A real-time analyzing system in qualifying process of magnetic hard disk drives

In hard disk quality control process, a set of m read/write heads are usually used to measure the width of head and bit error rate (BER) of a magnetic recording disk (MRD) under test to decide whether the disk under test is deflect. However, the quality of the read/write heads degrade for every read/write. To guarantee that the heads are still function, these heads must be changed after using them to test n disks. This paper proposes a statistical-based method to continuously using a head as long as it can be used to test the disks while guaranteeing that the testing is correct. For each disk under test, the width of head and BER are measured by the five heads. For each head, if either measured value is out of the respective acceptable range, the disk is graded as deflect by the head. The actual grade of the disk is based on the majority grades of the five heads. Then, degrade counts of the heads that are in the minority are increased by one. Continuously, if the degrade count of a head is greater than the threshold, the head is considered degraded and must be changed. The method was tested with 5 good heads and 5 bad heads using 10 good disks and 10 bad disks. The results showed that with 1, 2, 3 and 4 good heads, the accuracy of the test are 85%, 89%, and 96% respectively. This means that as long as not more than 2 heads that degrade at the same time, the method can correctly test the disks while keeping the heads as long as they do not degrade lower than the threshold.