PRML channel performance under the influence of medium noise in tape recording systems

Previous studies have shown that tape medium noise is dominated by surface roughness. Here, we use error rate analysis to evaluate the effects of surface roughness as well as particle size distributions in a partial-response maximum-likelihood (PRML) channel. We use: 1) a modified self-consistent model, incorporating a log-normal particle size distribution and a random packing of the particles to simulate the recording process and 2) a predeveloped error rate model to evaluate the system performance with fixed head/medium combinations. Surface roughness is introduced as a varying random head/tape spacing during both the recording and the playback processes. The impact of mean particle length, roughness variance, and roughness correlation length on system performance is examined for a partial response (PR4) channel. A surface roughness variance changing from 5 to 10 nm yields, at 200 kfci with a particle length of 60 nm, a change of error rate from 10/sup -10/ to 10/sup -5/. Because of the finite head read track width, roughness with shorter correlation lengths can be cross-track-averaged more readily and yields better system performance. We also found that larger particle size media have higher DC noise but lower roughness-induced transition noise. The shorter-particle-size media give better system performance, and the advantage increases with reducing roughness variance.