Published in IET Microwaves, Antennas & Propagation
Received on 9th September 2012
Revised on 6th May 2013
Accepted on 24th May 2013
doi: 10.1049/iet-map.2013.0036
ISSN 1751-8725
Phase compensation of intermodulation-induced
spurs in fractional-N frequency synthesisers
Hongyu Wang
Beijing Institute of Technology, 5 South Zhongguancun Street, Haidian District, Beijing 100081, People’s Republic of China
E-mail: wanghongyu@bit.edu.cn
Abstract: Intermodulation effect occurs when the nonlinearity of a device or system with multiple frequencies causes undesired
outputs at other frequencies. Intermodulation effect may cause high amplitude spurs on some dedicated channels in addition to the
spurs introduced by nonlinear distortion in the modulator of fractional-N synthesisers, which severely degrade the signal-to-noise
ratio. In this study, a phase compensation technique is proposed to suppress at least part of these fractional spurs – in particular the
most dominant sidebands. A small amount of controlled amplitude and phase voltage controlled oscillator (VCO) signal is
coupled onto the phase/frequency detector input to cancel-out the potential intermodulation-induced spurious component.
A prototype of fractional-N frequency synthesiser based on the proposed technique has been developed. The synthesiser has
measured −113 dBc/Hz in-band phase noise benefiting from the high speed sigma–delta noise shaper. A modest degrees of
VCO to phase detector cross-coupling suppressed the spurious components by 18 dB. A variety of measured results are
shown to demonstrate a substantial improvement in spectra purity.
.
1 Introduction
Fractional-N synthesisers can simultaneously achieve fast
tuning speed and fine frequency resolution [1–3]. The
advent of sigma–delta ( Σ –Δ) modulation has given the
potential for lower close-in phase noise floor in these
synthesisers [4–6]. However, they tend to suffer from a
particular set of discrete spurious components, the so-called
fractional spurs [7–12]. For several decad es these spurs
have been attributed to the nonlinear distortion present in
Σ–Δ modulators [13–15], as a result, attempts at improving
performance have concentrated on refining the design of the
modulator. An alternative mechanism [16–18] has been
recently reported. The sampling of two non-harmonically
related frequency components (the synthesiser output
frequency and the reference frequency) is able to produce a
family of spurs at exactly the same frequencies as these
fractional spurs. The intermodulation effect in fractional-N
synthesisers is analysed in [16, 17]. Brennan et al. [18]
provide a comparison of modelled and measured results to
verify the presence of intermodulation effect in a practical
system. This finding offers an explanation for the
disappointing discrete spur performance so often observed
in these synthesisers, which demands an effective solution
to improve such performance.
The aim of this work is to investigate a method to reduce
the impact of intermodulation-induced spurious tones.
A phase compensation technique is proposed to suppress at
least part of these fractional spurs – in particular the most
dominant sidebands. A generic module has been developed
by coupling a small amount of controlled amplitude and
phase voltage controlled oscillator (VCO) signal onto the
phase/frequency detector (PFD) input to cancel-out the
potential intermodulation effect. The measured results
demonstrate that the proposed technique can effectively
suppress close-in sideband spurs and is therefore capable of
enhancing spectral purity in wide bandwidth.
2 Intermodulation-induced spur distribution
There are two main frequencies existing in a fractional-N
synthesiser, one is the reference frequency f
ref
, the other is
VCO output frequency kf
ref
, where k is the average division
ratio of the synthesiser. The intermodulation occurs when
f
ref
and kf
ref
present at a same time in the loop. One
possible scenario of intermodulation occurs at the PFD
[19], as illustrated in Fig. 1. In this case, the potential
intermodulation frequencies [16] equal to
f
d
= f
ref
(m + kn) ; f
ref
(m + n(N +
a
)) (1)
where m and n are in tegers and ( N + α)=k is the division
ratio, with N the integer and α the fractional component,
named fractionality. The following conclusions can be
derived from (1): the intermodulation frequencies are
discretely distributed at an even space within the range of
[−f
ref
/2 f
ref
/2]. There is a value p which is the smallest
integer, satisfying pα equals to an integer. The space of
intermodulation frequencies Δf is equivalent to f
ref
/p. In the
worst case, the intermodulation frequency separations equal
to the desired channel spacing, which is a situation identical
www.ietdl.org
1064 IET Microw. Antennas Propag., 2013, Vol. 7, Iss. 13, pp. 1064–1071
&
The Institution of Engineering and Technology 2013 doi: 10.1049/iet-map.2013.0036