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

3, MARCH 2017 245

Six-Port Reflectometer Using Edge-Coupled

Microstrip Couplers
Debapratim Ghosh, Student Member IEEE, and Girish Kumar, Member, IEEE

Abstract— This letter presents the design of a six-port reflec- broadside-coupled stripline couplers for SPR design have been
tometer (SPR), which makes use of edge-coupled, coupled-line reported in [9] and [10] to obtain about 33% bandwidth.
microstrip couplers. By designing the couplers to have coupling An alternate approach makes use of multiple probes along
slightly larger than the isolation, good Q-point placement on the
complex −plane is obtained. The SPR Q-points are obtained by a transmission line where the probes are relatively weakly
a proposed two-step process using two standard terminations—a coupled to the main line. Two to seven probe systems have
matched load and an open circuit load. The designed SPR has been reported in [11]–[13]. The use 10-, 6-, and 3-dB couplers
been tested using various loads and the computed load reflection with tunable phase shifter is reported in [11]. Use of five and
coefficients have been compared against those of a standard seven microstrip and stripline probes have been reported in
commercial Vector Network Analyzer. The raw dynamic range
is 21–39 dB in the frequency range from 1.6 to 2.2 GHz and [12] and [13], respectively, making the designs overdetermined
the uncalibrated load reflection coefficients follow the expected and an adaptive approach is used to select three Q-points
values closely, with average phase error spread of only about best separated. A two-detector system is reported in [14],
2.7%–7.7%. The designed SPR is fabricated on FR4 substrate. which uses 3-dB dividers and delay lines, with the proposed
It achieves a reduction in size due to lesser number of components calibration to reduce error. To summarize, most reported
and is ideal for narrowband and low cost use.
techniques have emphasized the use of directional couplers
Index Terms— Coupled-line couplers, edge-coupled lines, with tight 3-dB coupling. These designs have made use of
network analyzer, power divider, six-port reflectometer (SPR). at least four to five such components. Alternatively, the use
I. I NTRODUCTION of multiple probes has also been reported, with the systems
relying on calibration and adaptive approaches to obtain low-
S IX-PORT reflectometers (SPRs) offer an alternative tech-
nique over conventional heterodyne schemes to measure
the complex reflection coefficient of a load ( L ) by measur-
error  L solution.
In this letter, a hybrid approach is proposed, which makes
ing only the power magnitudes at defined points in a six- use of edge-coupled microstrip couplers and 3-dB Wilkinson
port network [1]. The solution for the reflection coefficient divider. The detectors are arranged with coupled probes along
 = a/b at a particular frequency is obtained by the common a main line, and the probe couplers are designed for loose
intersection point(s) of three circles centered at Q-points on coupling and low isolation on a lossy substrate to get desired
the complex −plane, and radii determined by the load. magnitudes and good phase separation of Q-points. It is
The early SPRs used five hybrid waveguide couplers [2] and shown that without any calibration on the proposed SPR, good
this configuration has formed a basis for many subsequently dynamic range of 21–39 dB is obtained and the measured
reported designs. The SPR in [3] showed that minimal error magnitude and phase of the complex load reflection coefficient
distribution of Q-points is obtained using 3-dB waveguide  L follows the expected values closely. The proposed design is
couplers. An SPR using four 180° hybrids was reported in [4], shown to offer a good compromise between size, cost, dynamic
which obtained 27% bandwidth; however, the performance range, bandwidth, and error.
of the SPR over the desired band is not reported. An SPR
using a 6-dB coupler and four 3-dB 90° couplers was reported II. P ROPOSED S IX -P ORT D ESIGN U SING E DGE -C OUPLED
in [5], which shows a strong effect of coupler directivity on M ICROSTRIP C OUPLERS
measurement of  L . SPR configurations using four modified
180° and 90° 3-dB couplers have been reported in [6] and [7], In the proposed SPR, Ports 5 and 6 are for the load Device
respectively, showing improved bandwidth, but drawbacks are Under Test (DUT) and the RF input, respectively. The other
the use of adjustable terminations and increased size. Four 90° ports are for power detection with Port 2 as a reference
3-dB coupler-based SPR with additional delay lines has been detector. Though not an exact approximation, in terms of the
proposed in [8] with 20% bandwidth, but the in-band perfor- S-parameters of the SPR, the i th Q-point (i = 1, 3, 4) using
mance of the SPR is not reported. The use of 3- and 4.8-dB the expressions in [1] can be modified as Q i ≈ −Si6 /Si5 S56 .
A probe coupler is shown in Fig. 1(a). It is designed √ for
Manuscript received September 13, 2016; revised November 22, 2016; the wave from Port 6 to couple to the detector 2 times
accepted December 15, 2016. Date of publication February 13, 2017; date
of current version March 9, 2017. stronger than the wave√from Port 5, i.e., |Q i | ≈ 2 is obtained
The authors are with the Department of Electrical Engineering, IIT Bombay, (assuming |S56 | = 1/ 2). Thus, low isolation of the coupler
Mumbai 400076, India (e-mail: [email protected]; [email protected]). is obtained using a lossy FR4 substrate. Si6 and Si5 of a
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. single 16-dB edge-coupled microstrip coupler probe designed
Digital Object Identifier 10.1109/LMWC.2017.2661708 for 1.6 GHz shown in Fig. 1(b) indicate that 1.8 ≤ |Q i | ≤ 2.82
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.
246 IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS, VOL. 27, NO. 3, MARCH 2017

Fig. 4. (a) Linear magnitudes and angular (phase) separation of the three
Fig. 1. (a) Waves at the ith port of an SPR through a coupler and (b) Si6 ,
Si5 , and Sii of a 16-dB edge-coupled microstrip probe coupler with λ/4 at Q i of the SPR and (b) uncalibrated  L (dB and phase) of microstrip 0-,
1.6 GHz. 10-k, and 51- terminations measured by the SPR.

the SPR at its i th port, respectively [1]

b i = A i a + Bi b (1)

where Ai and Bi are the complex coefficients of the SPR

(Q i = −Bi /Ai ), and a and b, respectively, are the reflected
and incident waves at the load on Port 5 (ultimately,  L = a/b
is to be measured). Since Port 2 is used as a reference port, b2
must be independent of the load to obtain A2 → 0 ⇒ Q 2 → ∞.
In the first step to obtaining the Q i , a 50- matched termina-
tion of the Agilent 85033E kit [15] is connected on Port 5.
A good assumption for the matched load, as verified by
measurement, is that it has a  b. Equation (1) then simplifies
Fig. 2. Configuration of the proposed SPR using coupled microstrip couplers. to bi = Bi b. For a 50- load, Si6 and S56 of the above
SPR indicate bi and b, respectively. Thus, the Bi values are
obtained as
Si6
Bi = . (2)
S56
The Ai values are obtained by connecting an open termination
of the 85033E to Port 5 (alternatively, a short termination
Fig. 3. Photograph of the SPR (without detectors) fabricated on FR4 may be used, but it has greater parasitics than an open load
substrate. beyond 1.5 GHz). The actual open load reflection coefficients
√ (O = aO /bO ) as a function of frequency have been used
(for |S56 | = 1/ 2) from 1.6 to 2.2 GHz (conventional 3-dB to obtain Ai . When the open load is connected, it follows
couplers have coupling six to seven times the isolation, which from (1) that bO = S26 O /B (S O indicates the S-parameter
2 i6
gives very large |Q i |). The probe at the i th port has return loss when Port 5 has an open load). Once bO is known, the other
better than 20 dB, which ensures minimal power flow back to O are measured. The A values are then obtained in
three Si6 i
the main line. The angular separation between the three Q- terms of all known, measured quantities, including Bi from (2)
points is obtained by placing three identical 16-dB couplers by rearranging (1), as
along the main line. The complete SPR geometry is shown  
in Fig. 2. The reference detector at Port 2 has to be isolated O
1 Si6
from Port 5 for the reference power to be largely independent Ai = − Bi . (3)
O b O
of the load. This is achieved using a 3-dB Wilkinson power
divider designed at 1.9 GHz with isolation above 20 dB from From (2) and (3), the Q i values are obtained as Q i = −Bi /Ai .
1.6 to 2.2 √GHz. In addition, the 3-dB power divider ensures This approach is valid only if the Port 6 RF source is
|S56 | ≈ 1/ 2, which justifies the initial assumption. well matched. Fig. 4(a) shows the magnitudes and angular
separations, respectively, of Q 1 , Q 3 , and Q 4 in the frequency
III. S OLUTION FOR THE SPR Q-P OINTS range from 1.6 to 2.2 GHz; |Q i | ≥ 2 for nearly the entire
The SPR has been fabricated on glass-epoxy FR4 substrate frequency range. In addition, Fig. 4(a) also shows that the
(r = 4.4, h = 0.8 mm, and tan δ = 0.02), and Fig. 3 shows |Q i | values are all nearly equally spaced from each other
the photograph. Since it is not possible to physically measure by nearly 120° from 1.75 to 1.9 GHz. The minimum phase
Si5 for an arbitrary load at Port 5, the Q-points of the SPR Q i separation between any two |Q i | is better than 50° from
(i = 1 . . . 4) are obtained through a proposed two-step process, 1.6 to 2.2 GHz, which ensures less measurement error, since
by making use of two standard terminations: a 50  and an any lower separation would increase measurement inaccuracy
open load. If ai and bi denote the waves entering and leaving and error [1].
GHOSH AND KUMAR: SPR USING EDGE-COUPLED MICROSTRIP COUPLERS 247

V. C ONCLUSION
In this letter, an SPR using edge-coupled microstrip couplers
is proposed. The design comprises of a hybrid approach using
loose coupling and low isolation of microstrip couplers on
a lossy substrate and a 3-dB Wilkinson divider, to obtain
good Q-point placement and to isolate the reference power.
The procedure to obtain Q-points of the SPR is proposed
as a two-step process using two standard terminations. The
uncalibrated, measured  L for various loads using this SPR
is compared against those of a commercial VNA, and the
magnitude and phase results are found to be in close agree-
ment from 1.6 to 2.2 GHz, with raw load dynamic range of
21–39 dB. The SPR makes use of fewer components and is
smaller in size over those reported in [4] and [10]. Compared
with the multiprobe techniques in [12] and [13], this SPR is
not overdetermined, and does not require adaptive selection
of Q-points for a given frequency. The SPR achieves 32%
bandwidth, better than those reported in [4], [5], and [8].
This design is scalable to any frequency and is suitable for
Fig. 5. Uncalibrated  L measured by the SPR, compared with
Agilent E5071C VNA measurements for (a) 36 , (b) 100 , (c) 1 pF, narrowband applications, such as antenna, dielectric, and soil
and (d) printed monopole antenna. moisture testing [16].
R EFERENCES
IV. M EASURED R ESULTS
[1] G. F. Engen, “The six-port reflectometer: An alternative network
After obtaining Q i , the designed SPR was tested for its analyzer,” IEEE Trans. Microw. Theory Techn., vol. 25, no. 12,
ability to distinguish matched and mismatched loads and pp. 1075–1080, Dec. 1977.
[2] G. F. Engen, “An improved circuit for implementing the six-port
phases. A small 50- microstrip line terminated with: 1) a technique of microwave measurements,” IEEE Trans. Microw. Theory
0- load; 2) a 10-k load, and 3) a 51- load was connected Techn., vol. 25, no. 12, pp. 1080–1083, Dec. 1977.
to Port 5 of the SPR. The power at ports 1−4 was measured [3] E. J. Griffin, “Six-port reflectometer circuit comprising three directional
couplers,” Electron. Lett., vol. 18, no. 12, pp. 491–493, Jun. 1982.
using MAX2015 logarithmic detectors and the uncalibrated [4] A. S. Mohra, “Six-port reflectometer based on four 0°/180° microstrip
 L were computed. The measured  L are shown in Fig. ring couplers,” Microw. Opt. Technol. Lett., vol. 40, no. 2, pp. 167–170,
4(b). The | L | values of 0- and 10-k loads are very 2004.
[5] R. Uttamatharin, R. Phromloungsri, V. Chamnanphrai, P. Sirisuk, and
close to the expected value(s) of 0 dB, while the difference M. Chongcheawchamanan, “Design six-port reflectometer based on
open
between the measured phase angles ( L ) − ( short L ) is inductively-compensated coupled lines,” in Proc. Asia–Pacific Conf.
Commun., 2007, pp. 323–326.
in the range of 172°–185°, close to the ideal value of 180°. [6] J. J. Yao and S. P. Yeo, “Six-port reflectometer based on modified
The load dynamic range between 0 /10 k and 51- load hybrid couplers,” IEEE Trans. Microw. Theory Techn., vol. 56, no. 2,
varies from 21 to 39 dB in the range 1.6–2.2 GHz, which pp. 493–498, Feb. 2008.
[7] J. J. Yao, S. P. Yeo, and M. E. Bialkowski, “Modifying branch-line
indicates good ability to distinguish matched and mismatched coupler design to enhance six-port reflectometer performance,” in IEEE
loads. The threshold Minimum Detectable Signal (MDS) of the MTT-S Microw. Symp. Dig., Jun. 2009, pp. 1669–1672.
MAX2015 is around −55 dBm, which makes the system SNR [8] C. Li, H. Zhang, and P. Wang, “A novel six-port circuit based on four
quadrature hybrids,” Int. J. RF Microw. Comput.-Aided Eng., vol. 20,
better than 15 dB (with measured noise floor of −70 dBm) no. 1, pp. 128–132, 2009.
for the band from 1.6 to 2.2 GHz. [9] K. Staszek, S. Gruszczynski, and K. Wincza, “Design and accuracy
The performance of the SPR has also been tested using four analysis of a broadband six-port reflectometer utilizing coupled-line
directional couplers,” Microw. Opt. Technol. Lett., vol. 55, no. 7,
other loads: a 36- load, a 100- load, a 20 mm long, 50- pp. 1485–1490, 2013.
microstrip line loaded with a 1-pF capacitor, and a printed [10] K. Staszek, S. Gruszczynski, and K. Wincza, “Six-port reflectometer
providing enhanced power distribution,” IEEE Trans. Microw. Theory
monopole antenna. For each load, the magnitude and phase Techn., vol. 64, no. 3, pp. 939–951, Mar. 2016.
of  L were measured by recording the MAX2015 detector [11] S. P. Yeo and M. Cheng, “Improved four-port instrument using two
outputs at ports 1−4. The uncalibrated  L measured by the power detectors to measure complex reflection coefficients of microwave
devices,” Electron. Lett., vol. 32, no. 6, pp. 565–566, Mar. 1996.
detector outputs for the different loads are shown in Fig. [12] G. Madonna, A. Ferrero, and M. Pirola, “Design of a broadband
5. Each measurement has been compared with a standard multiprobe reflectometer,” IEEE Trans. Instrum. Meas., vol. 48, no. 2,
 L reading taken from a calibrated E5071C Vector Network pp. 622–625, Apr. 1999.
[13] K. Staszek, S. Gruszczynski, and K. Wincza, “Ultrabroadband mul-
Analyzer (VNA), as shown in Fig. 5. As seen from Fig. 5, tiprobe reflectometer,” Microw. Opt. Technol. Lett., vol. 57, no. 8,
the measured | L | follows the expected characteristic, with pp. 1968–1971, 2015.
[14] K. Haddadi, M. M. Wang, K. Nouri, D. Glay, and T. Lasri, “Calibration
average magnitude error of about 0.9–2.8 dB. The measured and performance of two new ultra-wideband four-port-based systems,”
phase ( L ) follows a similar trend as the expected phase, IEEE Trans. Microw. Theory Techn., vol. 56, no. 12, pp. 3137–3142,
with an in-band average measurement error (spread over the Dec. 2008.
[15] Keysight 85033E 3.5 mm Calibration Kit, Keysight Technol., Santa Rosa,
360° complex −plane) of 2.7%–7.7%. Thus, the designed CA, USA, 2014.
SPR, without any calibration, follows the expected trends and [16] V. Bilik, “Six-port measurement technique: Principles, impact, applica-
gives  L close to the expected values. tions,” in Proc. Int. Conf. Radioelectron., 2002, pp. 1–32.