Method and a test setup for measuring large-signal S-parameters

Abstract

The values of the parameters of a large-signal S-parameter model of a device-under-test are determined by using a frequency-offset probe-tone method and by measuring and processing the spectral components of the scattered voltage waves with frequencies equal to the fundamental frequency, the fundamental frequency plus the said frequency offset and the fundamental frequency minus the said frequency offset.

Description

FEDERALLY SPONSORED RESEARCH

[0002] Not Applicable

SEQUENCE LISTING OR PROGRAM

[0003] Not Applicable

BACKGROUND OF THE INVENTION

[0004] 1. Field of Invention

[0005] The present invention relates to the measurement of scattering parameters for microwave and radio-frequency (RF) devices-under-test (DUTs) under nonlinear operating conditions.

[0006] 2. Description of the Related Art

[0007] It was shown in J. Verspecht, M. Vanden Bossche and Frans Verbeyst, “Characterizing Components Under Large Signal Excitation: Defining Sensible ‘Large Signal S-Parameters’?!,” 49th ARFTG Conference Digest, pp. 109-117, June 1997 that the nonlinear behaviour of a radio-frequency or microwave device-under-test with a periodic excitation signal at the input in a near matched environment can accurately be described by a ‘Large-Signal S-parameter’ model as defined in the said reference.

[0008] One practical method to measure the said ‘Large-Signal S-parameter’ model for an actual DUT, restricted to fundamental frequency behaviour only, is described in Frans Verbeyst and Jan Verspecht, “Characterizing non-linear behavior,” European Patent Application ‘EP 1 296 149 A1’. With the said method the said ‘Large-Signal S-parameter’ model is identified by connecting to the output port of the DUT a matched load, an open, a short and a plurality of attenuators and delays and by each time measuring the fundamental spectral component of the incident and scattered voltage waves. The ‘Large-Signal S-parameters’ are then calculated by digitally processing the acquired data.

[0009] In Jon S. Martens, “Probe Tone S-parameter Measurements,” United States Patent Application Publication, Pub. No.: ‘US 2002/0196033 A1’, [0006]-[0007] is described how ‘Hot S22’, which is one part of the said complete ‘Large-Signal S-parameter’ model, can be measured by using a frequency-offset probe-tone method. With the said probe-tone method a one-tone excitation signal is applied at the input signal port of the DUT (this is called the “power tone”, its frequency is called the fundamental frequency), while a small one-tone signal with a frequency equal to the “power tone” frequency plus a small frequency offset is applied at the output signal port of the DUT (this signal is called the “probe tone”). ‘Hot S22’ is calculated by taking the ratio between the measured reflected voltage wave at the output signal port having a frequency equal to the fundamental frequency plus the said frequency offset and the measured incident scattered voltage wave at the output signal port having a frequency equal to the fundamental frequency plus the said frequency offset.

BRIEF SUMMARY OF THE INVENTION

[0010] In accordance with the present invention, a method and an apparatus are provided to determine the values of the parameters of a large-signal S-parameter model of a device-under-test by using a frequency-offset probe-tone method and by measuring and processing the spectral components of the scattered voltage waves with frequencies equal to the fundamental frequency, the fundamental frequency plus the said frequency offset and the fundamental frequency minus the said frequency offset. The said frequency offset can have a positive as well as a negative value. With the said probe-tone method a one-tone excitation signal is applied at the input signal port of the DUT (this is called the “power tone”, its frequency is called the fundamental frequency), while a small one-tone signal with a frequency equal to the “power tone” frequency plus a small frequency offset is applied at the output signal port of the DUT (this signal is called the “probe tone”). The small probe tone in the incident voltage wave at the output signal port of the device gives rise to spectral components in the reflected voltage waves at both DUT signal ports which are proportional to the real and imaginary parts of the complex value of the probe tone phasor. These spectral components appear at 2 frequencies: the fundamental frequency plus the said frequency offset and the fundamental frequency minus the said frequency offset. An apparatus measures the complex values of the phasors of the incident and the scattered voltage waves at the three frequencies of interest: the fundamental frequency, the fundamental frequency plus the said frequency offset and the fundamental frequency minus the said frequency offset. The 6 ‘Large-Signal S-parameters’ are calculated using the measured complex values of the said phasor quantities.

BRIEF DESCRIPTION OF THE DRAWINGS

[0011] FIG. 1 shows the naming convention of the incident and the scattered voltage waves

[0012] FIG. 2 shows a schematic test setup in accordance with the present invention

[0013] FIG. 3 shows a typical spectrum of the a1 voltage wave during the operation of the test setup

[0014] FIG. 4 shows a typical spectrum of the a2 voltage wave during the operation of the test setup

[0015] FIG. 5 shows a typical spectrum of the b1 voltage wave during the operation of the test setup

[0016] FIG. 6 shows a typical spectrum of the b2 voltage wave during the operation of the test setup

[0017] FIG. 7 shows an implementation of the invention in an harmonic balance simulator

[0018] FIG. 8 shows the simulated amplitude of S22 and Sacc22 as a function of the amplitude of a1

[0019] FIG. 9 shows the simulated phase of S22 and Sacc22 as a function of the amplitude of a1

DETAILED DESCRIPTION

[0020] A. Test Method

[0021] FIG. 1 depicts a microwave or radio-frequency (RF) device-under-test (DUT) (1) having an input signal port (2) and an output signal port (3). The electromagnetic stimulus and response are described by 4 travelling voltage waves a1, a2, b1 and b2. The voltage waves are defined in a characteristic impedance which is typically 50 Ohm. Voltage wave a1 is incident to the input signal port (2) of the DUT (1), voltage wave a2 is incident to the output signal port (3) of the DUT (1), voltage wave b1 is scattered by the input signal port (2) of the DUT (1) and voltage wave b2 is scattered by the output signal port (3) of the DUT (1). The aforementioned voltage wave definitions are in standard use with a vector network analyzer (VNA). For a linear DUT the relationship between the stimulus (a1, a2) and the response (b1, b2) is described by an S-parameter matrix which can be measured with a standard VNA. When the DUT is an active device, like e.g. a power amplifier, the aforementioned S-parameter description can typically only be used if the input signals are small compared to the operating range of the device. Under large signal stimulus conditions (large compared to the specified operating range of the DUT) the superposition theorem will no longer be valid and S-parameters can no longer be used.

[0022] So called “Large-Signal S-parameters” are used to describe DUT (1) behaviour under large signal operating conditions, as explained in William H. Leighton, Roger J. Chaffin and John G. Webb, “RF Amplifier Design with Large-Signal S-parameters,” IEEE Transactions on Microwave Theory and Techniques, Vol. MTT-21, No. 12, December 1973. It is shown in Jan Verspecht, Marc Vanden Bossche and Frans Verbeyst, “Characterizing Components Under Large Signalning Sensible ‘Large Signal S-Parameters’?!,” 49th ARFTG conference digest, June 1997 that a “Large-Signal S-parameter” description having 4 complex parameters (S11, S12, S21, S22), as it is used for a VNA and also in the aforementioned paper by W. H. Leighton et al., is incomplete and needs to be extended by adding two additional complex parameters, here called Sacc12 and Sacc22. These two additional parameters describe the relationship between the conjugate of the a2 incident voltage wave and the b1 and b2 scattered voltage waves. Sacc12 and Sacc22 are exactly equal to zero under linear operating conditions and are as such not measured by a VNA. The object of the present invention is a novel method and test setup to measure the 6 “Large Signal S-parameters” (S11, S12, S21, S22, Sacc12, Sacc22).

[0023] Consider a one-tone excitation a1 with frequency fc at the input signal port (2), with a large amplitude, and a one-tone excitation a2 having the same frequency fc at the output signal port (3), with a relatively small amplitude. The signal a2 at the output port (3) should be small enough to ensure a linear response of the DUT (1) versus a2. Under such operating conditions, as can be derived from the aforementioned paper by Jan Verspecht et al., the relationship between the response (b1, b2) and the stimulus (a1, a2) is given by:

b 1 (fc)=S11(|a1(fc)|)a1(fc)+S12(|a1(fc)|)a2(fc)+Sacc12(|a1(fc)|)p2a2*(fc)  (1)

b 2 (fc)=S21(|a1(fc)|)a1(fc)+S22(|a1(fc)|)a2(fc)+Sacc22(|a1(fc)|)p2a2*(fc)  (2),

[0024] where

[0025] p=ejφ(a1(fc)),

[0026] φ(x) represents the phase of complex phasor x,

[0027] |x| represents the amplitude of complex phasor x and

[0028] x* stands for the conjugate of phasor x.

[0029] S11, S12, Sacc12, S21, S22 and Sacc22 are the 6 “Large Signal S-parameters”. They are a function of the amplitude of the large signal component a1(fc).

[0030] A method to measure the “Large-Signal S-parameters” S22 and Sacc22 is described in Frans Verbeyst and Jan Verspecht, “Characterizing non-linear behavior,” European Patent Application ‘EP 1 296 149 A1’. With the said method the said “Large-Signal S-parameters” S22 and Sacc22 are measured by connecting to the output signal port (3) of the DUT (1) a matched load, an open, a short and a plurality of attenuators and delays and by each time measuring the fundamental spectral component of the incident and scattered voltage waves. The “Large-Signal S-parameters” S22 and Sacc22 are then calculated by digitally processing the acquired data. In contrast to the present invention, the aforementioned method described in the said European patent application by Frans Verbeyst et al. requires many manipulations which are hard to automate and only determines three “Large-Signal S-parameters”, namely S21, S22 and Sacc22.

[0031] The test setup of the present invention is illustrated in FIG. 2. A signal generator (4), called the “Power Tone Generator” is used to generate the necessary one-tone a1 signal. In contrast with the description in the aforementioned European patent application by Frans Verbeyst et al. one will not with the present invention apply different loads to the output signal port (3) of the DUT (1). Instead one will use a signal generator (5), called the “Probe Tone Generator”, for the generation of an a2 one-tone signal with a frequency slightly offset from fc and equal to fc+fd. In order for the method to be accurate the frequency difference fd, called the offset frequency, should be that small that the S-parameters do not vary across a frequency difference of fd. The frequency difference fd can have a positive as well as a negative value. In practice one typically uses an offset frequency of several kHz. Since fc is a microwave or RF frequency it is typically much higher than 100 MHz, such that the S-parameters do not significantly change for a kHz frequency change. This method is known as an “offset frequency probe tone” method and is described in J. Martens, “Probe Tone S-parameter Measurements,” United States Patent Application Publication, Pub. No. US 2002/0196033 A1, [0006]-[0007], Dec. 26, 2002.

[0032] The resulting spectra of the voltage waves a1, a2, b1 and b2 during one “offset frequency probe tone” experiment are illustrated in FIG. 3, FIG. 4, FIG. 5 and FIG. 6. For a1 (FIG. 3) there is only one spectral component present with a frequency fc and for a2 (FIG. 4) there is only one spectral component present with a frequency fc+fd. For b1 (FIG. 5) and b2 (FIG. 6) there are spectral components present with frequencies fc−fd, fc and fc+fd. Only the presence of the spectral components with frequencies fc and fc+fd is recognized and used in prior art. The existence of b1 and b2 spectral components with a frequency fc−fd is not recognized nor used in prior art, in contradiction with the present invention.

[0033] In the aforementioned patent application publication by J. Martens, the “offset frequency probe tone” method is used for the determination of “Hot S22” only. “Hot S22” corresponds to the “Large-Signal S-parameter” S22. The value of “Hot S22” is found by sensing and measuring the voltage waves b2(fc+fd) and a2(fc+fd) and by calculating the ratio of the two measured quantities:

S 22 =b 2 (fc+fd)/a2(fc+fd)  (3).

[0034] It is not uncommon to measure the values a1(fc), b1(fc) and b2(fc) with a similar setup in order to calculate the “Large-Signal S-parameters” S11 and S21 as illustrated in “Measurement of Large Signal Input Impedance During Load Pull,” Maury Microwave Corporation Application Note 5C-029, November 1998. The presence of the probe tone signal a2(fc+fd) is not necessary for the measurement of S11 and S21, which are calculated by using the following equations:

S 11 =b 1 (fc)/a1(fc)  (4)

and

S 21 =b 2 (fc)/a1(fc)  (5).

[0035] With the present invention one will use an “offset frequency probe tone” method and one will not only measure the components a1(fc), a2(fc+fd), b1(fc), b1(fc+fd), b2(fc) and b2(fc+fd), as is done in the described prior art, but one will include the measurement of the components b1(fc−fd) and b2(fc−fd). The mere existence of these frequency components is not recognized nor used in prior art. With the present invention one will measure the values of a1(fc), a2(fc+fd), b1(fc−fd), b1(fc), b1(fc+fd), b2(fc−fd), b2(fc) and b2(fc+fd) and one will calculate the values of the “Large-Signal S-parameters” S22, S11 and S21 as described in (3), (4) and (5). As will be explained in the following, the values of the “Large-Signal S-parameters” S12, Sacc12 and Sacc22 are given by:

S 12 =b 1 (fc+fd)/a2(fc+fd)  (6)

S acc12 =b 1 (fc−fd)/(p2a2*(fc+fd))  (7)

S acc22 =b 2 (fc−fd)/(p2a2*(fc+fd))  (8)

where

p=ejφ(a1(fc)).

[0036] The above results can be derived as follows.

[0037] First one rewrites equations (1) and (2) using a complex envelope representation for the signals a1, a2, b1 and b2 (using fc as the carrier frequency). Complex envelope representations are mixed time frequency domain representations of signals and are explained in equation (1) of the detailed description of the invention described in U.S. Pat. No. 5,588,142 “Method for simulating a circuit”. In short, X(t) is the complex envelope representation (with carrier frequency fc) of a signal x(t) if the following holds.

x (t)=Real(X(t)ej2πfct)  (9)

[0038] This results in the following equations.

B 1 (t)=S11A1(t)+S12A2(t)+Sacc12A2*(t)P2(t)  (10)

B 2 (t)=S21A1(t)+S22A2(t)+Sacc22A2*(t)P2(t)  (11)

where

P (t)=ejφ(A1(t)).

[0039] During a “frequency offset probe tone’ measurements A1(t) and A2(t) are given by the following equations.

A 1 (t)=K  (12)

A 2 (t)=Lej2πfdt  (13)

[0040] In the above equations K and L are 2 complex constants having a real and an imaginary part. Substituting (12) and (13) in (10) and (11) results in the following equations.

B 1 (t)=S11 K+S12Lej2πfdt+Sacc12L*ej2φ(k)e−j2πfdt  (14)

B 2 (t)=S21K+S22Lej2πfdtSacc22L*ej2φ(K)e−j2πfdt   (15)

[0041] Next equations (14) and (15) are converted into the frequency domain by applying a Fourier transform. The function δ(f) will be introduced. It is defined as being equal to 0 for all f different from 0 and with δ(0) equal to 1. This results in the following.

b 1 (f)=S11Kδ(f−fc)+S12Lδ(f−fc−fd)+Sacc12L*ej2φ(K)δ(f−fc+fd)  (16)

b 2 (f)=S12Kδ(f−fc)+S22Lδ(f−fc−fd)+Sacc22L*ej2φ(K)δ(f−fc+fd)  (17)

[0042] For the described “frequency offset probe tone” method K and L are given by the following.

K=a 1 (fc)  (18)

L=a 2 (fc+fd)  (19)

[0043] Substitution of (18) and (19) in (16) and (17) results in the following.

b 1 (f)=S11a1(fc)δ(f−fc)+S12a2(fc+fd)δ(f−fc−fd)+Sacc12a2*(fc+fd)p2δ(f−fc+fd)  (20)

b 2 (f)=S21a1(fc)δ(f−fc)+S22a2(fc+fd)δ(f−fc−fd)+Sacc22a2*(fc+fd)p2δ(f−fc+fd)  (21)

[0044] Subsequent substitution of f by fc, fc+fd and fc−fd leads to the results described in equations (3), (4), (5), (6), (7) and (8).

[0045] B. Test Setup

[0046] A typical test setup which implements the aforementioned method in practice is described in FIG. 2. A test set (6) is connected to the input signal port (2) of the DUT (1), the output signal port (3) of the DUT (1), a “Power Tone Generator” (4), a “Probe Tone Generator” (5) and a microwave and RF receiver (11). The a1 signal generated by the “Power Tone Generator” (4) is guided by the test set (6) to the input signal port (2) of the DUT (1) and the a2 signal generated by the “Probe Tone Generator” (5) is guided by the test set (6) to the output signal port (3) of the DUT (1). The test set senses the voltage waves a1, a2, b1 and b2 and guides the sensed voltage waves to a microwave and RF receiver (11). Four couplers (7), (8), (9) and (10) are typically used for sensing the voltage waves. Other possibilities for sensing the voltage waves are electrical and magnetic field probes and resistive bridges.

[0047] The output of the receiver (11) are the measured values of a1(fc), a2(fc+fd), b1(fc−fd), b1(fc), b1(fc+fd), b2(fc−fd), b2(fc) and b2(fc+fd). These values are transfered to a signal processor (12) for calibration purposes and for the calculation of the 6 “Large-Signal S-parameters” S11, S12, Sacc12, S21, S22 and Sacc22 according to equations (3), (4), (5), (6), (7) and (8).

[0048] In a typical experiment the amplitude of a1 (fc) and the frequency fc are swept across a range specified by the user.

[0049] In a practical test setup a1(fc−fd), a1(fc+fd), a2(fc−fd) and a2(fc) may be slightly different from zero due to parasitical reflections in the test set. In that case the use of equations (3), (4), (5), (6), (7) and (8) will result in a small error. This error can be corrected by using calibration techniques which make use of calculations based upon the 6 measured quantities a1(fc), a2(fc+fd), a1(fc−fd), a1(fc+fd), a2(fc−fd) and a2(fc).

[0050] To those skilled in the art it is clear that the presented novel method and test setup can easily be extended to the measurement of “Large-Signal S-parameters” for DUTs having more than two signal ports.

[0051] C. Simulation

[0052] The present invention for determining the 6 “Large-Signal S-parameters” of a DUT (1) can also be used in circuit and system simulators. The described “Large-Signal S-parameter” model can be used as a fast and accurate black-box behavioral model which can easily be derived from complex transistor circuits by simulating the test setup as described above. An example of the implementation of the present invention in a commercial harmonic balance simulator is shown in

[0053] FIG. 6. FIG. 7 shows the simulated amplitude of S22 and Sacc22 for a swept a1(fc) amplitude. FIG. 8 shows the simulated phase of S22 and Sacc22 for the same swept a1(fc) amplitude. Note that the amplitude Of Sacc22 tends to zero for a small amplitude of a1(fc). This is what one expects since the linear small-signal S-parameter model is valid under those conditions, which implies a non-existent or in other words zero-valued Sacc22.

[0054] D. Classic Linear S-Parameter Measurements

[0055] Classic linear S-parameters are a subclass of the said “Large-Signal S-parameters”. As such classic linear S-parameters can be measured by using the said “offset frequency probe tone” method on the condition that the amplitude of a1(fc) is kept small enough. The four S-parameters can be derived by using equations (3), (4), (5) and (6). The advantage of measuring classic S-parameters in this novel way is that all four S-parameters are determined with one measurement of the voltage waves, whereas prior art VNAs use a forward and a reverse measurement. It can as such be expected that a properly designed VNA based on the “offset frequency probe tone” method can be significantly faster and can have a simpler test set architecture than existing VNA instruments which are based on the execution of a forward and a reverse measurement which necessitates the use of switching one signal generator once to the input signal port (2) of the DUT (1) and once to the output signal port (3) of the DUT.

Claims

1. A method for characterizing a device under test (DUT) comprising the steps of: providing a first sinusoidal signal having a first frequency to a first signal port of the DUT as a power tone signal providing a second signal having a second frequency to a second signal port of the DUT as a probe tone signal, whereby the difference between the said second frequency and the said first frequency is equal to a third frequency determining the spectral components of the traveling voltage waves which are incident to the said DUT signal ports and the traveling voltage waves which are scattered by the said DUT signal ports and this for a frequency equal to the said first frequency, equal to the said second frequency and equal to the said first frequency minus the said third frequency

2. The method of claim 1, whereby the measured values of the said spectral components are being used for calculating the “large-signal S-parameters” of the DUT

3. The method of claim 1, whereby the said first signal and the said second signal are provided to the said DUT in a circuit simulator and whereby the said determination of the said spectral components is provided by performing a circuit simulation

4. A method for determining the scattering parameters of a device under test (DUT) comprising the steps of: providing a first sinusoidal signal having a first frequency to a first signal port of the DUT as a power tone signal providing a second signal having a second frequency to a second signal port of the DUT as a probe tone signal, whereby the difference between the said second frequency and the said first frequency is equal to a third frequency which is substantially smaller than the said first frequency measuring the spectral components of the traveling voltage waves which are incident to the said DUT signal ports and the traveling voltage waves which are scattered by the said DUT signal ports and this for a frequency equal to the said first frequency and equal to the said second fre-quency using the measured values of the said spectral components for calculating the scattering parameters of the said DUT