Low Power Upconversion Mixer for Medical Remote Sensing

This work presents the design of a low power upconversion mixer adapted in medical remote sensing such as wireless endoscopy application. The proposed upconversion mixer operates in ISM band of 433 MHz. With the carrier power of −5 dBm, the proposed mixer has an output inferred 1 dB compression point of −0.5 dBm with a corresponding output third-order intercept point (OIP3) of 7.1 dBm. The design of the upconversion mixer is realized on CMOS 0.13 μm platform, with a current consumption of 594 μA at supply voltage headroom of 1.2 V.


Introduction
Mixer is the lifeline of the upconversion process in a transmitter, coupled with the integrating voltage controlled oscillator (VCO) and power amplifiers (PA) circuit defining the transmitter. An upconversion mixer is tasked upon performing frequency translation from a baseband frequency to radio frequency (RF). The supply voltage headroom limitation aligned with the scaling of CMOS technology has initiated efforts upon the realization of low power architecture deemed crucial in portable wireless electronic devices in prolonging the battery lifetime. The limitation of voltage headroom and the degradation of the performance at high operating frequency arise concurrently with technology scaling [1].
The highlight of performance indicator for upconversion mixer design is the linearity and power consumption. Passive mixers do provide comparatively superior linearity performance in comparison to an active architecture, with a penalty paid in conversion loss, often requiring a mandatory larger carrier signal power. Large carrier signal power translates to higher power consumption, while low carrier signal power is undesirable for passive mixer integration [2,3]. Gilbert cell mixer is the default conventional double balanced architecture in radio frequency integrated circuit (RFIC) realization. The advantage of Gilbert cell mixer is in its superior port-to-port isolation and its ability in cancelling off undesired RF and local oscillator (LO) output signal feedthrough in the midst of providing higher conversion gain. In the evolution of technology scaling with a proportional lower supply headroom the fact that Gilbert cell mixers integrate high number of series stacked transistors limits the endeavor in achieving low power consumption, [4][5][6][7][8][9][10][11].
Various architectures had been proposed to combat the constraint of voltage headroom limitation [7][8][9]. Foldedswitching is a promising alternative in operating at low supply voltage abstaining significance effect on the gain and linearity performance [1]. However, with an additional integration of biasing current source, noise is added into the circuit, while the power consumption is ramped up. An alternative solution for low voltage headroom implementation is the switchedtransconductance, where the switching stage and transconductance stage are swapped in position respective to the conventional double-balanced mixer [4]. This architecture solves the setback of stacking transistors, while upholding the performances with lower power consumption. The switching pairs of the transconductance stage are turned on and off alternately. The principle of frequency translation is akin to double-balanced topology. The aspect ratio in the width of PMOS to the width of NMOS is twice in magnitude justifying different carrier mobility between PMOS and NMOS devices. At node ⟨ ⟩ of Figure 2, for a positive half cycle of LO signal, M 3 is switched off while M 1 is switched on, resulting in the transconductor to be switched on. The reversed operation is observed at node ⟨ ⟩. Push pull configuration of switching stage is adopted instead of single NMOS switch due to larger output swing and thus providing superior noise immunity.
This paper presents the design of low power switched transconductance mixer targeted towards portable medical devices at 433 MHz operating frequency, simulated in 0.13 m CMOS technology. The operating frequency of 433 MHz is selected in consideration of the in-body wireless communication losses. The range of frequency with least losses is found to be between 400 MHz and 900 MHz [12], which confirms the industrial, scientific, and medical (ISM) band with frequency ranging from 433.05 MHz to 434.79 MHz, and the medical implant communication services (MICS) with frequency of 402-405 MHz. ISM band is chosen as the preferred operating frequency over MICS although the latter is reserved by Federal Communications Commission (FCC) for implantable medical devices as MICS has low data rate up to a permissible level of 500 kbps.
The paper is organized as follows. Section 2 discusses the architecture and operation of the proposed upconversion mixer. The simulation result is presented in Section 3 in verifying the performance. Finally the conclusion is given in Section 4. The transconductance stage converts the input voltage to current signal, which is subjected to mix with LO or carrier signal in the switching stage. A modified structure of the double-balanced mixer in Figure 1 is proposed in Figure 2 which consists of switching pairs (M 1 -M 4 ), transconductance stage (M 5 -M 8 ), and load stage (M 9 , M 10 and ) coupled with the compensating common-mode feedback structure (CMFB). M 11 is the tail transistor integrating as current source. In comparison with the conventional architecture, the inverting switching stage and transconductance stage are swapped in position.

Mixer Architecture and Operation
CMFB structure is adapted where the PMOS transistor M 9 and M 10 operates as current source for the differential signals. Common mode current absorbed by the two PMOS load transistor increases the differential gain by and at the same time allowing wider voltage headroom. It is an encouraging alternative in integrating diode-connected PMOS load rather than resistor loading, even in the event that the advantages of CMFB comes with a trade-off of higher noise introduced to the circuit. The rational of this tradeoff is constructed, such that the noise figure is not paramount in upconversion mixer [5], and the significance of low power consumption in the application of portable medical devices is inevitable.
Transistor The comprehensive operation of the proposed mixer as illustrated in Figure 2 is depicted by the following mathematical expressions. Current through transistor M 5 is where 5,6 is the transconductance for NMOS transistor pair M 5 and M 6 , V bb = bb sin bb is the input baseband signal, and is the current due to instantaneous LO voltage. Similarly, current through transistor M 6 can be expressed as Current at node ⟨ ⟩ can then be obtained as Similarly, current at node ⟨ ⟩ is expressed as The differential mixer output current can be derived as The Scientific World Journal where sq[sin LO ] is the square wave function of LO signal and eff is the effective transconductance of the mixer. The differential mixer output signal is eventually obtained as From (7), the conversion gain of the mixer is observed to be Looking at low voltage perspective, the derivation of overdrive voltage for conventional mixer as in Figure 1 is given as follows: where ds is the overdrive voltage and th is the threshold voltage. Similarly, the supply voltage derived for the proposed mixer in Figure 2 can be derived as dd = ds 1 + ds 5 + ds 9 + tĥ1 .
Equation (10) shows that the proposed mixer requires one less overdrive voltage of a transistor for the operation, which leads to reduced DC power consumption.

Simulation Results
The proposed architecture of the upconversion mixer is simulated and verified using Cadence Spectra RF platform. Components parameters are shown in Table 1. The design is implemented in a 0.13 m standard CMOS technology at supply headroom of 1.2 V. The results verify that the upconversion mixer consumes a total current of 594 A which results in a dissipation of DC power of 0.71 mW. Low DC power consumption is essentially vital in the application of wireless medical device such as capsule endoscope. Figure 3(a) shows the input referred 1 dB compression point of −5.43 dBm and output referred 1 dB compression point of −0.5 dBm. Two-tone analysis is conducted to verify the linearity characteristic as illustrated in Figure 3(b). The input third-order intercept point (IIP3) of 2.0 dBm and output third-order intercept point (OIP3) of 7.1 dBm were achieved with an equivalent gain of 5.4 dB. This agrees with the general relationship of OIP3 = IIP3 + Gain, as also obtained in [1]. Table 2 enlists the performance comparison highlighting the proposed mixer respective to recent reported works. From Table 2, the proposed architecture highlights a lead in low power consumption and high linearity performance.
where is the mixer gain, OP 1 dB is the output referred 1 dB compression point, Freq is the operating frequency, DC is the total power consumption with baseband and LO signal, and LO is local signal power in dBm. The FOM originates from the fact that gain and OP 1 dB contribute to better performance of a mixer. At the same time, power consumption should be kept as low as possible in measuring a superior performance. Operating frequency is taken into account to obtain a fair comparison with other designs of different operating frequencies. Referring to the FOM comparison in Table 2, the proposed mixer shows a superior overall performance with a FOM of 5.31. The physical layout of the circuit is illustrated in Figure 4

Conclusion
This paper presented a low power, high linearity upconversion mixer for the portable medical devices at 433 MHz. The low power architecture integrates regulated bias with inverting switching input. Implemented in 0.13 m standard CMOS technology the mixer consumes supply headroom of 1.2 V while dissipating 0.71 mW of DC power.

Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.