A Fast and Accurate Automatic Matching Network

Designed for Ultra Low Power Medical Applications

Chan Wai Po F., De Foucauld E., Dal Molin R., Pons P., Pierquin R. Kerhervé E.
Vincent P., Hameau F., SORIN Group / ELA Medical IMS Laboratory
Morche D., Delaveaud C La Boursidière 351, cours de la Libération
CEA, LETI, MINATEC 92357 Plessis Robinson, France 33405 Talence, France
38054 Grenoble, France Regis.Pierquin@sorin.com eric.kerherve@ims-bordeaux.fr
francis.chan-wai-po@cea.fr

Abstract—We present a method to automatically match a system circuitry [1][2][3]. This automatic matching topology will
to the load variation accurately with a very short matching time solve the problem with high precision and single iteration,
for Ultra Low Power medical applications. A demonstrator was saving power consumed during tuning process.
fabricated and an experimental set-up in the Medical Implant
Communication System (MICS) frequency band of 402-405MHz In this paper, we identify the problems affecting
was realized including a pacemaker antenna prototype pacemaker telemetry in section II. We propose a fast and
immersed on a homogeneous lossy liquid. The matching network accurate solution using automatic matching network based on
consists of a low pass π structure with tunable components made vector detection in section III. We developed an experimental
of varactors. Methodologies for simplifying the matching setup to verify the concept. The measurements and results are
network design are introduced. The measurements show that a evaluated in section IV.
S11 up to -30dB is obtained with a matching time of 900μs.
II. EVALUATION OF LOSSES
I. INTRODUCTION An antenna for implanted device has been designed at the
New generation of pacemakers integrates RF transceiver to biomedical MICS frequency band of 402-405 MHz. It is
allow home monitoring, surgery report, full RF follow up at known that the pacemaker antenna input impedance is affected
home and clinic. Battery lifetime consideration in medical by human tissue properties (muscle, fat, skin) and body
applications imposes ultra low power constraint on the position [4][5][6]. Simulation and experimentations have been
transceiver design. Specifications are relaxed at the expense of conducted using homogeneous human model, hand position
saving power and the communication range is short. model and heterogeneous models as shown in Fig. 1. The
antenna input impedance variations are illustrated in Fig.2.
Unfortunately, the system is strongly affected by body
factors [4][5][6] like tissue properties and its thickness. The On the medical MICS band frequency, a -16dBm EIRP
position of body also impacts on the impedance of the transmitting power is limited by the regulations [7][8][9]. The
implantable antenna on near field condition. Due to the tissue maximum communication range between implant and explant
losses and impedance mismatches the transceiver in the is estimated to only 5.6 meters with a receiver sensitivity of
implant may no longer communicate with the explant.
To make sure that this scenario will not occur, we decided to
not over specify the design but to optimize the budget link
using an automatic matching network that reduces impedance
mismatches between the antenna and the radio frequency

(a) (b) (c)

Fig. 1. Effect of human body on the antenna (a) experimental human Fig.2. Antenna input impedance (a) measured homogeneous model (black
homogeneous model, (b) hand position impact, (c) human heterogeneous lines), (b) human+hand model (blue lines), and (c) simulated heterogeneous
simulation model. model (red lines).

This work is sponsored by Ela Medical (Sorin Group).

978-1-4244-3828-0/09/$25.00 ©2009 IEEE 673

 TABLE I Antenna
ESTIMATION OF OPTIMIZED PACEMAKER’S COMMUNICATION RANGE
C Tunable Matching Network
Budget Link Estimated losses Implant to Explant
PA
L
TX power -16 dBm v1 v2
Tissue Attenuation 13 dB -29 dBm C1 C2
Path loss 29 dB -58 dBm Attenuator
Fading 18 dB -76 dBm Source
Mixer Filter

Receiver sensibility -91 dBm

Estimated Range 5.6 meters RF IF Processor

TABLE II
Downconversion
INSERTION LOSSES VS LOAD LO module
LOAD 50+J*250 50+J*370 220+J*470
NO MN -8.60 dB -10.26 dB -10.26 dB Fig. 4. Architecture of the proposed automatic matching system
FIXED MN 0 dB -1.34 dB -4.41 dB

L C
Source Load C = 0,1pF C = 16pF C = 100pF

RS C1 50nH C2 RL Z = R+j*I
v1 v2
50Ω
16,9pF 5,2pF
R from 10Ω to 300Ω
I from -100Ω to 100Ω

Fig. 3. Fixed matching network

-91dBm as illustrated in Table I. Any mismatches will reduce Fig. 5. Phase and magnitude resolution versus detection capacitor C
the communication range.
Although it is difficult to estimate the worst case that could This architecture is easy to implement. To down convert
affect the antenna input impedance, Table II illustrates that the signals, IQ receiver can be reuse since the tuning process
more than 10dB can be lost when a 50Ω optimal output is done during the transmitting path. An attenuator is inserted
impedance power amplifier is connected to the characterized on the chain to reduce the signals’ power according to the
loads and more than 4dB when a non tunable π matching linearity of the down conversion module.
network designed to match 50+j250 Ω to 50Ω is inserted
between the power amplifier and the load as shown in Fig. 3.
B. Detection capacitor C
III. AUTOMATIC MATCHING NETWORK The detection capacitor C is used to extract (v1, v2).
Capacitors have the advantages to be easy to integrate and to
It is demonstrated on section II that it is necessary to present a high quality factor. The impact of this capacitor on
match the system and to develop an automatic matching the matching network quality factor is given by:
network in order to optimize the budget link. Ultra low power
consideration encourages us to develop a new concept able to ⎡ ⎤
match the system with a high precision and a very short tuning R s ⎢1 + 1 2⎥
time to save power. ⎣ (C R s ω p ) ⎦ (1)
Q= −1
Rl
A. Proposed solution Where Rs is the optimal load of the power amplifier source
Compared to [2] and [3], this solution shown in Fig. 4 is to match, RL the load and ωp the carrier pulsation.
coupler less for integration issue. Such devices would require
die size [12][13] or would be external at the desired frequency.
A tradeoff between the resolution simulated on Fig. 5 and
As demonstrated on [10], this vector method spends single the matching efficiency [11], which depends on the network
iteration to match the system with a high precision. Because quality factor (1), leads us to set the value of the detection
the tuning process is done during the transmitting path, the capacitor in a single-end topology to:
tuning time should be as short as possible in ultra low power
applications such in medical applications with the need to save 2 (2)
C=
power at the expense of performance. Rs * ω p
Vector data (v1, v2) are detected and down converted to a C. The processor
lower intermediate frequency. The goal is to extract the
antenna input impedance by analyzing the signals. The The goal of the processor is to analyze data in order to
processor calculates the resulting antenna input impedance calculate the antenna input impedance and to match the
and matches it to the optimal impedance of the RF front-end tunable network. The challenge consists in developing
transmitter but also of the receiver. methodologies that could ease calculations and reduce the

674
 C L

RS+j*XS RL+j*XL
C1 C2
Methodology
Steps : P2
rL+j*xL
Final control voltage
yp2= 1-j*x rS+j*xS Network Initial control voltage
transformation
rPL
x = f (rPS, rPL, l) => CPS & C1
rPS L Matching time = 900 μs
zp2=1/yp2 l
TMatching
zp1=zp2-j*l ωp
RPS RPL
yp1=1/zp1 => CPL & C2
CPS CPL
P1
Start Stop

Fig. 8. Automatic matching processing time

Fig. 6. Introduction to fixed L Matching network design methodology
loss tangent 0.02 and layer’s thickness 360μm. The antenna
processing time. Therefore, de-embedding procedure is used input impedance was varied by switching complex
to extract antenna input impedance and a matching network impedances before connecting the pacemaker antenna
design methodology introduced by Fig. 6 has been developed. prototype to the board. The tunable matching network is
The initial network is transformed into a standard π structure realized by using varactors based network which control
with a real source and load to be matched, and that is followed voltages are set by the microcontroller. The carrier frequency
by some basic calculations to reach the proper state is 403MHz and data are down converted to an intermediate
configuration of the network. frequency of 256 kHz.
IV. DEMONSTRATOR AND MEASUREMENTS Analog device’s microcontroller ADUC7026 has been
This section first describes the implementation of the MICS chosen to implement the matching software. It is an
frequency band demonstrator. The matching time, the value ARM7TDMI based controller with a CPU that clocks up at
of the extracted antenna complex impedance and the reflection 40.96MHz. The algorithm running on the processor calculates
coefficient are measured to evaluate the concept. the antenna input impedance, matches the system and controls
the varactors with a voltage resolution of 750µV/code.
A. The MICS frequency band experimental set-up
B. Matching time
Fig. 7 shows the experimental set-up including the MICS
frequency band demonstrator, a microcontroller board and The matching time evaluate the required time to match the
ELA medical pacemaker antenna immersed on a 402-405MHz system including the acquisition time, the antenna impedance
homogeneous human model liquid whose permittivity εr and calculation and the matching design to set the proper
conductivity σ are respectively 56.2 and 0.95 S/m. configuration of the system. Fig. 8 shows that 900μs matching
time is achieved with a microcontroller unit speed running at
The demonstrator was made using a Flame Retardant 4 40 MIPS. The matching processing ended with an update of
substrate (FR4) with a relative permittivity of 4.6, dielectric the varactors control voltage.

Antenna in a lossy
homogeneous fluid

Automatic matching ADUC 7026

network demonstrator Microcontroller

Fig. 7. Experimental set-up of the proposed 403MHz automatic matching system including the ELA Medical pacemaker antenna prototype immerged on a
homogeneous lossy fluid

675
 TABLE III
COMPARISON OF THE AUTOMATIC MATCHING PROCESS CURRENT POWER
CONSUMPTION
Our work De Mingo Ida
Matching time (ms) 0.9 150 100
PA consumption (mA) 0.1 / 2 0.1 / 2 0.1 / 2 (a)
Additional consumption (mA) 1 0 0.5
Matching Consumption (µA) 1 / 2.7 15 / 300 60 / 250 (b)
TABLE IV
ANTENNA INPUT IMPEDANCE CALCULATION RESULTS
Antenna impedance Calculated Characterized
Value (Ω) 54+j*262 50+j*250
Fig. 9. Measured reflection coefficient (a) before and (b) after tuning
C. Automatic matching process consumption
ACKNOWLEDGMENT
According to the ultra low power constraint of pacemaker
design, the power consumption of single iteration matching The authors would like to thank ELA Medical (Sorin
method has been strongly reduced compared to iterative ones Group) for sponsoring this work, providing liquids and
as illustrated in Table III. The current power consumption pacemaker samples. The antenna input impedance study and
depends on the power amplifier, the additional system and the characterization were realized in the context of the EURIMUS
matching time. This work would be more than 15 times and project EPADIMD EM91 (European Platform for Advanced
almost 100 times superior with a power amplifier whose active Implantable Devices).
current consumption is respectively 0.1mA and 2mA.
REFERENCES
D. Calculated antenna input impedance [1] Van Bezooijen A., De Jongh M., Chanlo C., Ruijis L.C.H., Van Straten
The precision of the antenna input impedance calculation F., Mahmoudi R., Van Roermund A.H.M., “A GSM/EDGE/WCDMA
depends on the capability to model the discrete components, adaptative series-LC matching network using RF-MEMS switches”,
IEEE journal of Solid-State Circuits, vol. 43, no 10, oct. 2008
the microwave effects of the transmission lines and the RF
[2] J. de Mingo, A. Valdovinos, A. Crespo, D. Navarro, and P. García ,
switches. Table IV shows that the antenna input impedance is “An RF electronically controlled impedance tuning network design and
calculated with a small error less than 10%. This error is its application to an antenna input impedance matching system,” IEEE
expected to be smaller in a fully integrated circuit with a better Transactions on Microwave Theory and Techniques, vol. 52, no. 2, pp.
ability to model the matching network. First, the transmission 489–497, Feb. 2004.
line length is short and neglected compared to the wavelength. [3] I. Ida, J. Takada, T. Toda and Y. Oishion, “An adaptive impedance
Second, the packaging effects on the system are reduced. matching system and its application to mobile antennas,” TENCON
2004 proceedings, vol. C, pp.543–546, 21–24 Nov. 2004.
E. Reflection coefficient [4] E.L. Firrao, A.J. Ennema, B. Nauta, “antenna behaviour in the presence
of human body”, ProRISC 2004, 15th Annual Workshop on Circuits,
The matching design is computed by the microcontroller. Systems and Signal Processing, 25-26 Nov 2004
It calculates the matching network that could match the [5] A. Cseh, “In-vivo biomedical application of antennas”, 16th IST mobile
calculated antenna impedance of 53+j262 Ω to the 50Ω and Wireless Communication Summit, pp. 1-4, July 2007
optimal load of the power amplifier. The measurements shown [6] A.J. Johansson, “Wave propagation from medical implants – Influence
in Fig. 9 demonstrate that a reflection coefficient up to -30dB of body shape on radiation pattern”, IEEE EMBS Conference, pp.
1409-1410, Oct. 2002
is obtained at the MICS frequency band with this vector
method. The reflection coefficient is extracted at the output of [7] H.S. Savci, A Sula, Z. Wang, N.S. Dogan, E. Arvas, “MICS
Transceivers : regulatory standards and applications”, IEEE
the power amplifier. SouthestCon 2005 proceedings, pp. 179-182, April 2005
[8] A.L. Johansson, “Performance of a radio link between a base station
V. CONCLUSION and a medical implant utilising the MICS standard”, IEEE EMBS
To improve the limited budget link of MICS band conférence, pp. 2113-2116, Sept. 2004
transceiver on a pacemaker, we developed a vector automatic [9] FCC Rules and Regulations, Subparts E and I, Part 95, Nov 2002
matching network. This method calculates the antenna input [10] F. Chan Wai Po, E. De Foucauld, C. Delavaud, P. Ciais and E.
Kerhervé, “A vector automatic matching network designed for wireless
impedance in order to match the whole system. The antenna medical telemetry”, IEEE Northeast Workshop on Circuit and System,
input impedance is extracted with good precision. pp. 89-92, June 2008
Matching time and matching process current power [11] Y. Han and D.J. Perreault, “Analysis and design of high efficiency
matching networks,” IEEE Transactions on Power Electronics, vol. 21,
consumption are highly reduced by making the process in no. 5, Sept. 2006.
single iteration. Matching design methodologies have been
[12] R. Mongia, I. Bahl and P. Bharta, “RF and Microwave coupled-line
developed to reduce the number of instruction. Matching time circuits”, Artech House, Inc. 1999
of 900μs is achieved to give a reflection coefficient of less [13] A.K. Settaluri, A. Weisshaar, C. Lim and V.H. Tripathi, “Design of
than -30dB at the MICS frequency band. compact multilevel folded-line RF couplers”, IEEE Transaction on
Microwave Theory and Techniques, vol. 47, no 12, pp 2331-2339,
The future work is to realize a fully integrated circuit at the December 1999
2.4GHz ISM frequency band.

676