See discussions, stats, and author profiles for this publication at: [Link]

net/publication/309637684

Combined RF and multiphase PWM transmitter

Conference Paper · October 2015

CITATIONS READS
5 1,350

3 authors:

Muhammad Fahim Ul Haque Ted Johansson

NED University of Engineering and Technology Uppsala University
11 PUBLICATIONS 55 CITATIONS 78 PUBLICATIONS 375 CITATIONS

SEE PROFILE SEE PROFILE

Dake Liu
Hainan University
249 PUBLICATIONS 2,886 CITATIONS

SEE PROFILE

All content following this page was uploaded by Dake Liu on 03 November 2016.

The user has requested enhancement of the downloaded file.

 2015 European Conference on Circuit Theory and Design (ECCTD)

Combined RF and multiphase PWM Transmitter

Muhammad Fahim Ul Haque, Ted Johansson and Dake Liu

Department of Electrical Engineering, Linköping University, Sweden.
Email:{muhha,ted,dake}@[Link]

Abstract—This paper presents two novel transmitter archi- multiphase quadrature transmitter architecture, are proposed
tectures based on the combination of radio-frequency pulse- in this paper, which further extend this idea. The advantages
width modulation and multiphase pulse-width modulation. The with both architectures are high power efficiency, small chip
proposed transmitter architectures provide good amplitude res- area, large dynamic range at high carrier frequency, and relaxed
olution and large dynamic range at high carrier frequency, band-pass filtering requirement.
which is problematic with existing radio-frequency pulse-width
modulation based transmitters. They also have better power
efficiency and smaller chip area compared to multiphase pulse- II. RF P ULSE - WIDTH M ODULATED T RANSMITTER
width modulation based transmitters. (RF-PWMT)
In this section, we will discuss the RF-PWMT in more
I. I NTRODUCTION detail. The block diagram of a typical RF-PWMT is shown in
In modern low-cost portable wireless communication sys- Fig. 1. In this architecture, the baseband signal is separated
tems, most of the radio circuitry is implemented in a few, into its phase φ(t) and amplitude a(t) components. The phase
or even a single CMOS chip. Highly linear power amplifiers φ(t) varies the pulse position of a pulse train to generate a
(PAs) for the amplification of high peak-to-average power pulse-position modulated (PPM) signal, where the pulse train
ratio (PAPR) signals with high efficiency for extending the frequency is equal to the carrier frequency (ωc ). The amplitude
battery life are required. PAs are broadly classified into two component a(t) varies the pulse width of the PPM signal
types: linear mode PAs, which are highly linear but have poor to generate an RF pulse-width and position-modulated (RF-
power efficiency, and switch mode PAs (SMPAs) which have PWPM) signal. The RF-PWPM signal is then amplified by the
good power efficiency but can only amplify constant envelope PA and finally band-pass filtered to generate the transmitted
signals [1]. signal. The RF-PWMT has the advantage of having smaller
chip size [5] and high power efficiency [6]. However, it has
Several transmitter architectures that amplify non-constant small dynamic range at high carrier frequency [5].
envelope signal using SMPAs are being studied, such as
polar modulation [2], load modulation [3], outphasing [4], and III. M ULTIPHASE P OLAR PWM T RANSMITTER
pulse-width modulation (PWM) [5]. For PWM transmitters, (MP-PPWMT)
the popular architectures are the radio frequency pulse-width
modulated transmitter (RF-PWMT) [5] [6] and the polar pulse- In this transmitter architecture, shown in Fig. 2, the base-
width modulated transmitter (PPWMT) [7]. band signal is also first separated into its phase φ(t) and
amplitude a(t) components. The carrier is phase-modulated by
The RF-PWMT changes the amplitude of the SMPA output φ(t) to generate a phase-modulated carrier. The a(t) is used
by varying the pulse width of the input pulse train having fre- to generate multiple PWM signals which are time-shifted by a
quency equal to the carrier frequency [5] [8]. The advantages fixed delay. These multiple PWM signals are up-converted by
of RF-PWMT are small chip area and high power efficiency, being multiplied by the phase-modulated carrier. The multiple
but it has small dynamic range and poor amplitude resolution. up-converted signals are then amplified by separate PAs and
The PPWMT varies the pulse width of the pulse train, combined to generate a multiphase PWM signal. The multi-
based on the envelope of the baseband signal, where the phase PWM signal is finally band-pass filtered to generate the
pulse-train frequency is at intermediate frequency. The pulse- transmitted signal.
width modulated pulse train is then multiplied by the phase- The advantage over PPWMT is relaxed band-pass filtering.
modulated carrier to generate a carrier-based PWM signal By increasing the number of time-shifted PWM signals, the
(CPWM). The CPWM signal is amplified and filtered to gener-
ate the transmitted signal. The advantages of PPWMT are large
dynamic range at high carrier frequency and smaller chip area, a(t)
while having low amplitude resolution and requiring strict band I

pass filtering. The amplitude resolution can be addressed by RF-PWM

PA

CORDIC BPF
using multilevel PPWMT [9], but using multiphase PPWMT φ(t) Pulse
Q
(MP-PPWMT) solves both problems [10], however with larger Position RLoad
Mod.
chip area and higher combiner losses. In section III, we will
discuss multiphase PPWMT in further detail.
The authors previously proposed a combined RF and mul-
tilevel PPWMT [11]. Two new architectues, a combined RF Fig. 1. Block diagram of an RF pulse-width modulated (RF-PWM) trans-
and multiphase PPWMT architecture and a combined RF and mitter.

978-1-4799-9877-7/15/$31.00 2015
c IEEE.
band-pass filtering requirements are further relaxed [10]. The frequency (ωIF ). These multiple PWM signals are similar,
advantage over RFPWMT is large dynamic range and good except they are shifted in time by a fixed delay (TD ), given
amplitude resolution at high carrier frequency. as
T
The disadvantages are combiner losses and large chip area TD = (3)
M
because of the presence of a combiner [12].
where T is the pulse-repetition period of the individual PWM
IV. C OMBINED RF AND M ULTIPHASE PWM signals and M is the total number of PWM signals. Then the
T RANSMITTER multiple PWM signals are added to generate a composite PWM
signal. The normalized composite PWM signal is expressed as
To overcome the deficiency of both previously discussed τ
techniques, we are suggesting a new concept where the mul- vc =
T
tiphase PWM signals, having pulse-repetition frequency equal
∞  ∞  nπτ    (4)
to intermediate frequency, are added and then combined with 2 2nmπ
+ sin cos nωIF t −
an RF pulse train using an RF-PWM modulator. In this way, m=0 n=1
M nπ T M
the PA combiner is not needed and the dynamic range and
efficiency will be improved. where τ is the pulse width of individual PWM signals. The
relation between τ and a(t) is given as
In the next two subsections, we describe the implemen-
tation and analytical analysis for two different transmitter τ = a(t)T. (5)
architectures; a PWM polar transmitter and a PWM quadrature
transmitter The composite PWM signal in (4) is generated from the ad-
dition of M time-shifted PWM signals, hence its first harmonic
A. Combined RF and Multiphase PWM Polar Transmitter is at M ×ωIF , giving relaxation in the final band-pass filtering
(CRFM-PPWMT) requirements. However, the composite PWM signal in (4) have
The conceptual block diagram of a combined RF and M amplitude levels which cannot be amplified by SMPA. To
multiphase PWM polar transmitter (CRFM-PPWMT) is shown achieve this, the amplitude of the composite PWM signal is
in Fig. 3. In this architecture, we first separate the baseband first pre-distorted to overcome the distortion of the RF pulse-
signal into its phase φ(t) and amplitude a(t) components. The width modulator, and then translated to the pulse width of an
phase component generates an RF pulse-position modulated RF-PPM signal by the RF pulse-width modulator. This results
(RF-PPM) signal by varying the pulse position of RF pulse in a combined RF and multiphase PWM signal (CRFM-PWM)
train, which has a frequency equal to ωc . The normalized RF- having only two amplitude levels. The pre-distorted composite
PPM signal is given as PWM signal and normalized CRFM-PWM signal are given as
  (6) and (7), respectively
1 2 π 2πTP
vp (t) = + sin cos ωc t − Tc
2 π 2 Tc vc = sin−1 (πvc ) . (6)
∞     (1) π
 2 nπ 2nπTP
+ sin cos nωc t −
n=2
nπ 2 Tc
The CRFM-PWM signal is then amplified by the SMPA
where TP and TC are the pulse-position and pulse-repetition and finally band-pass filtered to generate the transmitted signal.
frequency of the PPM signal. The relationship between φ(t)
The CRFM-PWMP transmitter amplifies the multiphase
and TP is given as
PWM signal with only one SMPA, hence no combiner is
φ(t)Tc required, resulting in smaller chip area and high power effi-
TP = . (2)
2π ciency. It also retains the multiphase PWM advantages of large
dynamic range and good amplitude resolution at high carrier
The amplitude component generates multiple PWM signals frequency, and relaxed final band-pass filtering.
having pulse-repetition frequency equal to an intermediate

a(t) Multi
Multiple PWM Generation ADDER
Phase
a(t) I PWM
I PWM
0xTD Delay

Band Pass CORDIC RF-PWM BPF

Combiner
Filter
CORDIC PWM Pulse
MxTD Delay RLoad
Q φ(t) RLoad
Position ClassD
SMPAs Mod. PA
Q
φ(t) Phase
Modulator

Fig. 3. Conceptual block diagram of a combined RF and multiphase polar

Fig. 2. Block diagram of a multiphase polar PWM (MP-PPWM) transmitter. PWM transmitter (CRF-PPWMT).
  
vc τ 2πTP
vU = + cos ωc t −
Tc T Tc
        (7)
nπτ 
M −1
  2 ∞ ∞
2nmπ 2πTP 2  2kπTP
+ sin cos nωIF t − cos ωc t − + sin (kvc ) cos kωc t −
m=0 n=1
M nπ T M Tc kπ Tc
k=2

TABLE I. S IMULATED DYNAMIC RANGE FOR AN RF-PWMT AND THE

B. Combined RF and Multiphase PWM Quadrature Transmit- PROPOSED CRFM-PPWMT FOR DIFFERENT CARRIER FREQUENCIES
ter (CRFM-PWMQT)
Carrier frequency RFPWMT CRFM-PPWMT
We can also apply the same concept with an IQ architec- [GHz] [dB] [dB]
ture. The block diagram of the CRFM-PWMQT is shown in 1 16.69 26
Fig. 4. In this architecture, the baseband signal is separated 2 9.791 26
into two positive components IP and QP , and two negative 3 6.468 26
components IN and QN , which generate their individual
composite PWM signals, given as
τIP,IN,QP,QN
vCIP,CIN,CQP,CQN = The advantage of the CRFM-PWMQT over the CRFM-
T PWMPT is easy digital implementation, as it does not require
∞  ∞  nπτ   
2A IP,IN,QP,QN 2nmπ a phase modulator block.
+ sin cos nωt −
m=0 n=1
M nπ T M
(8)
V. S IMULATION R ESULTS
where τIP = IP (t)T , τIN = IN (t)T , τQP = QP (t)T and
τQN = QN (t)T . The system simulations of dynamic range and instanta-
neous efficiency were done in Cadence/ADS using 2.5 volt
The composite PWM signals shown in (8) have M ampli-
thick-oxide transistors in standard 130 nm CMOS technology.
tude levels. The signal is first pre-distorted and then translated
The EVM and spectral mask simulations were done using
to the pulse width of their respective IP , IN , QP and QN
MATLAB.
pulse trains to generate the respective CRFM-PWM signals.
The pulse trains have pulse-repetition frequency equal to the Table I compares the dynamic range of an RF-PWM trans-
carrier frequency, however they are shifted in time. mitter (RF-PWMT) and the proposed combined RF multiphase
The IP , IN , QP and QN pre-distorted composite PWM polar PWM transmitter (CRFM-PPWMT) using 1 watt Class
signals are given by (9), whereas the normalized IP , IN ,QP D PAs having supply voltage of 2.5 volt. The CRFM-PPWMT
and QN CRFM-PWM signals are given by (10). in this simulation uses two phase-shifted PWM signals having
 πv  pulse-repetition frequency (PRF) of 50 MHz, and a minimum
vDIP,DIN,DQP,DQN = sin−1
CIP,CIN,CQP,CQN
(9) duty cycle of 5%. From Table I we observe that dynamic
2 range of the RF-PWMT decreases when increasing the carrier
frequency from 1 to 3 GHz, while it stays the same for the
The maximum pulse width of all CRFM-PWM signals is CRFM-PPWMT. A similar decrease in RF-PWMT dynamic
one fourth of their respective time period, hence combining IP , range at increased carrier frequency was also observed in [5],
IN , QP and QN CRFM-PWM results in a signal having two however using a 1 Watt Class E PA in standard 65 nm CMOS.
amplitude levels. This signal is power-amplified by an SMPA
and finally filtered to generate the transmitted signal. The power efficiency performance of a multiphase polar
PWM transmitter (MP-PPWMT), using two phases, two 0.5 W
class D PAs having supply voltage of 2.5 volt, and a Wilkinson
I IP Multiple Adder RF-PWM
combiner, and the same CRFM-PPWMT as in previous para-
PWM graph is shown in Fig. 5 in which the instantaneous efficiency
S
E IN Multiple
at 1 GHz carrier frequency is plotted against normalized
Adder RF-PWM
P
A
PWM
M
power. The CRFM-PPWMT exhibits better efficiency than the
R
A
U
X
BPF MP-PPWMT. Moreover the better efficiency of the CRFM-
QP Multiple
T
PWM
Adder RF-PWM CLASS D RLoad PPWMT at back-off power make it a more suitable solution
O PA
R for modulated signal with high PAPR.
Q QN Multiple Adder RF-PWM
PWM
The EVM of the proposed quadrature CRFM-PWMQ
transmitter using two-phase I and Q PWM with a PRF of 80
Phase MHz and minimum pulse resolution of 0.195 ns was simulated
Shifter for 802.11g baseband input signal. The simulation result shows
maximum EVM of 2.8% which complies with 802.11g EVM
Fig. 4. Conceptual block diagram of a combined RF and multiphase PWM requirements. The spectral mask is shown in Fig 6, which also
quadrature transmitter (CRFM-PWMQT). complies with the 802.11g mask requirement.
 vDIP,DIN,DQP,DQN τCIP,CIN,CQP,CQN
vU IP,U IN,U QP,U QN = + cos (ωc t − θU IP,U IN,U QP,U QN )
Tc T
M
 −1 
∞  nπτ   
2 CIP,CIN,CQP,CQN 2nmπ
+ sin cos nωIF t − cos (ωc t − θU IP,U IN,U QP,U QN )
m=0 n=1
M nπ T M
∞
 2
+ sin (kvDIP,DIN,DQP,DQN ) cos (kωc t − θU IP,U IN,U QP,U QN )
nπ
k=2
where θU IP = 0, θU IN = π, θU QP = π/2, θU QP = 3π/2
(10)

VI. C ONCLUSION range compared to a RF-PWM based-transmitter and bet-

ter power efficiency compared to multiphase PWM based-
A combined RF and multiphase polar PWM transmitter transmitter (MP-PPWMT). They can also be integrated on-chip
(CRFM-PPWMT) and combined RF and multiphase PWM with smaller areas compared to other architectures.
quadrature transmitter (CRFM-PWMQT) and their system
equations are proposed in this paper. The simulation results R EFERENCES
show that the proposed transmitters exhibit large dynamic
[1] T. Johansson and J. Fritzin, “A review of Watt-level CMOS RF power
amplifiers,” IEEE Trans. Microw. Theory Tech., vol. 62, no. 1, pp. 111–
124, Jan 2014.
[2] W. Sander, S. Schell, and B. Sander, “Polar modulator for multi-mode
cell phones,” in Proc. CICC, Sep 2003, pp. 439–445.
[3] F. Raab, “High-efficiency linear amplification by dynamic load modu-
lation,” in IEEE MTT-S Int. Microw. Symp. Dig., Jun 2003, pp. 1717–
1720.
[4] J. Fritzin, C. Svensson, and A. Alvandpour, “A +32 dbm 1.85 GHz
class-D outphasing RF PA in 130nm CMOS for WCDMA/LTE,” in
Proc. ESSCIRC, Sep 2011, pp. 127–130.
[5] J. Walling and D. Allstot, “Pulse-width modulated CMOS power
amplifiers,” IEEE Microw. Mag., vol. 12, no. 1, pp. 52–60, Feb 2011.
[6] N. Singhal, N. Nidhi, A. Ghosh, and S. Pamarti, “A 19 dBm 0.13
μm CMOS parallel class-E switching PA with minimal efficiency
degradation under 6 dB back-off,” presented at the RFIC Symposium,
Jun 2011.
[7] M. Taromaru, N. Ando, T. Kodera, and K. Yano, “An EER transmitter
architecture with burst-width envelope modulation based on triangle-
wave comparison PWM,” presented at PIMRC, Sep 2007.
[8] H. Song and R. Gharpurey, “Digitally intensive wireless transmitter
architecture employing rf pulse width modulation,” presented at DCAS,
Fig. 5. Instantaneous efficiency plot of a two-phase polar PWM transmitter Oct 2014.
(MP-PPWMT) and a combined RF and two-phase polar PWM transmitter [9] J.-H. Chen, H.-S. Yang, and Y.-J. Chen, “A multi-level pulse modulated
(CRFM-PPWMT). polar transmitter using digital pulse-width modulation,” IEEE Antennas
Wireless Propag. Lett., vol. 20, no. 5, pp. 295–297, May 2010.
[10] J.-H. Chen, H.-S. Yang, H.-C. Lin, and Y.-J. Chen, “A polar-transmitter
architecture using multiphase pulsewidth modulation,” IEEE Trans.
Circuits Syst. I, vol. 58, no. 2, pp. 244–252, Feb 2011.
[11] M. Ul Haque, T. Johansson, and D. Liu, “Combined RF and multilevel
PWM switch mode power amplifier,” presented at NORCHIP, Nov
2013.
[12] J. Long, “Monolithic transformers for silicon RF IC design,” IEEE J.
Solid-State Circuits, vol. 35, no. 9, pp. 1368–1382, Sept 2000.

Fig. 6. Simulated spectral mask of the proposed CRFM-PWMQT for 802.11g

baseband input.

View publication stats