to a maximum value.

The conditions are now explained in more detail with reference to FIG. 3. 3 shows the measurement pulses along the time axis a, the correction pulses along the time axis b and the superimposed pulses along the time axis c. Fig. 3 shows the case where the frequency of the transmitter approaches the lower limit, where gi is the penultimate pulse and g2 is the last pulse that was transmitted with the frequency fmin. G3 denotes a pulse that may be sent later outside the working area. The correction pulses ki, k2 etc. are sent regularly after an appropriately selected time interval. According to the third condition explained above, the correction pulse sequence must be terminated after the last measurement pulse g2, this termination not taking place simultaneously with the occurrence of the measurement pulse gi, but only after a certain number n correction pulses have been sent in accordance with the time interval between the measurement pulses gi and g2 have been. If the lowest measuring pulse frequency is fmin, the maximum time interval between the measuring pulses in this case is tma) !, d. H. the relationship tmax = l / fmin applies at the lower limit of the working range. If the correction frequency has the value fk, the number n of correction pulses during this time interval is equal to tmax * fk or fk / fmjn. The broken vertical line S in FIG. 3 indicates that after the last measuring pulse g2, only n correction pulses are sent. 3 shows the corrected pulse sequence which is present immediately before the measurement variable converter stops along the time axis c.

In this context, it is pointed out that fmin generally represents the lowest point of the rectilinear part of the characteristic curve of the transducer. As already indicated, in certain cases measurement pulses are actually sent at a lower frequency than fmjn, which then occur in a non-linear area outside the working area. In particular, in the case of flow rate meters for media with high viscosity, when the flow rate is reduced, the transducer can emit pulses which deviate from the linear characteristic curve and lie on the characteristic part leading to the end point f'min in FIG. 2. In the practical application of the invention, however, fmin is always set so that this point lies at the end of the straight part of the characteristic. This principle also does not change when frequency multiplication takes place, i. H. if the output signal of the transducer is multiplied by a factor u. The correction frequency fk must also be multiplied by the factor n so that the ratio n = fk / fmm remains unchanged.

4 shows a block diagram of an electronic circuit for carrying out the method according to the invention. The measurement variable converter designated by the reference number 1 supplies a measurement pulse sequence, the frequency of which is proportional to the measurement variable, for example proportional to the volume flowing through a flow meter per unit of time. The output of the transducer 1 is connected to the input of a correction unit 2 and to the control input of a gate 3. A pulse generator 4 for generating correction pulses is connected to the second input of the gate 3. The four circuit components are shown as separate units for the sake of simplicity, but can be combined with one another in various ways, for example in the form of a single integrated circuit.

In accordance with the above explanations, the measurement pulses from the measurement variable converter 1 and the correction pulses from the generator 4 are superimposed on one another in the correction unit 2 and then arrive at a counter or adder (not shown). However, the impulses are not simply superimposed; rather, the correction unit 2 together with the gate circuit ensures that the conditions for the superimposition specified above are met.

First of all, the “separation condition” is to be dealt with, according to which it must be ensured that a measuring pulse and a correction pulse, even if both occur simultaneously, are not combined into a single pulse and fed to the downstream counter or adder. The correction circuit ensures a separation of the two simultaneously occurring pulses and for this purpose contains a memory in which simultaneously occurring pulses are stored and output one after the other. If a pulse is currently being output when another pulse flows to the correction unit, the new pulse is put in a queue in the memory. This new pulse is only output when a certain time interval has elapsed after the output of the previous pulse, as a result of which sufficient separation of the pulses is achieved and the adder can reliably detect both pulses.

The gate 3 ensures compliance with the two remaining conditions, since it coordinates the two pulse sequences in such a way that correction pulses are only transmitted continuously if a measuring pulse has occurred beforehand, and on the other hand the number of correction pulses is limited when the distance between the Measuring pulses has risen to a value which is below the lower limit of the measuring range. On the one hand, the gate now only allows correction pulses to pass if it has previously been triggered by a measuring pulse. On the other hand, the gate interrupts the correction pulse sequence if a measuring pulse is missing, for example if, according to FIG. 3, the measuring pulse g2 is not followed by a measuring pulse gi. However, should one or more delayed measuring pulses follow, then the gate only allows a predetermined maximum number of correction pulses for each such subsequent measuring pulse. The gate can be controlled either on the basis of a time measurement or on the basis of a pulse count.

In the case of a gate 3 controlled on a time basis, each measurement pulse causes the gate to open for a specific maximum time interval tmax, which corresponds to the reciprocal value of the lower limit frequency fmjn. If the measuring pulse frequency is above fmin, the gate remains open continuously.

4th

5

10th

15

20th

25th

30th

35

40

45

50

55

60

65

5

The situation explained above is shown in FIG. 5, in which the measuring pulses are shown in line a and the opening intervals tmax in line a. The correction pulses are shown in line b, while the output pulses of the correction unit are shown in line c. 5 that the measuring pulse gi opens the gate, so that it passes a subsequent correction pulse ki. The next measuring pulse g'2 only occurs, for example, because of a brief standstill of the transducer only after the gate is locked again. The measuring pulse, 0 g '2 now opens the gate again, so that the following correction pulse k2 is let through again. The subsequent correction pulse k3 is not allowed through, however, since the gate is already blocked again. It is assumed that the measurement pulse g3 does not occur at all. , 5th

6 shows the mode of operation of the gate controlled on the basis of pulse counting, which continuously transmits at most a predetermined number n of correction pulses,

where n = fk / fmin and has the value 3 in the present case.

As in FIG. 5, the measurement pulses are shown in line a, the correction pulses in line b and the superimposed pulses in line c. The measuring pulse g2 opens the gate, which then passes three correction pulses, whereupon the gate is blocked, as indicated by the line S, since no further measuring pulses occur. However, should a measurement pulse g3 25 occur later, this in turn triggers three correction pulses, as shown in dashed lines in FIG. 6 to the right of line S. In this context, it should be pointed out that both in FIG. 3 and in FIG. 6, for reasons of clarity and for a better understanding, the correction frequency has been chosen much higher than the measurement pulse frequency. In fact, however, the two frequencies are of the same order, i.e. H. n 1 and in some cases even f ^ is smaller than fg.

In the case described last, an approximation solution is obtained because the number of pulses passed by the gate is by definition equal to n = fk / fmin and this quotient must be rounded up to the next positive integer. However, the gate can be modified without great difficulty so that the gate passes an average of a number of pulses corresponding to the setpoint n 40. For this purpose, a memory is assigned to the gate, which ensures that the gate passes a number N or K of pulses, which is slightly larger than n and once smaller than n, so that the average number passed by the gate converges to n after impulses.

A simple numerical example illustrates this principle. Assume the quotient fk / fmin. d. H. the desired number n of correction pulses is 2.6. The value 2.6 now corresponds to 13/5 or the integer 2 plus 3/5. If *> the gate is controlled in such a way that in each case three correction pulses are transmitted in three successive opening intervals during three intervals and only two correction pulses each during two intervals, then the average value for a longer sequence of measuring pulses is obviously 2.6.

A circuit will now be explained with reference to FIG. 7,

which meets the three overlay conditions listed above. The operation of the circuit according to FIG. 7 differs from the operation of the circuit according to FIG. 4, since the circuit according to FIG. 7 is made up of standard components implemented in an integrated circuit. In the circuit according to FIG. 7, the measurement pulses are not directly superimposed on the correction pulses, rather secondary measurement pulses are triggered by the measurement pulses, which originate from the same pulse source that also supplies the correction pulses. The secondary measurement pulses and the correction pulses wer615 542

the added and processed in the manner explained below to a corrected output signal.

The transducer 60 shown in FIG. 7, which is, for example, a flow meter of the type mentioned above, provides an analog signal which is appropriately amplified and converted into a pulse-shaped signal which occurs on the output line 62. The signal occurring on the output line 62 is then applied to a so-called shock generator 64, which has the property that it delivers a series of pulses, for example ten pulses, in each case after receipt of an input pulse. The generator 64 receives the pulses for such a pulse series from a pulse generator 66, which is fed by a crystal oscillator 68, which supplies a high base frequency of 1 MHz, for example. The pulse generator 66 has the property that firstly it reduces the frequency of the pulses received by the oscillator 68, for example by a factor of 10, and secondly it delivers pulse trains with the reduced frequency to a number of output terminals, the pulse trains being staggered in time. Such a pulse generator can be, for example, a known decade decoding counter. A pulse train, which in the present case has a frequency of 100 kHz, is supplied to the pulse generator 64 for chopping from an output terminal of the pulse generator 66 via the line 70. The shock generator 64 therefore delivers a pulse series each time a measurement pulse is received on the output line 72, which in the present example comprises 10 pulses, the pulses within the series having a frequency of 100 kHz. The pulse series is fed to an OR gate 74 via the output line 62.

Another pulse sequence, which likewise has a frequency of 100 kHz, is taken from another output terminal of the pulse generator 66 via the line 76; however, is offset in time from the pulse train taken via line 70. The pulse train taken off via line 76 is used to generate the required correction signals. To generate the correction signals, two gate arrangements were explained above with reference to FIGS. 5 and 6, one of which operates on a time basis and the other on a count basis. In the embodiment according to FIG. 7, the pulse train taken off via line 76 is fed to an AND gate 78, which is opened as a function of the measuring pulses occurring on line 62, since the measuring pulses not only serve the shock generator 64, but also a monostable flip -Flop 80 are supplied, which after each measurement pulse opens the AND gate 78 for a predetermined period of time for the passage of the pulse sequence present on the line 76. The monostable flip-flop 80 ensures that the first and third of the three conditions mentioned above are met, i. H. no correction pulses are generated,

before a measurement pulse has not occurred and the correction pulses can no longer occur indefinitely after no more measurement pulses are generated.

The pulse sequence passed by the AND gate 78 reaches a frequency divider 82 which reduces the frequency of the pulse sequence, in the present case reduced by a factor of 100, so that the pulse sequence occurring on line 84 has a frequency of 1 kHz. The line 84 leads to the input of a multiplier 86, on which a multiplier ensuring the desired frequency can be set in a known manner with the aid of decadal dial wheels. The pulses occurring on the output line 88 of the multiplier 84, the frequency of which is, for example, 10 fk, are supplied to the OR gate 74 and added there to the pulse series generated by the pulse generator 64 as a function of the measurement pulses.

In the manner explained above, a corrected measurement pulse fre corresponding to the sum fg + fk is thus obtained

615 542

sequence fg corr, where fk is to be set such that the line representing fg corr intersects the coordinate zero point in the manner shown in FIG. 2. The value of fk adapted to the respective measuring device is set on the multiplier 86 taking into account the fact that the pulses occurring on line 72 represent the value 10 fg. The signal combination occurring on the output line 90 of the OR gate 74 is therefore 10 (fg + fk).

The second condition required for the correction of the measurement pulses, namely that the measurement pulses and the correction pulses must always be separated from one another in such a way that they do not coincide when added, is automatically ensured in the circuit according to FIG. 7 in that the measurement pulses or rather those of the Measuring pulses dependent pulses and the correction pulses come from the same source, namely from the high-frequency oscillator 68, the pulses of which are divided into the same but temporally staggered pulse sequences, two of which are used to generate the secondary measuring pulses or correction pulses.

The superimposed pulses occurring at the output 90 of the OR gate 74 are fed to a further multiplier 92, on which a dimension factor K can be set as a multiplier for adaptation purposes, so that the pulses occurring at the output 94 of the multiplier 92 have such a frequency K • 10 (fg + fk) have that each pulse provided for counting corresponds exactly to the desired measuring unit, for example a flow rate of 11 per minute. Since the pulses can be close together (although they never coincide because the minimum pulse spacing is always one microsecond in terms of the frequency of the common pulse source (1 MHz), some counters, in particular electromechanical counters, are not suitable for counting the pulses, so that the pulses occurring at the output 94 of the adder 92 are passed to a pulse shaper 9b / ugc in which, on the one hand, the time interval between incoming pulses and, on the other hand, the width of the pulses can be stretched 5 without changing their total number Down counting method is possible and the pulse shaper 96 is therefore based on a known up / down counter. The output of the pulse shaper 96 is connected via a line 98 to the input of a counter driver stage 100. The pulses occurring on the line 98 have such a distance and duration that it can be abilities can be detected by a mechanical counter 102 or an electronic counter 104.

In the circuit according to FIG. 7, the frequency of the correction signal is modified in a special way, for example by the pulse generator 66 and by the frequency divider 82, and a common pulse source with a frequency of 1 MHz is also provided. The frequency values listed above can of course be modified according to the respective 20 conditions within the scope of the invention.

A major advantage of the circuit arrangement according to FIG. 7 is that the correction pulse frequency can be modified in a simple manner by compensating for the various disturbance variables influencing the measured value 25. In the case of flow meters, for example, temperature fluctuations cause corresponding fluctuations in the viscosity of the flow medium and measurement errors resulting therefrom can be corrected by changing the frequency of the correction pulses as a function of the temperature, which can take place automatically. Furthermore, the frequency can be modified with regard to manufacturing tolerances between different transducers in such a way that optimum accuracy is achieved with each individual instrument.

Claims (11)

615 542 PATENT CLAIMS 1.Procedure for correcting the characteristic curve of an analog-digital converter that responds to a physical measurement variable and thereby delivers a measurement pulse sequence with a corresponding frequency, wherein the characteristic curve representing the measurement pulse frequency as a function of the measurement variable is essentially a straight line within the working range of the converter that is due to the condition of the transducer when extended beyond the working range of the transducer, the coordination axis representing the measuring pulse frequency intersects outside the coordinate zero point, in which method of the measuring pulse sequence a correction pulse sequence is superimposed and the correction pulse frequency is dimensioned such that the resulting characteristic curve, which is the pulse frequency of the resultant superposition Represents pulse sequence as a function of the measured variable, through which the coordinate zero point runs, characterized in that a) a basic pulse sequence is generated with a predetermined frequency, b) the basic pulse sequence is divided into at least two sub-pulse sequences, the pulses of which are offset in time from one another, c) correction pulses are separated from the one sub-pulse sequence under the control of the measuring pulses, d) under the control of the measurement pulses, a predetermined number of secondary measurement pulses is derived from the second sub-pulse sequence for each measurement pulse, and e) the correction pulses and the secondary measurement pulses are superimposed on the resulting corrected pulse sequence, with the frequency being dependent on the graphical representation The characteristic curve runs through the coordinate zero point. 2. The method according to claim 1, characterized in that that when the correction pulses are separated from the one sub-pulse sequence, the number of correction pulses generated per measurement pulse is limited if the measurement pulse frequency falls below the lower limit of the working range. 3. The method according to claim 2, characterized in that to limit the number of correction pulses, these are separated only during a period of time that is not greater than the time interval between successive measurement pulses at the lower limit of the working range. 4. The method according to claim 2, characterized in that the number of correction pulses is limited to a value n which is not greater than the quotient of the correction pulse frequency and the measurement pulse frequency at the lower limit of the working range. 5. The method according to claim 4, characterized in that correction pulses are sent alternately in a number corresponding to the integers N and K, where N and K are immediately above or immediately below the number n and thereby the average value of the correction pulses generated per measurement pulse is the same n is. 6. The method according to claim 1, characterized in that the corrected pulse sequence resulting from the superposition of the correction pulses and the secondary measurement pulses is stretched to increase the time interval between the pulses and to spread the pulses themselves, without changing the total number of pulses. 7. Device for carrying out the method according to claim 1, characterized by a pulse generator (68) for generating pulses with a base frequency, a pulse generator (66) fed by the pulse generator (68) for generating at least two sub-pulse sequences (70, 76), whose pulses are offset in time from one another, a first evaluation (80,78) and converter device (82,86), which can be controlled by each measuring pulse (62) and then parts of the one sub-pulse sequence to generate n * fk correction pulses (88 ), where n is a positive number and fk is the correction frequency, a second evaluation and conversion device (64) which responds to each measurement pulse (62) and thereby parts of the other sub-pulses (70) to generate secondary measurement pulses (72) a frequency of n • fg, where fg is the measurement characteristic frequency, and by an addition device (74) for generating a corrected pulse train (90) from the secondary Measuring pulses (72) and the correction pulses (88), the corrected pulse train having a frequency of n • fg corr = n • fk + n • fg, thus fg corr = fk - fg, where fg corr is the corrected measurement characteristic frequency, and the value fk is set with the aid of the first evaluation and conversion device (86) such that the resulting characteristic curve runs through the coordinate zero point. 8. The device according to claim 7, characterized in that the first evaluation device, after receiving a measurement pulse, passes correction pulses during a time interval which is not greater than the time span between successive measurement pulses at the lower limit of the working range. 9. The device according to claim 7, characterized in that the first evaluation device transmits correction pulses after receipt of a measurement pulse, the number of which is on average not greater than the quotient of the correction pulse frequency and the measurement pulse frequency at the lower limit of the working range. 10. The device according to claim 7, characterized in that one of the evaluation and converter devices contains switching means which respond to a measurement pulse and then pass a number of successive pulses of the relevant sub-pulse train. 11. The device according to claim 7, characterized in that the first evaluation and converter device contains a logic element which is acted upon by a sub-pulse sequence and can be controlled by a monostable circuit which responds to a measurement pulse and the logic element thereupon for a predetermined period of time Passage of the one sub-pulse sequence, which are subsequently converted in a controllable frequency converter into correction pulses of a corresponding correction frequency, the second linking and converting device contains a generator which responds to a measuring pulse and then supplies a number of successive pulses of the second sub-pulse sequence and the adding device a link contains, to the inputs of which both the generator and the frequency converter are connected and which supplies a pulse signal formed by adding the two frequencies, which is proportional to the corrected measurement is pulse frequency, and the adder further includes a multiplier and a pulse shaper to generate a proportionally adapted and stretched output pulse version of the logic element of the adder.