Many standard tests for write nonlinearities such as nonlinear transition shift and partial erasure assume a linear read process. Nonlinear effects from readback with a magnetoresistive head can make the results of these tests difficult to interpret. We show how a polynomial can be used to approximate the inverse transfer function and thus linearize the read channel. Write nonlinearities can be measured more easily by performing tests on the linearized channel. We also show how linearizing the channel for PRML detection is limited by amplification of noise over parts of the inverse transfer curve with high differential gain.
展开▼
