DE10233604A1 - Sensor signal acquisition method in which sensor measurements are captured as phase shifts or phase delay times by use of parametric amplification technology within sensor networks or bridge circuits - Google Patents

Abstract

Method for capture of sensor values using the phase change between signals of relatively high frequency in a sensor networks in which parametric amplification technology is used with the phase times in transferred signal. The invention also relates to a corresponding instrument and alternating current bridge measurement circuit.

Description

The use of sensors is essential Part of today's technical development. electronic Sensors are mostly constructions where one is not physical parameters to be recorded directly, e.g. the temperature or air pressure, another, well-measurable parameter in known Way influenced.

Electronically relatively easy to grasp Parameters are currents and voltages, but also the values of capacitors, inductors and especially of resistance. Correspondingly often sensors are therefore resistance constructions or inductive, capacitive or other constructs (e.g. mechanical constructions) or combinations thereof, whose value (whichever that is) dependent in a known manner changed by the size to be measured and thus the measurement of these quantities indirectly allowed.

So there is of temperature, of of strength a magnetic field, stretch, or some other physical Parameter dependent Resistances. So you find a level from capacities dependent on humidity, distance and other parameters and inductors (or resistors).

If in a (mostly linear) network such components are included, then the primary parameter can be set via the Properties of the network are determined when taking a measurement are accessible.

The usual evaluation using electronics such construct parameters are varied, but often also by turn marked indirect processing; so tensions or streams measured directly, but also I / U or U / I conversion to measure the resistors, inductors or capacities, (without or with bridge circuits, often used with quite complex adjustment) and / or other methods. In principle, it is also possible this about general properties of a network in which such a sensor component has been installed to record and evaluate; also phase measurements belong to.

The up to date of the Technology achievable resolution a phase measurement, usually realized as a time measurement but only in a few cases one for Sensor applications sufficient accuracy. A phase measurement also applies (with evaluation) as a rather cumbersome procedure.

In connection with the evaluation of phase relationships on electronic sensor bridges were precisely working, General-purpose, electronic amplifier technologies searched and developed.

The invention realizes a highly sensitive Method for measuring the phase position of a periodic signal across from the phase position of a given reference signal. The for this purpose found technology for reinforcement of very small phase differences between two higher frequencies Signals using a closely adjacent reference frequency, is as a parametric (to a certain extent also as a non-optical interferometric) amplifier technology that can be used to measure physical quantities (e.g. through sensors).

The change in phase response of a signal (generally a sinusoidal signal) with a relatively high frequency is used when passing through the network and the phase difference is then transformed into a much larger phase time between two superimposed or mixed signals. Highly sensitive phase time measurements down to the ps range (picoseconds, 10 ^–12 s) are possible.

In a certain sense, the technical design results in a universal circuit for sensor technology (cf. 23 and description), whereby (highly precise) digitization is given quite simply (implicitly) with which an interface to a microcontroller can be realized. The developed method therefore also provides a new type of electronic bridge amplifier for AC-excited measuring bridges and sensors.

The one to capture a physical Size frequently used The arrangement is a bridge circuit. Often, these measuring bridges are not when a sensor module is used to be built up explicitly, but are themselves a direct component the sensors, e.g. when using strain gauges within commercial Pressure, force or bending sensor designs.

This fact (as a default the use of a measuring bridge) was on the lookout for new processes and development consider

The one with such a sensor construct achievable changes the resistance values are generally very small. The most Signal quantities setting sensor output must therefore generally still electronic processed, further processed and / or reinforced. There is for that quite highly specialized amplifier modules, mostly in IC form, which generally corresponds to the application still need to be connected.

When further processing the signals obtained with such a measuring bridge by means of such a module, care must generally be taken that the properties of the measuring bridge (or the sensor construct) cannot be falsified by tapping the bridge voltage measurement itself. However, measuring bridges operated in the short circuit have already been described.

The search therefore had to take into account sensor constructions that are intended for a voltage-related processing technique (internal resistance of the measuring arrangement with R _inside → ∞), as well as those sensors whose “tapping techniques” are specified for further processing in terms of current (R _inside → 0) ,

This is generally with today's amplifier techniques (Operational amplifier in the following shorter as "op-amp" referred to) no problem. But when using inductive or capacitive components, e.g. in a bridge, In practice, there are often considerable problems. Frequency compensations are not always satisfactory Linearization of a bridge circuit is not always an easy task, etc .; the ones to be hit activities usually reduce the sensitivity of the bridge again.

The arrangements according to the invention arose primary in the construction of sensors with built-in bridge circuits, which is conventional at first, but then also through a more general arrangement had to be realized. In another important sensor arrangement should be a spatial Distance can be detected by a signal transit time measurement and a another aimed at a flow measurement. Various procedures and Orders were examined, some with success; but only presented the solutions according to the invention a sufficiently high sensitivity and resolution and as Side effect - the easy digitization of the measured variables and the easy to implement Interface to a microcontroller safely.

In the preferred technique becomes one when calculating addition theorems of the trigonometric Functions primary process to be understood as a mathematical transformation process, namely the translation the phase property of a sine function in the envelope of the overlay result or technically simulated in the LF component of a mixed signal and used. A phase shift in a high-frequency sine signal namely under certain circumstances with an overlay; one Blend or subsample to almost the same phase shift in the low frequency envelope transformed ("almost" means e.g. with beat formation "down to a factor of 2").

This "parametric gain effect" has hardly been considered in technical applications.

Requirement of the to be shown Methods, as will be shown, is the generation of two in frequency as possible sinusoidal signals close together. The generation of such defined Signals may be however, also given as given, since such signals with today's generator technologies may also different are to be generated and interpreted suitably from case to case have to; the applies especially when using very high frequencies. Important for the method according to the invention is not primary the technology used here to generate f + df the way this Generating frequencies is a separate topic. This Frequencies can therefore regarded as given in the context to be seen here become. However, one of the methods described is preferred used, derived from the simulation of the addition theorems can be.

Will be one of these two signal frequencies used for measurement in one of the sensor arrangements described below and then the two given, to be compared in their phase position Sinusoidal signals superimposed with the second sinusoidal signal (additive or subtractive), then e.g. two beats (or also when scanning or Mix - like shown below - two sampled LF signals, or two LF mixed signals).

Because of the special properties of the addition theorems of trigonometric functions, that the phase position in a single component (here the phase of the RF signal) is found in a sum or overlay component, this phase time (based on the much lower envelope frequency) is around one has grown many times over. (As shown below, other, but above all digital, or square-wave signals are also suitable) This shows, for example, the relationship that can be found in almost every manual in mathematics sin (α) + sin (β) = 2cos [(α - β) / 2] sin [(α + β) / 2] , if it is interpreted in the following sense: sin (ω + Δ) t + sin (ωt + φ) = 2cos [(Δt - φ) / 2] sin [ωt + (Δt + φ) / 2] ,

The factor cos [(Δt - φ) / 2] here is particularly important. The phase position φ / 2 lying in the low frequency Δ / 2 of the cosine component represents the same phase position introduced or caused by the HF, that is to say by sin (ωt + φ) in the envelope curve. However, φ / 2 is in this NF cosine component a much longer phase time! In this sense, therefore, a beat generation represents a parametric amplification of a phase difference in the HF that arises, for example, due to signal propagation times. The double (or also multiple) beatings generated in the arrangements according to the invention allow the measurement of the HF phase shift to the much easier one Phase measurement in the envelopes (or LF mixed signals) ren.

In the sense that a temporally very small quantity (which, for example, only needs to be identified in a HF signal for ps, ie ^10-12 s) has a very strong effect on the phase position of the resulting superposition envelope through a technical simulation of the mathematical process ( For example, with a corresponding frequency selection to be "amplified" to more than 10 ⁶ times, that is to say 1 μs, or more, that is to say that a very short time is very strongly translated, ie amplified, this process is based on the technically already existing one Term, referred to here as "parametric gain".

Then this procedure can always for measurements, e.g. also for sensor applications, if the coding a primary, physical parameters in the phase position of a sinusoidal signal he follows; and that can almost always be achieved.

For an inventive, technical development this process, which is in itself clear, is not only knowledge this fact of parametric amplification in itself. DC overweight and very clear is the need to work out that - in addition to the generation of frequency as close as possible neighboring sinusoidal signals - first the construction also according to the invention an envelope phase difference measurement (i.e. the exact measurement of the phase positions of two beats to each other) allows to realize the highly accurate method because of an inaccuracy in this phase position determination, of course, part of that with the parametric gain the accuracy gained would be destroyed again.

Avoiding this problem is particularly easy with a replica of the one described here analogue procedures on digitally working techniques.

To determine the phase position between two sinusoidal signals are therefore initially (as the first method) assumed that beats should be formed and e.g. the minima (or maxima) of the two beatings in relation to one another are. That this is not always easy, especially when the overlapping Signals of different amplitudes are in practice such circuits to understand at any time and will soon addressed.

When training the parametric reinforcement themselves, provided that with standard technology (e.g. arrangements using analog techniques, UHF transceiver chips, etc.) at least three, basically different ways possible. once the path just mentioned of a direct simulation of two beatings, the phase relationship of which is to be evaluated.

und für das abgetastete phasenverschobene SignalThe second way uses subsampling of a signal:
For example, is there a sine function Asin (2πft + φ), which should have the phase shift φ relative to another (e.g. Asin (2πft) as a reference with the same frequency f) and which has a sampling rate (sampling frequency f +) that is close to the natural frequency df) is scanned (which in itself means a violation of Shannon's sampling theorem), then results with the sampling support points at t = T λ = λ (T - dT) + t 0 = λ / (f + df) + t 0 , λ = 0, 1, 2, ..., for the sampled reference signal

and for the sampled out-of-phase signal

As you can see from it, remains also in this case the phase difference between the two RF signals, between Asin (2πft) and Asin (2πft + φ), in the phase differences of those obtained by the (subsample) receive low-frequency signals with the frequency df.

In this case there is also a parametrically amplified time difference (with f = 1MHz and df = 10Hz, for example, dT / T = 10 · 10 ^–12 !).

The same can be done as a third possibility by mixing the two signals, one with respect to the achievable or given possibilities for high quality parametric gain mostly overlooked Property of the signals generated or generated during mixing.

In the expression

sin (ω + Δ) t sin (ωt + φ) = ½ {cos [(Δt - φ)] - cos [2ωt + (Δt + φ) / 2]} is again a parametric in the phase position of cos [(Δt - φ)] as the lower-frequency component To see amplification of the phase position. The separation of the two parts forming the right expression by means of a filter is relatively easy given a suitable frequency selection due to the high frequency spacing.

When using very high frequencies this way will be the better if not the only possible one. This is generally not that much of a hindrance because they are sophisticated Techniques from GHz practice are available.

This process can also be seen as an intermediate step between extremely high frequencies and the later preferred digital technologies with the limited speed of the digital technologies. In the 8th and 9 and the following related explanations, this intermediate stage property is shown.

9 shows two UHF sources ( 76 ) and ( 77 ). The signal from the generator ( 76 ) is used for the measurement here (above the measuring block ( 75 ) indicates how a measuring bridge would be installed in this arrangement). The signal used for the measurement is thus available twice (once before and once after the measuring block ( 75 ) or in the form of the two branch tap signals of the measuring bridge, which is excited by the UHF1).

These two UHF1 signals are now transmitted to the second UHF2 (generator ( 77 )) in the mixers ( 80 ) respectively. ( 81 ) mixed, and in the blocks ( 82 ) and ( 83 ) low pass filtered, resulting in the two intermediate frequencies ZF1 and ZF2.

Because of sin (ω + Δ) t · sin (ωt + φ) = ½ {cos [(Δt - φ)] - cos [2ωt + (Δt + φ) / 2]}, from which the filtering ( 82 ) respectively. ( 83 ) only the NF component remains, i.e. ½cos [(Δt - φ)], the phase difference between the two UHF1 signals is retained in the respective IF, ie there is the same phase shift between the two IF signals, which also exists between the two UHF1 signals exist (in front of and behind the test section). This applies because the measurement UHF1 signals are mixed with exactly the same UHF2 signal; this is an important condition in the realization.

The additional of the mixing and filter stages caused phase shift with the same structure in both Branches turn out the same and therefore appear in the phase difference also not on. A possibly different phase response of the amplifiers (through Parameter spread) is by bridging the Measuring distance easily measurable. This calibration measurement is a technical one So mistakes can be easily corrected.

In 9 is in ( 79 ) from the ZF1 with the techniques that lead to the 10 . 11 and 13 are described, specifying the difference frequency df (in 9 by the generator ( 78 ) generates) the frequency f + df adjacent to ZF1 = f. The superimposition, subsampling or repeated mixing of ZF1 and ZF2 (ZF2 has the same frequency as ZF1 but a different phase position caused by the measuring arrangement) with this frequency ZF1 ± df in ( 84 ) respectively. ( 85 ) generates the low-frequency signals to be evaluated in their phase relationship with ( 86 ) respectively. ( 87 ), which now reflect the phase relationship between the UHF signals with high precision, here in a two-stage process, parametrically enhanced.

To mix down two very high frequencies in a lower intermediate frequency range a mixture well suited. However, it should be noted that e.g. the two UHF signals, their phase shifts to each other is to be measured, mixed with exactly the same signal (i.e. also, in the layout equal signal distance from a common source point); This is the only way to guarantee phase difference maintenance in the intermediate frequency with parametric gain and thus also the preservation of the measured variable in the further course!

According to one of the applications examined 8th For example, the two phase-shifted UHF signals (from the generator ( 73 ) generated at the entrance of the measuring section ( 62 ) lying and at the exit of the measuring section ( 62 ) with a phase shift) by mixing with a UHF reference ( 74 ) in the mixing stages ( 64 ) and ( 63 ) mixed down to an intermediate frequency (IF) of approx. 1 MHz, ie the mixed signal through the stages ( 65 ) (low pass) filtered and also with a ( 65 ) each contained Schmitt trigger stage prepared for digital further processing.

With the in 8th and 19 The digital method shown below, the signal parameters (the signal positions) were then prepared for a processor interface.

The steps ( 63 ), ( 64 ) and ( 65 ) additional phase shifts that occur due to the same structure cancel each other out in the phase difference of the signals.

Since there is a phase angle φ on all Steps refer to the period of 2π and the reduction factors for the ratio φ / 2π doesn't matter play (phase difference and the time of a period are in transformed in the same way) in this case was a step by step Processing caused errors (e.g. not exactly 1MHz as IF frequency, just "about 1MHz") Completely negligible.

The two IF signals were used in ( 67 ) a digital sine / cosine representation (amount function of the signals to control a circuit according to 13 ) formed with half the frequency, the low-frequency sine-cosine signals were specified directly by DAC functions of a microcontroller ( 69 ).

In ( 68 ) the neighboring frequency to this half IF was formed; this reduction by a factor of 2 also had to be done on the right side ( 66 ). The now possible formation of the digital beat described below with the D-FFs ( 70 ) and ( 72 ) and a simple preparation in ( 71 ) allows maximum accurate phase time measurement.

The implementation of these methods according to the invention to digital signals that are easy to generate and evaluate digital methods now only represents the last step in the Development of the method. Although shown in the required Mathematics almost exclusively Sinus signals occur for the description of the method, especially for the Representation of the parametric gain, and definitely in this form are also suitable for implementation, they are actually special those used according to the invention digital signals and methods (for combinations as "semi-digital Procedure "), which is a very simple and inexpensive implementation allow the method.

At this point it makes sense to show that to generate an arbitrarily close to ω, that is to say at ω ± dω Angular frequency, this with a constant, rigid frequency spacing f or angular frequency spacing dω (or even with a fixed ratio the frequencies of Δ / ω, or df / f e.g. generated by a digital divider), again one of the trigonometric ones Formulas are reproduced, here e.g.

sin (α ± β) = sin (α) cos (β) ± cos (α) sin (β) ,

In this printout, the goal to be achieved is to the left of the equal sign, the right shows which components are required and how they are to be calculated. This is again translated into the application-related frequency considerations: sin (ω ± Δ) t = sin (ωt) cos (Δt) ± cos (ωt) sin (Δt)

Obviously you need to simulate the right side, sine and cosine functions: On the one hand that of the high frequency, that is sin (ωt) and cos (ωt), on the other hand the desired frequency difference through the two functions cos (Δt) and sin (Δt) specified.

The simulation can be simulated relatively easily - at least for frequencies that are not too high - with operational amplifiers (OP amps). Classic standard circuits are used (cf. 10 . 11 and 13 ). Digital or mixed technologies can be used in medium frequency ranges, only mixed technologies in high UHF and SHF ranges.

With mixed techniques, other forms of realization can be shown using parametric amplification (cf. 8th and 9 in connection with eg a measuring bridge insert).

The use of a mixed technique is but in some circumstances almost unavoidable: in some cases e.g. no particular freedom in choosing the one in a measurement frequency to be used. Define the given laws quite strict the use of frequency bands e.g. in 2.45 GHz, 433 MHz or 868MHz range. Force other applications out of technical establish the use of very high frequencies, e.g. with a range radar.

The phase shift between two Signals with such high frequencies are often themselves direct measurement parameters (e.g. when measuring a distance resulting phase difference of a signal). This It is therefore important to evaluate the phase position between signals.

The generation methods described closely adjacent frequencies are in the lower and middle frequency range still relatively easy to use, but in the upper frequency range often very difficult to control. There is therefore a need to Methods from very high frequency ranges down as well also from areas that are still well manageable up to the UHF area to transform.

The path of a method of transformation from top to bottom has already been made 8th respectively. 9 described; the opposite way remains:
If a given UHF is mixed once with f and once with f + df, then after high-pass filtering of the two UHF Mich signals, frequencies that are relatively close together are also obtained in the UHF range. Since df can be kept arbitrarily small with the method described in the same way, the frequency difference in the UHF is also quite small. (Depending on the filter quality used, the frequency f should not be chosen too low).

21 indicates again the diversity of the given possibilities and also shows possibilities for the generation of high neighboring frequencies.

21a and 21b show the interchangeability of the frequencies used in the measurement. 21a shows the use of the UHF ( 160 ) in a measuring arrangement ( 162 ), the mixture with a ZF ( 161 ) in the mixers ( 163 ) and ( 164 ) and filtering the mixed signal (low or high pass, in ( 165 ) and ( 166 )), which causes two UHF frequencies with the phase position generated by the measuring arrangement in the UHF. This arrangement can be used sensibly when channel separation is required for multiple measurement arrangements.

21b conversely shows the use of the ZF ( 167 ) as a measurement signal in a measurement arrangement ( 169 ), the mixture with a UHF ( 168 ) in ( 170 ) and ( 171 ) and filtering of the mixing result (low or high pass in ( 172 ) and ( 173 )), which in turn has two UHF frequencies with that generated by the measuring arrangement in the IF Phase position. This arrangement makes sense if the signals are to be impressed on UHF, but the measurement must take place in the IF range.

21c shows the use of a UHF1 ( 174 ) in a measuring arrangement ( 176 ), the mixture with a second UHF2 ( 175 ) in ( 177 ) and ( 178 ) and filtering of the mixing result with low pass ( 179 ) and ( 180 ), which generates two IF frequencies with the phase position generated by the measuring arrangement in the UHF1. This has been shown in the presentation 8th and 9 already used.

21 d represents a way of generating the neighboring UHF frequencies: With one of the techniques described in ( 183 ) two adjacent frequencies f and f + df from the specifications f ( 181 ) and df ( 182 ) generated. F and f + df are used as reference signals, from which two identical PLL circuits generate the two UHF frequencies with the frequencies nf and n (f + df). The PLL construction is known as such; the phase comparison is to be carried out once with the frequency f and once with the frequency f + df.

21 e represents another way of generating the neighboring UHF frequencies: from a single frequency ( 190 ) are in ( 191 ) and ( 192 ) by simple division generates two reference frequencies f / n and f / (n + 1), from which two PLL circuits again generate two UHF frequencies with the frequencies mf / n and mf (n + 1) (the case m = n).

So there are enough variations for generating one or two UHF signals and the measurement using one of these UHF or one ZF, always while maintaining the by phase shift caused by the measuring arrangement. The upward mix to generate the UHF signals is a rather cumbersome Process.

You will generally rather the reverse Prefer path, i.e. the measurement e.g. realize with a UHF frequency, then the two UHF signals (one used as a reference, which by Definition without a phase shift is one with a measurement-related one parameter-dependent Mix phase shift) in an intermediate frequency range (the phase remains as shown) and then with continue working on these two signals.

As mentioned, it is not even so necessary, to keep the frequency difference between UHF1 and UHF2 particularly stable, if the generation of f + df from the specifications f and df the Frequency distance df can be kept stable. This is relevant to the invention!

22 represents two digital methods for generating neighboring frequencies for applications in which relatively low frequencies can be used. 22a shows the generation of a frequency f / n from a generator frequency f ( 435 ) by reduction in a first counter ( 433 ). (The stage ( 426 ) is only used for further subdivision and can be used as a divider with the division factor without restriction 1 be accepted).

Counter 2 ( 410 ) is switched by 1 with each period of the frequency f / n. The comparator ( 428 ) compares the two counter values and sets the RS-FF ( 434 ) every time the two counter values are equal. If you simply consider the edges of the signal f / n (output from ( 433 )) and this comparator signal (output from ( 428 )), then the flank positions shift when the counter is switched up 2 ( 410 ) against each other by one cycle. This means that there are adjacent frequencies within the scope of the counter accuracy.

22b shows the generation of neighboring frequencies by simply dividing a common generator frequency ( 437 ). The resolutions with a phase measurement that can be achieved with these simple divider generators in the arrangements according to the invention, however, leave something to be desired compared to the other possibilities described and can only be used for very simple problems.

Adjacent frequencies are preferably generated here by simulating the addition theorems. 10 . 11 and 13 show only a few ways to approximate an addition theorem with simple techniques (there are many more!).

10 shows the direct, analog simulation of the formula of an addition theorem. The low-frenuent specification is ( 412 ), the high-frequency at ( 411 ). The required sine and cosine signals are at ( 413 ), ( 414 ), ( 415 ) and ( 416 ) to disposal. The multiplication is in ( 417 ) or in ( 418 ) simulated, the difference between the products is in an op-amp circuit ( 422 ) realized. ( 422 ) provides (after filtering) the neighboring frequency for the high-frequency specification (i.e. to sinωt).

The 13 represents a simplified form in which the multiplication stages have been replaced by switch functions.

The low-frequency sine or cosine functions can be easily generated by RC, RL or LC elements, for example, and are used for the simple OP amp circuit with ( 150 ) respectively. ( 151 ) fed. The two switches ( 152 ) respectively. ( 153 ) determine, depending on whether they are open or closed, whether the 150 ) respectively. ( 151 ) lying signals can be amplified positively or negatively. If the switch is closed, the following amplifier stage inverts the applied signal; if it is open, the circuit is not inverting.

Are digital sine or cosine simulations (digital signals with the correct phase relationship to each other) then results in ( 156 ) the signal Cos (NF) · sign [sin (HF)] and at ( 157 ) the signal Sin (NF)) · sign [cos (HF)], where sign (...) simply represents the sign of the function, here the sine / cosine function.

Sine and cosine function of the lower The frequency, as specified, is thus retained, that of the HF component are correct in terms of phase and sign using simple rectangular functions have been replicated.

The difference (equivalent would be a summation) of these two signals ( 156 ) and (157) using the op amp stage ( 158 ) delivers the signal at (159) cos (2π (df) t) · sign [sin (2πft)) - sin (2π (df) t) · sign [cos (2πft)] ,

A selective filter with a center frequency on the present HF, downstream at (159), the higher frequency Remove signal components from this signal; with this it results as first (usable) approximation the simulation of an addition theorem. The (after filtering) The result obtained is sin (2π (f + df) t) with a good approximation.

20 below shows an example of the signal at the output for two different times ( 159 ) in the 13 Without filtering (only a limited frequency response of the OP amp used), the right overlay signal was obtained during a measurement ( 89 ) (with a hardly possible phase measurement), after a simple RC filtering at the output ( 159 ) the left overlay signal ( 88 ). As you can see, the circuit is after 13 after additional filtering a quite well pronounced, approximately sinusoidal superposition signal can be achieved.

A third possibility of generating neighboring frequencies suggests 11 on. In this circuit, in the area ( 400 ) all required low-frequency sum and difference signals from sinusoidal specifications for ( 402 ) and ( 403 ) calculated in advance and periodically sampled by a switch arrangement. The sequence of the sampling is determined by a control stage, the frequency of the sampling by a clock ( 404 ) specified. After preparation and filtering in the area ( 401 ) there is again a frequency adjacent to f = clock frequency / 4.

As a third option 11 an arrangement suitable for DSP or microcontrollers (see below).

• – zunächst benötigt man mindestens zwei Signale mit einer je nach Anwendung zu wählender Frequenzlage und einer möglichst geringen Frequenzdifferenz. Bei Bedarf wird durch Mischtechniken der Frequenzbereich geeignet aufwärts oder abwärts transformiert (vgl. 9 und 21).
• – Dann wird eines dieser Signale mit einer so gewählten Frequenz zur Messung (Erregung, Aufschaltung, Weiterleitung, also zur beliebigen Ausbildung einer sensorisch geprägten Phasendifferenz) verwendet (was ein Referenzsignal und ein Messsignal erzeugt) und das Signal mit der anderen Frequenz
• – entweder zur Ausbildung der Überlagerungssignale mit Referenz und Messung
• – oder zur Abtastung von Referenz- und Messsignal
• – oder zur Bildung eines Mischsignals mit Referenz und Messung

genutzt.This means that there are two signal frequencies with the frequency f and f + df. Ultimately, the following part of the description now only represents a "playful examination" with the previously described possibilities of parametric amplification in various forms (apart from the description of the digital beat generation that follows). But the same constructs of ideas are always available, which are later used using the digital processes will be even easier to use:

• - First you need at least two signals with a frequency position to be selected depending on the application and the lowest possible frequency difference. If necessary, the frequency range is suitably transformed upwards or downwards using mixing techniques (cf. 9 and 21 ).
• - Then one of these signals with a frequency selected in this way is used for measurement (excitation, activation, transmission, that is to say for the arbitrary formation of a sensor-shaped phase difference) (which generates a reference signal and a measurement signal) and the signal with the other frequency
• - either to form the overlay signals with reference and measurement
• - or for sampling the reference and measurement signal
• - or to form a mixed signal with reference and measurement

used.

All of this may be applied in several stages and possibly several times and in each case related to each other (cf. 8th or 9 ).

Before working out some applications is to be shown now, as well as in a resistance-only one measuring bridge the actual measured variable in one Phase shift can be encoded. Generate capacitors and inductors in such networks the phase shifts between sinusoidal (or other) signals a little different - and are therefore still far better for a phase coding is suitable.

The different constructs are supposed to with some examples compared with each other and with a "normal case" become.

l shows the known basic circuit of a measuring bridge. Two of the four resistors R1 ( 6 ), R2 ( 7 ), R3 ( 8th ), and R4 ( 9 ) divide the applied voltage U _err ( 3 ). This happens once in the right branch, consisting of R3 ( 8th ) and R4 ( 9 ), and on the other in the left branch, consisting of R1 ( 6 ) and R2 ( 7 ). In general, a measured value (current or voltage) is measured as the difference between the centers tapped, eg in l ( 1 ) and ( 2 ), in 2a ( 11 ) and ( 18 ).

In general, in such an arrangement, at least one of the resistors R1..R4 will have a value dependent on the primary measured variable (temperature, pressure, light, .., etc.). The corresponding resistance in the other branch can be used to tune the bridge. The bridge is coordinated when the one with the instrument ( 5 ) measurable voltage U _{mess is} zero, this voltage measurement should take place without current (internal resistance of the instrument ( 5 ) R _inside → ∞).

For a measuring bridge that is not loaded in this way, the following also applies U mess = U err (R 2 R 3 - R 4 R 1 ) / [(R 1 + R 2 ) (R 3 + R 4 )] , (I) that the tuning of the bridge (in the tuned bridge U _mess = 0) R 2 R 3 = R 4 R 1 or R 1 / R 2 = R 3 / R 4 requires.

If (for comparison, but ultimately arbitrary) it is assumed that R, which is the resistance dependent on the target measured variable, R ₂ for tuning the bridge is variable, but is kept constant after tuning and, moreover, the matching condition is fulfilled, then the sensitivity of the bridge to a small change in R ₁ , i.e. ∂R ₁ , given by ∂U mess = –U err R 2 / (R 1 + R 2 ) 2 ∂R 1 (II)

This expression describes the change of the measured value U _mess when R ₁ changes to ∂R. ₁ This sensitivity analysis will be used in the following to compare the different arrangements.

The maximum sensitivity of this bridge arrangement to such a change in resistance dR ₁ is even when R ₂ = R ₁ = R ∂U mess = - [U err / (4R)] ∂R , (III)

2 shows modified conditions. Since bridge sensors are often predefined sensor arrangements, the bridge itself has been completely retained, while in the following the interest can also relate to the individual branch.

A trigonometric function (here a sine function with an angular frequency to be determined) feeds the "upper node" of the measuring bridge ( 16 ), the tension at the "lower node" ( 10 ) is a cosine function. Both voltages have the (approximately) the same amplitude and the same frequency.

The right one ( 18 )) and left ( 11 ) Branch center are in the arrangement of the 2a , due to the subsequent I / U converter stages, - virtually - to GND potential. The operational amplifiers ( 14 ) and ( 39 ) in 2a adjust their output so that the branch middle nodes ( 11 ) respectively. ( 18 ) are virtually at the potential at the non-inverting input of the OP amps. Strictly speaking, this will not really be zero in the specific case, but the offset voltage of the amplifier. The choice of the feedback resistance allows the specification of a suitable gain (in 2a R6 or R5, 2 B would be changed accordingly as an amplifier stage with v> 1).

Since the center tap of the bridge is GND in this case, the current flowing into the node (left node ( 11 ) considered) I = (U 1 / R 1 ) sin (wt) - (U 2 / R 2 ) cos (wt) = (R 1 R 2 ) -1 SQRT [(U 1 2 R 2 2 + U 2 2 R 1 2 )] sin [wt - arctan ((U 2 R 1 ) / (U 1 R 2 ))] (Herein, SQRT [...] denotes the expression √ ..., "the root of ....").

Alternatively, the center taps of the bridge can be connected directly to a secondary amplifier as a voltage source (in the arrangement of 2 B the OP-Amp stages are designed as simple high-resistance buffer amplifiers (impedance converters) for the voltages at the center taps; these can be as Amplifier stage with v> 1).

The (simple) center tap ( 38 ) is the adjusting voltage U = U 2 cos (wt) + R 2 [U 1 sin (wt) - U 2 cos (wt)] / (R 1 + R 2 ) = [1 / (R 1 + R 2 )] SQRT [(U 2 R 1 ) 2 + (R 2 U 1 ) 2 ] sin [wt + arctan ([U 2 R 1 ] / [R 2 U 1 ])]

The phase expression at the at the branch center tapped voltage shows (apart from the sign) the same relationship as before when looking at the currents that flow into the nodes located on virtual GND was seen (only the sign of the phase changes; a fact that may be useful in some applications).

As mentioned, the sensitivity of the phase position as a function of a change in the resistance R1 should be used to compare the arrangements:
It is because of ∂φ = ∂ / ∂R 1 {Arctan ((U 2 R 1 ) / (U 1 R 2 ))] DR 1 = {(U 1 R 2 U 2 ) / [(U 1 R 2 ) 2 + (U 2 R 1 ) 2 ]} DR 1

If you set U ₁ = U ₂ = U and R ₂ = R ₁ = R again, then this applies to both a current and a voltage tap ∂φ = ∂R / (2R)

The same applies to the right side of the bridge, shown here only for the current tap: I = U 1 / R 3 sin (wt) - U 2 / R 4 cos (wt) = (R 3 R 4 ) -1 SQRT [(U 1 2 R 4 2 + U 2 2 R 3 2 )] sin [wt - arctan ((U 2 R 3 ) / (U 1 R 4 ))]

Because this side here as constant is viewed (R3 = R4 = const), the sensitivity analysis is unnecessary. (Especially since the result is already known; the resistance designations are only to be exchanged).

Looking at - expanding this representation - the phase shifts (depending on the resistance values) between the two sine functions at the node taps ( 11 ) and (18), one finds (again valid for both current and voltage tapping) under the assumption just made that here the right summand is constant or that it is independent of R ₁ and that here again U ₁ = U ₂ = U, R ₂ = R ₁ = R also applies to the sensitivity of this phase difference to a change in R ₁ ∂Δφ = ∂R / (2R) applies.

The conditions of the AC-excited measuring circuits in 2 are thus those of the measuring circuit in l same if you concentrate on a phase measurement; The tuning condition is R ₄ R ₁ - R ₂ R ₃ = 0 in all cases and the estimate of the maximum bridge sensitivity is the same in all cases (except for the factor U _mess / 4).

The advantage of the circuits according to 2 is that there is an AC excitation of the bridge. A small AC voltage is easier to amplify because slow DC voltage drifts - for example an offset drift of the operational amplifiers used - can easily be separated by means of RC elements and then no longer play a role. Bridge voltages can still be detected in the μV range, even if the offset voltages of the operational amplifiers should drift in the mV range.

On the other hand, with arousal the arrangement with pure sine functions through the use of filters with low bandwidth the noise can be kept low because the noise voltage of such an arrangement from the bandwidth of the transmitted Signals dependent is.

The generated by such stages additional With the same structure, phase shifts increase in the phase difference again on each other.

First of all it shows that also a measuring bridge with excitation with sinusoidal signals, which are at the excitation nodes by a certain phase amount differ, the phase shift of the signals at (and between) the Brückenzweigabgriffen for one Measured value acquisition can be suitable.

Other arrangements for generating a parameter-dependent phase position than the bridge arrangements described up to here are possible in many ways in the context seen here:
5 shows the use of any dipole ( 45 ) in a bridge arrangement consisting of an all common impedance ( 45 ) with resistor R ₁ in series and here resistors R ₃ to R ₄ . The sinusoidal excitation of the bridge circuit is as usual with ( 43 ) and ( 44 ) fed. As with the simple bridge, the branch taps ( 47 ) and ( 48 ) Sinusoidal signals, the phase positions of which change in relation to one another as a function of the two-pole circuit, which in turn depends on at least one of the RLC components contained therein. The measurement of the phase position of the two sinusoidal signals ( 47 ) and ( 48 ) then follows your measuring method according to the invention, by scanning, superimposing or mixing.

( 47 ) respectively. ( 48 ) practically represent the signals that are present in the representations before or after the measuring block (e.g. in 14 before and after ( 240 ) respectively. ( 250 )), in 8th before and after ( 62 ), in 9 before and after ( 75 ), in 19 before and after ( 425 ).

6 shows the use of a general four-pole arrangement. A sine signal function is included at the input of the four-pole ( 49 ) and also forms the reference ( 50 ) to the one at the exit ( 52 ) of the four-pole ( 51 ) resulting sine function. The measurement of the phase relationship between these signals provides information about a physical quantity that one of the RLC components of the four-pole ( 51 ) influenced in a known manner. If this four-pole consists, for example, of an RC element with temperature-dependent resistance, then the temperature can be measured in this way.

7 shows an arrangement with which, for example, the phase response difference of one from the generator ( 53 ) generated, and at ( 54 ) emitted US signal (US = ultrasound), but also a UHF signal along a route (here ( 55 ) assumed, for example, as a tube or waveguide) can be measured.

The one here in the middle of the route at ( 54 ) emitted sound is transported to the ends of the route and there ( 56 ) respectively. ( 57 ) recorded and reinforced ( 58 ) respectively. ( 59 ). Depending on the transit time difference of the signal transport on this route, there is a certain phase difference between 60 ) and ( 61 ) recorded and pending signals. Ia the phase position of the signals will be approximately the same at the ends.

An overlapping movement of the medium (air, liquid, solid, etc.), possibly even a movement of the arrangement itself, but creates an additional transit time difference, resulting in a change in phase position between the sinusoidal signals. For example, the medium flows or flows in the channel ( 55 ), this results in an additional phase difference between the signals depending on the flow velocity of the medium ( 60 ) and ( 61 ).

Using the described parametric gain a measurement is satisfactorily accurate even if it is light or e.g. UHF signal can be used. Remarkably results there is a phase shift between the ends even when moving the measuring arrangement, even if electromagnetic waves or light be used.

If the signal transit times directly for the formation of a phase shift (e.g. with a route or Distance measurement) should be used when using of light or electromagnetic radiation due to the extreme Only set very small phase shifts at high speed. (The two signals arrive practically simultaneously).

Using the ones shown above parametric gain are such small phase shifts with relatively simple ones Techniques measurable.

Let the frequency used for a measurement be 50MHz, the difference frequency 5Hz. Then the gain factor is 10 ⁷ . Thus, a 1ns phase shift in the HF now becomes 10ms in the "sampling beat"! If you evaluate this 10ms with a resolution of 0.02us (50MHz as a normal TTL generator clock), this results in a distinguishability of 1/50000 in the original phase time of 20ns, ie 0.04ps (assuming the same resolution, but based on the 1ns of the phase shift in the original signal), which corresponds to the time in which light covers the distance of approx. 0.3μm.

Even if there are some "edge effects" to consider and to be mastered, which really hardly allow such a factor, is this a previously unachievable and quite as "extreme" reinforcement to be designated or Resolution!

In the beat evaluation described above it is assumed that two sinusoidal functions for the purpose of parametric Phase time gain overlaid become; that happened twice. The phase relationship of the two beats to each other provides information about the present phase difference between the separately available sine functions.

When scanning with an adjacent one Frequency and reference and measurement signal were sampled and generated so two LF sinusoidal signals in their phase relationship the phase position the HF reflected

When (multiplicative) mixing the The resulting sum of the two signals is the NF component isolate and evaluate.

With these methods now stand practically the possibilities according to the invention to disposal. Extremely accurate phase measurements can can be realized with the methods described.

Already knowing about the existing method for phase measurement initiated with such accuracy i.a. new design principles of sensor arrangements, of which some using the example of measuring bridges should be displayed.

3 shows a normal measuring bridge again. The excitation principle is again the sine excitation on the upper branch and the cosine excitation on the lower branch. The phase position of the sinusoidal signals at the center taps of the branches is known here (seen individually or in their difference from one another), as described above. The removal of the signal at the center taps can still be done according to the 2 by means of appropriate (reinforcing) OP amp stages in idle or short circuit. This additional component is not shown here (not only, but above all) for simplification.

In this arrangement the 3a a sampling of these values is carried out with a sample rate lying close to the excitation frequency. For the control ( 29 ) the scanner ( 30 ) respectively. ( 31 ) a sine function is specified, but this can be seen here as a digital function with a phase position that corresponds to the sine function. This corresponds to the use of phase measurement by subsampling.

Again, the goal can be achieved by using mixer levels instead of the scanner ( 3b ). This is recommended if very high frequencies are used. The low-frequency signals ( 40 ) respectively. ( 39 ) result after mixing ( 36 ) respectively. ( 37 ) and filtering ( 38 ).

This approach is particularly interesting when, for example, effects on very small structures at extremely high frequencies have to be evaluated (including when developing fast sensors for gases, if the effects themselves are in the order of the wavelength of the frequency used). 9 shows the use of such a mixture and only then the "normal frequency stage". In 9 Above is shown how such a measuring bridge is to be introduced into the arrangement.

The applications implemented here The sampling techniques used are usually sample hold stages or bucket chain circuits with a limited frequency range. However, a mixture is preferably only used if such high frequencies are used that these scanning techniques can no longer be realized with simple technology.

As shown above, arise at (34) and (35) two signals with the difference frequency df and df / 2, the phase relationship of which depends on the phase relationship of the signals dependent on the center taps is, but in their measure of time Represent multiples of the phase difference time of the HF. Further down it is shown how the phase relationship of such beat signals to each other easy to evaluate when using digital processing techniques are and especially the problem of accurate phase measurement also between the beat signals then shows itself as no longer existing.

In the end, however, with everyone Applications described here always the phase difference of the RF signals measured at the center taps with very high accuracy.

A comparison of the sensitivity of all the bridge arrangements presented here shows that they are comparable (apart from a factor of the order of magnitude 2 ... 4 or U _err / 2). Regardless of whether you use a conventional bridge arrangement with conventional further processing or amplification, or an arrangement of the type described here in context, the sensitivity of the basic parameter, here the phase time, there the tapped voltage value, is roughly the same compared to the primary sensor size.

An advantage in the arrangement according to the invention is that the resulting measurand i.a. not from the Excitation voltage amplitude dependent is; however, a certain minimum signal strength must be ensured his.

Or in other words: a difference in sensitivity in the methods is not available. If instead of the evaluation the voltage difference tapped on a (DC-excited) bridge the evaluation of the phase difference is used, then lie comparable sensitivities. The crucial question is which of the two sizes (voltage or current or phase) during further processing with greater accuracy can be recorded and evaluated better and more precisely and with which Effort that happens in each case.

A current or voltage measurement is now easily possible (ie using common techniques) with an accuracy of approx. 10 ^–6 . Given the accuracies of a phase measurement of better than 10 ^-8 that can be achieved with the new methods according to the invention, there are some reasons for such a phase measurement.

Figure 18 shows the principle of a realized evaluation option for an arrangement 7 : G ₁ ( 277 ) and G ₂ ( 276 ) are the two generators for generating the two, here very closely adjacent frequencies f and f + df.

One of the two signal frequencies, here it is (chosen arbitrarily) fs = f + df is applied to the US encoder, e.g. a piezo transducer ( 278 ) given and thus in both directions "on the route ( 275 ) and ( 274 ) sent ". This signal is sent to the receivers ( 279 ) or (280) a phase shift dependent on the length of this distance.

Here, for example, a pipe ( 271 ) assumed, in which the signal can spread and through which any medium flows. When this medium moves at speed v, the speed of sound changes differently due to superimposition in both directions (once increased by v, once decreased). 279 ) respectively. ( 280 ) received and (in ( 272 ) respectively. ( 273 ) amplified) signals therefore show a phase shift dependent on v.

The US signal frequency is thus available three times; once before sending (exit of the Generator G ₂ ( 276 )) and twice after passing through the measurement section in both directions.

These signals are now with the second frequency f _{g of} the generator G ₁ ( 277 ) in the summers ( 269 ), ( 268 ) and ( 267 ) overlaid (again, sampling and mixing are equally possible here), ie 283 ), ( 282 ) and ( 281 ) three beat signals shown in ( 261 ), ( 262 ) and ( 265 ), which in their mutual phase shift reflect the phase shift that has occurred in the US HF when crossing the route, plus a device-related delay to each other and to the starting point. These phase shifts in the beat represent the parametrically amplified transit time difference of the US signal frequency. This is now much easier and, moreover, can be measured with high precision.

The beat ( 265 ) with its phase position "before transmission" practically represents the reference for the other beats. The phase shift ( 270 ) results from the running time for the sound radiation in the direction of the movement of the medium, the phase shift ( 266 ) results from the difference between the two speeds of sound. Since the technically determined phase shifts of the pickups on the right and left (including amplification arrangement etc.) will be approximately the same, these effects cancel each other out in the phase difference; ie: if the velocity v = 0, then the phase difference ( 266 ) Be zero. This allows the measuring arrangement to be easily calibrated.

However, a few “boundary conditions” in such simple-looking arrangements have to be observed or can either be fulfilled by the arrangement or modified in the description:
The complete beatings so far tacitly assumed in the illustrations ( 261 ), ( 262 ) or ( 265 ) are normally never given because when adding up in ( 268 ) or ( 267 ) generally cannot work with signals of the same amplitude. By bridging the route, the passing signal is damped. That from the amplifiers ( 273 ) or ( 272 ) the signal given will be smaller (possibly larger) than that of ( 278 ) activated. Even after "suitable" amplification, these two levels will generally be different.

But that's not the beat Completely, the zero crossing "smeared" so to speak and there are therefore corresponding zero crossing detectors or detection strategies to develop.

The impact of this effect is through 15 shown. On the left is the full beat ( 226 ), which can be found at ( 269 ) in 18 can be realized relatively easily because the two signals from the generators ( 277 ) and ( 276 ) can be generated with the same amplitude. The zero crossing ( 228 ) of the beat signal in 15 left is sharp.

In the 15 on the right is the image of a superimposition of signals with different amplitudes. After your reception, the signals are available with an uncertain amplitude. If the phase measurement is to be determined, for example, by the position of the minima in the two beatings, as in 18 is shown, that would be the signal of the

15 on the right associated with some difficulties. This can be mastered better with the following methods.

If you are looking for other options for generating a parametric phase-time gain, you should first look for suitable other signals. 16 shows, for example, a signal processing which is the same as the beat formation if only digital signals are used for this purpose.

16 shows again for the comparison above the formation of a complete beat, ie the summed (or subtracted) sine functions at ( 200 ) and ( 201 ) have the same amplitude. In the periods ( 202 ) and ( 219 ) these sine signals are shown with different time scales. In the period ( 202 ) individual vibrations are shown; above the smaller frequency (larger period), below the slightly larger (period difference dT ( 203 )). In the period ( 219 ) individual sine waves are no longer recognizable due to the selected scale.

These two signals are used at ( 200 ) and ( 201 ) an addition level ( 204 ) and as a sum (or as a difference) at ( 205 ) provided. The result is the beat with maximum at ( 222 ). The aim for a phase measurement is to detect the minimum as accurately as possible (( 221 ) Zero crossing) of this beat or of the maximum in order to change the phase position to another beat - as eg in 10 shown - to be able to determine.

Below this beat in 16 Four individual vibration relationships are shown with a position assignment to the beat.

The phase position of the two sine waves in the range ( 206 ) is such that they virtually cancel each other out (provided the same amplitudes); the result is therefore the minimum of the beat at ( 221 ).

Due to the different period of the two added sinusoidal signals, the individual vibrations continuously shift towards each other. ( 207 ) shows an intermediate state with an assignment to the amplitude of the beat increasing there.

The shift between the sinusoidal signals continues, the amplitude of the sum function continues to increase and reaches ( 222 ) by the at ( 208 ) now the same position of the two sine functions the maximum. From now on the amplitude of the beat decreases again and reaches 209 ) darge placed intermediate stage, with decreasing amplitude again the next minimum of the beat.

16 shows (almost) the same behavior of digital signals below.

In the area ( 213 ) respectively. ( 220 ) two digital signals are shown, as is comparable in the areas ( 202 ) respectively. ( 219 ) has been shown with the sine functions. The periods of these digital signals are the same as the periods of the sine signals shown above, so the time difference between the two periods is again dT ( 210 ). In the area ( 220 ) these individual vibrations are again not recognizable due to the time scale.

These two digital signals are now once - at ( 211 ) - to the D input of a D-F1ipFlop (D-FFs) ( 214 ) given the other digital signal ( 212 ) secondly used as a clock for this D-FF. The output of the D-FF is the Q output ( 215 ) and supplies the digital signal shown there (marked by ( 216 ), ( 217 ), ( 218 ) and ( 223 )).

Above this representation of this signal (Q output of the D-FF ( 215 )) the time assignments of individual vibrations for the digital signals are shown, the markings ( 224 ) and ( 225 ) make the assignment to the comparable time coordinates, which are also used above for the sine functions. The respective assignment of the two digital signals to the D-FF is as described on the D-FF ( 214 ) shown: the lower signal is used as a clock input, the upper signal is at the D input of the D-FF.

As can be seen by comparison, the phase positions of the signals in the sections ( 206 ), ( 207 ), ( 208 ) and ( 209 ) the same for the sinusoidal signals as for the digital signals; the digital signal is 1 (high) if the value of the sine function is> 0 and 0 (low) if the sine value is <0.

The phase relationship of the two digital signals at ( 206 ) is such that the positive edge at the clock input ( 212 ) from D-FF (trace to ( 225 ) below, edge time indicated by arrow) just the high signal of the at the D input ( 211 ) lying signal "caught", that is, on the Q output ( 215 ) of the D-FF ( 214 ) will lay.

In the illustration, this is precisely the flank ( 216 ) produce. Due to the shorter period of the lower signal, the pos. Clock edge continues into the high range of the upper signal. ( 207 ) shows an intermediate stage of this phase position; it corresponds to the sine phase position shown above and now generates the state time and signal point ( 223 ) at the exit ( 215 ) of the D-FF.

The edge of the clock instant is continuously delayed due to the smaller period compared to the signal at the D input ( 211 ) of the D-FF ( 214 ). In the area ( 208 ) the phase position of the two digital signals is shown again - comparable to the sinusoidal signals shown above - and this phase position now generates the negative edge ( 217 ) in the signal at the output ( 215 ) of the D-FF. So it goes now with a low signal at the D input - continue.

At ( 218 ) the next positive edge occurs at the output of the D-FF. The digital signal at the output of the D-FF thus corresponds (almost) completely to beat generation with sine functions and allows the formation of a parametric gain in exactly the same way, but now with digital signals.

If you compare the time course of the beat above ( 222 ) with the digital signal of the D-FFs output of the "digital beat" shown below, then you can see:

• 1. The zero crossing of the beat ( 221 ) (caused by the phase position of the sine signals at ( 206 )) corresponds to the edge ( 216 ) in the digital signal (caused by the phase relationship of the digital signals at ( 206 )).
• 2. The maximum of the beat ( 222 ) (caused by the phase position of the sine signals at ( 208 )) corresponds to the edge ( 217 ) in the digital signal (caused by the phase relationship of the digital signals at ( 208 )).
• 3. The next zero crossing of the beat corresponds to the edge ( 218 ) in the digital signal.
• 4. The time in the beat from the zero crossing ( 221 ) until the next zero crossing corresponds to half a period of the envelope in this beat, but a full period in the "digital beat signal". Compared to the analog beat, the temporal translation factor in the digital form is therefore only half as large.

Apart from this factor 2, the digitally generated beat does not differ from that generated with sine functions. This output signal generated from D-FF from two high-frequency digital signals ( 215 ) is therefore referred to below as digital beat.

Also the phenomenon of parametric gain through Undersampling can be digitally reproduced or interpreted in this way: one considers the effect of the edge at the clock input of the D-FF as a "scanning process" of the function value at the D input, the analogy is immediately apparent.

A D-FF ( 214 ) can therefore implement parametric phase time amplification or undersampling in the digital world, just as summing or subtraction levels or sampling can do this in the analog world with sine functions. That implies that the effect of the phase shift of a digital high-frequency signal on the digital beat is the same as in the analog world Comparison and is easy to understand in a realization; (in fact you can show that mathematically again).

The advantage of digital implementation is, however, that with digital signals no problems arise due to different amplitudes of the signals and also a zero crossing through the one form of the edges in the digital beat (here the positive edge ( 216 ) respectively. ( 218 )) is present immediately and directly, the respective maximum through the other form (here the neg. flank ( 217 )).

The operational process of digital Beat formation by the D-FF delivers directly and immediately by the edge times the place of a maximum or minimum equivalent, which in the analog Order only by an additional Detection operation can be determined. (They also examined Possibilities, for beat simulation instead of D-FFs e.g. AND or XOR gates with following Using integration shows such serious disadvantages that a closer Description can be omitted here).

The direct use of digital signals in the measuring section is sometimes, but not always possible. For realization Any application is therefore either the present Must map sine functions digitally (e.g. by comparator or Schmitt trigger functions from the sinusoidal signals with the frequency f + df) or at predefined digital signals, which may be used for the measurement (e.g. in a linear network) as such need sinusoidal signals, still have to win from the digital ones (e.g. through filtering). This depends on the individual solution approach differently.

With this variety of methods developed so far, the task of phase measurement can be tackled. 14 shows the current state of affairs and presents the basic solution structures for measuring the phase delay using RF signals.

In this 14 the basic arrangements for the analog (top) and digital (bottom) form of beat training are shown. For the primary task here, the sensory arrangement is only determined by the blocks ( 240 ) respectively. ( 250 ). The use of a measuring bridge is again the same as in 9 shown to assign such a block.

On the left are (arbitrarily) two of the representations ( 231 ) respectively. ( 241 ) shown to generate adjacent frequencies f and f + df; this could also be specified here simply by specifying two generators. That in each case on the measuring arrangement ( 240 ) respectively. ( 250 ) switched signal can before filtering ( 232 ) respectively. ( 233 ) be present, or so, as shown here, the already filtered. Which frequency is used, i.e. f or f + df can also be freely selected.

For an arrangement working in the superposition with sine functions ( 14 above) is the signal used for the superimposition, which "could be deformed" at various points, possibly beforehand for the superimposition in the summation stages ( 234 ) and ( 235 ) (as always, differentiators, sampling or mixing are possible again) beforehand using appropriately narrow filters ( 232 ) and ( 233 ) to bring them back to a pure sinus shape and of course to be strengthened accordingly ( 238 ).

If one uses in this arrangement (instead of the summation levels ( 234 ) and ( 235 )) the described alternative possibility of subsampling (the totalizers ( 234 ) respectively. ( 235 ) are to be replaced by sampling stages), or a mixture (in this case multiplicative stages with the following filtering should be used instead), then the filtering shown may possibly also be omitted.

The outputs of the stages ( 234 ) and ( 235 ) provide the two parametrically amplified, in terms of their phase relationship to each other, but generally not complete beat or superposition signals or the signals obtained by undersampling or by mixing. In general, this requires a preparation ( 237 ) necessary. How this should look depends on the general conditions of the respective application and is only to be determined on a case-by-case basis.

In general, a demodulation of the beat or superposition signal will be useful in the case of heterodyne signals, filtering in the case of a mixture, and then the conversion of the low-frequency signal obtained into a digital form (for example, Schmitt trigger stages). Since digital evaluation is preferred here, this stage ( 237 ) not investigated further; this functionality of the block (only by the diode symbol and by the hysteresis (e.g. of a Schmitt trigger) 237 ) indicated.

This stage ( 237 ) is, however, the one with an incomplete heterodyne signal (cf. 15 right) would have to solve the problems of accurate minimum detection. And this will not even happen with the digital solution.

14 shows the same arrangement below as before, only now assuming digital beat formation. As shown above when comparing the digital and analog beat formers, there is a D-FF (instead of the summation levels) 244 ) respectively. ( 245 ) for use. If it is necessary or inevitable, the measurement on the channel ( 250 ) with sinusoidal signals, then the sinusoidal signals at the ends of the line must be converted to a digital form beforehand (or vice versa, the sensor measurement signals can be obtained from the digital signal specification, for example by filtering). This should be modified on a case-by-case basis.

(Here technical limits become visible, because the D-FFs used and all other levels na must be able to process the frequencies used. If the frequencies are too high, only the method of mixing shown above remains).

In itself, the is also interchangeable Signal origin for Clock signal and D input of the D-FFs, at least if this is an exact one 1: 1 pulse pause duration of the pulse-ratio to have. If this is not the case, you can use a D-FF signal processing reduced to half the frequency enforce this 1: 1 ratio. Since the D-FF only responds to the clock edge, this ratio is sufficient for the signal to request that feed the D-entrance.

Here in the 14 Below is the frequency generation stage ( 241 ) in which the RF signal is already available digitally for use in switch control (cf. 13 and description). The associated (sine) signal shows such a 1: 1 ratio and is therefore also used for D control ( 243 ) of the D_FFs ( 244 ) used.

The measurement signal connected to the measurement section is therefore only used in relation to the transit time property of a signal edge for transit time or phase measurement (this therefore corresponds more to the digital sampling mentioned above); a fact that has proven to be a very useful property for the measurement arrangement or evaluation in most applications. The filtering shown here is therefore often ( 242 ) is no longer necessary to form pure sine signals.

The last thing that remains is the evaluation ( 239 ) respectively. ( 249 ) in the 14 to describe.

17 shows the simple, digital processing of the phase information between the two beat signals which is preferably used here by means of a start-stop counting process by means of the corresponding signals and the digital logic required for this.

The signals ( 252 ) and ( 253 ) (this designation is used both for the management of the logic at the top right, as well as for the signal representation on the left) show the digital beat signals as they are used by the D-FFs of the 14 ( 244 ) respectively. ( 245 ) are delivered or as they result from the beat signals after processing in ( 237 ) be available. Since the phase information is in the edge distances of the digital beat signals, these distances ( 17 , ( 255 ) respectively. ( 256 )) to eat.

The XOR gate ( 258 ) generates a high signal from these signals in the event of inequality ( 251 ) whenever, even if signal ( 253 ) is high. (The goal of the measurement here is the distance 255 capture; ( 256 ) can be replaced by exchanging the corresponding signals ( 252 ) and ( 253 ) be measured). If a high-frequency clock at ( 254 ), then in the time of the distance ( 255 ) this clock at the output of the AND gate combination ( 257 ) at ( 236 ) provided.

The number of these cycles thus provides a digital value for the phase position between the beat signals and can be processed as desired. As with the in 17 shown signals, for example, direct microprocessor control is possible at any time, these signals represent an ideal interface to a processor or computer stage, the construction of which will depend on the processor, microcontroller or DSP (digital signal processor) used.

19 shows the realization of a bridge universal evaluation using parametric amplification with digital processing in a microcontroller control.

The prerequisite for this configuration of the arrangement is that the microcontroller ( 441 ) can provide two digital-to-analog converter (DAC) channels and that a counter function with an external clock input (input for the line ( 409 )) is available. The DAC outputs generate the two signals b ( 449 ) and 2a ( 448 ) in the area ( 450 ) are available both positively and negatively through simple inverting amplifiers.

By scanning in ( 97 ) these voltages on the virtual GND nodes ( 99 ) and form after filtering ( 429 ) a frequency that is close to the frequency given by the scan sequence period. The sampling sequence itself is determined by the frequency of the generator ( 405 ) and the control level ( 442 ) to the switch block ( 97 ) fixed; the sampling sequence frequency here is ¼ of the generator frequency, due to the four switches to be controlled one after the other. This "working frequency" stands at ( 443 ) to disposal; after filtering, the adjacent frequency is in ( 429 ) as a measurement signal.

This signal is used in the measurement ( 425 ) used and thus experiences in ( 425 ) - eg an arrangement for measuring the transit time - a phase shift. Between the signals at ( 446 ) and (447) the phase shift to be measured now exists.

(Would ( 425 ) a measuring bridge, then for example from ( 429 ) to generate a measuring signal from a 90 ° phase shift beforehand, which would be used together to excite the bridge. Instead of the signals ( 446 ) and ( 447 ) the second taps of the bridge - prepared accordingly - would have to be used in this case.

The counter function of the DSP must be able to process the clock speed of the generator G ₂ (400); if this generator frequency is too high for the microcontroller, then the line ( 409 ) between AND gate ( 424 ) and controller input a fast counter for reduction can be switched and this counter reading should also be readable by the controller via a free port.

In some cases the generator G ₂ ( 405 ) can save completely. Modern con trollers often operate at frequencies well above 2MHz; the associated clock management is shaded on the processor with a quartz ( 428 ), itself part of the controller functions and can generally be coupled outwards. In some applications, this frequency is sufficient and can therefore completely replace the G ₂ generator.

In the order of the 19 is this generator frequency so high that the quartz clock of the processor ( 441 ) is not sufficient for the generation of the measuring frequency, but the generator frequency ( 405 ) from the external input of an internal timer and from the control stage ( 442 ) can still be processed (with a processor with a quartz ( 428 ) or working frequency of e.g. 16MHz and a G ₂ frequency of e.g. 100MHz).

The levels in this 10 With ( 406 ), ( 407 ), ( 408 ), ( 423 ) and ( 424 ) or the signals ( 409 ), ( 430 ) correspond to the signals of the representation of the 17 , The signal ( 409 ) in 19 corresponds to the signal ( 236 ) in 17 and returns the phase shift duration in the form of a count.

The signal ( 430 ) corresponds to the signal ( 253 ); the falling edge triggers an IRQ (interrupt request). At the point in time when this interrupt is triggered, the counting process (running in the DSP) has been completed, all settling processes have come to a standstill and the internal count value of the counter (possibly together with the value of the mentioned external, additionally at ( 409 ) is switched on as a digital value of the phase measurement.

How to 17 for controlling the D-FFs ( 407 ) and ( 406 ) three signals necessary, here ( 443 ), ( 444 ) and ( 445 ). ( 446 ) and ( 447 ) are the signals before and after passing through the measurement section ( 425 ), here the frequency f + df (as output of the filter ( 429 )).

This frequency f + df (here approx. 25 MHz) is determined by the arrangement ( 450 ) and the DAC values supplied by the microcontroller (outputs labeled 2a ( 448 ) and b ( 449 )) in connection with a scanning mechanism, which in one of ( 442 ) fixed order, these voltages across the resistors ( 98 ) scans. The exits 2a ( 448 ) or b ( 449 ) are sine or cosine functions of low frequency (df = 25Hz), which the microcontroller generates by generating a counter value (e.g. also a timer function), an associated lookup table (each counter value - as the address of a memory - becomes two output values, just a sine value and cosine value assigned) and by outputting these values via the DAC outputs at ( 448 ) and ( 449 ) provides.

The values are also required in their inverted form. For this, the inverting amplifiers are at ( 450 ) intended; however, if two additional DAC channels are free in the controller, the inverting amplifiers can also be saved.

The ADC values are sampled by the resistors ( 98 ) which, when the switch is closed (in ( 97 ), Switch 1 ... 4 ) a current proportional to the voltage in the virtual GND nodes ( 99 ), which is converted back into voltage by the following OP-Amp stage.

The electronic filter ( 429 ) is only necessary if for the measurement on the measuring section ( 425 ) a sine function is absolutely necessary. The signals that the (inverting, equipped with a Schmitt trigger function) gates ( 427 ) and ( 431 ) are the signals (reconstructed in the same way) that occur once before and once after passing through the measuring section ( 425 ) are tapped and which after this run compared to the (input signal ( 446 ) respectively. ( 445 )) with a phase difference (to be measured here).

This phase difference (measured via the digital beat function) results in the phase difference measured value just described, which is detected by the microcontroller. 19 shows that in this arrangement only very few components (apart from the rather complex controller or a DSP) are necessary. The effort described above for the development illustration is therefore only really necessary in the rarest of cases.

• – Wegsensor (US-Wandler am Anfang und Ende eines Rohres)
• – Distanzmessung mit UHF zwischen Funkpartnern
• – Wegsensor (LED und Photodiode am Anfang und Ende einer Strecke)
• – Temperatursensor (RC-Glied mit temperaturabhängigem Widerstand, nach Frequenzuntersetzung mit D-FFs um den Faktor 2)
• – Drucksensoren (RC-Glied mit druckabhängigem Widerstand)
• – Drucksensoren (mit gegebener Brückenanordnung)
• – Stromabhängige Brücke zur Messung sehr kleiner Ströme

u.a.mWith the in 10 shown arrangement (as controller ( 441 ) is used in DSP) the following sensors succeed almost without additional measures instead of ( 425 ) to use:

• - displacement sensor (US converter at the beginning and end of a pipe)
• - Distance measurement with UHF between radio partners
• - displacement sensor (LED and photodiode at the beginning and end of a section)
• - Temperature sensor (RC element with temperature-dependent resistance, after frequency reduction with D-FFs by a factor of 2)
• - Pressure sensors (RC element with pressure-dependent resistance)
• - pressure sensors (with given bridge arrangement)
• - Current-dependent bridge for measuring very small currents

etc.

A certain universality is thus achieved. If all the components described that are required for the highly precise phase measurement described are combined in a block, then there is a block configuration as in 23 indicated before. Very little configuration modifications are necessary to enable other applications.

If one covers measuring bridges in the application with one, then four external pins are required to connect the bridge. Connections for power supply, Display elements, control and communication connections are regardless of that pins that are necessary or are used for ease of use.

The same pin arrangement can therefore be used as a bridge amplifier ( 23a ) or in runtime measurements ( 23b ) or in four-pole arrangements ( 23c ) are used. Only the pins for the power supply and an interface for the transfer of measured values (e.g. an SPI interface) are essential, so that such a construction can already be accommodated in an 8-pin IC housing.

Claims (19)

Method and device for measuring a phase shift or a phase delay of periodic signals, or for measuring the mutual phase position of periodic signals with a relatively high frequency, hereinafter referred to as measurement signal or measurement frequency, characterized in that the phase position between the Signals of the high frequency (measurement signals) are processed by a technique, referred to here as "parametric amplification", in such a way that the small phase times or phase positions of the high frequency are transformed into a much larger phase time or phase position of low-frequency signals in which the Phase times or phase positions can be determined much more easily and precisely, this “parametric amplification” consisting of a combination of technical measures, in that a signal that is as close as possible in frequency to the high frequency of the measurement signal is also referred to here as a reference signal t is generated as a reference frequency and then either - the measurement signals provided with the phase shift are both simultaneously superimposed with the same reference signal (e.g. additive or subtractive) and then the phase relationship of the two resulting superimposition signals is evaluated, or - the measurement signals provided with the phase shift both are sampled with the same reference frequency, so-called undersampling is used and then the phase relationship of the two signals resulting from this undersampling is evaluated with respect to one another, or - the measurement signals provided with the phase shift are both mixed with the same reference signal (e.g. multiplicative) The respective low-frequency component is isolated from the two resulting mixed signals by filtering, and then the phase relationship between these two low-frequency signals is evaluated with respect to one another. Method and device according to claim 1 for measuring the phase position between signals of a very high frequency, characterized in that a cascade-shaped chain of elements of the arrangement 1 is built up, each cascade working as an arrangement according to claim 1 - preferably using mixed signal formation with low-pass filtering - that is, using the "parametric amplification" from the two higher measurement signal frequencies with the phase position to be measured, each two significantly lower-frequency components ( or intermediate frequency components) with a significantly longer phase time, which can be used in the next cascade stage as higher-frequency measurement signals and thus the phase times in each stage are further enlarged (amplified or transformed), the phase evaluation between the resulting, constantly lower-frequency signals only in the last stage. Procedure and device according to claim 1 and 2 for measuring the phase relationship between signals relatively high or very high frequency characterized in that for the realization of the parametric amplification (with a cascading at the lowest level of the phase time evaluation) rectangular or digital signals are used so that the overlay or subsampling using F1ipFlops (preferably using D-F1ipFlops) in the form of a "digital Beat ", for example by emitting a signal of frequency f to control the D input, the other signal of the frequency f + df to control the clock input (or frequency-wise too vice versa) is used, these digital signals may first from sinusoidal Signals (e.g. through Schmitt trigger stages). Procedure and device according to claim 1 to 3 for measuring the phase position between signals relatively high or very high frequency characterized in that to simulate an overlay or a mixture of simple gate functions (XOR, AND, etc.). Procedure and device for generating closely adjacent frequency for use according to claim 1 to 4 characterized in that the generation of sinusoidal signals with closely adjacent frequency from given or generated high-frequency sine and cosine signals, when implemented according to claim 2 also from one of the intermediate frequencies created during mixing (ZF) (from the e.g. by 90 ° degree Phase shifter techniques won the sine and cosine components can be) and from simultaneously predetermined, low-frequency sine and cosine signals by replicating one of the formulas of the addition theorems for trigonometric Functions. Procedure and device for generating closely adjacent frequency for use according to claim 1 to 4 characterized in that the two signals are adjacent Frequency from a common clock frequency source through reduction with the factor n or (n + k) with a counter. Method and device for generating closely adjacent frequency for use according to claim 1 to 4, characterized in that a first counter generates the frequency f / n from a high (generator) frequency f, a second counter in each period of this reduced frequency f / n a comparison value in a second counter is changed by the value 1 and a comparator uses this comparison value to compare the values of the counter 1 and the counter 2 generate an edge which, compared to the edges of the signal with frequency f / n, has the period of a frequency f / (n + 1). Procedure and device for generating closely adjacent frequency for use according to claim 1 to 4 characterized in that for the generation of signals with closely adjacent frequencies required low-frequency sine and cosine signals by a counter or a counter function, also e.g. can be replicated in a microprocessor in that these counter values used to generate the NF signals, e.g. by being as Address for a look-up table is used in which the values of the required sine and cosine signals are stored digitally, these values one after the other and read out periodically and then by means of analog-digital conversion (ADC) the NF signals required according to the method be generated Procedure and device for generating closely adjacent frequency for use according to claim 1 to 4 characterized in that for the generation of signals with closely adjacent frequencies, higher frequencies required by PLL generators are generated, either two NF generators with different Frequencies are used as reference generators or, at same reference frequencies by different divisions of the VCO generator frequencies are included in the phase comparison, or by taking both measures be combined. Procedure and device according to claim 1 to 4 for sensory physical detection Parameter characterized in that the measurement over the depend on the parameter to be measured Frequency response of an electronic network takes place, which - preferred but not necessarily exclusively - from passive elements (R-, L, C components), the evaluation using the Measurement of the phase shift for exactly one signal frequency occurs because one used for this purpose sinusoidal or rectangular Measurement signal when passing through the network a reproducible and experiences specific phase shift, and thus from the physical Parameter dependent Phase shift at the output of a network relative to the phase position of the signal at the input or at another intermediate stage of the network is recorded. Procedure and device according to claim 1 to 4 and claim 10 for sensory detection physical parameters characterized in that the used Measurement signals from non-sinusoidal, e.g. digital or square wave signals, e.g. only by filtering are obtained, these then sinusoidal signals before and after the passage through the measurement network e.g. through a Schmitt trigger function be converted back into digital signals, and in this form the phase position between the signals before and after the run through the network or a stretch in a digital beat be evaluated. AC excited bridge arrangement using the claims 1 to 4 and 10 to 11, characterized in that a knot of bridge excited with a signal (preferably a sinusoidal signal) of frequency f the opposite Node with a phase-shifted signal (preferably a cosine signal) the same frequency, is excited and at the branch centers the voltages there are sampled at the frequency f + df or with a second signal of frequency f + df (additive or subtractive) overlaid, rectified and smoothed be, which creates two low-frequency signals, their phase relationship to each other is evaluated as the measuring signal of the measuring bridge. AC excited bridge arrangement using the claims 1 to 4 and 10 to 12, characterized in that a knot of bridge excited with a signal (preferably a sinusoidal signal) of frequency f the opposite Node with a phase-shifted signal (preferably a cosine signal) the same frequency, is excited and at the branch centers the voltages there are multiplied by the frequency f + df mixed and then be filtered with a low-pass filter, creating two low-frequency Signals are generated whose phase relationship to each other is the measurement signal of the measuring bridge is evaluated. AC excited bridge arrangement using the claims 1 to 4 and 10 to 13 characterized in that previously the voltages lying at the branch center node with an impedance converter (Voltage follower) can be tapped or by a current-voltage conversion, where the branch center taps with your I-U converter virtually open GND are drawn. AC excited bridge arrangement using the claims 1 to 4 and 10 to 14 characterized in that the for measurement used sinusoidal Signals through suitable amplification and e.g. digitized using a Schmitt trigger function and then the two measurement signals to the D input and the reference frequency on the two clock input from two D-F1ipFlops, for the symmetrization of the signals possibly also after one, again with D-FFs generated reduction by a factor of 2 and so the two D-FFs digitally simulate a beat. Procedure and device according to claim 1 to 14 for the sensory detection of distances or path differences using ultrasound, light or electromagnetic Waves, characterized in that by a corresponding geometric receiver arrangement the signal of a transmitter is recorded by at least two signal receivers and the possibly existing signal delay differences between the received signals can now be measured with high precision Can express phase shift between the respective measurement signals and a conclusion on location, on the transport speed of a medium (e.g. at Using ultrasonic wind speed), on a flowing or allow media speed. Structure of a hybrid arrangement or an integrated circuit (e.g. as an FPGA or ASIC) for use as a universal bridge amplifier or as a general sensor amplifier using claims 1 to 16, characterized in that this amplifier has at least 4 pins, one of which is a (in frequency Possibly freely selectable) high-frequency sine signal, a second one supplies a high-frequency cosine signal (e.g. from the sine signal generated by a 90 ° phase shift technique) that can be used for a network to be connected, e.g. a bridge arrangement, and two pins that serve as inputs for evaluation the phase relationship of two signals applied to these pins can be used according to claims 1 to 4 and 10-16, with the at least two pins for power supply and for example two pins for a communication channel (for example SPI or I ² C) an 8-pin housing a complete, universal sensor amplifier is given. Evaluation of the phase position between two digital beat signals in an application using according to claim 1 to 17 thereby characterized by between two digital beat signals a simple XOR link the period during which the two signals are recorded Distinguish their position from one another, moreover not necessarily possibly one side (i.e. either the proportion after a positive or a negative edge in one of the two signals) Use of an AND gate for selected the evaluation and then for the duration of a different location determined in this way, e.g. again by an AND function, a cycle to a counting function (Counter or a microprocessor function) and through which itself resulting count value a direct digital value at each end of a beat period for the Phase relationship between the two digital beat signals and therefore for the phase relationship of the RF signals used in the measurement is obtained is. Procedure and device for sensory detection of a physical parameter using a simple RC, RL or LC element in which either R, L or C depends on the physical parameter characterized by that the phase response of this RLC arrangement for data acquisition under Use of the method according to claim 1 to 18 is evaluated.