Abstract— As integrated circuit technology advances,
implantable biotelemetry devices will be able to handle more
complex functions. Their flexibility and performance can be
increased by incorporating multi-rate and multi-modulation
schemes. The Class-E transmitter has been a primary choice so
far in implantable devices as it provides low-power design with
high-power efficiency. This paper investigates the design
procedures of class-E amplifier for different modulation
schemes in biotelemetry systems. A Class-E transmitter circuit
that produces OOK, FSK and PSK modulated signals has been
designed, optimized and analyzed in terms of a second order
system for general implantable electronics.
lthough the Class-E amplifier is commonly used for
wireless power and data transfer in biomedical
telemetry systems, it has mostly been optimized for
power signals (pure sinusoidal signals) in literature
,. The focus of this paper lies in applying the
Class-E amplifier to different modulated signals in
biomedical telemetry devices. The paper also shows that a
class-E system can be analyzed better with the second order
Two transmission standards which are of particular
interest are the Medical Implant Communication Service
(MICS) standard and 27 MHz ISM standard. The MICS
standard has recently been allocated by international
communication authorities to provide small size, low-power,
faster data transfer and a longer communication range for
future implantable devices. The standard specifies a
frequency range between 402MHz and 405MHz, with a
maximum bandwidth of 300kHz at any one time, and a
maximum transmission power of 25µW . The ISM
standard involves a number of allowable frequencies, one of
which is 27 MHz. This frequency was of particular interest
due to the fact that some of the recent implantable designs
have been targeting some frequencies around this value to
accommodate a higher data rate transmission . Thus,
transmitter circuit given in herein is being implemented for
these two medical bands.
II. ANALYSIS OF CLASS-E AMPLIFIER
The reason why the Class-E amplifier is used in
implantable biotelemetry is because of the fact that when
used as a modulator, it eliminates the need for a mixer,
which is a power hungry block and used in many
Manuscript received April 16, 2007. This work was supported in part by
the Australian Research Council (ARC) under Discovery Projects Grant.
A. N. Laskovski and M. R. Yuce are with the School of Electrical
Engineering and Computer Science, University of Newcastle, Callaghan,
NSW 2308, Australia (e-mail: firstname.lastname@example.org ).
Fig. 1 The Class E Amplifier
conventional modulators. Fig. 1 shows a generic schematic
of a class-E oscillator. The MOS device (i.e. M1 transistor)
acts as a switch and is driven by a digital signal at the carrier
frequency fc. The capacitor C2 connected in series with L acts
as a DC blocking capacitor and its value should be small
enough to be nearly resonant with L at the carrier frequency
fc . For an inductive link, the inductor L is used as a
physical coil to transfer data and power to the secondary side
through a magnetic coupling. The resistor R represents the
parasitic resistor of the inductor and the reflected resistance
from the loading of the secondary site. When an antenna is
used, this load resistor will be the impedance of the antenna.
The transmitter power can be variable by changing the power
supply VDD or the resistor at the load R. Thus any required
power level specified by medical standards can easily be
adjusted using these components.
The load network, characterized by C1, C2 and L converts
the digital input signal into a sinusoidal output with a zero
DC offset. By considering the load network as a second
order system, it is possible to gain an understanding of how
to manipulate the output of the circuit to achieve as high an
output as possible with a less power consumption. The
transfer function of the load network can be calculated as:
where vin is the voltage across the capacitor C1.
A digital square wave is represented as a series of positive
and negative step inputs, and after conducting an inverse
Laplace transform, the output signal for one half period is
determined to be:
Analysis of Class-E Amplifier with Mixed Data Modulation
Anthony N. Laskovski and Mehmet R. Yuce, Member IEEE
Fi g. 2. A low-power transmitter design using Class-E amplifier for PSK, FSK and OOK signals
where ωn is the natural frequency and ζ is the damping
factor. The output signal and therefore efficiency may be
manipulated by selecting a particular damping coefficient (ζ)
and damped frequency (ω = 2πfc) where
Simulation results based on the effect of these parameters
will be presented in Section IV.
According to , the efficiency of the Class E amplifier is
the RMS power seen at R divided by the input DC power, as
shown in (4). It assumes that the output voltage is sinusoidal.
Using (2), the efficiency can be given by (5).
LC RCCR LC
Since the RMS voltage of PSK and FSK signals are the same
as that of a sinusoidal signal, (5) has also been used to
determine the efficiency of the PSK and FSK modulated
signals. In the case of OOK, the efficiency is simply halved.
III. A CLASS-E TRANSMITTER WITH MIXED MODULATED
In this section, we present a transmitter circuit using a
Class-E amplifier that provides a low-power implementation
of three different modulated signals: On-Off Keying (OOK),
Phase-Shift Keying (PSK) and Frequency Shift Keying
(FSK). The PSK modulated signal is generated by applying
the carrier and the data signals to the gate of transistor M1
through an exclusive-OR as shown in Fig 2. The data is
directly applied to M3 for an OOK modulated signal. FSK
can be achieved by generating a variable frequency signal at
the input of the Class-E amplifier with a digital voltage
controlled oscillator. Note also that the circuit does not need
to be altered to allow different data rates (multi-rate
IV. CIRCUIT SIMULATIONS
Simulations were conducted on the basic class-E amplifier
of Fig.1 in order to determine the optimal circuit parameters
for the proceeding modulation simulations. The circuit is
simulated using Berkeley BSIM4v4 spice models. Two
parameters were examined closely based on the theory of
second order systems. The first parameter to be considered
was the damping coefficient ζ∈(0,1). A transfer function
tuned to 403MHz with a load resistance of 500Ω was
simulated in MATLAB under varying conditions, and step
response plots of two of the outermost coefficients are
presented in Fig. 3. The higher curve represents a damping
coefficient of 0.9, meaning that the signal is over-damped.
The lower curve is the result of a damping coefficient of 0.1,
meaning that it was under-damped. This explains why the
under-damped signal oscillated for several cycles of its
oscillation frequency. It is evident that the damping
coefficient makes a considerable difference to the output
signal of the Class E circuit.
Fig. 3. A Step response with damping ratios of 0.1 and 0.9 applying to the
higher and lower curve respectively.
This effect was further investigated through a series of
circuit simulations in SABER with a digital clock input
signal. The load resistance R was kept constant at 500Ω, as
was the forced oscillation frequency of 403MHz. The
relationship between the damping coefficient and circuit
efficiency is presented in Fig. 4, showing that the lower the
damping coefficient, the higher the efficiency of the circuit.
As an under-damped system with periodic step inputs,
superposition occurs cycles after each change in the digital
carrier signal. Due to the fact that there is no change in
phase, the oscillations constructively superimpose to
increase the overall output power of the signal and therefore
higher circuit efficiency at lower values of ζ.
The next parameter considered was the output resistance R
and the MOSFET channel lengths and widths. R was varied
while ζ was kept constant at 0.1 and the forced oscillating
frequency at 403MHz. The simulations were conducted with
three different transistor settings. From the results plotted in
Fig 5, it is evident that the efficiency of the class E circuit
peaks according to the relationship between R and the
parameters of the transistors in use. In reality R will vary
according to transmission conditions, however the circuit
can be optimized for a particular value of R which is likely
to be the most common load resistance to the class-E
Fig. 4. The effect of the damping factor ζ on efficiency (η) with R =500Ω,
VDD= 3V, lchannel=500nm, widthp-channel=20µm, widthn-channel=10µm.
Fig. 5. The effect of the load resistance R on efficiency η with a constant ζ
of 0.1 and VDD of 3V and varying the scale of the MOSFET channel widths,
widthp-channel/widthn-channel while keeping the channel length 500nm.
It was mentioned in section II that C1 was approximately the
same as C2, and that C2 serves the purpose of blocking DC
volts from entering the output. A number of simulations
have been conducted to show a relationship between C2 and
C1. C1 was kept constant at 0.157 pF and C2 was varied as
shown in Fig. 6. It may be observed that when C2 is close to
the value of C1 a maximum transfer of energy takes place.
This is because the voltage at the junction between C2 and L
approaches zero. As C2 exceeds C1 the efficiency drops and
then plateaus to a constant value, meaning that C2 needs to
be the same value as C1 (which is in resonance with L) to
successfully drive the average voltage across C1 to 0V at the
highest frequency. Fig. 6 also shows the variation of C1
while keeping C2 constant. This plot shows the efficiency
rapidly decrease when C1 is increased to be higher than C2.
This occurs because as C1 exceeds the value of C2, C2 is no
longer large enough to drive the average voltage between L
and C2 down to zero, meaning that the excess DC energy is
dissipated through R, which is detrimental to the efficiency
of the circuit.
Fig . 6. Variations in C1 and C2 with VDD=3V, ζ=0.1, R=500Ω, L=982nH,
MOSFET channel length 500nm, widthp-chan/widthn-chan=20µm/10µm. a) The
effect of varying C1 while keeping a constant C2 at 0.157pF. b) The effect
of varying C2 while keeping a constant C1 at 0.157pF.
An important factor in complying with the MICS standard
is for the transmission signal to remain below 25µW. Thus,
the effect of VDD was investigated. It is evident from Fig. 7
that for lower values of VDD, circuit efficiency is highly
dependant on the input power, but as VDD exceeds a certain
value, its value becomes less significant. By comparing Fig.
5 with Fig. 7 it is evident that balancing VDD and R is
Fig. 7. The effect of the damping factor VDD on efficiency (η) with R
=500Ω, ζ= 0.1 (C1=C2=0.157pF, L=982nH), lchannel(n,p)=500nm, widthp-
important in controlling the efficiency of the circuit and also
the output power such that it complies with the MICS
It was predicted in Section II that PSK would have an
almost identical efficiency to an un-modulated signal. The
Class E transmitter given in Fig. 2 was simulated in PSK and
OOK (plots shown in Fig. 8 and 9 respectively) modes to
indicate the efficiency of each method.
The simulation results show that the efficiency of the un-
modulated circuit (with VDD = 3V, ζ= 0.1 and R =500 Ω) is
66.9%, while the efficiency of the PSK circuit is 25.9%. The
difference may be explained by the presence of an XOR
gate, which comprises 20 additional transistors. The
efficiency of the OOK signal was simulated to be 32.7%,
which is just less than half the efficiency of the un-
modulated circuit. Considering the fact that the XOR was
not in operation during the OOK or the un-modulated
circuit, the results are close to what was expected. An FSK
signal was simulated with the two signal frequencies
269MHz and 403MHz. The efficiency of this circuit was
29%. As shown in Fig 10, there is a noticeable difference
between the two frequencies; however the circuit has been
optimized for 403MHz. This is the reason why the efficiency
is not similar to the un-modulated value of 66.9%.
The transmitter consumes 75 µW at 0.8 VDD (with the
output signal, Pout = 2.25 µW) and 2.2 mW at 3V (Pout =
941µW) when operating at the MICS band. To meet the
MICS regulation for the transmitted power, the VDD should
be kept low. In the case of 27 MHz ISM, the power
consumptions 9.16 µW (Pout = 0.243µw), 18.5 µw (Pout =
0.592µW), 479 µW (Pout = 44µW), 4.22 mW (Pout =
300µW) for supply voltages of 0.8V, 1V, 3V and 5 V
respectively. The damping factor was ζ = 0.08 and data rate
was 300 Kbps.
A data transmitter using class-E oscillator for telemetry
has been optimized with a second order system. A circuit has
been designed to verify the effect of the damping factor on
the operation of Class E. A transmitter circuit is designed
based on the recently available MICS band. Generation of
modulations schemes such as OOK, FSK and PSK have
been discussed. When modeling the load network of the
amplifier as a second order system, several parameters were
noticed to make a considerable difference to the output of
W. Liu, et al., “A neuro-stimulus chip with telemetry unit for retinal
prosthetic device,” IEEE J. Solid State Circuits, vol. 35, pp. 1487-
1497, Oct. 2001.
M. Ghovanloo, K. Beach, K. D. Wise, and K. Najafi, “A BiCMOS
wireless interface chip for micromachined stimulating microprobes,”
in Proc. IEEE EMBS Special Topic Conf. Microtechnology in
Medicine and Biology, May 2002, pp. 277-282.2.
P. R. Troyk and G. A. DeMichele, “Inductively-coupled power and
data link for neural prostheses using a class-E oscillator and FSK
modulation,” in Proc. IEEE 25th EMBS Conf., Sept. 2003, pp. 3376-
M. Ghovanloo and K. Najafi, “A wideband frequency shift keying
wireless link for biomedical implants,” IEEE Trans. Circuits Syst. II,
vol. 51, pp. 2374- 2383, Dec. 2004.
K. Sohee and O. Scholz, “Implantable active telemetry system using
microcoils,” in IEEE-EMBS 2005. 27th Ann. Intern. Conf. of Eng. in
Medicine and Biology Society, pp. 7147 – 7150, 2005.
M. Zhou, W. Liu, G. Wang, M. Sivaprakasam, M. R. Yuce, J.
Weiland, M. Humayun, “ A Transcutaneous data telemetry system
tolerant to power telemetry interference,” in the 28th Annual
International Conference of the IEEE Engineering in Medicine and
Biology Society (IEEE EMBC 2006),September 3, 2006.
M. K. Kazimierczuk “Class E tuned power amplifier with
nonsinusoidal output voltage,” IEEE J. of Solid State Circuits, vol
21- August, 1986.
 F. H. Raab, "Effects of circuit variations on the class E tuned power
amplifier," IEEE J. of Solid State Circuits ,vol. 13, pp. 239-247, 1978.
 FCC Rules and Regulations, “MICS Band Plan,” Table of Frequency
Allocations, Part 95, Jan. 2003.
 Australian Communications Authority, Radio Frequency Planning
Group, “Planning for Medical Implant Communications Systems &
Related Devices”, Proposals Paper, Oct. 2003.
Fig. 8. A simulation plot of the output of the PSK circuit (403MHz)
Fig. 9. A simulation plot of the output of the OOK modulated circuit.
Fig. 10. Output plots of a FSK signal. a) Time transient signal. b)