IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS— II: EXPRESS BRIEFS, VOL. X, NO.

X, JULY 2012 1

A Low In-Band Radiation Superregenerative

Oscillator
Pere Palà-Schönwälder, Member, IEEE, Jordi Bonet-Dalmau, Member, IEEE, F. Xavier Moncunill-Geniz,
Francisco del Águila-López and Rosa Giralt-Mas

Abstract—This paper describes a superregenerative voltage-

controlled oscillator as a building block for superregenerative
receivers where most of the oscillator spectrum components are
outside the reception frequency band. This allows to overcome
one of the main drawbacks of superregenerative receivers, Fig. 1. Equivalent circuit of a SRO (when C(t) is constant) or an
namely the potential interference to nearby receivers operating SRVCO. The constant conductance G0 includes the source resistance and the
at the same frequency due to oscillator reradiation. We perform losses in the tank circuit while the periodic time-varying conductance Ga (t)
a qualitative analysis of the solution of the circuit equations, de- determines, among other properties, the stability or instability periods of the
scribe the most relevant parameters for design and provide some circuit [13].
numerical simulation results. Experimental results on a proof-of-
concept implementation validating the described principle and a wireless sensor networks [12], one of the currently most
discussion of the observed behavior are provided. promising fields of application of SR receivers. This undesired
Index Terms—Radio receiver, superregenerative receiver, os- phenomenon is due to the fact that the main building block of
cillator reradiation, low power. the SR receiver, the SR oscillator (SRO), generates RF bursts
of appreciable amplitude at a frequency f0 which is very close
to f . In conventional receivers, reradiation is minimized by
I. I NTRODUCTION designing the low-noise amplifier with the additional constraint

T HE superregenerative (SR) receiver [1], [2] has proven

to be a suitable architecture for low-cost and low-power
wireless data links. This is due to a significantly reduced com-
of exhibiting high reverse isolation. The required isolation may
be achieved making use of a cascode structure, but this places
a lower limit on the power supply voltage. Other alternatives,
plexity, especially in comparison with conventional receivers. such as making use of two cascaded amplifier stages, also
This receiver has been traditionally used in remote control result in increased power consumption.
systems, short distance telemetry and wireless security, where In this paper we propose a new idea to overcome the
the drawbacks of this receiver, mainly its excessive reception problem of oscillator reradiation. We design a SR voltage-
bandwidth in comparison with conventional receivers, are controlled oscillator (SRVCO) and provide a control signal
more than compensated by the reduction in complexity. In such that the SRVCO frequency is f during the sensitivity
the last years, the SR receiver has been revisited and several period of the quench cycle. Once the oscillator amplitude starts
integrated implementations have been reported [3]–[5], some growing, the oscillator frequency is increased (or decreased).
of its drawbacks have been eliminated and its field of applica- The envelope of the generated waveform carries the required
tion has been broadened, with papers reporting applications to information for signal demodulation (as in a conventional SR
spread-spectrum [6] and UWB communications [7], [8]. On receiver) while the signal energy is shifted to frequencies
the other hand, applications of these receivers beyond classic higher (or lower) than f , strongly reducing the interference
ASK modulation have been reported [9], [10] and the SR to other receivers operating at f . The spectral energy shifting
principle has even been extended to broadband amplifiers and effect combined with the bandpass responses of the matching
mixers [11]. circuits and the antenna, which are all typically narrowband
The design of SR receivers intended to detect signals at around f , contribute to significantly reducing the restrictions
a carrier frequency f requires paying special attention to the on isolation of the low-noise amplifier.
potential problem of oscillator reradiation, especially (although
not only) through the antenna, which may interfere with II. VOLTAGE -C ONTROLLED S UPERREGENERATIVE
nearby receivers operating at or near the same frequency. O SCILLATOR
For instance, this is a significant problem when developing Figure 1 depicts an equivalent circuit of several possible
implementations of an SRO (if C(t) is constant) or an SRVCO
Manuscript received xx, 2012; revised xx, 2012. Work supported by the (if C(t) is time-varying). This circuit is described by the
Spanish Dirección General de Investigación under Grant TEC2009-09924.
P. Palà-Schönwälder, J. Bonet-Dalmau, F. del-Águila-López and R. Giralt- differential equation
Mas are with the Department of Electronic System Design and Programming 2Ċ(t) + G0 − Ga (t)
(DiPSE), School of Engineering of Manresa (EPSEM), Universitat Politècnica v̈o (t) + v̇o (t)+
de Catalunya (UPC), 08242 Manresa, Spain (e-mail: pere.pala@upc.edu). C(t)
F. Xavier Moncunill-Geniz is with the Department of Signal Theory and 1
(1)
Communications (TSC), Universitat Politècnica de Catalunya (UPC), 08034 L + C̈(t) − Ġa (t) 1
vo (t) = i̇(t).
Barcelona, Spain. C(t) C(t)
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS— II: EXPRESS BRIEFS, VOL. X, NO. X, JULY 2012 2

0.2

−0.2

Fig. 2. Waveforms of C(t) and G(t). The shaded area depicts the sensitivity (a)
period. 0

−20

In the linear mode of operation, the response of a classical −40

SRO with constant capacitance value, C(t) = C0 , to an input
−60
signal described by
−80
i(t) = Ipc (t) cos(ωt + φ), (2) (b)

where ω = 2πf , φ is a constant phase term, and pc (t) is a 1.1

normalized pulse envelope, may be approximately written as
[13]
1
vo (t) = IK|H(ω)|p(t) cos(ω0 t + φ + ∠H(ω)), (3)
0
where H(ω) is a bandpass function centered at ω0 = 2πf0
(c)
and the unity normalized oscillation envelope p(t) is given by
Rt Fig. 3. Simulation results. 3a) SRVCO waveform vo (t) [V] for constant
−ω0 ζ(λ) dλ C(t) = C0 . 3b) The envelope of vo (t) in dBV for constant C(t) = C0
p(t) = e tb
(4)
(bottom) and time-varying C(t) (top). 3c) Instantaneous pulsation of vo (t).
with p
ω0 = 1/ LC0 (5) C(t) and G(t) have a duty cycle of 50% and exhibit rise
and fall times which are 5% of the quench period. Figure 3a
and
G0 Ga (t) shows the generated signal vo (t), Fig. 3b shows the envelope
ζ(t) = (1 − ). (6) of vo (t), expressed in dBV to explicitly show the exponential
2C0 ω0 G0
dependence with time, and Fig. 3c shows the instantaneous
From (3) it follows that the oscillator signal vo (t) has a frequency of vo (t), computed from the zero crossings of vo (t).
strong frequency content around ω0 during the whole SRO Next, we discuss the observed behavior.
operation. In normal operation conditions, ω ≈ ω0 . On the Considering that the signal from the previous quench cycle
other hand, the peak gain K of the superregenerative oscillator, is negligible, the output signal up to t = t1 (see Fig.
which depends on several factors (most importantly, the time 2 and Fig. 3) corresponds to that of a conventional SRO
tb during which the circuit is unstable), is usually sufficiently with constant capacitance C(t) = C0 with the normalized
high to produce an output signal vo (t) with peak amplitudes oscillation envelope, p(t), given by (4). In this interval, the
of some hundreds of millivolts. These relatively high signal instantaneous frequency is constant and given by (5).
levels in the close vicinity of the input frequency ω may be Note also that if t1 is outside the sensitivity period, the
reradiated, interfering with nearby receivers. evolution for t > t1 is essentially independent of the input
Furthermore, in the classical SRO, vo (t) is only influenced signal. As a consequence, the overall frequency dependence of
by the input signal during a short time interval, the sensitivity the SRVCO will be determined by the frequency dependence
period, centered around the instant where the net conductance during the sensitivity period, which is given by |H(ω)|, as in
G(t) = G0 − Ga (t) of the circuit becomes negative [13], [13], with the center frequency given by (5).
which is taken as t = 0 in Fig. 2. The signal generated during In the time interval t2 ≤ t ≤ tb , C(t) = C1 . As
the sensitivity period together with the circuit parameters a consequence, the SRVCO signal is expected to have an
determine the future evolution of the circuit variables, meaning envelope similar to (4) substituting C0 by C1 in (5) and (6).
that there is no dependence on the input signal outside the Hence, most of the SRVCO signal energy will be located
sensitivity period. around
The solution of (1) for arbitrary variations of Ga (t) and
C(t) is not trivial and is beyond the scope of the current paper. p
ω1 = 1/ LC1 (7)
However, to gain insight into the circuit operation we may
consider the waveforms depicted in Fig. 2, where C(t) varies which is the value of the instantaneous frequency in this
between C0 and C1 and G varies between G0 and G1 . interval.
Figure 3 depicts the results of the numerical integration of Note that, since C1 < C0 , the slope in this interval (Fig.
the circuit equations taking normalized values L = 1, C0 = 1, 3b ) is higher than the slope for t < t1 , producing a higher
C1 = 0.8, G0 = 0.07, G1 = −0.02 and a sinusoidal input total gain. Choosing C1 > C0 the SRVCO would exhibit a
signal with I = 10µA and ω = 1, which correspond to lower gain than with fixed C(t) = C0 and most of the signal
reasonable values of a real design conveniently normalized. energy would be at frequencies lower than ω. The extra gain
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS— II: EXPRESS BRIEFS, VOL. X, NO. X, JULY 2012 3

Fig. 4. Schematic of the implemented SRVCO and envelope detector. VQ (t):

quench signal. VC (t): VCO control signal. VE (t): envelope detector output.

Fig. 6. SRVCO waveforms. Top: envelope detector output. Center: Varicap

control signal. Bottom: Quench signal.

RBW 100 kHz Marker 1 [T1 ]

VBW 100 Hz -93.90 dBm
Att 10 dB SWT 15 s 840.000000000 MHz

Fig. 5. RF input and output signal coupling.

-60

-70 0
achieved may be a good reason to choose C1 < C0 .
During the time interval t1 ≤ t ≤ t2 the SRVCO output -80 -10

frequency (Fig. 3c) changes as in a conventional VCO, since

-90 -20 1
the rate of change of C(t) is slow compared to the gener-
ated frequency. A soft transition is observed in the SRVCO
Center 880 MHz 15 MHz/ Span 150 MHz
envelope during this interval.
(a)
RBW 100 kHz Marker 1 [T1 ]
VBW 100 Hz -92.97 dBm
III. E XPERIMENTAL RESULTS AND DISCUSSION Att 10 dB SWT 15 s 840.000000000 MHz

To validate these ideas, a proof-of-concept SRVCO in the

800 MHz European ISM band has been built with discrete -60

components on an FR4 board following the schematic depicted

-70 0
in Fig. 4. As may be seen, the SRVCO is based on a common-
collector Colpitts oscillator based on a BFP405 transistor with -80 -10

a BB131 varicap diode controlling the resonant frequency of

-90 -20 1
the selective network. Component values are RV = 5.6 kΩ,
CB = 2.5 pF, C = 0.47 pF, L = 5.6 nH, CA = 1.5 pF,
Center 880 MHz 1 5 MHz/ Span 150 MHz
C1 = 2.7 pF, C2 = 1 pF. The envelope detector is also built
(b)
with a BFP405 transistor and allows further signal processing
and final bit decision in other stages. Component values are Fig. 7. Measured spectra for an SRVCO tuned to 840 MHz (indicated by the
marker and by the superimposed resulting frequency response) and providing
C0 = 12 pF, RB = 750 kΩ and CE = 82 pF. RE = 1.8 kΩ a frequency shift of 66 MHz. 7a) Spectrum in the absence of signal. 7b)
in both circuits and the VCC =3.3 V power supply is provided Spectrum in the presence of a CW signal at 840 MHz.
through conventional bypass and filtering elements.
Since we wished to investigate the inherent performance of
the SRVCO concept, no previous amplifier stage was provided. the center of the receiver sensitive passband at 840 MHz and
Also, as the objective is to make repeatable measurements, but provides a maximum frequency shift of +66 MHz. The bottom
there is no need to measure absolute sensitivity figures (there trace (trace #1) depicts the applied quench voltage vQ (t). This
is no reason to expect differences with previously reported signal is inversely related to the conductance G(t): The circuit
ones), we have used the simple ad-hoc test setup depicted is stable (G(t) > 0) for low values of the quench signal. For
schematically in Fig. 5. The RF input signal and the output this experiment a sawtooth quench signal was chosen, resulting
to the spectrum analyzer are loosely coupled to the SRVCO in a narrower receiver bandwidth, as predicted by conventional
making use of MCX connectors soldered on the ground plane theory [2], [13]. The quench frequency is 1 MHz, which is
closely to the SRVCO with no direct connection to the circuit within the modulation bandwidth of the VCO.
and using the center pins of the connectors as small antennas. The observed behavior closely matches the predictions. As,
Figure 6 shows the relevant signal waveforms of the pro- in this case C1 < C0 , the peak gain increases with decreasing
totype. The top trace (trace #2) is the envelope detector C1 , i.e. with increasing varicap control voltage.
output showing the receiver operating at the threshold of the Figure 7 shows the measured signal spectrum when the
logarithmic mode. The varicap control signal is the center SRVCO is tuned to 840 MHz, as indicated by the markers
trace (trace #3) in this figure. The adjustment shown places and the superimposed frequency response of the receiver. The
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS— II: EXPRESS BRIEFS, VOL. X, NO. X, JULY 2012 4

0
subfigures depict the signal generated by the SRVCO a) in the
absence of input signal (noise-driven response) and b) in the −10
presence of a CW signal at 840 MHz. These spectra exhibit
the characteristic shape of a SR receiver tuned to 840 MHz −20
but with the spectral peaks shifted upwards approximately
−30
+66 MHz. Note that the in-band radiated power density is −10 −5 0 5 10 15 20 25
attenuated by more than 35 dB (see Fig. 7b) compared to
what would be obtained with a classical SRO where the peak (a)
located at 906 MHz would be at 840 MHz. On the other 0
hand, the resulting out-of-band attenuation will be strongly
dependent on several factors which will be unique to each −10
design: the antenna Q, the antenna matching circuit bandwidth,
the combined frequency response of the LNA and the corre- −20

sponding LNA-VCSRO coupling circuit and, most important,

−30
the particular relative frequency shift α = ω1 /ω0 . Assuming −10 0 10 20 30 40 50 60 70
α = 1.08 and moderate values of Q (in the range of 10 to
15) in each of these bandpass terms, the combined response (b)
achieves ∼ 20 dB of attenuation to be added to the reverse Fig. 8. Measured frequency responses for an SRVCO operating with a)
+33 MHz frequency shift, b) +66 MHz frequency shift. : τ = 175 ns, ◦:
isolation of the LNA. τ = 203 ns, 4: τ = 230 ns, with τ = tb − (t1 + t2 )/2 (see also Fig. 2 ).
The measured receiver frequency response also matches that
of a conventional receiver, as expected. However, as the time
point t1 (where the capacitance starts changing) is decreased of arbitrary variations of the control signal of the voltage-
(which may be attractive to achieve more overall gain), an controlled oscillator and the development of design guidelines
increasing part of the sensitivity period includes a zone where to meet specific criteria are still open areas for further research.
the receiver is sensitive to higher frequencies. So, the shape
of the frequency response also gives a useful hint to indirectly R EFERENCES
determine the end of the sensitivity period. [1] E. H. Armstrong, “Some recent developments of regenerative circuits,”
An experimental demonstration emphasizing this effect is in Proc. IRE, Aug. 1922, vol. 10, pp. 244–260.
[2] J. R. Whitehead, Super-Regenerative Receivers. Cambridge, U.K.: Cam-
presented in Fig. 8a and Fig. 8b, which correspond to an bridge Univ. Press, 1950.
SRVCO adjusted to provide +33 MHz and +66 MHz shifts in [3] M. Pelissier, J. Jantunen, B. Gomez, J. Arponen, G. Masson, S. Dia,
the spectral peak, respectively. In these figures we show the J. Varteva, and M. Gary, “A 112 Mb/s full duplex remotely-powered
im- pulse-UWB RFID transceiver for wireless NV-memory applications,”
measured frequency responses for three different relative phase IEEE J. Solid-State Circuits, vol. 46, no. 4, pp. 916–927, Apr. 2011.
shifts between C(t) and G(t). There are steps of 10° phase [4] Vouilloz, A.; Declercq, M.; Dehollain, C., ”A low-power CMOS super-
shift, corresponding to 27.8 ns, between traces. As the figure regenerative receiver at 1 GHz,” Solid-State Circuits, IEEE Journal of ,
vol.36, no.3, pp.440-451, Mar 2001
caption shows, we have referenced these traces to tb instead [5] Jia-Yi Chen; Flynn, M.P.; Hayes, J.P., ”A Fully Integrated Auto-Calibrated
of the origin, as tb is much better determined for the sawtooth Super-Regenerative Receiver in 0.13-µm CMOS,” Solid-State Circuits,
quench used in the experiments. IEEE Journal of , vol.42, no.9, pp.1976-1985, Sept. 2007
[6] F. Xavier Moncunill-Geniz, P. Palà-Schönwälder, C. Dehollain, N. Joehl
A detailed analysis of the resulting frequency response and M. Declercq, “A 2.4-GHz DSSS superregenerative receiver With a
is still open to current research and will be investigated in Simple Delay-Locked Loop,” IEEE Microw.Wireless Comp. Letters, vol.
15, no. 8, 2005, pp. 499-501.
detail elsewhere but the qualitative effect corresponds to the [7] M. Anis, R. Tielert, N. Wehn, ”Low power complementary-Colpitts self-
qualitative predictions outlined before. quenched super-regenerative ultra-wideband (UWB) bandpass filter in
CMOS technology,” in Proc. IEEE MTT-S Micr. Symp. Dig., 2008, pp.
1047-1049.
IV. C ONCLUSION [8] F. Xavier Moncunill-Geniz, J. Bonet-Dalmau, P. Palà-Schönwälder, F. del-
Águila-López and R. Giralt-Mas, “A High-Resolution UWB IR Superre-
In this letter we have presented a superregenerative voltage- generative Receiver Front End With an SRD Quench Shaper,” in IEEE
controlled oscillator. We have shown how the simultaneous Trans. Circuits Syst. II, Exp. Briefs, vol. 59, no. 2, pp. 93-97, Feb. 2012.
[9] L. Hernandez, S. Paton, “A superregenerative receiver for phase and
control of the quench signal and the oscillation instanta- frequency modulated carriers,” in Proc. IEEE ISCAS, 2002, pp. III-81
neous frequency allows retaining the essential properties of - III-84.
a conventional SR receiver with the advantageous ability to [10] P. Palà-Schönwälder, F. Xavier Moncunill-Geniz, J. Bonet-Dalmau,
F. del-Águila-López and R. Giralt-Mas, “A BPSK superregenerative
modulate the locally generated RF signal away from the receiver: Preliminary results,” in Proc. IEEE ISCAS, 2009, pp. 1537-1540.
reception frequency. Experimental results on a 800 MHz ISM- [11] P. Palà-Schönwälder, F. Xavier Moncunill-Geniz, J. Bonet-Dalmau,
band receiver validate the proposed approach, demonstrating F. del-Águila-López and R. Giralt-Mas, “Baseband superregenerative
amplification,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 56, no.
a significant reduction of in-band radiated power, lowering the 9, pp. 1930–1937, Sep. 2009.
interference to nearby receivers in the same frequency band. [12] Ayers, J.; Mayaram, K.; Fiez, T.S.; , ”An Ultralow-Power Receiver for
The choice of capacitance variation influences the envelope Wireless Sensor Networks,” Solid-State Circuits, IEEE Journal of , vol.45,
no.9, pp.1759-1769, Sept. 2010 .
of the generated RF bursts and may also be exploited to [13] F. Xavier Moncunill-Geniz, P. Palà-Schönwälder and O. Mas-Casals,
achieve a higher gain compared to a SR receiver operating “A generic approach to the theory of superregenerative reception,” IEEE
at a fixed frequency. The exact investigation of the effects Trans. Circuits Syst. I, Reg. Papers, vol. 52, no. 1, pp. 54–70, Jan. 2005.