308 IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS, VOL. 27, NO.

3, MARCH 2017

Dual Frequency Millimeter-Wave Illumination

of a Detector Diode
Robert A. Smith, Senior Member, IEEE, and Zahra A. Pour, Senior Member, IEEE

Abstract— In this letter, a GaAs beam lead diode is illuminated

by two fixed W-band millimeter-wave frequencies to assess the
difference frequency response at the X-band. Measured results
are compared with an expected mixed frequency response,
utilizing a Volterra series expansion solution, which considers
the reactive component influences under nonconjugate matching
conditions. A nonlinear model is assessed with the solution
approach to predict potential performance. A nonconjugate
matched condition provides good agreement with less than 2-dB
variation across the central slice of the response.
Index Terms— Harmonic analysis, millimeter-wave measure- Fig. 1. Experimental setup using Millitech W -band 6-in GOAs in the
ments, nonconjugate matching, nonlinear detection. transmitter system driven by an Agilent 8257 signal generator and Narda
X-band antenna and Agilent E4440A spectrum analyzer in the receive system
to form a test bed for millimeter-wave experiments.
I. I NTRODUCTION

M UCH of the published work on the nonlinear response

of electromagnetically illuminated objects focused on
single-frequency harmonics below the microwave region and
considered large scattering objects [1]–[3]. Multiple fre-
quency excitations results were consistent with harmonic
and intermodulation product techniques under ideal analysis
assumptions in [4] and [5] that included conjugate matching
conditions. More recently, studies have been performed for
radio frequency identification and remote sensing applications
using dual frequency illumination and frequency modulation
to excite harmonic responses [6], [7]. For many tracking Fig. 2. Scattering object that shows the embedded mixer diode in the
waveguide with an aperture area of a × b. The internal waveguide surface is
applications, existing approaches are not practical due to the copper. The waveguide top surface and the panel with the depth parameter
size of the antenna and limited read ranges. Millimeter-wave are shown partially transparent to aid in viewing the mixer diode location.
approaches can scale antennas to smaller sizes and a frequency
mixing approach provides a method to obtain a tracking signal data from the spectrum analyzer and control movement of an
that is not obscured by a clutter of the transmission signal and automated positioning system that located the scattering object
potentially detectable at longer ranges. in discrete reference plane positions. The system transmitted
This experiment illuminates a scattering object with dual y-directed linearly polarized continuous wave output.
millimeter frequencies at 77 and 87 GHz to assess the The two 15.24-cm diameter Gaussian optical
10-GHz difference frequency response. Section II details the antennas (GOAs) were aligned so that the beams overlapped
experimental approach. Section III explains the analytical at the reference plane that was sized to capture the span of
methodology, which is based on a realistic model of the the Gaussian beams. The output power of each transmitter
nonlinear scattering object, with corresponding results. It is was 14.9 dBm for the 77-GHz system and 14.7 dBm for the
followed by some concluding remarks. 87-GHz system. The GOA diameter establishes the far field
for the 77-GHz system at 11.9 m and the 87-GHz system at
II. E XPERIMENTAL A PPROACH 13.5 m. The transmitters are 2.74 m from the reference plane,
which set the reference plane in the near-field region. The
The experimental setup is shown in Fig. 1, where a laptop automated positioning system moved the scattering object
computer was used to enable the transmit system, record through the 20-cm width and 15-cm high reference plane
Manuscript received September 7, 2016; accepted November 16, 2016. Date in increments of 5 and 10 mm, respectively. The antennas
of publication February 15, 2017; date of current version March 9, 2017. did not change positions as the scattering object was moved
The authors are with the Department of Electrical and Computer Engineer- through the reference plane.
ing, University of Alabama in Huntsville, Huntsville, AL 35899 USA (e-mail:
[email protected]; [email protected]). The scattering object shown in Fig. 2 contains an MA40415
Digital Object Identifier 10.1109/LMWC.2017.2661703 beam lead GaAs Schottky mixer diode [8]. The conductive
1531-1309 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
SMITH AND POUR: DUAL FREQUENCY MILLIMETER-WAVE ILLUMINATION OF A DETECTOR DIODE 309

Fig. 4. Circuit representation of detector diode [9], where Rs and

C D are the diode series resistance and junction capacitance, respectively,
C p and L are the package capacitance and lead inductance, respectively, and
K NLp are the Volterra series expansion elements of the junction resistance.
Z A is the antenna impedance calculated to be 79.7 + j60.1  at 77 GHz and
110.9 + j143.2  at 87 GHz.

Fig. 3. Measured 10-GHz power responses from the scattering object. The the nonlinear elements. IOSC is used to calculate R j that is
peak value is −110.9 dBm at the center of the Gaussian profile. Responses used in the calculations of K NLp .
below −125 dBm represent the noise floor of the spectrum analyzer.
In (1), the n parameter of the term V D(n) represents an
order and not a power term. For example, (2) represents the
traces beyond the beam lead were cut to isolate a 2-mm second-order difference frequency diode voltage associated
dipole structure with an embedded nonlinear component from with the second term of (1), where H1 (x) is the first-order
the detector circuit in the sacrificed device that provided the transfer function of the circuit in Fig. 4. The viable difference
waveguide structure. The aperture size was selected to propa- frequency terms are generated at even orders and have contri-
gate the 10-GHz difference frequency signal in the waveguide. bution from artifacts in the first-order terms. For brevity, the
For the air-filled rectangular waveguide shown in Fig. 2, the transfer functions and the order terms are not provided
cutoff frequency of the TE10 mode is 9.49 GHz.
α 2 V D(2) α 4 V D(4)
The measured difference frequency power response from i D = Iosc αV D + Iosc + Iosc
the mixer diode at the reference plane is shown in Fig. 3 with 2! 4!
a maximum value of −110.9 dBm. In subsequent plots, the α 6 V D(6) α 8 V D(8)
+ Iosc + Iosc (1)
referenced central slice contains data from the 0-cm vertical 6! 8!
∗ E
E 77
measurement position across the horizontal axis. (2) 87 ∗
VD = H1 ( f 77 )H1( f 87 )e j 2π( f 87− f77 )t . (2)
2
The transfer function effective nonlinear factors are
III. A NALYTICAL M ETHODOLOGY AND R ESULTS (2) (4)
5.9−3 − j 2.8−3 for V D and two terms for V D which are
(6) (8)
In this section, the circuit-model of the nonlinear scattering 21.4−6 − 10.1−6 and 69.3−6 − j 32.7−6. The V D and V D
object is presented and the analytical approach is described. nonlinear factors are at least two orders of magnitude smaller
The analytical results are compared with the measured results. and the incident voltage multipliers in the higher order transfer
Fig. 4 shows a circuit model of the GaAs beam lead detector functions do not overcome the small nonlinear factors. There-
diode, where the component values and package parameters fore, i D is set by the first three terms of (1) which are 1.2,
are defined in [8] and [9]. The antenna impedance, Z A , is 0.84, and 0.44 mA, respectively.
calculated from finite-length dipole antenna equations for the The calculation of the incident electric fields, E 77 and E 87 ,
W -band incident fields and includes the Ohmic loss resistance, at the waveguide aperture requires near-field solutions for the
R L , of the dipole structure [10]. The linear resistance, K NL1 , electric fields using electric and magnetic vector potentials
and nonlinear resistance elements, K NLp , are derived using a and paraxial conditions. Since the GOA aperture is not in a
Volterra Series expansion [11]; however, under initial excita- ground plane, we assume electric current density, Js , and a
tion, the nonlinear resistance is the junction resistance, R j , magnetic current density, Ms , exist on the aperture surface
of the diode as defined in [8]. The remaining identifiers in and are zero outside of the aperture [12]. For brevity, the
Fig. 4 are useful for defining the node voltage equations for the resulting equations are not provided. In the near-field analysis,
circuit. For the 10-GHz response signal, the antenna elements power was calculated at the reference grid horizontal and
of the model are represented by the small dipole impedance vertical positions and validated with measurements at each grid
equations since the response wavelength at 10 GHz is 15 times location using a W -band horn antenna. The maximum electric
the dipole length. field and input powers at the W -band horn antenna aperture
Equation (1) represents the Volterra series expansion of was calculated to be 1.03 V/m and 1.4 mW for the 77-GHz
the diode current equation into the nonlinear resistance ele- system and 0.87 V/m and 1 mW for the 87-GHz system.
ments [11], where IOSC is the initial oscillator drive current, The incident mixer diode power is the power incident at
V D is the voltage across the diode, and α = 34.9 V−1 for the waveguide aperture times the waveguide effective aperture.
a GaAs diode. The value of IOSC is calculated from the Equation (3) defines the waveguide gain used to determine
average power incident on the mixer diode divided by the the effective aperture, where a and b are the aperture dimen-
incident voltage across the mixer diode without considering sions, λ is the appropriate incident wavelength, η is the free
310 IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS, VOL. 27, NO. 3, MARCH 2017

the series resistance provides a lower bound result that aligns

well with the measured results.
The results that align best to the measured results are
highly dependent on the capacitance values and the inclusion
of inductance in the model. Without the package features
of inductance and capacitance, the results predict received
power that is well above the measured results. If the package
capacitance is not set to the low tolerance then the results
are higher by up to 20 dB. As shown in Fig. 5, the junction
capacitance of 0.075 pF in conjunction with the package
capacitance of 0.01 pF aligns well to the measured data and
the adjustment of the series resistance provides fine-tuning.

IV. C ONCLUSION
The experiment has measured a difference frequency
Fig. 5. Comparison of measured difference frequency power response to cal- response from a dual frequency illumination scattering object.
culated results using Table I diode parameters cases. The central peak values An analytical approach using a Volterra series expansion with
of the curves are as follows. Measured: −110.9 dBm. Case 1: −124.1 dBm. transfer model representations of the diode equivalent circuit
Case 2: −121.3 dBm. Case 3: −109 dBm; Case 4: −110.9 dBm.
shows good agreement with the measured results. An impor-
TABLE I tant aspect of the analytical approach is the consideration
D IODE PARAMETER C ASE S TUDY D EFINITIONS [8], [9] on nonconjugate matching for power transfer relationships at
the nonlinear component. The experiment demonstrates that
the reactive components of the equivalent circuit for a diode
model must be included in the analytical model in order
to successfully predict reasonable results at millimeter-wave
frequencies.

R EFERENCES
[1] R. O. Harger, “Harmonic radar systems for near-ground in-foliage
nonlinear scatterers,” IEEE Trans. Aerosp. Electron. Syst., vol. AES-12,
no. 2, pp. 230–245, Mar. 1976.
space impedance, and ηTE is the transverse-electric waveguide [2] A. A. Gorbachev, S. V. Lartsov, S. P. Tarakankov, and E. P. Chigin,
impedance at the appropriate incident frequency [13] “Amplitude characteristics of nonlinear scatterers,” J. Commun. Technol.
Electron., vol. 41, no. 6, pp. 515–519, 1996.
32 ab ηTE [3] S. A. Novikov and D. V. Zelentsov, “Radar study of nonlinear scat-
GW G = . (3) tering in L-band microwave range by semiconductor objects,” in Proc.
π λ2 η SIBCONVERS, 1999, pp. 403–405.
[4] T. Sarkar and D. Weiner, “Scattering analysis of nonlinearly loaded
A nonconjugate matched power transfer analysis of the antennas,” IEEE Trans. Antennas Propag., vol. AP-24, no. 2,
circuit shown in Fig. 4 is performed using the calculated power pp. 125–131, Mar. 1976.
at the mixer diode to obtain the incident E 77 and E 87 fields [5] M. A. Flemming, F. H. Mullins, and A. W. D. Watson, “Har-
monic radar detection systems,” in Proc. IEE Int. Conf. Radar, 1977,
used in the transfer functions and diode voltage terms in (2) pp. 552–554.
and the higher order diode voltage terms. Next, the diode [6] C. Mandel, C. Schuster, B. Kubina, M. Schüßler, and R. Jakoby, “Dual
current at the intermodulation frequency can be determined frequency selective multiple access with quasi-chipless/powerless RFID
mixer tags,” IEEE Microw. Wireless Compon. Lett., vol. 24, no. 8,
from (1), where IOSC is calculated from the relationship of pp. 572–574, Aug. 2014.
the power at the diode to the incident electric field. The [7] G. J. Mazzaro, A. F. Martone, and D. M. McNamara, “Detection of RF
current from (1) creates a radiated field at 10 GHz from the electronics by multitone harmonic radar,” IEEE Trans. Aerosp. Electron.
Syst., vol. 50, no. 1, pp. 477–490, Jan. 2013.
small dipole. Since the receive antenna is in the far-field of [8] M/A-COM, Lowell, MA, USA. GaAs Schottky Mixer Diodes
the radiating dipole, the scattered power is translated to the MA40400 Series Datasheet V3.00, accessed on Aug. 2016. [Online].
receiver system using the radar range equation. Available: http://pdf.datasheet.technology/datasheets-1/m_a-
com_technology_solutions/MA40415.pdf
Fig. 5 shows the comparison of the analytical results against [9] M/A-COM, Lowell, MA, USA. AG314 Principles, Application and
Fig. 3 measured data across the central peak of the response. Selection of Receiving Diodes, accessed on Aug. 2016. [Online]. Avail-
The nominal diode parameters are C D = 0.065 + 0.01 pF able: http://cdn.macom.com/applicationnotes/AG314.pdf
[10] C. A. Balanis, “Linear wire antennas,” in Antenna Theory: Analysis and
and Rs = 4.5 + 1.5  as defined in [8] and L = 0.1 nH Design, 3rd ed. New York, NY, USA: Wiley, 2005, ch. 4, secs. 4.3–4.5,
and C p = 0.02 + 0.01 pF as defined in [9]. No tolerance pp. 162–179.
was provided for the inductance. Table I shows the diode [11] J. J. Bussgang, L. Ehrman, and J. W. Graham, “Analysis of non-
linear systems with multiple inputs,” Proc. IEEE, vol. 62, no. 8,
parameters for four cases that bound the tolerance bands and pp. 1088–1119, Aug. 1974.
lead to a best-fit solution. Case 3 shows that the maximum [12] C. A. Balanis, “Aperture antennas,” in Antenna Theory: Analysis and
junction capacitance and the minimum package capacitance, Design, 3rd ed. New York, NY, USA: Wiley, 2005, ch. 12, sec. 12.2,
pp. 653–660.
along with nominal values for the remaining parameters, [13] S. J. Orfanidis, (Mar. 3, 2016). Electromagnetic Waves and Antennas.
provide a reasonable comparison. Case 4 shows that lowering [Online]. Available: http://www.ece.rutgers.edu/~orfanidi/ewa