This application claims priority to foreign French patent application No. FR 2313794, filed on Dec. 7, 2023, the disclosure of which is incorporated by reference in its entirety.
The invention lies in the field of electronic instrumentation. It relates more particularly to a method and a device for measuring a complex impedance of an electrical element.
The concept of complex impedance generalizes that of resistance for sinusoidal signals at a given frequency f. In the case of an electric dipole, the complex impedance Z is defined by
where U is the phasor (complex number) representing the amplitude and phase of the voltage across the terminals of the dipole and I is the phasor representing the amplitude and phase of the current flowing through it. More generally, in the case of an N-port circuit (the dipole corresponding to the case N=1), it is possible to define an impedance
In other words, the impedance Zij is the (complex) ratio between the phasor representing the voltage across the terminals of the port “i” and the phasor representing the current entering (or exiting, depending on the convention adopted) the port “j” when the current entering all other ports is zero. The various terms Zij form the impedance matrix of the multiport element. Hereinafter, the term impedance and the symbol “Z” will be used to designate both the impedance of a dipole and a term Zij of the impedance matrix of a multiport.
Being a complex number, the impedance Z may be broken down into a complex part and an imaginary part −Z=R+jX, where “j” here denotes the imaginary unit—or, in terms of modulus and phase: Z=|Z|ejφ, where |Z| is the ratio between the RMS values of voltage and current and φ is their phase offset.
Generally speaking, impedance varies with the frequency of the electrical signals under consideration. To characterize an electrical element, it is therefore necessary to measure its (or their) impedance(s) in a more or less wide frequency band. Z (f), |Z(f)| and φ(f) are therefore written to designate, respectively, a complex impedance, its modulus and its phase as a function of the frequency f.
Many techniques have been developed to measure the phase of an impedance, φ(f), as a function of frequency.
(Angrisani 2001) discloses a measurement method in which a resistor of known value is connected in series to the element to be characterized and an excitation sinusoidal signal is applied to said element through this resistor. The impedance of the element to be characterized is able to be determined based on the measurement of the voltage u (t) of the excitation signal and the voltage v (t) of a node located between the known resistor and the element to be characterized. More particularly, two methods are proposed for determining the phase of said impedance:
Either the zero crossings of the signals u (t) and v (t) are detected, after having filtered these two signals by way of finite impulse response predictive filters in order to limit the impact of noise on zero detection;
Or the phase offset of the two signals is calculated based on their internal product and their mean squared value.
(Schröder 2004) also determines the phase of a complex impedance by measuring the phase offset between two voltage signals. This phase offset measurement may be carried out by detecting zero crossings of said signals, or else through analytical calculation based on amplitude and phase parameters of said signals, determined by interpolation.
The solutions proposed by (Schröder 2004) and (Angrisani 2001) implement a phase measurement based on a threshold comparator or zero crossing that requires fast measuring electronics, this not being compatible for example with a microcontroller-based solution, but rather requiring an ASIC or an FPGA. In addition, the solution in (Angrisani 2001) requires implementing a specific power sensor and carrying out relatively complex calculations.
The invention aims to overcome at least some of the aforementioned drawbacks of the prior art. More specifically, it aims to make it possible to measure the phase of the impedance of an electrical element in a particularly simple way, implementing a device of low complexity, which may be based notably on a microcontroller or an FPGA.
According to the invention, this aim is achieved by way of a method in which the phase of the impedance of an electrical element is determined by digital processing based on a first signal representative of a voltage between two terminals of the electrical element and a second signal representative of a current through the electrical element. The digital processing involves generating a phase-offset replica of the first or the second signal. Ideally, the phase offset Φ of the replica should be 90° (or, equivalently, π/2 rad), this being able to be achieved easily at a given frequency, but requiring a complex implementation if it is desired to be able to perform a frequency scan to determine φ(f) over a spectral band of interest having a significant width (for example, a bandwidth greater than or equal to 10% of the center frequency of the band). The invention circumvents this difficulty by using a fixed delay instead of a constant phase offset. This delay corresponds to a phase offset Φ of 90° only for a frequency f0 belonging to the band of interest, and to a phase offset of 90°+ΔΦ(f) for frequencies other than f0. This leads to an error in the measurement of φ(f)|f≠f
One subject of the invention is therefore a method for measuring the phase of a complex impedance of an electrical element, comprising the following steps:
According to particular embodiments of such a method:
Steps a) to h) may be repeated a plurality of times for a plurality of frequencies fex within a spectral band, the time offset introduced in step f) being constant and equal to one quarter of a period corresponding to a frequency included in said spectral band.
Said spectral band may have a relative width Δf/fm, where Δf is the difference between the highest and the lowest frequency of the band and fm is its average frequency, greater than or equal to 10%.
Step h) may comprise:
As a variant, step h) may comprise:
In step d), the first and the second analog signal may be sampled and converted to digital format at the same rate.
Another subject of the invention is a device for measuring the phase of a complex impedance of an electrical element, comprising:
According to some particular embodiments:
The device may also comprise a generator for generating an oscillating excitation signal having an oscillation frequency that is able to be varied in controlled fashion within a spectral band, said delay line being configured to introduce a constant time offset equal to one quarter of a period corresponding to a frequency included in said spectral band.
Said spectral band may have a relative width Δf/fm, where Δf is the difference between the highest and the lowest frequency of the band and fm is its average frequency, greater than or equal to 10%.
The digital circuit may be configured to:
As a variant, the digital circuit may be configured to:
The device may also comprise a clock configured to clock said first and second analog-to-digital converters at the same rate of acquisition and conversion of said first and second analog signals.
Other features, details and advantages of the invention will become apparent on reading the description given with reference to the appended drawings, which are given by way of example, and in which, respectively:
In
The device of
The digital circuit CN comprises a delay line LR that generates a replica Ûi of the digital signal Ui, delayed by a known delay T. In the embodiment of
Therefore, for an excitation frequency
the phase offset is equal to 90° (or π/2 rad).
Consideration is given to an excitation signal at the frequency fex=f0. The two digitized signals Uv and Ui may be written as:
As explained above, for fex=f0, the delay line LR introduces a phase offset of π/2 rad. Therefore:
The digital circuit CN therefore calculates the product of the first digital signal Ui and the delayed replica of the second digital signal, Ûi, so as to generate a third digital signal Mn. It also calculates the product of the first digital signal Ui and the second digital signal (without a phase offset) Ui so as to generate a fourth digital signal Md:
The third and the fourth digital signal both comprise a component that oscillates at the frequency 2f0 (terms in 2ω0t) and a DC component. These signals are filtered by digital low-pass filters FPB1, FPB2 in order to recover the DC components. The filtered signals are expressed by (ignoring the amplitude factor ½):
An estimate {circumflex over (φ)}1 of the value of φ1 is obtained by calculating the arctangent of the ratio of the filtered third digital signal and the filtered fourth digital signal:
When fex≠f0, however, {circumflex over (φ)}1 is no longer a good estimate of φ1, and therefore of the phase of the complex impedance of EL, because the phase offset introduced by the delay line LR is no longer equal to 90°.
For example, consideration is given to a case where fh=100 MHZ (sampling period of TH=10 ns) and the frequency band of interest is between 8 MHz and 9 MHz, sampled with a step of 1 MHz. In this frequency range, a delay corresponding to a phase offset of 90° would be from 31.25 ns to 27.77 ns. It is then chosen to implement the delay line LR by way of three flip-flops clocked at the frequency fh of 100 MHz, therefore introducing a constant delay of 30 ns. This delay corresponds to a phase offset of 90° for an excitation signal at 8.33 MHz.
For signals at a frequency fex≠f0=8.33 MHz, the phase offset Φ assumes a value other than 90°. It is said that Φ(fex)=90°+ΔΦ, where ΔΦ is the phase offset error.
With a generic phase offset Φ=90°+ΔΦ (or, in radians, Φ=π/2+ΔΦ), the delayed replica of the second digital signal may be written
Applying the same method as for the case of Φ=90° gives:
It is possible to write {circumflex over (φ)}1=φ1+Δφ1, where Δφ1 is an estimation error that depends both on the initial estimate {circumflex over (φ)}1 and on the frequency fex. For example,
However, the above equation makes it possible to calculate this estimation error, and therefore to correct it by adding a corrective term: φ1={circumflex over (φ)}1+Φcmp with Φcmp=−Δφ1. Specifically, the spectral band of interest is discretized into N−1 intervals F0=[f0; f1), F1=[f0; f1), . . . . FN−1=[fN−1; fN]; similarly, the angular range (−180°; 180°] is discretized into M−1 intervals (Φ0=({circumflex over (φ)}10=−180°; {circumflex over (φ)}11], . . . ΦM−1({circumflex over (φ)}1M−1; {circumflex over (φ)}1M]. A discrete value of the corrective term Φcmp(i,j) is calculated for each pair (Fi, Φj) so as to form a double-entry lookup table LUTDE. Each time a new phase {circumflex over (φ)}1 is estimated, the digital circuit CN identifies the interval Φj containing it, as well as the interval Fi containing the excitation frequency fex, extracts the corresponding corrective term Φcmp(i,j) from the lookup table and adds it to {circumflex over (φ)}1 in order to find a best estimate {circumflex over (φ)} of the phase of the complex impedance of the element EL.
To reduce the memory occupancy of the double-entry lookup table LUTDE, it is possible to replace it with two single-entry lookup tables, as in the embodiment of
The addition of a first corrective term equal to −sin(ΔΦ) originating from a first lookup table LUTA makes it possible to obtain a first intermediate value equal to tan(φ1)cos(ΔΦ). The latter is multiplied by a second corrective term equal to
originating from a second lookup table LUTB, so as to obtain a second intermediate value equal to tan(φ1). The calculation of the arctangent of this second intermediate value provides the expected estimate {circumflex over (φ)} of the phase of the complex impedance of the element EL.
The two corrective terms depend solely on ΔΦ, which in turn is solely a function of the excitation frequency fex. Consequently, the two lookup tables LUTA and LUTB may have a single input (in other words, vectors of values) and receive, at input, a value representative of said frequency.
This solution is not preferred due to the need for an additional analog-to-digital converter and a PLL, thereby increasing its complexity, its cost and its power consumption. Additional complexity stems from the fact that the converters ADC2 and ADC3 have to have very close performance in terms of linearity and gain, otherwise significant errors will be introduced. Moreover, the PLL that generates sh′ requires a relatively long time (several tens of μs) to stabilize on a new phase, thereby slowing down the rate of acquisition of the measurements.
Regardless of the embodiment under consideration, the digital circuit CN, CN′ may comprise a microprocessor, in which case some or all of the functionalities of the circuit are implemented in the form of software, or logic circuits based for example on an FPGA. In particular, the “delay line” LR may be implemented using flip-flops or else be emulated by software instructions. The lookup tables LUTDE, LUTA, LUTB may be stored in dedicated memory devices or in specific locations in a single memory. It will also be noted that the arctangent may be calculated by way of a lookup table.
The invention has been described with reference to particular embodiments, but variants are possible. For example:
The delay line LR is used to generate a delayed replica of the first digital signal Uv, rather than a delayed replica of the second digital signal Ui as described above.
The digital-to-analog converters ADC1, ADC2 might not be clocked by one and the same clock signal, provided that resynchronization is carried out during the digital processing.
The discretization of the spectral band of interest and/or that of the angular range (−180°; 180°] for implementing the one or more lookup tables might not be uniform in order to minimize the maximum residual error after applying the one or more corrective terms.
It is also possible to modify the frequency of the clock signal sh by way of a PLL as a function of the frequency fex so as to reduce the phase offset error of the replica Ûi. This makes it possible to simplify the correction of an estimation error for the phase of the complex impedance—for example by allowing use of coarser discretization of the spectral band and/or of the angular range (−180°; 180°], and thus by reducing the size of the one or more lookup tables. However, this simplification is offset by the use of a PLL with the associated drawbacks (slowdown, increase in cost and complexity).
Number | Date | Country | Kind |
---|---|---|---|
2313794 | Dec 2023 | FR | national |