A Low-Power Active Self-Interference Cancellation Technique ... - MDPI

Journal of Low Power Electronics and Applications

Article

A Low-Power Active Self-Interference Cancellation Technique for SAW-Less FDD and Full-Duplex Receivers Saheed Tijani

ID

and Danilo Manstretta *

ID

Department of Electrical, Computer and Biomedical Engineering, University of Pavia, 27100 Pavia, Italy; [email protected] * Correspondence: [email protected]; Tel.: +39-382-98-5943 Received: 29 September 2017; Accepted: 10 November 2017; Published: 13 November 2017

Abstract: An active self-interference (SI) cancellation technique for SAW-less receiver linearity improvement is proposed. The active canceler combines programmable gain and phase in a single stage and is co-designed with a highly-linear LNA, achieving low noise and low power. A cross-modulation mechanism of the SI canceler is identified and strongly suppressed thanks to the introduction of an internal resistive feedback, enabling high effective receiver IIP3. TX leakage of up to −4 dBm of power is suppressed by over 30 dB at the input of the LNA, with benefits for the entire receiver in terms of IIP3, IIP2, and reciprocal mixing. The design was done in a 40 nm CMOS technology. The system, including receiver and active SI canceler, consumes less than 25 mW of power. When the canceler is enabled, it has an NF of 3.9–4.6 dB between 1.7 and 2.4 GHz and an effective IIP3 greater than 35 dBm. Keywords: blocker; cross-modulation; distortion cancellation; diversity; FDD; full-duplex; IIP3; SAW-less; self-interference

1. Introduction The evolution of mobile communication technologies ranging from GSM to the future 5G has continuously required increased data rates and quality of service and, in general, the miniaturization of devices and costs reduction. Off-chip Surface Acoustic Wave (SAW) filters are typically used to reject out-of-band interferers and Self-Interference (SI) signals (or transmitter leakage) but cause a significant form factor burden, even more so when the device operates over a large number of bands and uses multiple antennas. Also, the reconfigurability of receivers with SAW filters is difficult to achieve. Much attention is now focused toward SAW-less receivers, and several architectures have been proposed that are capable of tolerating large SI signals. There are various challenges that arise with the SAW-less receivers. For instance, when operating in Frequency Division Multiplexing (FDD) mode, where the transmitter (TX) and receiver (RX) operate simultaneously in different frequency bands, the TX leakage could inter-modulate with the out-of-band (OOB) blockers, thereby resulting in in-band third-order and second-order intermodulation, as well as generate reciprocal mixing. Furthermore, wideband TX noise leakage in the receive band can severely degrade receiver sensitivity [1]. Transceivers operating in Full Duplex (FD) mode, where the TX and RX overlap in time and frequency, are inherently SAW-less and pose an even greater SI challenge since a very large (in the order of 110 dB [2]) TX-RX isolation must be ensured, which also requires very high in-band linearity. In a single-antenna transceiver, a circulator can be used to provide isolation between transmitter and receiver. Alternatively, isolation can be achieved using separate antennas for transmitter and receiver. This basic isolation can be improved using RF cancelation architectures, either based on an RF canceller [3,4] or on an auxiliary transmitter [5]. The use on an auxiliary transmitter allows to J. Low Power Electron. Appl. 2017, 7, 27; doi:10.3390/jlpea7040027

www.mdpi.com/journal/jlpea

J. Low Power Electron. Appl. 2017, 7, 27

2 of 17

achieve very high cancellation levels and bandwidths even in the presence of strong power amplifier nonlinearity. However, it does not address the issue of transmitter noise leaking into the receiver. With a typical TX noise level in the order −160 dBc/Hz [6], if the transmitter leakage is sufficiently low (e.g., −20 dBm or less), this is not a major concern. If instead the transmitter leakage approaches 0 dBm, sensitivity degradation becomes large, and an RF canceller is potentially more attractive. A major issue of RF cancellation is given by the limited cancellation bandwidth, which results from the substantial delay (in the order of ns) between the TX signal and the SI at the RX input. Several board-level demonstrators have been implemented using off-the-shelf components, reporting strong cancellation levels over broad bandwidths. In [3], using a circulator showing multiple coupling paths with delays from 0.5 ns to 2 ns, 33 dB cancellation over 20 MHz and 18 dB cancellation over 100 MHz was reported using an RF canceller based on a two-tap delay line followed by programmable vector modulators under self-adaptive control. In [4], a three-tap RF canceller achieved 41 dB cancellation over 80 MHz bandwidth. Moving from the above mentioned solutions—based on off-the-shelf components mounted on a printed circuit board or connected using cables—to fully-integrated solutions, several implementation challenges related to component integration (e.g., RF delay lines) and dynamic range limitations are faced, typically leading to much smaller form factors but inferior cancellation levels and bandwidth. Prior works have proposed several integrated architectures to substitute the off-chip filters with tunable integrated solutions. Early works have advocated the use of passive hybrid transformers (HT) [7,8]. The HT allows us to connect TX and RX to the antenna port while ensuring isolation between TX and RX. The main limitation of HTs is that 3 dB loss is incurred between the TX and the antenna as well as between the antenna and the RX, degrading power efficiency and sensitivity. Furthermore, in typical HT configurations, the TX signal appears as a large common-mode signal at the LNA input due to parasitic coupling capacitors. A large common-mode leakage could saturate the LNA. The common-mode leakage could also cross-modulate with a jammer, thereby degrading the receiver linearity. The noise of the broadband common-mode signal could also appear at the RX signal thereby degrading the receiver sensitivity. The author [7] proposed a differential HT that would cancel out this common mode leakage signal, but this would nearly double the power consumption and area and additional loss from the balun that is needed to convert the transceiver differential signal to single ended at the antenna port. In [8], the issue of common-mode coupling due to parasitic capacitor was not addressed. The author relied on the linearity of the LNA design, which tolerated blocker of up to −15 dBm. Due to the frequency variability of the antenna impedance, transmitter-to-receiver isolation is in the order of 50 dB for a signal bandwidth of 6 MHz as illustrated on the TX/RX isolation plot. Other single antenna self-interference cancellation works have also been proposed in [9–16]. The use of a dual antenna architecture can significantly relax the design challenges since the limited coupling provides isolation between TX and TX without incurring any signal loss. However, the usual isolation between a pair of antennas in a mobile device is typically in the order of 20–30 dB [17] in contrast to a HT-based transceiver. Hence, a highly linear and tunable self-interference canceler (for FDD and FD) is required to further attenuate the TX leakage. Other works have advocated the replacement of the off-chip filters with SI cancellation techniques [11,18] or the use of N-path filters [13,19] to either achieve FDD or FD operations. The SI canceler must have variable gain and it must be able to track and adjust its phase over the entire 360◦ phase shift. In the two-point cancellation architecture proposed in [18] the LNA is based on the noise-cancelling approach and is able to tolerate up to 2 dBm TX leakage power while achieving high effective OOB IIP3. However, the TX leakage is effectively cancelled only after down-conversion and recombination between common-gate and common-source paths. As a result, it does not address SNR degradation due to IIP2 and reciprocal mixing. An N-path filter based canceler, which was designed for both FDD and FD applications has been proposed in [20] using dual antenna architecture. The use of passive mixer switches in the N-path filter results in large power dissipation and the N-path LO is prone to reciprocal mixing. With 30 dB isolation between the TX and RX, up to −4 dBm TX leakage power was tolerated by the filter. However, the effective IIP3 is only improved by 10 dB for

J. Low Power Electron. Appl. 2017, 7, 27

3 of 17

J. Low Power Electron. Appl. 2017, 7, 27

3 of 17

more than dB of leakage which highlights an intrinsic from the by 10 dB for25more than 25 dBsuppression, of leakage suppression, which highlights anlinearity intrinsiclimitation linearity limitation canceler. A low-power mixer-first receiver with passive mixer based self-interference cancellation is from the canceler. A low-power mixer-first receiver with passive mixer based self-interference proposed in is [21] for dualinantenna handlesFD. a SIItofhandles up to −a10 25 dB of and suppression cancellation proposed [20] for FD. dualItantenna SI dBm of upand to −10 dBm 25 dB of is achieved over a limited bandwidth. Moreover, the effective IIP3 improvement is limited. The SI suppression is achieved over a limited bandwidth. Moreover, the effective IIP3 improvement is cancellation is done after the mixer, hence, reciprocal mixing is an issue for this architecture. A passive limited. The SI cancellation is done after the mixer, hence, reciprocal mixing is an issue for this canceler based on an analog FIR filteroninan the RF side FIR filter the baseband was architecture. A passive canceler based analog FIRand filteranother in the RF side andinanother FIR filter in proposed by [22] for dual antenna FD systems. The FIR filter taps consist basically of a true-time-delay the baseband was proposed by [21] for dual antenna FD systems. The FIR filter taps consist basically circuit, a buffer and acircuit, variable gain amplifier. A 40 MHz cancellation bandwidth was achieved over a of a true-time-delay a buffer and a variable gain amplifier. A 40 MHz cancellation bandwidth 50 dB leakage cancellation, however at the expense of high power consumption and noise degradation was achieved over a 50 dB leakage cancellation, however at the expense of high power consumption which scales with the number of FIRwith taps.the number of FIR taps. and noise degradation which scales This paper paper describes describes an an active active SI SI cancellation cancellation technique technique for for aa SAW-less SAW-less diversity diversity receiver, receiver, This extending the conference publication [23]. In this work, the design is extended including the extending the conference publication [22]. In this work, the design is extended including the downdown-conversion mixers and baseband trans-impedance amplifiers (TIAs). Furthermore, an expanded conversion mixers and baseband trans-impedance amplifiers (TIAs). Furthermore, an expanded analysis of of the the system system requirements, requirements, performance reported. The The active active analysis performance analysis analysis and and design design details details are are reported. canceler was designed to adjust both its gain and phase to produce a signal that adds destructively to canceler was designed to adjust both its gain and phase to produce a signal that adds destructively thethe SI signal, acting as aas programmable vector modulator. The use the of SICthe in SIC the dual antenna system to SI signal, acting a programmable vector modulator. Theofuse in the dual antenna further increases the TX-RX isolation from below 30 dB to above 50 dB, as obtainable in a SAW-based system further increases the TX-RX isolation from below 30 dB to above 50 dB, as obtainable in a system. Thissystem. approach therefore, improves theimproves effective receiver linearity. It was shown It inwas [17] shown that in SAW-based This approach therefore, the effective receiver linearity. a typical multi-antenna configuration, group delay the of the SI path dueoftothe antenna coupling in [17] that in a typical handset multi-antenna handsetthe configuration, group delay SI path due to is typically in the order 2–3 ns and that, this limits the achievable cancellation to about 20 dB overtoa antenna coupling is typically in the order 2–3 ns and that, this limits the achievable cancellation bandwidth of over 10 MHz. Improved cancellation be achieved integrating an analog delay about 20 dB a bandwidth of 10 MHz.bandwidth Improved may cancellation bandwidth may be achieved line in the canceller as proposed in [22] or using an additional digital cancellation path, as advocated in integrating an analog delay line in the canceller as proposed in [21] or using an additional digital Figure 1, due to space limitations issues related to limited cancellation bandwidth as a result of larger cancellation path, as advocated in Figure 1, due to space limitations issues related to limited antenna group delay areas left for future work. This paper is delay organized as follows: describes cancellation bandwidth a result of larger antenna group are left for futureSection work. 2This paper the system architecture and the system level requirements of the receiver chain. Section 3 describes is organized as follows: Section 2 describes the system architecture and the system level requirements thethe circuit implementation active the SI canceler and the receiverofchain. Section 4 discusses of receiver chain. Sectionof3the describes circuit implementation the active SI canceler and the the simulation results. Section 5 concludes the paper. results. Section 5 concludes the paper. receiver chain. Section 4 discusses the simulation

Figure 1. System block diagram. Figure 1. System block diagram.

2. System Architecture and Requirements 2. System Architecture and Requirements The proposed receiver with active SIC is shown in Figure 1. The analog RF canceler senses the The proposed receiver with active SIC is shown in Figure 1. The analog RF canceler senses the TX signal from the PA and cancels out the modulated TX leakage signal at the input of the LNA. An TX signal from the PA and cancels out the modulated TX leakage signal at the input of the LNA. auxiliary receive path is used to sense the broadband TX noise and cancel it out in the digital domain, An auxiliary receive path is used to sense the broadband TX noise and cancel it out in the digital as reported in [244]. Digital cancellation also allows to improve the cancellation level provided by the domain, as reported in [24]. Digital cancellation also allows to improve the cancellation level provided RF canceller and to extend the cancellation bandwidth. This issue however, is not covered in this by the RF canceller and to extend the cancellation bandwidth. This issue however, is not covered in work due to space limitations. The choice of cancelling the leakage signal at the LNA input relaxes this work due to space limitations. The choice of cancelling the leakage signal at the LNA input relaxes the stringent linearity requirement of the entire receive chain, including the LNA, both in terms of the stringent linearity requirement of the entire receive chain, including the LNA, both in terms of IIP2 IIP2 and IIP3, and reciprocal mixing is strongly reduced. However, the noise of the canceler is directly and IIP3, and reciprocal mixing is strongly reduced. However, the noise of the canceler is directly seen seen at the receiver input port and, therefore, the canceler must be designed with lower noise contribution. Canceling the SI after the LNA, e.g., as proposed in [255], could result in LNA clipping and receiver desensitization at large SI levels. Similar considerations apply to the SI architecture

J. Low Power Electron. Appl. 2017, 7, 27

4 of 17

J. Low Power Electron. Appl. 2017, 7, 27

4 of 17

at the receiver input port and, therefore, the canceler must be designed with lower noise contribution. Canceling the[21], SI after theSIC LNA, e.g., as proposed in [25], Furthermore, could result inthe LNA clipping and proposed in where is carried out at baseband. introduction of areceiver downdesensitization at large levels. Similar considerations apply to the SI architecture proposed in [22], conversion mixer in theSI SIC path may cause reciprocal mixing with the LO phase noise, further where SIC is out at baseband. Furthermore, the introduction of a down-conversion mixer in increasing thecarried RX noise. the SIC path may cause reciprocal mixing with the LO phase noise, further increasing the RX noise.

2.1. Receive Chain 2.1. Receive Chain 2.1.1. Requirements for FDD FDD 2.1.1. Requirements for In aa dual dual antenna antenna mobile mobile terminal, terminal, the the isolation isolation between between TX TX and and RX RX antennas antennas can can be be around around 20 20 dB dB In to 30 dB [17]. Here, we assume 25 dB TX-RX antenna isolation as a typical value. For the FDD scenario, to 30 dB [17]. Here, we assume 25 dB TX-RX antenna isolation as a typical value. For the FDD scenario, we consider consider as asaareference reference4G 4GLTE LTEband band2 2with with2020MHz MHz channel bandwidth [266]. With maximum we channel bandwidth [26]. With the the maximum TX TX power of 23 dBm, the maximum TX leakage signal arriving at the RX antenna is −2 dBm. Removing power of 23 dBm, the maximum TX leakage signal arriving at the RX antenna is −2 dBm. Removing external filters the receiver, OOB blockers as high as −15 present at the receiver external filtersin infront frontofof the receiver, OOB blockers as high asdBm −15may dBmbemay be present at the input together with the −2 dBm TX leakage. This can cause strong in-band intermodulation if the receiver input together with the −2 dBm TX leakage. This can cause strong in-band intermodulation receiver IIP3 is not sufficiently high. In the presence of a strong OOB blocker, the receiver sensitivity if the receiver IIP3 is not sufficiently high. In the presence of a strong OOB blocker, the receiver of −92 dBmof(for modulated wanted signals) can be degraded by 9 dB for MHz channel sensitivity −92QPSK dBm (for QPSK modulated wanted signals) can be degraded by 20 9 dB for 20 MHz bandwidth [266]. Considering a signal-to-noise ratio of −1 dB for QPSK, the maximum channel bandwidth [26]. Considering a signal-to-noise ratio of −1 dB for QPSK, the maximumnoise noise ++ distortion level level is is − −82 has the the maximum maximum allowed allowed distortion 82 dBm. dBm. Assuming, Assuming, pessimistically, pessimistically, that that the the receiver receiver has noise figure figure of of 9.8 9.8 dB, dB, which which barely barely meets meets the the sensitivity sensitivity requirement, requirement, the the maximum maximum third-order third-order noise intermodulation level is −82.6 dBm. Hence, in the worst case, i.e., when the OOB blocker appears at at intermodulation level is −82.6 dBm. Hence, in the worst case, i.e., when the OOB blocker appears a frequency offset from the receiver that is exactly twice the TX-RX frequency spacing, the effective a frequency offset from the receiver that is exactly twice the TX-RX frequency spacing, the effective receiver IIP3 IIP3 (as + (−15 = 31.8 dBm.dBm. SuchSuch a high receiver (as defined defined in in [17]) [18]) should shouldbe belarger largerthan than−2 −2dBm dBm + (−+1582.6)/2 + 82.6)/2 = 31.8 a value is nearly impossible to achieve using standard LNA design techniques and represents high value is nearly impossible to achieve using standard LNA design techniques and represents aa challenging target target even even for for SIC SIC FDD FDD receiver receiver architectures. architectures. Assuming Assuming an SIC of of 20 20 dB dB in in front front of of challenging an RF RF SIC the receiver, the receiver IIP3 requirement is decreased by the same amount to a more feasible value the receiver, the receiver IIP3 requirement is decreased by the same amount to a more feasible value of 11.8 11.8 dBm. dBm. Nonetheless, Nonetheless, the also very very important important and and can can easily easily of the nonlinearities nonlinearities of of the the canceller canceller are are also become the the bottleneck bottleneck for for the the entire entire receiver. receiver. To Toillustrate illustratethis thisissue, issue,we werefer referto toFigure Figure2.2. become

Figure Figure 2. 2. Canceler Canceler feedback feedback linearization. linearization.

The canceler sees TX) and a large blocker signal (VBK ) BK at )the sees the the large large TX TXsignal signalatatthe theinput input(V(V a large blocker signal (V at TX ) and output. This This causes two types of nonlinearities. First, due to the canceler gain, the TX signal the output. causes two types of nonlinearities. First, duenonlinear to the nonlinear canceler gain, the generates products inproducts the vicinity of vicinity the TX band. mode, TX and RX TX signal intermodulation generates intermodulation in the of theIn TXFD band. Insince FD mode, since bands coincide, this is the main nonlinearity mechanism affecting the receiver effective IIP3. Second, TX and RX bands coincide, this is the main nonlinearity mechanism affecting the receiver effective due to the canceler reverse isolation, its output signals, including its nonlinear are IIP3. Second, due tolimited the canceler limited reverse isolation, its output signals, includingproducts, its nonlinear modulated by the blocker at the canceler output (i.e., at the receiver input). This creates crossproducts, are modulated by the blocker at the canceler output (i.e., at the receiver input). This creates modulation products between the TXthe signal theand blocker, that degrades the SIC receiver cross-modulation products between TX and signal the blocker, that degrades the SIC effective receiver IIP3 in FDD Wemode. express canceler using a Taylor series expansion follows: effective IIP3mode. in FDD Wethe express thecharacteristic canceler characteristic using a Taylor series as expansion as follows:

(1) V = (α V + α V +2 α V )(13 +V ) Vo = α1 VTX + α2 VTX + α3 VTX (1 + λVB ) (1) and we assume that the TX signal is given by two tones (V = V cos(ω t) + V cos(ω t)) and the and we assume that = theVTX signal given by nonlinear two tones mechanisms (VTX = VA cos the (ω1 t) + Vabove 2 t)) andtwo blocker is CW (V cos(ω t)).is The two described generate A cos(ω blocker CW (VBKnonlinear = VB cosproducts: nonlinear mechanisms described above generate (ωB t)). The distinct is 3rd order thetwo intermodulation between the TX input tones (Vo,IM3) two and the cross-modulation between the TX tones and the blocker (Vo,XM3). V

,

= (3⁄4)α V (cos(2ω − ω )t + cos(2ω − ω )t)

(2)

J. Low Power Electron. Appl. 2017, 7, 27

5 of 17

distinct 3rd order nonlinear products: the intermodulation between the TX input tones (Vo,IM3 ) and the cross-modulation between the TX tones and the blocker (Vo,XM3 ). Vo, IM3 = (3/4)α3 V3A (cos(2ω1 − ω2 )t + cos(2ω2 − ω1 )t)

(2)

Vo,XM3 = (1/4)α2 λV2A VB (cos(2ω1 − ωB )t + cos(2ω2 − ωB )t + 2 cos(ω1 + ω2 − ωB )t)

(3)

To mitigate the cross-modulation product, which is the limiting factor in FDD mode, a feedback path (β) is introduced from the canceler output to its input. Hence, the canceler input signal becomes VIN = VTX − βVB , and an additional cross-modulation term is generated as follows: V0 o,XM3 = (1/4)α3 βV2A VB (cos(2ω1 − ωB )t + cos(2ω2 − ωB )t + 2 cos(ω1 + ω2 − ωB )t)

(4)

Setting the feedback factor properly (β = λα2 /α3 ) the two cross-modulation products cancel each other: Vo,XM3 + V0 o,XM3 ∼ = 0. Notice that since both the canceler gain (α1 ) and the feedback factor (β) are small, the loop gain is much smaller than one and hence no stability issue arises. 2.1.2. Requirements for FD In full duplex transceivers, the simultaneous transmission and reception of signals at the same time and frequency makes self-interference cancellation an even greater challenge. For this reason, FD transceivers typically target communication links over shorter distances, where lower transmitted power levels and receiver sensitivities can be tolerated. In this work, we envision a dual antenna FD system with a transmitted power of 10 dBm and a sensitivity of −82 dBm. Assuming 25 dB antenna isolation, the TX leakage is −15 dBm and the total cancellation needed to have the SI below the RX sensitivity is (82 − 25 + 10 =) 67 dB. The use of an RF canceler would not be able to achieve this cancellation due to nonidealities and low resolution constraints in the canceler design. Assuming 27 dB of cancellation in the RF domain, the further 40 dB would be done in digital domain (e.g., using an auxiliary path). Notice that both the TIA and the ADC must have a dynamic range (DR) greater than 40 dB. The effective receiver in-band IIP3 requirement can be computed assuming a 3 dB sensitivity degradation when the transmitted power is maximum, i.e., −15 + (−15 + 82)/2 = 18.5 dBm. 2.2. Self-Interference Canceler Architecture Prior works have proposed different ways of generating the SI cancellation signal. In [18], the SI canceller consists of a variable phase shifter followed by a programmable gain amplifier. Cascading the two functions degrades noise and linearity performances and raises the power consumption. An alternative solution to generate the cancellation signal requires the use of a passive RC-CR quadrature splitter followed by two programmable gain amplifiers [27]. A single-stage RC-CR can be used since a high precision quadrature is not required for this application. Nonetheless, the quadrature splitter loads the driving stage and increases the noise. In this work, the generation of the cancellation signal was achieved by merging the quadrature splitter with the variable gain functions. The canceller acts as two transconductors that generate and in-phase current signal (I) and a quadrature current signal (Q), as shown in Figure 3. Both I and Q are generated by independently programmable transconductors. Hence, the vector sum of these currents can be set to be equal in magnitude and opposite in phase to the SI current at the receiver input. An important observation from [27] is that the quadrature precision has relaxed requirements. In fact, assuming that the vector signals (I and Q) generated by the transconductors are not in perfect quadrature but have a phase error ∆Φ, as illustrated in Figure 4a. The result is that the sum vector will be I + Q0 instead of I + Q. Since both I and Q are independently programmable, it will be possible to adjust the magnitude of I and Q0 such that the desired vector is generated. However, the phase error will reduce the maximum magnitude that can be generated from A to A0 = Acos(∆Φ). So, if a 5% range reduction is acceptable, the maximum quadrature error will be arccos(0.95) = 18.2◦ .

loads the driving stage and increases the noise. In this work, the generation of the cancellation signal was achieved by merging the quadrature splitter with the variable gain functions. The canceller acts as two transconductors that generate and in-phase current signal (I) and a quadrature current signal (Q), as shown in Figure 3. Both I and Q are generated by independently programmable transconductors. Hence, vector sum of these currents can be set to be equal in magnitude6 and J. Low Power Electron. Appl. 2017,the 7, 27 of 17 opposite in phase to the SI current at the receiver input.

J. Low Power Electron. Appl. 2017, 7, 27

6 of 17

quadrature but have a phase error ΔΦ, as illustrated in Figure 4a. The result is that the sum vector will be I + Q′ instead of I + Q. Since both I and Q are independently programmable, it will be possible to adjust the magnitude of I and However, the phase (a)Q′ such that the desired vector is generated. (b) error will reduce the maximum magnitude that can be generated from A to A′ = Acos(ΔΦ ). So, if a Figure 3. (a) Canceler block design concept; (b) Canceler gain adjustment. Figure 3. (a) Canceler design quadrature concept; (b) Canceler gain 5% range reduction is acceptable, theblock maximum error will beadjustment. arccos(0.95) = 18.2°. An important observation from [277] is that the quadrature precision has relaxed requirements. 2.3. Magnitude and Phase Errors on Cancellation In fact, assuming that the vector signals (I and Q) generated by the transconductors are not in perfect The required resolution resolutionof ofthe thecanceler cancelerdepends dependsonon the desired cancellation level. In practice, the desired cancellation level. In practice, the the cancellation level is also severely limited by the frequency selectivity of the coupling between the cancellation level is also severely limited by the frequency selectivity of the coupling between the TX TX antennas. shown in [17], Reference consideringsource a typical platform size ofa 6typical cm bymobile 10 cm, andand RX RX antennas. As As shown in [Error! notmobile found.], considering aplatform pair of planar antennas, commonly used for mobile wireless commonly applications, show typical size ofinverted-F 6 cm by 10 cm, a pair of planar inverted-F antennas, used fora mobile coupling of −20 dB and a group delaycoupling in the order of 2dB ns.and Foraagroup broadband this of delay wireless applications, show a typical of −20 delaycanceler, in the order 2 ns.results For a in a minimum cancellation of 20results dB over 14 MHz bandwidth. Hence, for over a broadband canceler such broadband canceler, this delay in a minimum cancellation of 20 dB a 14 MHz bandwidth. as the one in canceler this work, it isasreasonable to determine the work, required resolution based Hence, forconsidered a broadband such the one considered in this it iscanceler reasonable to determine on target cancellation level of 27 dB. on a target cancellation level of 27 dB. thearequired canceler resolution based If the canceler is affected by aa finite ∅), the resulting finite magnitude magnitude error error (V (Vε)) and and phase phase error error ((∅), relative cancellation error is computed as follows: cancellation error is computed as follows: s   (5) ε = −1 + cos∅ Vε 2 − sin 2∅, − sin ∅, (5) ε = −1 + cos ∅ ± VSI is the cancellation signal from the canceler, ∅ is the where V is the self-interference signal, V phase V error, ε is the magnitudesignal, error Vand V the is cancellation the residualsignal leakage The maximum where self-interference fromsignal. the canceler, ∅ is the canc is SI is the magnitude errors forerror 20, 25 and cancellation aresignal. plottedThe in Figure 4c. Ifmagnitude a canceler phase error,and ε is phase the magnitude and Vε30isdB theofresidual leakage maximum architecture based a phase by a variable gaininwas chosen, requiredarchitecture phase and and phase errors foron20, 25 andshifter 30 dB followed of cancellation are plotted Figure 4c. Ifthe a canceler magnitude resolution 25 dB by cancellation would 3.2° and 5.6% respectively. While the based on a phase shifterfor followed a variable gain wasbe chosen, the required phase and magnitude magnitudefor accuracy of 5.6% is easily achievable, a phase shifter wit better than resolution is not resolution 25 dB cancellation would be 3.2◦ and 5.6% respectively. While the 3.2° magnitude accuracy trivial. chosen architecture basedshifter on two quadrature vectors significantly relaxes theThe resolution of 5.6%The is easily achievable, a phase wit better than 3.2◦ resolution is not trivial. chosen requirements. In fact, assuming each vectors I and Qsignificantly vector to berelaxes programmable with requirements. n bits, the minimum architecture based on two quadrature the resolution In fact, achievableeach cancellation level isto[288] assuming I and Q vector be programmable with n bits, the minimum achievable cancellation level is [28]  2 SIC = 20log 2n (6) SICdB = 20 log √√2 (6) 2 with n = 5 bits a SIC greater than 27.1 dB can be ensured. with n = 5 bits a SIC greater than 27.1 dB can be ensured. 12

20dB

Mag. Error [%]

10

25dB

30dB

8 6 4 2 0 0

(a)

(b)

1

2 3 Phase Error (°)

4

(c)

Figure 4. 4. (a) range reduction due to to quadrature phase error; (c) Figure (a)Quadrature Quadraturephase phaseerror; error;(b) (b)Canceler Canceler range reduction due quadrature phase error; Cancellation in dB with corresponding output vector magnitude and phase error. (c) Cancellation in dB with corresponding output vector magnitude and phase error.

3. System Implementation The system is designed in 40 nm CMOS technology. Figure 5 shows the full system design that includes active self-interference canceler, LNTA, mixer, and TIA.

J. Low Power Electron. Appl. 2017, 7, 27

7 of 17

3. System Implementation The system is designed in 40 nm CMOS technology. Figure 5 shows the full system design that 7 of 17 canceler, LNTA, mixer, and TIA.

J. Low Poweractive Electron. Appl. 2017, 7, 27 includes self-interference

System architecture: architecture: Active Active self-interference self-interference (SI) (SI) canceler, canceler, Low Noise Transconductance Figure 5. System Amplifier (LNTA), (LNTA), passive passive mixer, mixer,and andtransimpedance transimpedancefilter. filter.

3.1. Low Noise Transconductance Amplifier Amplifier (LNTA) (LNTA)Design Design The LNTA, shown in Figure 5, was designed based on the topology described in detail in [1] and is used in this work work primarily because of its low noise and and high high linearity. linearity. High linearity and low noise is achieved achievedusing usinga transformer a transformer based complementary common-gate stagecurrent with reuse current reuse based complementary common-gate stage with between between NMOS and PMOS input transistors class-AB. Thetransformer three-coil transformer actsand as NMOS and PMOS input transistors working inworking class-AB.inThe three-coil acts as a balun a balun and provides a dual differential signal and to the NMOS andpairs. PMOS input pairs. The provide cascode provides a dual differential signal to the NMOS PMOS input The cascode stages stages provide high output impedance which makes suitable for proper operation with the currenthigh output impedance which makes it suitable for itproper operation with the current-mode mixer. mode mixer. The main difference with respect to [1] is that the gates of the input transistors are The main difference with respect to [1] is that the gates of the input transistors are not connectednot to connected toofthe of thebut transformer are simply cross-coupled. A similar was the primary theprimary transformer are simplybut cross-coupled. A similar approach wasapproach also followed also in [8]. This allows eliminate thethat external balun that was required in [1]. Crossin [8].followed This allows to eliminate the to external balun was required in [1]. Cross-coupling increases coupling increases significantly the immunity to common mode signals (including second-order significantly the immunity to common mode signals (including second-order distortion terms) that distortion terms) appear the input of the LNApresent since they equally at the appear directly at that the input of directly the LNAatsince they are equally at theare gate and atpresent the source of gate and at the source of the input devices. As a result, composed second order nonlinearity, which the input devices. As a result, composed second order nonlinearity, which is the dominant term in is term in the CG LNA third-order distortion, vanishes, boosting LNA IIP3. This thethe CGdominant LNA third-order distortion, vanishes, boosting the LNA IIP3. This allowsthe to reduce the LNA allows reduce LNA current from 6an to 4IIP3 mAofwhile achieving anasIIP3 of around 14 dBm, currentto from 6 tothe 4 mA while achieving around 14 dBm, shown in Figure 6a. as shown in Figure 6a.

J. Low Power Electron. Appl. 2017, 7, 27 J. Low Power Electron. Appl. 2017, 7, 27

8 of 17 8 of 17

3 -10

Fund

2

NF [dB]

Pout [dBm]

-60 -110

1

IM3

-160

14dBm

0

-210 -30

-20

-10

0

10

20

1.7

1.9

2.1

Freq [dB]

Freq [GHz]

(a)

(b)

2.3

2.5

Figure 6. (a) LNA IIP3; LNA frequency. Figure 6. (a) LNA IIP3; (b)(b) LNA NFNF vs.vs. RFRF frequency.

For the same reason, any common-mode second-order distortion term injected by the canceler For the same reason, any common-mode second-order distortion term injected by the canceler is is also suppressed and does not degrade the receiver IIP3. The noise figure of the LNTA is also also suppressed and does not degrade the receiver IIP3. The noise figure of the LNTA is also improved improved thanks to cross-coupling. Neglecting transformer losses, the noise factor of the crossthanks to cross-coupling. Neglecting transformer losses, the noise factor of the cross-coupled common coupled common gate is NF = 1 + [29]. Shown in Figure 6b is the simulated NF of the LNA, gate is NFCG = 1 + γ2 [29]. Shown in Figure 6b is the simulated NF of the LNA, including transformer including transformer losses. It has a NF of 1.6–2.3 dB in the range between 1.7 and 2.4 GHz. losses. It has a NF of 1.6–2.3 dB in the range between 1.7 and 2.4 GHz. 3.2. Active Self-Interference Canceler 3.2. Active Self-Interference Canceler The schematic proposed active is shown Figure consistsof oftwo two parallel The schematic of of thethe proposed active SICSIC is shown in in Figure 5. 5.It Itconsists parallel transconductance amplifiers (AMP I and AMPQ) with independently programmable gains and transconductance amplifiers (AMPI and AMPQ ) with independently programmable gains and ◦ phase quadrature output phases.To Toachieve achievethe the 360 360° I and AMPQ with quadrature output phases. phase coverage, coverage, aareplica replicaofofthe theAMP AMP I and AMPQ inverted polarities is used, but only one of the two instances is active at any given time.time. Each with inverted polarities is used, but only one of the two instances is active at any given transconductance amplifier consists of an array of identical elements (slices) that can Each transconductance amplifier consists of an array of identical elements (slices) that can bebe independently enabled and disabled, providing a total transconductance gain ms, digitally independently enabled and disabled, providing a total transconductance gain of of 2020 ms, digitally programmable with 5-bits of resolution. The cancellation current produced by these amplifiers programmable with 5-bits of resolution. The cancellation current produced by these amplifiers is is injected directly secondary terminals LNTA input transformer. The canceler is designed injected directly at at thethe secondary terminals of of thethe LNTA input transformer. The canceler is designed cancel dBmofofSISIsignal signal the LNTA input and consumes a maximum 6 mA. Each slice to to cancel upup to to −4−4dBm atat the LNTA input and consumes a maximum of of 6 mA. Each slice consists of a degenerated complementary transconductance stage. Resistive and capacitive consists of a degenerated complementary transconductance stage. Resistive and capacitive degeneration degeneration toquadrature generate nearly quadrature signals. allow to generateallow nearly signals. reality, gate drain-capacitance determines a dominant right-hand side zero gain In In reality, thethe gate drain-capacitance determines a dominant right-hand side zero in in thethe gain of of AMP I (GmI) while the finite transistors transconductance determines a dominant pole in the gain of AMPI (GmI ) while the finite transistors transconductance determines a dominant pole in the gain of AMP(G Q (GmQ), causing a large phase deviation from ideal quadrature. The expressions of GmI and GmQ AMP Q mQ ), causing a large phase deviation from ideal quadrature. The expressions of GmI and are GmQ are   gmI gm s s GmI = 1 − (7)(7) 1− Gm = 1 + gm ωZ ω 1 +I Rgm c R sCc Gm = sC Gm Q= 1 +s ωsp 1+ ω

(8) (8)

where ωz = gmI /C0 gd and ωp = gmQ /Cc at the frequency of interest. ωz is typically at a much where ω = gm ⁄C′ and ω = gm ⁄C at the frequency of interest. ω is typically at a much higher frequency compared with ωp . As a result, a significant phase error results that creates a tilt in higher frequency compared with ω . As a result, a significant phase error results that creates a tilt in the canceler output vector constellation and reduces the covered range. Adding an explicit capacitance the canceler output vector constellation and reduces the covered range. Adding an explicit Cgd in parallel to the intrinsic gate-drain capacitance of AMPI lowers ωz , eventually canceling out the capacitance C in parallel to the intrinsic gate-drain capacitance of AMPI lowers ω , eventually quadrature phase error (phase compensation). The phase difference between AMPI and AMPQ before canceling out the quadrature phase error (phase compensation). The phase difference between AMPI and after phase compensation is plotted in Figure 7. and AMPQ before and after phase compensation is plotted in Figure 7.

J. Low Power Electron. Appl. 2017, 7, 27 J. Low Power Electron. Appl. 2017, 7, 27

9 of 17 9 of 17

J. Low Power Electron. Appl. 2017, 7, 27

9 of 17

Figure 7. 7.Quadrature error compensation. Figure Quadrature error compensation. Figure 7. Quadrature error compensation.

Figure 8 shows normalized vector gain of active canceler for all combinations Figure 8 shows thesimulated simulated normalized vector gain thethe active canceler for combinations Figure 8 shows thethe simulated normalized vector gain ofofthe active canceler forall all combinations of of the control bits before and after the phase error compensation. of the control bits before and after phase error compensation. the control bits before and after the phase error compensation.

(a)

(b)

Figure 8. Canceler constellation: (a) before compensation; (b) after compensation.

(b)compensation. Figure 8. Canceler(a) constellation: (a) before compensation; (b) after Figure 8. Canceler 3.2.1. Canceler Noise Analysisconstellation: (a) before compensation; (b) after compensation.

3.2.1. Canceler Noise Analysis

The canceler output noise is added directly at the input of the receiver and must be kept low to

3.2.1. Canceler Noise Analysis The canceler output noise is added directly the receiver based and must be kept achieve low overall NF. Both AMPI and AMPatQ the add input noise of proportionally on their transconductance, with noise AMP contributing less noiseof thanks to the capacitive (noiseless) low to achieve lowoutput overall NF.Qis Both AMP and at AMP add noise proportionally based on their The canceler added directly the input the receiver and must be kept low to I relatively Q degeneration. The signal at the receiver input determines the cancelling condition set by the transconductance, withSIAMP contributing relatively less noise thanks to the capacitive (noiseless) achieve low overall NF. Both AMP I and AMP Q add noise proportionally based on their Q magnitude The (Gm) SI andsignal phase at (θ)the of the canceler transconductance (i.e., the gain settings of AMP I and degeneration. receiver input determines the cancelling condition set by the transconductance, with AMPQ contributing relatively less noise thanks to the capacitive (noiseless) AMPQ). The noise current injected at the input of the LNTA from the canceller, neglecting the magnitude (Gm )The and SI phase (θ) at of the canceler the gain settings of AMP and degeneration. signal receiver transconductance input determines (i.e., the cancelling condition set Iby the feedback resistors, is given as follows: AMP ). The noise current injected at the input of the LNTA from the canceller, neglecting the feedback magnitude (Gm) and phase (θ) of the canceler transconductance (i.e., the gain settings of AMPI and Q = 8KTγ σG + G , i (9) resistors, given as follows: AMPQ). isThe noise current injected , at the input of the LNTA from the canceller, neglecting the feedback resistors, is given as follows:   where σ = σ = (σ ≈ 1 for γ close to 1). The noise factor expression can be simplified ωo 2 i σGσG ++ G , (9)(9) mI (θ) n,canc = Grelation i , =(G8KTγ m) 8KTγ and phase inmQ further as a function of the magnitude ωp ,to the self-interference signal as expressed in (10).

where σ =γ +gm R σ = γ+gm R (σ ≈ 1 for γ close to 1). The noise factor expression can be simplified 8KTγG f(θ),The noise factor expression can be simplified (10) where σ = 1+gm IRcc σ = 1+gm IRcc (σ ≈ i1 ,for γ=close to 1). I I m) and phase (θ) in relation to the self-interference signal as further as a function of the magnitude (G further as a function of the magnitude (Gm ) and phase (θ) in relation to the self-interference signal as is the excess noise factor accounting for the different contributions where f(θ) = cosθ + sinθ expressed (10). expressed inin (10). 1 and. from AMPI and AMPQ. When AMPQ is 2 OFF f(θ) = 8KTγG 8KTγG f(θf(θ) ), ≪ 1 when AMPI is off, where ω ≫ (10) iin,canc = f(θ), (10) m , ω . The canceler noise factorreferred to its input is thus calculated as F =1+ , . The noise  ωo where θ ++sin θω is ) = =cos is the the excess excess noise noisefactor factoraccounting accountingfor forthe thedifferent differentcontributions contributions wheref(θf(θ) cosθ sinθ p from AMPI and AMPQ . When AMPQ is OFF f(θ) ≈ 1 and. f(θ)  1 when AMPI is off, where from AMPI and AMPQ. When AMPQ is OFF f(θ) 1 and. f(θ) ≪ 1 when AMPI is off, where ω ≫ ω . The canceler noise factor referred to its input is thus calculated as F

=1+

,

. The noise

J. Low Power Electron. Appl. 2017, 7, 27

10 of 17

J. Low Power Electron. Appl. 2017, 7, 27

10 of 17

i2

ωp  ωo . The canceler noise factor referred to its input is thus calculated as Fcanc = 1 + n,canc 2 . 4KTRs Gm current of current the canceler the input of the receiver expressed by: i , + , by:=i28KT(G = The noise of thereferred cancelertoreferred to the input of theisreceiver is expressed n,canc,ref 1 2n 1 2 )2n] , where n is the LNA input transformer turn ratio, R is the R (G+m +RR1 )f(θ)/[( 8KT + R1FQ )f(θ)R/[( RLNA ) ], where n is the LNA input transformer turn ratio, RLNA is the FI differential LNA input impedance and RRFI ,, R arethe thecross crosscoupled coupledfeedback feedbackresistances resistancesfor forAMPI AMPI RFQ are AMPQ respectively. Gm = Rs = = 50 50 Ω, the maximum maximum noise noise and AMPQ respectively. For For n n= = 0.7, 0.7, the canceler Gm = 20 mS 20 mS and Rs Ω, the contribution to the system when AMPI = 11 is is aa factor factor of of 0.58. 0.58. The The entire entire noise noise contribution AMPI is fully ON, for ff(θ) (θ) = therefore calculated calculated as as FFTOT = = FFRX + +G factor of the system is therefore G2(F − 1), 1), where where G is the the canceler canceler gain. gain. (Fcanc − Figure 99shows showsthe thesimulated simulatedand andcalculated calculated noise figure Canceler LNA where Figure noise figure of of thethe Canceler andand LNA onlyonly where the 2G (F the calculated is expressed as:LNA NF+Canc = 10 =log 10log [F + − 1)], where F = 1.44 calculated noisenoise is expressed as: NF [ F + G ( F − 1 )] , where F = 1.44 at canc LNA 10 LNA 2 GHz from Figure in Section 2atGHz from Figure 6b 6b in Section 3.1.3.1. 4 LNA Only

Sim. LNA+Canc

Calc. LNA+Canc

NF [dB]

3

2

1 0

120

240

360

Phase [deg] function of of the the phase phase shift. shift. Figure 9. Noise figure of LNA with canceler as function

3.2.2. Canceler Canceler Linearization Linearization 3.2.2. Due to to the the finite finite output output conductance conductance of of the the canceler canceler transistors transistors (MN1-4 (MN1-4 and and MP1-4), MP1-4), the the blocker blocker Due signal at at the the canceler canceler output output generates generates cross-modulation cross-modulation products products with with the the TX TX signal, signal, limiting limiting the the signal effective receiver IIP3. To address this issue, a feedback linearization technique is introduced, as effective receiver IIP3. To address this issue, a feedback linearization technique is introduced, as anticipated in section II. The feedback resistors (RFI and RFQ in Figure 5) were sized in such a way anticipated in section II. The feedback resistors (RFI and RFQ in Figure 5) were sized in such a that that a small fraction of the blocker signal appears inputs, generating generating way a small fraction of the blocker signal appearsatatthe theAMPI AMPIand and AMPQ AMPQ inputs, intermodulation products that cancel out with the input-output cross modulation terms, improving intermodulation products that cancel out with the input-output cross modulation terms, improving the receiver effective IIP3. Simulation results will be reported in the next section. the receiver effective IIP3. Simulation results will be reported in the next section. Trans-Impedance Amplifier Amplifier (TIA) (TIA) 3.3. Baseband Trans-Impedance current mode mode passive passive mixer mixer driven driven by aa 25% 25% duty-cycle duty-cycle LO LO was was used, used, followed followed After the LNA a current shown in Figure Figure 5. The TIA introduces a pole at 22 MHz to attenuate large OOB by a baseband TIA, as shown interferers. The TIA is based on a three-stage operational transconductance amplifier amplifier (OTA). (OTA). The OTA and is is designed in design is based based on on aa similar similarconcept conceptas asthe theone oneinin[30] [30]but butitithas hasa awider widerbandwidth bandwidth and designed 40 nm (versus 28 nm in [30]), which makes the design more challenging. In order to achieve wide in 40 nm (versus 28 nm in [30]), which makes the design more challenging. In order wide bandwidth for better linearity instead of using Miller compensation capacitors, stability is ensured by introducing zeros zeros both both within within the the OTA OTA and and in in the the feedback feedback network. network. The The first first stage of the TIA consists introducing of PMOS PMOS telescopic—cascode telescopic—cascode amplifier designed to have high gain in order amplifier designed to have high gain in order to ensure high linearity, while the the second second stage stage is is aa full full differential differential pair pair designed designed with both moderate moderate gain and distortion, distortion, and while the third stage work is aa complementary complementaryclass classAB, AB,to toensure ensurevery veryhigh highoutput outputvoltage voltageswing. swing. Analysing the stability of the TIA, the following critical poles and zeros are are identified. identified. The first stage has hasaadominant dominantpole, pole,ωω p,OTA,1 at 32 32 MHz MHzand andaahigh highfrequency frequency zero, ωz,OTA,1==1/(R 1/(RccCcc).). There is stage zero, ωz,OTA,1 p,OTA,1 also a second pole at high frequency, ω ωp,OTA,2 ≈ G m,L /C L,1 that must be well above the loop the unity-gain p,OTA,2 ≈ Gm,L /CL,1 that must be well aboveunity-gain bandwidth. To extend the bandwidth of the second stage, astage, feed-forward path is introduced through loop bandwidth. To extend the bandwidth of the second a feed-forward path is introduced Rz and Cz. introduces a zero-pole doublet, doublet, in whichin the pole, the ωp,OTA,3 zCz) at is placed through Rz This and Cz. This introduces a zero-pole which pole,atω1/(R 1/(R ) is z Cztwo p,OTA,3 times the zero frequency, and the zero is used to cancel out the pole introduced by the second stage placed two times the zero frequency, and the zero is used to cancel out the pole introduced by the finite output resistance, ro,2 and its load capacitance, CL,2. Simultaneously, the pole of the second stage, ωp,OTA,3 is compensated for by the zero of the first stage ωz,OTA,1. A second dominant pole at 5 MHz, ωp,OTA,5 is created through the feedback network and the third stage at 1/(Rd//(RF + ro)Cin), where RF is

J. Low Power Electron. Appl. 2017, 7, 27

11 of 17

second stage finite output resistance, ro,2 and its load capacitance, CL,2 . Simultaneously, the pole of the second stage, ωp,OTA,3 is compensated for by the zero of the first stage ωz,OTA,1 . A second dominant J.J.Low 11 LowPower PowerElectron. Electron.Appl. Appl.2017, 2017,7,7,27 27 11of of17 17 pole at 5 MHz, ωp,OTA,5 is created through the feedback network and the third stage at 1/(Rd //(RF + ro )C ),feedback where Rresistor, feedback resistor, Rd is theseen driving resistance seen looking thestage mixer and ro the resistance looking to and third in feedback F is the R the resistor, Rddisisthe thedriving driving resistance seen looking tothe themixer mixer andrrooisisto third stagefinite finite output resistance. In fact, a large C in (20 pF in this design) helps to filter out the blockers at high is third stage finite output resistance. In fact, a large C (20 pF in this design) helps to filter out the output resistance. In fact, a large Cin (20 pF in this design) helps to filter out the blockers at high in frequencies. With two the stable. RRininisisnot aasmall valued blockers at high frequencies. Withpoles, two dominant poles,not thebe loop would be stable. Rresistance is a small frequencies. With twodominant dominant poles, theloop loopwould would not be stable. small valued resistance in placed in introduce frequency zero, ω z,OTA,3 at 1/R ininC inin (around 700 MHz), valued resistance placed withaC to introduce a high frequency zero, ω at 1/R placed in series series with with CCinininto toseries introduce a high high frequency zero, ω z,OTA,3 at 1/R C (around 700 MHz), in z,OTA,3 in Cin which is placed before the GBW (at 1.5 GHz) to ensure stability. From the loop gain simulations which is placed before the GBW (at 1.5 GHz) to ensure stability. From the loop gain simulations (around 700 MHz), which is placed before the GBW (at 1.5 GHz) to ensure stability. From the loop gain shown has aaDC around 1.5 GBW aaphase of shownin inFigure Figure11, 11, theloop loop has DCgain gain of around 70 dB, 1.5GHz GHz GBW and phasemargin margin of simulations shown in the Figure 11, the loop hasof a DC gain70 ofdB, around 70 dB, 1.5and GHz GBW and a phase ◦ 85°. To ensure a stable DC quiescent point, a common-mode feedback circuit is introduced, and its 85°. To ensure stable aDC quiescent point, a common-mode feedback feedback circuit is introduced, and its margin of 85 . To aensure stable DC quiescent point, a common-mode circuit is introduced, output isisconnected to of stage shown Figure output connected tothe thePMOS PMOS ofthe thelast last stage OTA, asOTA, shown in Figure 10. and its output is connected to the PMOS of the lastOTA, stageas asin shown in10. Figure 10.

(a) (a)

(b) (b)

(a) operational transconductance amplifier (OTA); (b) TIA Figure 10. (a) Schematic Schematic of the the three-stage three-stage operational transconductance amplifier (OTA); (b)schematic. TIA FigureFigure 10. (a)10. Schematic of the of three-stage operational transconductance amplifier (OTA); (b) TIA schematic. schematic.

Figure 11. Loop gain magnitude and (TIA). Figure 11. Loopgain gainmagnitude magnitudeand andphase phaseof ofthe thetrans-impedance trans-impedance amplifiers (TIA). Figure 11. Loop phase of the trans-impedanceamplifiers amplifiers (TIA).

J. Low Power Electron. Appl. 2017, 7, 27

12 of 17

J. Low Power Electron. Appl. 2017, 7, 27  J. Low Power Electron. Appl. 2017, 7, 27 

12 of 17  12 of 17

4. Simulation Results 4. Simulation Results  4. Simulation Results The SIC receiver was designed in 40 nm CMOS technology and has a 1.8 V voltage supply. The SIC receiver was designed in 40 nm CMOS technology and has a 1.8 V voltage supply. The  The SIC receiver designed in 40 nm of CMOS and has 1.8 V voltageonly) supply. The The active canceler drawswas a maximum current 3 mAtechnology (AMPI only) or 6amA (AMP depending Q active canceler draws a maximum current of 3 mA (AMPII only) or 6 mA (AMP only) or 6 mA (AMPQ Qonly) depending on active canceler draws a maximum current of 3 mA (AMP  only) depending on  on the self-interference power level and its phase. The LNTA draws a current of 4 mA, while the TIAs the self-interference power level and its phase. The LNTA draws a current of 4 mA, while the TIAs the self‐interference power level and its phase. The LNTA draws a current of 4 mA, while the TIAs  drawdraw 1.8 mA each. The total power dissipation is 25 mW. The simulated performance of the canceler 1.8 mA each. The total power dissipation is 25 mW. The simulated performance of the canceler draw 1.8 mA each. The total power dissipation is 25 mW. The simulated performance of the canceler  alongside the entire receiver chain is is reported Figure12a 12ashows shows cancellation of SI the SI versus alongside the entire receiver chain reportedas asfollows. follows. Figure cancellation of the versus alongside the entire receiver chain is reported as follows. Figure 12a shows cancellation of the SI versus  frequency when the the SI is cancellation 2 GHz. 30ofdB of cancellation frequency when SIprogrammed is programmedfor foroptimum optimum cancellation at at 2 GHz. 30 dB cancellation over a over frequency when the SI is programmed for optimum cancellation at 2 GHz. 30 dB of cancellation over a  a bandwidth of 250 MHz is achieved when no delay is introduced between the TX signal and the SI. bandwidth of 250 MHz is achieved when no delay is introduced between the TX signal and the SI. The bandwidth of 250 MHz is achieved when no delay is introduced between the TX signal and the SI. The  30 dB cancellation bandwidth shrinks to 10 MHz with a delay of 2 ns. Figure 12b also shows the The 30 dB cancellation bandwidth shrinks to 10 MHz with a delay of 2 ns. Figure 12b also shows the 30  dB  cancellation  bandwidth  shrinks  to  10  MHz  with  a  delay  of  2  ns.  Figure  12b  also  shows  the  ◦ cancellation of the SI as a function of the coupling phase over the entire 360°. cancellation of the SI as a function of the coupling phase over the entire 360 . cancellation of the SI as a function of the coupling phase over the entire 360°.  -20

-30

CANC OFF CANC ON

Pout [dBm]

CANC ON w/ 1nS delay CANC ON

w/ 2nS delay CANC ON w/ 2nS delay

-70

‐70 -90

-30

SI Canc. [dB] SI Canc.  [dB]

Pout [dBm]

-50

‐50

w/ 1nS delay

CANC ON w/o delay CANC ON

‐90 -110

delay 1.9 w/o 2.0

1.8

‐110

‐20

CANC OFF

‐30

1.8

1.9

‐30 -40

‐40 -50

‐50 -60

‐60

-70

2.1

‐70 0 0

2.2

Freq 2.0 [GHz] 2.1

2.2

90

180

270

[deg] 90 Phase180

360

270

360

Phase [deg] 

Freq [GHz] (a)

 

(b)

(a) 

(b) 

Figure 12. Self-interference cancellation (a) over RF frequency; (b) as a function of canceler phase.

Figure 12. Self-interference cancellation (a) over RF frequency; (b) as a function of canceler phase. Figure 12. Self‐interference cancellation (a) over RF frequency; (b) as a function of canceler phase.  The simulated receiver NF is plotted in Figure 13. When the receiver NF is 4 dB, the noise The simulated receiver NF in Figure Whenthe  the receiver NFis is dB, the noise contribution of the canceler is is around 16%, LNA13. contributes around 42%NF  and the remaining The  simulated  receiver  NF  is plotted plotted  in the Figure  13.  When  receiver  4  4dB,  the  noise  blocks (Mixer and TIAs) is 2%, asaround  illustrated in Figure 13c.contributes  The system NF with the turned off contribution of the canceler around 16%, the LNA contributes around 42%42%  andcanceler the remaining blocks contribution  of  the  canceler  is  16%,  the  LNA  around  and  the  remaining  is shown in Figure 13a. The system NF without canceler has a NF of 3–3.7 dB in the range of 1.7 and (Mixer and TIAs) 2%, as illustrated in Figure 13c. The system NF with the canceler turned off is shown blocks (Mixer and TIAs) 2%, as illustrated in Figure 13c. The system NF with the canceler turned off  2.4 GHz. the canceler is turnedcanceler ON, it goes 3.9ofdB to 4.5dB dBin asthe a function of 1.7 the and coupling in is shown in Figure 13a. The system NF without canceler has a NF of 3–3.7 dB in the range of 1.7 and  Figure 13a.When The system NF without hasfrom a NF 3–3.7 range of 2.4 GHz. phase. The pattern of the noise shown in Figure 13b is consistent with the analysis in Section 3. When When the canceler is turned ON, it goes from 3.9 dB to 4.5 dB as a function of the coupling phase. 2.4 GHz. When the canceler is turned ON, it goes from 3.9 dB to 4.5 dB as a function of the coupling  AMPI, with degenerated resistor, is turned ON more noise is generated, while when only AMPQ is The pattern of the noise shown in Figure 13b is consistent with the analysis in Section 3. When AMPI , phase. The pattern of the noise shown in Figure 13b is consistent with the analysis in Section 3. When  turned ON the noise decreases.

AMP I, with degenerated resistor, is turned ON more noise is generated, while when only AMP Q is  with degenerated resistor, is turned ON more noise is generated, while when only AMPQ is turned 5 ONturned ON the noise decreases.  the noise decreases. 5

NF [dB]

4 3

4

4

5

NF [dB]NF [dB]

NF [dB]

5

3 2 1

2

1.6

1.8

2

2.2

2.4

3

4

2

3

1

2 0

1 1.6

1.8

(a) 2

1

2.2

120

240

360

Phase [deg]

Freq [GHz] 2.4

0

(b)

120

240

Phase [deg]

Freq [GHz]

(a) 

360

 

(b) 

(c) Figure 13. RX NF (a) vs. input frequency; (b) with (c)  and without canceler as a function of canceler phase; (c) Noise summary of the system.

Figure  13.  RX  NF  (a)  vs.  input  frequency;  (b)  with  and  without  canceler  as  a  function  of  canceler  Figure 13. RX NF (a) vs. input frequency; (b) with and without canceler as a function of canceler phase; phase; (c) Noise summary of the system.  (c) Noise summary of the system.

J. Low Power Power Electron. Appl. Appl. 2017, 2017, 7, 7, 27 J. J. Low Low Power Electron. Electron. Appl. 2017, 7, 27 27

13 of 17 17 13 13 of of 17

SI SI

Blk Blk HD3 HD3[dB] [dB]

Pout Pout[dBm] [dBm]

CANC CANC IM3 IM3

Freq [dB] [dB] Freq

100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0

60 60 50 50 40 40 30 30 20 20 10 10 0 0 -10 -10 -20 -20 -30 -30 -40 -40

LNA+CANC w/FB LNA+CANC w/FB LNA+CANC w/o FB LNA+CANC w/o FB

Control bits Control bits

(a) (a)

TXTXLeak. Leak.Suppr. Suppr.[dB] [dB]

Figure 14 14 shows shows the the result result of of aa two-tone two-tone test, test, with with one one from from the the TX TX and and one one from from the the blocker blocker at at Figure Figure 14 shows the result of a is two-tone test, with14a. onePlotted from the TX and14b one from thebetween blocker 100 MHz offset. The signal spectrum shown in Figure in Figure is the ratio 100 MHz offset. The signal spectrum is shown in Figure 14a. Plotted in Figure 14b is the ratio between at 100 MHz offset. Theas signal spectrum is shown in Figure as 14a.function Plotted of incanceler Figure 14b the ratio blocker and IM3 IM3 (HD3) well as as the TX TX leakage leakage attenuation gainissetting setting for blocker and (HD3) as well the attenuation as aa function of canceler gain for between blocker andatIM3 (HD3) asinput. well as the TX leakagewithout attenuation as a resistors function the of canceler gain −6 dBm TX leakage the receiver For the canceler feedback peak in HD3 −6 dBm TX leakage at the receiver input. For the canceler without feedback resistors the peak in HD3 setting forcorrespond −6 dBm TXtoleakage the receiver input. For the canceler feedback resistors does not not where at optimum cancellation is achieved. achieved. Atwithout the point point of optimum optimum SI does correspond to where optimum cancellation is At the of SI the peak in HD3 does not correspond to where optimum cancellation is achieved. At the point of cancellation, the the HD3 HD3 is is dominated dominated by by canceler canceler nonlinearities nonlinearities which which is is aa product product of of the the crosscrosscancellation, optimum SI between cancellation, thesignal HD3 at is the dominated by canceler nonlinearities which isinput. a product of modulation the TX canceler input and the blocker at the LNA Adding modulation between the TX signal at the canceler input and the blocker at the LNA input. Adding the cross-modulation between the TX signal at the canceler input and the blocker at the LNA input. the feedback feedback resistors resistors re-aligns re-aligns the the peak peak in in HD3 HD3 with with the the peak peak in in SI SI cancellation. cancellation. the Adding the feedback resistors re-aligns the peak in HD3 with the peak in SI cancellation.

(b) (b)

Figure 14. 14. Two-tone Two-tone test: test: (a) (a) spectrum spectrum with with blocker blocker at at 10 10 MHz MHz offset; offset; (b) (b) with with SI SI at at the the RX RX input input as as a Figure Figure 14. Two-tone test: (a) spectrum with blocker at 10 MHz offset; (b) with SI at the RX input as aa function of of canceler canceler control control bits. bits. function function of canceler control bits.

50 50

40 40

40 40

30 30

30 30

CANC ON CANC ON CANC OFF CANC OFF RX Gain RX Gain

20 20

20 20

10 10

10 10

00

00

11

10 100 10 Freq. [MHz] [MHz] 100 Freq.

(a) (a)

-10 -10

Fund Fund

-60 -60

Canc OFF OFF Canc

Pout Pout[dBm] [dBm]

50 50

RX RXGain Gain[dB] [dB]

IIP3 IIP3[dBm] [dBm]

Figure 15a 15a shows shows the the receiver receiver IIP3 IIP3 vs vs the the self-interference self-interference frequency frequency for for both both in-band in-band and and OOB OOB Figure Figure 15a shows the is receiver IIP3The vs the self-interference frequency forf both(blocker) in-band are andswept OOB signals while the canceler disabled. two tones f (TX leakage) and signals while the canceler is disabled. The two tones f (TX leakage) and f (blocker) are swept signals while the canceler is disabled. Thethe twointermodulation tones fRF1 (TX leakage) and fRF2 (blocker) are swept over a range of frequencies such that term is constantly observed at ff over (= over a range of frequencies such that the intermodulation term is constantly observed at (= a2frange of frequencies such that the intermodulation term is constantly observed at f 2f − fRF2 (= ) IM3(dashed RF1line) − f ) equal to 1 MHz and the LO is at 2 GHz. When the canceler is disabled the 2f − f ) equal to 1 MHz and the LO is at 2 GHz. When the canceler is disabled (dashed line) the equal to IIP3 1 MHz andfrom the LO is atin-band 2 GHz.toWhen the canceler is disabled (dashed the receiver receiver ranges dBm 18 dBm dBm when and are placed line) 100 MHz MHz and 199 199 receiver IIP3 ranges from 44 dBm in-band to 18 when ff and ff are placed 100 and IIP3 ranges from 4 dBm in-band to 18 dBm when f and f are placed 100 MHz and 199 MHz RF2 RF1 is limited MHz respectively respectively away away from from the the LO. LO. The The in-band in-band IIP3 IIP3 by the the TIA. TIA. When When the the canceler canceler is is MHz is limited by respectively away from the LO. The in-band IIP3in-band is limited by the TIA. When the canceler is enabled enabled the receiver IIP3 ranges from 36.3 dBm to 34.8 dBm OOB. enabled the receiver IIP3 ranges from 36.3 dBm in-band to 34.8 dBm OOB. the receiver IIP3 ranges from 36.3 dBm in-band to 34.8 dBm OOB.

-110 -110

Canc ON ON Canc IM3 IM3

-160 -160 -210 -210

14dBm 14dBm -30 -30

-10 -10

10 10

Pin [dBm] [dBm] Pin

30 30

36dBm 36dBm 50 50

(b) (b)

Figure 15. 15. Receiver Receiver gain gain and and effective effective IIP3 IIP3 before before and and after aftercancellation: cancellation: (a) (a) vs. vs. two-tone two-tone frequency frequency Figure Figure 15. Receiver gain and effective IIP3 before and after cancellation: (a) vs. two-tone frequency spacing; (b) vs. input power for a frequency spacing of 45 MHz. spacing; spacing; (b) (b) vs. vs. input input power power for for aa frequency frequency spacing spacing of of 45 45 MHz. MHz.

The effective IIP3 improvement is strictly related to the cancellation. Ideally, for for every 1 dB of SI The effective IIP3 improvement is strictly related to the cancellation. Ideally, Ideally, for every 1 dB of SI cancellation, the effective IIP3 should improve by 1 dB, i.e., the IM3 should decrease with the the square square cancellation, the theeffective effectiveIIP3 IIP3should shouldimprove improveby by1 1dB, dB,i.e., i.e.,the theIM3 IM3 should decrease with cancellation, should decrease with the square of of cancellation. In plots in Figure 15a, the SI cancellation changes with frequency from 32 dB in-band of cancellation. plots Figure 15a,the theSISIcancellation cancellationchanges changeswith withfrequency frequencyfrom from 32 32 dB dB in-band in-band cancellation. InIn plots inin Figure 15a, to 28 28 dB dB OOB. OOB. This This is is slightly slightly more more than than the the minimum minimum cancellation cancellation of of 27 27 dB dB that that is is guaranteed guaranteed by by the the to

J. Low Power Electron. Appl. 2017, 7, 27

14 of 17

J.J.Low LowPower PowerElectron. Electron.Appl. Appl.2017, 2017,7,7,27 27

14 14ofof17 17

to 28 dB OOB. This is slightly more than the minimum cancellation of 27 dB that is guaranteed by canceler cancelerresolution. resolution.However, However,ititisisnoticed noticedthat thatthe theeffective effectiveOOB OOBIIP3 IIP3improvement improvementisismuch muchless lessthan than the canceler resolution. However, it is noticed that the effective OOB IIP3 improvement is much less the thecancellation. cancellation.This Thisisisdue dueto tothe thecanceler cancelernonlinearity, nonlinearity,ultimately ultimatelylimiting limitingthe theeffective effectivereceiver receiverIIP3. IIP3. than the cancellation. This is due to the canceler nonlinearity, ultimately limiting the effective receiver This This isis shown shown in in Figure Figure16, 16, where where the thecancellation cancellation isis varied varied by by adjusting adjusting the the canceler canceler control control bits. bits. IIP3. This is shown in Figure 16, where the cancellation is varied by adjusting the canceler control When When the the cancellation cancellation isis above above 18 18 dB dB the the effective effective IIP3 IIP3 remains remains constant, constant, limited limited by by the the canceler canceler bits. When the cancellation is above 18 dB the effective IIP3 remains constant, limited by the canceler intrinsic intrinsicnonlinearities. nonlinearities. intrinsic nonlinearities.

Effective EffectiveOOB OOBIIP3 IIP3 IIP3 [dBm] [dBm] IIP3

50 50 40 40 30 30 20 20 10 10 00 00

10 20 10 20 SISICANC CANC[dB] [dB]

30 30

Figure Figure 16. Effective OOB IIP3 vs. SI cancellation. Figure16. 16.Effective EffectiveOOB OOBIIP3 IIP3vs. vs.SI SIcancellation. cancellation.

Additional critical system performance test that full-duplex Additionalcritical critical system performance test that demonstrates demonstrates the system systemfunctionality full-duplex Additional system performance test that demonstrates the systemthe full-duplex functionality is carried out by considering an in-band two-tone test with both tones applied at is considering carried out by an in-band two-tone test with bothattones appliedinput, atthe the isfunctionality carried out by an considering in-band two-tone test with both tones applied the canceler canceler the TX The are placed at MHz and offset cancelerinput, input, representing the TXsignal. signal. Theattones tones are placed at66offset MHzfrom and10 10MHz MHz offset from representing therepresenting TX signal. The tones are placed 6 MHz and 10 MHz the LO and thefrom IM3 the the IM3 at canceler disabled, IIP3 isisequal When theLO LO2and and theWhen IM3falls falls at22MHz. MHz. Whenthe thethe canceler disabled, the IIP3 equal to5.3 5.3dBm. dBm. When falls at MHz. the canceler is When disabled, IIP3 isisis equal to 5.3the dBm. When the to canceler is turned the canceler is turned on and set to the proper setting for SI cancellation, the IIP3 improves to 21 dBm, the canceler is turned and set proper setting for SI cancellation, IIP3 to 21 on and set to the properon setting fortoSIthe cancellation, the IIP3 improves to 21 the dBm, asimproves is illustrated bydBm, the as is illustrated by the simulation results in Figure 17. One interesting observation is that the result as is illustrated by the simulation results in Figure 17. One interesting observation is that the result simulation results in Figure 17. One interesting observation is that the result shows that the output IM3 shows that output IM3 canceler becomes dominant at In in SI shows that the the outputdominant IM3 of of the the canceler becomes dominant at the the RX RX input. input. In fact, fact, in the the SI of the canceler becomes at the RX input. In fact, in the SI cancellation full-duplex system, the cancellation full-duplex system, IIP3 isisgiven the canceler plus due cancellation full-duplex system, the effective receiver IIP3 given by the canceler IIP3 plus33dB, dB,gain due effective receiver IIP3 is given bythe theeffective cancelerreceiver IIP3 plus 3 dB, due toby the LNA inputIIP3 transformer to the LNA input transformer gain (−3 dB). The IIP3 of the canceler is thus, important and must be to the LNA input transformer gain (−3 dB). The IIP3 of the canceler is thus, important and must be (−3 dB). The IIP3 of the canceler is thus, important and must be designed to be sufficiently high to designed to be sufficiently high to guarantee both FDD and full-duplex operations. designed both to beFDD sufficiently high to guarantee both FDD and full-duplex operations. guarantee and full-duplex operations. 2020 00

Pout [dBm] [dBm] Pout

-20 -20

Fund Fund

-40 -40 -60 -60 -80 -80

IM3 IM3 21 21dBm dBm

-100 -100 -120 -120 -15 -15

-5-5

55 Pin Pin[dBm] [dBm]

1515

2525

Figure 17. IIP3 with two-tone cancellation atatreceiver receiver (RX) input. Figure17. 17.IIP3 IIP3with withtwo-tone two-tonecancellation cancellationat receiver(RX) (RX)input. input. Figure

Another receiver with the canceler isisthe point Anotherperformance performanceevaluation evaluation of the receiver with thethe canceler the 1dB dB1compression compression point Another performance evaluationof ofthe the receiver with canceler is 1the dB compression 1dB ). With the canceler disabled, the receiver, due to the presence of SI saturates, thereby desensitizing (P canceler disabled,disabled, the receiver, to thedue presence SI saturates, desensitizing (P1dB).(PWith).the point With the canceler the due receiver, to theofpresence of SIthereby saturates, thereby 1dB the RX signal. The gain 1 dB compression plot due to an in-band SI is shown in Figure 18a. The the RX signal. The gain 1 dB compression plot due to an in-band SI is shown in Figure 18a. 18a. The desensitizing the RX signal. The gain 1 dB compression plot due to an in-band SI is shown in Figure simulation result is obtained with the SI placed at 10 MHz away from the LO (at 2 GHz) while the Rx simulation result is obtained with the SI placed at 10 MHz away from the LO (at 2 GHz) while the Rx The simulation result is obtained with the SI placed at 10 MHz away from the LO (at 2 GHz) while signal at in-band PP1dB dBm. When the signal isplaced placed at33MHz MHz offset. The in-band 1dBreferred referred tothe thereceiver receiver inputisis−25 −25 dBm. When the the Rx is signal is placed at 3offset. MHzThe offset. The in-band P1dB to referred to the input receiver input is − 25 dBm. canceler is turned ON, with more than 30 dB of SI cancellation, P 1dB pushed to 4 dBm. For the OOB canceler is turned ON, with more than 30 dB of SI cancellation, P 1dBis is pushed to 4 dBm. For the OOB When the canceler is turned ON, with more than 30 dB of SI cancellation, P1dB is pushed to 4 dBm. PP1dB SI MHz away LO and isiskept at 1dBevaluation, evaluation, the SIisisplaced placed 100 MHz away from the LOfrom andthe theRx Rxsignal signal kept at33MHz. MHz. The For the OOB P1dBthe evaluation, the100 SI is placed 100from MHzthe away the LO and the Rx signal is keptThe at receive signal is swept until saturation occurred at 1.7 dBm referred to the RX port while the canceler receive signal is swept until saturation occurred at 1.7 dBm referred to the RX port while the canceler 3 MHz. The receive signal is swept until saturation occurred at 1.7 dBm referred to the RX port while isisdisabled. disabled.When Whenthe thecanceler cancelerisisturned turnedon, on,the thePP1dB 1dBis ispushed pushedto to55dBm, dBm,as asplotted plottedin inFigure Figure18b. 18b.

J. Low Power Electron. Appl. 2017, 7, 27

15 of 17

the canceler is disabled. When the canceler is turned on, the P1dB is pushed to 5 dBm, as plotted in 15 of 17 FigureJ. Low 18b.Power Electron. Appl. 2017, 7, 27

(a)

(b)

Figure 18. Receiver P1dB compression with and without the canceler: (a) In-band compression at 11

Figure 18. Receiver P1dB compression with and without the canceler: (a) In-band compression at MHz; (b) Out of band compression at 100 MHz as function of input power. 11 MHz; (b) Out of band compression at 100 MHz as function of input power.

This work is benchmarked against prior works focussing on SI cancellation for FDD and fullduplex systems [12,17,19,20] as against shown inprior Tableworks 1. The focussing system achieves noise figurefor andFDD high and This work is benchmarked on SIlow cancellation effective IIP3 thanks to the improvement of the canceler cross-modulation distortion, while full-duplex systems [12,18,20,21] as shown in Table 1. The system achieves low noise figure and dissipatingIIP3 low thanks power. TX cancellation isofperformed in front of the LNA achieving more than high effective to leakage the improvement the canceler cross-modulation distortion, while 30 dB of cancellation over 80 MHz bandwidth. This solution addresses also the issues of receiver IIP2 dissipating low power. TX leakage cancellation is performed in front of the LNA achieving more than and reciprocal mixing. Among prior works, only [17], which is tailored towards FDD applications, 30 dB of cancellation over 80 MHz bandwidth. This solution addresses also the issues of receiver IIP2 achieves similar IIP3 values and higher SI power handling capabilities. However, the TX leakage is and reciprocal mixing. Among prior works, only [18], which is tailored towards FDD applications, effectively cancelled only after down-conversion and recombination between common-gate and achieves similar IIP3paths, values and higher handling capabilities. However, the TXmuch leakage common-source making it proneSItopower IIP2 and reciprocal mixing. Furthermore [17] has is effectively cancelled only after down-conversion anddissipation recombination common-gate higher power dissipation and higher NF. In [12] power is verybetween low but the maximum SI and common-source paths, makingtoitbe prone IIP2be and reciprocal Furthermore [18] has much is only −20 dBm. However, fair, ittomust noticed that allmixing. of the other works reported in the report measuredand results, while in this work we are presenting is only simulation highertable power dissipation higher NF. In [12] power dissipation very low but results. the maximum SI is only −20 dBm. However, to be fair, it must be noticed that all of the other works reported in the table Table 1. Performance comparison. report measured results, while in this work we are presenting only simulation results. This Work [17] b JSSC ’14 [19] b ISSCC ’15 [20] b ISSCC ’15 FD/FDD FDD FDD/FD FD Table 1. Performance comparison. 40 nm 65 nm 65 nm 65 nm Technology CMOS CMOS CMOS CMOS b b ISSCC ’15 ’14 GHz [20] b ISSCC ’15 [21] Frequency This Work 1.5–2.5 GHz [18] JSSC 0.5–1.5 0.8–1.4 GHz 0.15–3.5 GHz NF w/o canc 3 dB 4.2–5.6 dB 4.8–5.8 dB 6.3FD dB FDD/FD FD/FDD FDD FDD/FD NF w/canc 3.9–4.6 dB 5–6.4 dB 5.3/7 dB 10.3–12.3 dB 40 nm 65 nm 65 nm 65 nm Technology IP3 RX 14 dBm 12 dBm 17/−22 dBm 16.2 dBm a CMOS CMOS CMOS CMOS Eff. IIP3 35 dBm 33 dBm 27/2 dBm 19 dBm a Frequency 1.5–2.5 GHz 0.5–1.5 GHz 0.8–1.4 GHz 0.15–3.5 >30 dB 30 dB 20 dB 27 dB GHz SI Cancellation /250 c–10 d MHz 4.2–5.6 dB /25dB MHz NF w/o canc 3 dB 4.8–5.8 6.3 dB Max TX Leak. +2 dBm dBm 1.5 dBm NF w/canc 3.9–4.6 dB−4 dBm 5–6.4 dB 5.3/7−8dB 10.3–12.3 dB 14 mW RX 83 mW RX 72 69 mW RX 23–56 mW Power IP3 RX 14 dBm 12 dBm 17/−91 22mW/path dBm 16.2 dBm a 11 mW canc mW canc FDD/FD

[12] b ESSCIRC ’15 FD 40 nm CMOS b ESSCIRC ’15 [12] 1–3 GHz 4.8–5.8 FD dB 5.3/7 dB 40 nm 18 dBm CMOS N/A 301–3 dB GHz /4 4.8–5.8 MHz dB −20 dBm 5.3/7 dB 11 mW RX 18 dBm 0.6 mW/Cac

a Eff.aIIP3 dBm dBm delay between 27/2 dBm 19 dBmd with In-band IIP3; b 35 measured results; c33 without transmitter (TX)/RX; 2 ns group N/A delay.

>30 dB SI /250 c –10 d MHz Cancellation 5. Conclusions Max TX Leak.

−4 dBm

30 dB

20 dB /25 MHz

27 dB

30 dB /4 MHz

+2 dBm

−8 dBm

1.5 dBm

−20 dBm

A highly linear and low noise receiver front-end is proposed for dual antenna systems (for Full 14 mW RX 83 mW RX 72 69 mW RX 23–56 mW 11 mW RX Power and FDD) that is based on an active self-interference canceler with a 5-bit resolution to duplex 11 mW canc mW canc 91 mW/path 0.6 mW/Cac a In-band IIP3; d with implement an bactive programmable vector To meet the FD requirements, the SI measured results; c without delaymodulator. between transmitter (TX)/RX; 2 ns group delay. cancellation can be improved in the digital domain [10]. The proposed canceler strongly attenuates the TX leakage of up to −4 dBm of power at the input of the LNA, with benefits for the entire receiver 5. Conclusions in terms of IIP3, IIP2 and reciprocal mixing. A cross-modulation mechanism of the canceler was and strongly suppressed thanks to resistive feedback. The for effective IIP3 of the receiver Aidentified highly linear and low noise receiver front-end is proposed dual OOB antenna systems (for Full was improved fromis14based dBm on withanthe canceler disabled to 35 canceler dBm with the acanceler enabled. to duplex and FDD) that active self-interference with 5-bit resolution

implement an active programmable vector modulator. To meet the FD requirements, the SI cancellation

J. Low Power Electron. Appl. 2017, 7, 27

16 of 17

can be improved in the digital domain [10]. The proposed canceler strongly attenuates the TX leakage of up to −4 dBm of power at the input of the LNA, with benefits for the entire receiver in terms of IIP3, IIP2 and reciprocal mixing. A cross-modulation mechanism of the canceler was identified and strongly suppressed thanks to resistive feedback. The effective OOB IIP3 of the receiver was improved from 14 dBm with the canceler disabled to 35 dBm with the canceler enabled. Compared with previous implementations, lower power, lower noise, and higher IIP3 with respect to self-interference was achieved. Author Contributions: Saheed Tijani is the main author and responsible for writing the paper. Danilo Manstretta was responsible for the review and supervising of the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3.

4.

5.

6.

7. 8.

9.

10.

11. 12.

13.

Fabiano, I.; Sosio, M.; Liscidini, A.; Castello, R. SAW-less analog front-end receivers for TDD and FDD. IEEE J. Solid-State Circuits 2013, 48, 3067–3079. [CrossRef] Bharadia, D.; McMilin, E.; Katti, S. Full Duplex Radios. In Proceedings of the ACM SIGCOMM, Hong Kong, China, 12–16 August 2013. Huusari, T.; Choi, Y.S.; Liikkanen, P.; Korpi, D.; Talwar, S.; Valkama, M. Wideband Self-Adaptive RF Cancellation Circuit for Full-Duplex Radio: Operating Principle and Measurements. In Proceedings of the 2015 IEEE 81st Vehicular Technology Conference (VTC Spring), Glasgow, UK, 11–14 May 2015; pp. 1–7. Tamminen, J.; Turunen, M.; Korpi, D.; Huusari, T.; Choi, Y.; Talwar, S.; Valkama, M. Digitally-controlled RF self-interference canceller for full-duplex radios. In Proceedings of the 2016 24th European Signal Processing Conference (EUSIPCO), Budapest, Hungary, 29 August–2 September 2016; pp. 783–787. Kiayani, A.; Anttila, L.; Valkama, M. Active RF cancellation of nonlinear TX leakage in FDD transceivers. In Proceedings of the 2016 IEEE Global Conference on Signal and Information Processing (GlobalSIP), Washington, DC, USA, 7–9 December 2016; pp. 689–693. Roverato, E.; Kosunen, M.; Cornelissens, K.; Vatti, S.; Stynen, P.; Bertrand, K.; Korhonen, T.; Samsom, H.; Vandenameele, P.; Ryynänen, J.; et al. 13.4 All-digital RF transmitter in 28 nm CMOS with programmable RX-band noise shaping. In Proceedings of the 2017 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 5–6 February 2017; pp. 222–223. Abdelhalem, S.H.; Gudem, P.S.; Larson, L.E. Hybrid Transformer-Based Tunable Differential Duplexer in a 90-nm CMOS Process. IEEE Trans. Microw. Theory Tech. 2013, 61, 1316–1326. [CrossRef] Fabiano, I.; Ramella, M.; Manstretta, D.; Castello, R. A +25 dBm IIP3 1.7–2.1 GHz FDD Receiver Front-End with Integrated Hybrid-Transformer in 28 nm CMOS. IEEE Trans. Microw. Theory Tech. 2017, 65, 4677–4688. [CrossRef] Calderin, L.; Ramakrishnan, S.; Puglielli, A.; Alon, E.; Nikoli´c, B.; Niknejad, A.M. Analysis and Design of Integrated Active Cancellation Transceiver for Frequency Division Duplex Systems. IEEE J. Solid-State Circuits 2017, 52, 2038–2054. [CrossRef] Zhou, J.; Reiskarimian, N.; Krishnaswamy, H. 9.8 Receiver with integrated magnetic-free N-path-filter-based non-reciprocal circulator and baseband self-interference cancellation for full-duplex wireless. In Proceedings of the 2016 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 31 January–4 February 2016; pp. 178–180. Zhang, T.; Suvarna, A.R.; Bhagavatula, V.; Rudell, J.C. An Integrated CMOS Passive Self-Interference Mitigation Technique for FDD Radios. IEEE JSSC 2015, 50, 1176–1188. [CrossRef] Zhang, T.; Chen, Y.; Huang, C.; Rudell, J.C. A Low-Noise Reconfigurable Full-Duplex Front-End with Self-Interference Cancellation and Harmonic-Rejection Power Amplifier for Low Power Radio Applications. In Proceedings of the 43rd European Solid-State Circuit Conference (ESSCIRC), Leuven, Belgium, 11–14 September 2017. Reiskarimian, N.; Dastjerdi, M.B.; Zhou, J.; Krishnaswamy, H. Highly-linear integrated magnetic-free circulator-receiver for full-duplex wireless. In Proceedings of the 2017 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 5–9 February 2017; pp. 316–317.

J. Low Power Electron. Appl. 2017, 7, 27

14. 15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27. 28.

29.

30.

17 of 17

Yang, D.; Yuksel, H.; Molnar, A. A Wideband Highly Integrated and Widely Tunable Transceiver for In-Band Full-Duplex Communication. IEEE J. Solid-State Circuits 2015, 50, 1189–1202. [CrossRef] Ramakrishnan, S.; Calderin, L.; Niknejad, A.; Nikoli´c, B. An FD/FDD transceiver with RX band thermal, quantization, and phase noise rejection and 64 dB TX signal cancellation. In Proceedings of the 2017 IEEE Radio Frequency Integrated Circuits Symposium (RFIC), Honolulu, HI, USA, 4–6 June 2017; pp. 352–355. Ayati, S.A.; Mandal, D.; Bakkaloglu, B.; Kiaei, S. Adaptive integrated CMOS circulator. In Proceedings of the 2016 IEEE Radio Frequency Integrated Circuits Symposium (RFIC), San Francisco, CA, USA, 22–24 May 2016; pp. 146–149. Zhou, J.; Chakrabarti, A.; Kinget, P.; Krishnaswamy, H. Low-Noise Active Cancellation of Transmitter Leakage and Transmitter Noise in Broadband Wireless Receivers for FDD/Co-Existence. IEEE J. Solid-State Circuits 2014, 49, 3046–3062. [CrossRef] Luo, C.; Gudem, P.S.; Buckwalter, J.F. 0.4–6 GHz,17-dBm B1dB, 36-dBm IIP3 Channel-selecting, Low-noise Amplifier for SAW-less 3G/4G FDD Receivers. In Proceedings of the IEEE RFIC Symposium, Phoenix, AZ, USA, 17–19 May 2015. Zhou, J.; Chuang, T.H.; Dinc, T.; Krishnaswamy, H. Reconfigurable Receiver with >20 MHz bandwidth self-interference cancellation suitable for FDD, co-existence and full-duplex applications. In Proceedings of the 2015 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 22–26 February 2015; pp. 342–343. Van den Broek, D.J.; Klumperink, E.A.M.; Nauta, B. A self-interference-cancelling receiver for in-band full-duplex wireless with low distortion under cancellation of strong TX leakage. In Proceedings of the 2015 IEEE ISSCC, San Francisco, CA, USA, 22–26 February 2015. Zhang, T.; Najafi, A.; Su, C.; Rudell, J.C. 18.1 A 1.7-to-2.2 GHz full-duplex transceiver system with >50 dB self-interference cancellation over 42 MHz bandwidth. In Proceedings of the 2017 IEEE International Solid-State Circuits Conference (ISSCC), San Francisco, CA, USA, 5–6 February 2017; pp. 314–315. Tijani, S.; Manstretta, D. A SAW-less receiver front-end with low power active self-interference canceler. In Proceedings of the 2015 IEEE International Conference on Electronics, Circuits, and Systems (ICECS), Cairo, Egypt, 6–9 December 2015; pp. 89–92. Montanari, D.; Silvestri, L.; Bozzi, M.; Manstretta, D. Antenna coupling and self-interference cancellation bandwidth in SAW-less diversity receivers. In Proceedings of the 2016 46th European Microwave Conference (EuMC), London, UK, 4–6 October 2016; pp. 731–734. Montanari, D.; Castellano, G.; Manstretta, D.; Castello, R. A 0.7–2 GHz Auxiliary Receiver with Enhanced Compression for SAW-less FDD. In Proceedings of the 43rd European Solid-State Circuit Conference (ESSCIRC), Leuven, Belgium, 11–14 September 2017. Yang, X.; Babakhani, A. A 4.6–5.35 GHz transceiver with 38 dB on-chip self-interference cancelation at 10 kHz offset frequency. In Proceedings of the 2015 IEEE Radio Frequency Integrated Circuits Symposium (RFIC), Phoenix, AZ, USA, 17–19 May 2015; pp. 239–242. ETSI. LTE (E-UTRA) User Equipment (UE) Radio Transmission and Reception, 3GPP TS 136.101 ver. 12.5.0, Release 12. November 2014. Available online: http://www.etsi.org/deliver/etsi_ts/136100_136199/136101/ 12.05.00_60/ts_136101v120500p.pdf (accessed on 21 August 2017). Kim, H.; Woo, S.; Jung, S.; Lee, K.-L. A CMOS Transmitter Leakage Canceller for WCDMA Applications. IEEE Trans. Microw. Theory Tech. 2013, 61, 3373–3380. [CrossRef] Van den Broek, D.J.; Klumperink, E.A.M.; Nauta, B. An In-Band Full-Duplex Radio Receiver with a Passive Vector Modulator Downmixer for Self-Interference Cancellation. IEEE J. Solid-State Circuits 2015, 50, 3003–3014. [CrossRef] Zhuo, W.; Embabi, S.; de Gyvez, J.P.; Sanchez-Sinencio, E. Using capacitive cross-coupling technique in RF low noise amplifiers and down-conversion mixer design. In Proceedings of the 26th European Solid-State Circuits Conference, Stockholm, Sweden, 19–21 September 2000; pp. 77–80. Pini, G.; Manstretta, D.; Castello, R. Highly linear TIA for SAW-less FDD receivers. In Proceedings of the 2016 IEEE Asian Solid-State Circuits Conference (A-SSCC), Toyama, Japan, 7–9 November 2016; pp. 117–120. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).