Abstract
Recently, the FCC has released a very wide unlicensed spectrum allocation from 3.1 to 10.6 GHz. The allowed emission in this spectrum is very low, so the use of this spectrum allocation limits itself to relatively short range applications. The fact that CMOS technology now reaches speeds of tens of GHz, opens up a whole new area of interesting possibilities to create cheap and wide band radio technology. One application here, is short-range radar. Thanks to the wide spectrum allocation the radar is able to send impulses rather than bursts of a carrier wave. This makes processing of the received signal much easier than the matched filters which are required in the carrier wave burst case.
In this master thesis we present two related sampling techniques for radar applications which use mostly digital circuitry and which can achieve high sampling rates. We have called these circuits, which are partially based on the Suprathreshold Stochastic Resonance (SSR) principle, swept threshold and stochastic resonance samplers. Although the samplers are mostly digital, which makes them perfect for CMOS, they are not clocked. We discover that for the case where the input signal contains much noise, typical for radars, the crudeness of these simple samplers does actually not have a very detrimental effect on the signal processing. A radar implementation in 90 nm CMOS using these samplers is presented, which is shown to reach sampling rates of about 23 GHz.