German S. Feyh

An EPR4 read/write channel with digital timing recovery

This paper presents a device that uses digital interpolation of asynchronously sampled data to perform timing recovery. Analog-to-digital conversion (ADC) and finite impulse response filter (FIR) latency do not contribute to loop delay, moving most of the pulse equalization to the digital FIR. By eliminating the variable frequency oscillator, crosstalk is eliminated. Very low oversampling rates are shown to be sufficient, enabling data recovery at a low rate loss. Calibration modes for low-pass filter tuning and the flash ADC are used to improve performance of the analog channel. The 0.35-/spl mu/m CMOS device is specified for data rates up to 245 Mbps.

Subsampled digital retiming for optical disk

As profit margins and market windows shrink, it becomes desireable to design as much of the optical storage electronics with digital CMOS. By using CMOS processes, the drive designer ensures the cheapest silicon possible. Unfortunately, the analog CMOS designer faces a greater challenge when designing in CMOS. For highest yield and quickest design cycles the percentage of drive functions implemented with digital blocks should be as high as possible.

Inverse eigenvalue problem for sinusoidal frequency estimation

The estimation of the frequencies of sinusoids in noise is a very common problem. This paper addresses the estimation of sinusoids in a low SNR environment. This sinusoidal frequency estimation problem can be used to find the carrier frequencies and baud rates of communication waveforms after some appropriate nonlinearity. If the underlying signal model is sinusoids in white Gaussian noise and we use the forward/backward prediction framework, then the forward/backward prediction equations force a Toeplitz/Hankel structure on the data matrix. If there are M distinct sinusoids in the data and no noise, then the data matrix has rank M. Cadzow and Wilkes (1991) enhance a noisy data matrix by enforcing both the structure and the rank of the data matrix, before solving for the coefficient vector of the prediction problem. Besides the Toeplitz/Hankel structure, the estimated singular values of the data matrix are also enforced. Using more information extracted from the original data matrix extends the threshold to lower SNR values.