Details

Abstrakt in Englisch

This work presents the system and circuit design, modelling and optimization of a frequency modulated continuous wave (FMCW) indoor positioning radar operating at the 2.4 and 5.8 GHz industrial, scientific and medical (ISM) bands. Various non-ideal effects on FMCW radar ranging precision and accuracy are modelled and investigated such as crystal oscillator (XO) tolerance, receiver (RX) thermal noise, fractional-N phase locked loop (Frac-N PLL) phase noise and systematic chirp nonlinearity.
Optimization techniques are proposed to the Frac-N PLL for highly linear wideband chirp generation. Verilog simulations as well as measurement results confirm a low root mean squared (RMS) chirp nonlinearity error of 14.6 kHz. A single-ended low noise amplifier (LNA) design alleviates the need for an external passive balanced-to-unbalanced (balun) converter. A gain interpolating variable gain amplifier (VGA) architecture is used and simple techniques are proposed to improve the input signal handling capability and VGA linearity. A baseband detector architecture is introduced with a detection accuracy of ±0.15 dB. The high accuracy is attributed to the use of an inherent process, voltage and temperature (PVT) cancellation concept. A nonlinear model for the automatic gain control (AGC) loop relying on simple and readily available components from the “analogLib” and “functional” libraries is built in the CADENCE design environment. The model provides insights into system level parameters such as AGC loop bandwidth, phase margin, settling time as well as estimating the AGC dynamic range and received signal strength indicator (RSSI) voltage vs. input power. Measured in lab conditions, the developed transceiver (TRX) chip achieves a ranging precision of 0.3 and 5.2 mm in primary and secondary radar configurations respectively. Indoor ranging precision is measured to be between 1.2 and 1.5 cm while the RMS indoor ranging error is at 17 cm. When used for a practical indoor positioning scenario, the system achieves an RMS tracking error of 10 cm which competes with more sophisticated state-of-the-art multiple- input multiple-output (MIMO) based local positioning systems incorporating a larger number of transmitters (TXs), RXs and base stations (BSs).