This is an author-created, un-copyedited version of the article C. D. Martino, Z. Hu, L. Galatro, G. Sarris and M. Spirito, "Power level control of mm- wave test benches for accurate small and large-signal DUT measurements," 2016 88th ARFTG Microwave Measurement Conference (ARFTG), Austin, TX, USA, 2016, pp. 1-4. IEEE is not responsible for any errors or omissions in this version of the manuscript or any version derived from it. The definitive publisher-authenticated version is available online at http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7839723 Power Level Control of mm-Wave Test Benches for Accurate Small and Large-signal DUT Measurements C. De Martino, Z. Hu, L. Galatro, G. Sarris and M. Spirito Electronics Research Laboratory, Delft University of Technology Mekelweg 4, 2628CD Delft, Netherlands Abstract In this contribution we describe the accuracy improvement achievable using a vector corrected, power calibrated test-bench when measuring mm-wave non-linear devices. The absence of automatic level control in commercially available network analyzer frequency extenders can result in AM distortion of the transfer characteristic. Moreover, when large- signal mm-wave test-benches are employed (i.e., using multipliers and power meters) the lack of vector correction causes transfer characteristic errors. The usage of power levelled mm-wave VNA test-benches allows to properly measure the true small-signal s-parameter response as well as accurately correct for the impedance mismatch present, providing increased correlation between small and large-signal test-benches response. The accuracy limitation of classical approaches and the improvement achieved by the levelled mm-wave VNA test-bench are validated on a 140GHz power amplifier. Index Terms — Power control, Power calibration, S- parameters measurements, mm-wave, sub-mm-wave. I. INTRODUCTION The increasing interests in millimeter-wave (mm-wave) systems for commercial applications, such as automotive and high data rate communication (falling under the umbrella of 5G), is pushing for higher quality mm-wave measurements and device models. Improving the model predictive capabilities is the path to minimize the redesign attempts, thus making the developing phase of a product commercially feasible. The mm-wave frequency range is characterized by the extensive use of frequency multiplication stages to generate signals. The testing and instrumentation field also makes use of frequency multiplication in the up- and down-converting extension modules of state-of-the-art VNAs. The frequency multiplier chains used in commercial VNA test-set extenders provide a strongly non-linear power relation between the input and output, nevertheless, as it was shown in [1] accurate power control at the mm-wave port is feasible up to the sub- mm-wave frequency range. Despite this, several contributions in the field still suffer from error arising from a non-accurate power level control during scattering parameter measurements, see Fig. 1 a), and errors arising from the simple scalar corrections of large-signal power-meter based setups, see Fig. 1 b). In this paper we analyze how these errors can be traced to the absence of amplitude power level control of mm-wave VNA extenders, and we describe how the use of full vector correction, using VNA based large-signal setups, allows to achieve an high correlation between the small-signal and large-signal test-benches response. a) b) Fig. 1: a), Small-signal model-hardware correlation of the PA from [2], b) measured (circles) and simulated (solid lines) S-parameters of the three-stage 150 GHz amplifier, measured with VNA and large- signal setup, from [3]. The paper is organized as follows, first an analysis of the AM distortion in the transfer characteristics of non-linear device excited by non-levelled drive signal is described, together with the limitation of commonly employed scalar mm-wave large-signal setups. Then, the proposed VNA based power controlled small-signal and large-signal test bench of [1] is briefly described. Finally, a 140GHz power amplifier characterization is used as a test vehicle to reproduce some of the shortcomings that can be encountered in mm-wave test- benches and highlight the high correlation between small- signal and large-signal setups when full vector correction is employed. II. MM-WAVES TEST-BENCHES SHORTCOMINGS A. Small-signal The small-signal characterization of power amplifiers, aims to look at input matching, reverse isolation and the small- signal gain. The latter is often evaluated simply looking at the S 21 at a sufficient back-off level, i.e., the linear region of Fig. 2. When the amplifier is composed by several stages, which is often the case at mm-wave frequencies due to the limited gain per stage, the compression mechanism of the gain curve can