Simple and accurate extraction methodology for RF MOSFET valid up to 20 GHz A.F. Tong, K.S. Yeo, L. Jia, C.Q. Geng, J.-G. Ma and M.A. Do Abstract: An accurate and simple parameter extraction technique for deep submicron MOSFETs using a conventional MOSFET model with three terminal resistances for the gate, source and drain, as well as a simple substrate coupling network and a non-reciprocal capacitor is proposed. This extraction technique utilises both Z and Y parameter analysis on the proposed small-signal equivalent circuit. RF simulation is carried out using the BSIM4 DC core model with the extracted extrinsic components. Analytical equations are derived for all the RF parameters and a linear regression technique is used to extract the parameters. Transcapacitance is utilised in the model to ensure charge conservation and good fitting of the Y 21 and Y 12 parameters. The extracted and optimised RF parameter values are in close agreement, which implies that little or no optimisation is required using this technique. Hence, this extraction methodology can be implemented easily for RF MOSFET modelling. Excellent agreement has been obtained between the simulated and measured results up to 20 GHz. 1 Introduction As CMOS technology improves, both the unity current gain frequency (f T ) and unity power gain frequency (f max ) increase while the minimum noise figure (NF min ) decreases [1]. Typical f T and f max values for a 0.18 mm technology are 50 GHz and NF min is about 0.35 dB at 2 GHz operating frequency [1]. This shows that CMOS technology is a viable technology choice for RF integrated components for the wireless communication market [2] . Owing to the fast growth in the RF wireless communication market, the design cycle time for RF circuits must be minimised. This can only be achieved if circuit designers have accurate RF MOSFET models. Therefore, the time required to develop an RF model and the accuracy of its parameter extraction technique become very important. Up to now, the most commonly used method is a macromodelling approach. In this approach, sub-circuits are added to the BSIM core model to model the RF parasitic components of the MOSFET [3, 4] . The sub- circuit components are usually extracted from the measured S-parameters of the transistor using Z or Y parameter analysis on the proposed small-signal equivalent circuit [5– 8] . There have been many different extraction techniques reported [5–8] for RF MOSFET modelling. In [5] , no substrate resistances are included in the small-signal equivalent circuit. Thus fitting of the output admittance will be difficult. The method in [6] requires S-parameter measurements up to about 40 GHz. High frequency measurement of the S-parameter is very difficult and requires very good calibration of equipment and measure- ment of de-embedding structures. Furthermore, there is no substrate network in its equivalent circuit. The approach in [7] is by direct extraction on its equivalent circuit using Y- parameter analysis in the linear and saturation region. Its substrate-coupling network is obtained by local optimisa- tion to fit the equivalent substrate admittance. Therefore, no analytical equations are derived for substrate-related components. In [8] , all the parameters are extracted using a linear regression approach by performing Y-parameter analysis on the small-signal equivalent circuit. In this technique, the source and drain resistances are omitted in the equivalent circuit for simplicity and ease of parameter extraction. As a result the accuracy of the extracted values for the gate resistance R g and the transconductance are compromised. Furthermore, without the source and drain resistances, the small-signal model gives inaccuracy in noise simulations. To circumvent the above-mentioned problems with the existing extraction techniques [5–8] , a new RF parameter extraction technique using the small-signal equivalent circuit in Fig. 1 is presented. It includes three terminal resistances G R g C gd C gs C sd C jd R d R subd R s S D C m C m = C dg - C gd dV gs dt g m V gs g ds Fig. 1 New small-signal RF equivalent circuit A.F. Tong and C.Q. Geng are with Advanced RFIC (S) Pte. Ltd., 2 International Business Park, #10-11, The Strategy, Tower 1, Singapore 609930 K.S. Yeo, L. Jia, J.-G. Ma and M.A. Do are with the Center for Integrated Circuits & Systems, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798 r IEE, 2004 IEE Proceedings online no. 20040778 doi:10.1049/ip-cds:20040778 Paper first received 23rd February and in revised form 24th May 2004 IEE Proc.-Circuits Devices Syst., Vol. 151, No. 6, December 2004 587