Original Publication Date: 1992-Dec-01
Included in the Prior Art Database: 2002-Jan-18
In the past, Bum-In and High Temperature over Operating Life (HTOL) Testing was performed by sequential down-loading of test code. This was done by A clock pulse that would be fed into a counter with a binary output of 13 lines. This has a total of 8,192 possi- ble addresses. The binary counter output was fed into EPROM's where the actual program would reside. The counters would start by presenting the address "Zero" to the EPROM and, in turn, the EPROM presents the data for that address to it's output. The counters count up in a contiguous manner until they reach their last address and roll over to the "Zero" address and con- tinue or until a reset signal is asserted. The EPROMs would output Code to the DUT's and a feedback signal from the DUT's were checked to verify: 1.) Did the code get accepted by the device? 2.) Did the device to acknowledge the instructions? 3.) Did the code put a programmable gate into an output low/high state? This method fails to truly verify operation of a CMMU. A CMMU cannot "obey" commands, it just routes Addresses and Data in a more efficient manner. This makes valid testing Cacheable Memory Manage- ment Units fairly diicult. In the past we checked a Bus Feedback pin (C2) with an osdloscope to make sure that it was a static high. If any part should have the signal 'C2' go low, all parts would permanently latch the status, until either the driver was Reset, or the Burn-In Cycle was ended. Because of this, it was necessary to create and develop the RISC-l Driver. A part could cat- astrophically fail and not put a low pulse on C2. Simply testing a device for a static signal is not a valid test for functionality of a device. These conditions could all occur and not identify a non-functional part.