DRAM operating states and burn in

Abstract

Observation of system level DRAM operations can provide insights for developing improved reliability screens such as burn-in. We instrumented a dual rank 4GB RDIMM to probe command bus traffic from the memory controller to the DRAMs on a server mother board. Internal DRAM operating states are derived by analyzing the command bus traffic. Analysis showed that when the controller uses an open page policy with bank interleaving, the active idle state can become the dominant bias configuration in activated arrays. This result can be applied to the development of an efficient burn in stress.

Signal Transmission Analysis of Burn-In Socket for Fine-Pitch Ball Grid Array Package

“…Burn-in is a process in which semiconductors or electronic devices are operated repeatedly at elevated temperature and voltage prior to being placed into operation [1][2], and thus, initial failures in the semiconductor package will be detected by the burn-in test process [3]. A burn-in socket is an electromechanical component that provides a separable electrical interconnection between the semiconductor package and the printed circuit board (PCB) located on the test equipment.…”