Modeling of low-frequency noise in GaInP/GaAs hetero-bipolar transistors

Heymann, P.; Rudolph, Matthias; Doerner, Ralf; Lenk, F.

doi:10.1109/mwsym.2001.967295

Cited by 9 publications

(1 citation statement)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In general, measurements for n different source-resistance values allow the separation of up to n independent noise sources, but for the simplified HBT equivalent circuit the voltage sources at base and emitter cannot be separated. From the simulation point of view, both contribute the same noise at the output, but from physical considerations one can conclude that the emitter layer acts as the relevant LF noise source [1]. In Fig.…”

Section: /F-noise Hbt Descriptionmentioning

confidence: 91%

Modeling and Measurement of Phase Noise in GaAs HBT Ka-Band Oscillators

Lenk

Schott

Heinrich

2001

2001 31st European Microwave Conference

View full text Add to dashboard Cite

 Accurate oscillator phase-noise simulation is a key problem in MMIC design, which is not solved satisfactory so far and needs further investigation. In this paper, a Ka-band MMIC oscillator with GaInP/GaAs HBT and on-chip resonator is treated as an example. Measured phase noise reaches -90 dBc/Hz and below at 100 kHz offset. To evaluate phase-noise prediction, the circuit is simulated using different commercial simulation tools and HBT models. Considerable differences in simulation results are observed. I. MotivationLow phase-noise oscillators are key components in communication as well as sensor systems for the emerging markets in the 20 GHz to 80 GHz range. In order to meet the low-cost requirements, monolithically integrated solutions (MMICs) with on-chip resonators are highly favorable. This demands for both transistors with low 1/f-noise and accurate CAD tools for circuit optimization. Regarding the transistor, the GaInP/GaAs HBT is particularly suited since its 1/fnoise is considerably lower than for GaAs HEMTs, but the frequency potential is higher than that of SiGe HBTs. Phase-noise prediction, on the other hand, involves too much uncertainties so far to be a reliable basis for circuit design. Further investigations are required on this su bject.The purpose of our paper is to contribute results in this field. As an example, we study a GaAs-HBT-based 36 GHz oscillator. Attention focuses on the phase noise at 100 kHz offsetfrequency. From the measurements, it can be seen that this frequency is in the transition region between 1/f 3 and 1/f 2 phase-noise behavior. Phase-noise modeling, therefore, needs to be composed of the following issues [6]:• Modeling and parameter extraction of the 1/f-noise of the active element used in the circuit.Measuring 1/f-noise of the GaInP/GaAs-HBTs at different source impedances and bias states, we found two independent low-frequency (LF) noise sources. The two noise sources are implemented in an in-house HBT large-signal model. The results are compared to simulations with a standard Gummel-Poon model, which includes only a single LF noise source.• Calculation of the conversion from LF noise into oscillator phase-noise. The simulations of the 36 GHz oscillator are performed using 3 different commercial design tools (MDS and ADS of Agilent and Serenade of ANSOFT). II. 1/f-Noise HBT DescriptionIn standard HBT models (e.g. Gummel-Poon) only the base current includes 1/f-noise. However, measurements of the spectral current density of 1/f-noise at the output of GaInP/GaAsHBTs with different source resistances R s show that this description is not sufficient. In the lowfrequency range, all of the extrinsic reactive elements can be neglected. Hence, the equivalent circuit of the HBT simplifies to a Y matrix with R be , and β and an additional Z matrix containing the parasitic resistors R b , R c , and R e (see Fig. 1) The 1/f-noise model should be consistent with the high-frequency noise model, so we suppose voltage sources at the extrinsic resistors and current sources a...

show abstract

Section: /F-noise Hbt Descriptionmentioning

confidence: 91%