The MWJ Besser Webinar was held August 17th, 2010. The following are several of the questions asked by the audience to Dr. Darwish.
To view the entire webinar,

Q: Why is GaAs HBT is popular for cell phone front end PAs?

AD: HBT are attractive for many reasons. First, they offer a single supply solution so for cost sensitive applications (such as cell phones) the expense of a negative power supply can be eliminated. Second, HBT offer improved linearity which is important for many of today’s modulation schemes. Third, the power density (output power per unit area) in HBTs is high because they are vertical transport devices as opposed to MESFETs and pHEMTs which are lateral transport devices. Last, the off state leakage current in HBTs is low which extends the battery life.

Q: Can you predict IP3 given S-Parameters to design LNA?

AD: The third-order intercept (IP3) is a large-signal parameter, which is linked to saturation current, maximum voltage, class of operation, etc. The S-parameters are small-signal parameters that are not directly related to output power, IP3, or efficiency. Thus, in general, one cannot predict IP3 from S-parameters.

Q: What are some of the common pitfalls in active device modeling?

AD: The most common pitfall is to use the default simulator’s optimization settings. Simulators, during optimization, give equal weight to S11, S21, S12 and S22. However, in typical transistors, S21 is larger than S11 and S22, both of which are much larger than S12. This leads to a good match for S21, a poor match for S11 and S22, and an unacceptable match for S12. A better approach is to define optimization goals which are normalized such that all S-parameters are given equal weight. It is also critical to match the magnitude and phase of the S-parameters. Second, typical lumped element models are ‘point-like’ models and do not take into account the distributed nature of the actual transistor. Thus, a common mistake is to try to match the model to the measurement without de-embedding the input and output feeding structures of the active device.

Q: Is MESA resistor nonlinear vs TaN resistor?

AD: Sputtered thin film resistors such as Tantalum Nitride (TaN), and Nickel Chrome (NiCr) usually offer improved linearity and lower temperature coefficients compared with mesa resistors.

Q: Are stability factors K and delta valid for multi-stage amplifiers? If not why not?

AD: They are valid for multi-stage amplifiers. However, it is not enough to check the stability at the input/output of the entire amplifier. One should check the stability of each stage to ensure that no internal oscillations occur.

Q: What is odd mode oscillation, how is it different than bias ports oscillations and what are some techniques to avoid either?

AD: The odd mode oscillation is an oscillation mode where two (or more) parallel transistors have anti-phase oscillation, similar to a push-pull or differential mode operation. The standard stability factors (such as K, and B1) calculated by CAD packages only relate to even mode oscillations. Thus, an amplifier having K > 1, and B1 > 0, can still experience odd mode oscillations. One can easily dampen and stop odd mode oscillations by placing bridging resistors between the parallel transistors (see RFIC and MMIC design and technology, by Ian D. Robertson, Stepan Lucyszyn).