SECTION 7

RAM ERROR DETECTION

1.0 Specification

RAM Error Detection and Corrective Action

1.1 Non-volatile Memory (NVM)

256 bytes of non-volatile RAM are maintained in the CPU by the standby power supply. Errors can occur within this RAM due to loss of battery voltage to the ECM or due to electrical noise originating within or external to the ECM This software feature provides a means by which loss of the contents of the NVM is detected and appropriate remedial action taken.

1.1.1 Memory Loss Detection

Loss of non-volatile memory due to loss of standby voltage to the CPU, NVM is tested by a checksum test on the three diagnostic memory cells (MALFFLG1, MALFFLG2, and MALFFLG3) located in non-volatile RAM. In operation, each time the system software stores information in one of the diagnostic memory locations, a checksum calculation is made on all three diagnostic locations, and the result Plus 1, is stored in the checksum NVM location (DCKSUM). Upon each reset initialization sequence, the same checksum calculation is performed and the result compared with the contents of DCKSUM. If the two differ, it is concluded that an NVM loss has occurred.

1.1.2 Memory Loss Remedial Action

Upon detection of a non-volatile memory loss, the ECM software performs the following operations:

1.The three diagnostic bytes are cleared, and the checksum byte is set equal to one.

2.All memory utilized as NVM by the ECM software, except the block learn table and idle air control variables are zeroed.

3.The block learn table and idle air control variables shall be initialized to starting values.

4.Upon detection of the NVM loss, the ECM shall set the NVM fail flag (Bit 6 of CLCCMW) for use by the IM to indicate this condition. This NVM fail flag is reset on any power-up sequence where an NVM alteration is not detected.

1.2 RAM Refresh

RAM cells may change state if the cell is not periodically refreshed. To prevent this possibility, the software reads and stores all active RAM at a rate of not less than once each 100 msec.