Active Wideband-Impedance Load Pull Measurements

Introduction

---

When working with modulated signals, for a well controlled linearity behavior of the DUT, the reflection coefficients offered to the DUT should ideally be constant (not vary versus frequency) within the modulation bandwidth at the fundamental, as well as in all related frequency bands at baseband and harmonic frequencies. This situation is approximated in real circuit implementations, where the matching networks are placed directly at the reference planes of the active device.

In conventional load-pull setups, however, the actual physical impedance is always located at some distance from the DUT, which is much larger than for any practical matching network. This distance, as well as any physical length within the tuning element itself (such as the position of the probe in mechanical tuners), yields very large electrical delays causing rapid phase changes of the reflection coefficients versus frequency.

It is clear that these large phase deviations represent nonrealistic circuit conditions and will cause measurement errors such as IM3 asymmetry, spectral re-growth and EVM degradation. In general, maintaining the reflection coefficients constant over frequency is getting more and more difficult with the increase in modulation bandwidth of communication signals, not only in practical circuits, but definitely in load pull measurement setups.

Wideband Impedance Control

---

To overcome the aforementioned problems of losses and electrical delay in conventional load-pull characterization systems, while being able to work with realistic wideband communication signals, a novel open-loop system was developed employing wideband signal generation and signal acquisition (see MT2000 product page). When the nonlinear DUT is excited with a user-defined modulated signal, it generates signals in the baseband, fundamental and higher harmonic frequency bands. By measuring the device reflection coefficient at every frequency, the waves to be injected are estimated at every iteration. When the required reflection coefficient versus frequency (at every controlled band) is achieved, the iteration has converged and the large-signal parameters (power added efficiency, output power, intermodulation distortion, etc.) are measured. To address these needs.

Figure 1. Source and load reflection coeficients at the device reference plane in the fundamental (2.1225– 2.1575 GHz) and harmonic (4.245–4.315 GHz) frequency range, with electrical delay (open symbols) and without electrical delay (?lled symbols).

Figure 2. Measured output power spectral density (dBm/ Hz) versus frequency [GHz] of an NXP GEN 6 LDMOS device (gatewidth 1.8 mm) in the proposed load–pull setup. (a) At the fundamental frequency band using a 3 kHz resolution bandwidth. (b) At the second harmonic frequency band using a 6 kHz resolution bandwidth. The measurement is shown for the two cases with (dashed line) and without electrical delay (drawn line). The re?ection coef?cients offered to the DUT are given in Figure 1.

Sitemap | Contact Us | Privacy Policy | Terms & Conditions

Copyright © 2012 - Maury Microwave Corporation. All rights reserved.

ISO: 9001:2008
AS9100:2009 Rev C
Registered