This section outlines the general procedure used for testing and troubleshooting memory devices. As will be seen in the paragraphs to follow, troubleshooting memory devices is far more complex than troubleshooting other digital building blocks. The procedure outlined earlier for digital building blocks such as logic gates, flip-flops, counters, registers, arithmetic circuits, etc., is not valid for testing memory devices such as RAM and ROM. One thing that is valid is that it is equally important fully to understand the operation of the system before attempting to troubleshoot. You must also remember that there is a lot of digital circuitry outside the memory device that is a part of the overall memory system. This may include a decoder circuit and some combinational logic.

The most common method of testing a RAM system involves writing known patterns of 0s and 1s to each of the memory locations and then reading them back to see whether the location stored the pattern correctly. This way, both READ and WRITE operations are checked. One of the most commonly used patterns is the 'checkerboard pattern' where all memory locations are tested with a 01010101 pattern and then with a 10101010 pattern. There are many more patterns that can be used to check various failure modes in RAM devices. No check, however, guarantees 100% accuracy. A chip that passes a checkerboard test may fail in another test. But if the chip fails in the checkerboard test, it is certainly not good.

RAM check is performed automatically. In the majority of computers and microprocessor-based systems, every time the system is powered, the CPU runs a memory-test program that is stored in the ROM. The operator can also execute this memory-test routine on request. The system displays some messages after the test is over. After that, remedial action can be initiated.

ROM devices cannot be checked by writing and reading back known patterns of 0s and 1s, as was done in the case of RAM devices. ROM is a 'read-only memory' device and its testing should basically involve reading the contents of each location of the ROM and then comparing them with what it is actually supposed to contain.

ROM testing is done with the help of a special instrument that can be used to read the data stored in each location of the ROM. It cannot be tested, like a RAM, by writing some pattern of 0s and 1s and then reading them back. One of the methods is to read data in each location and produce a listing of those data for the user to compare with what the ROM is actually supposed to store. But, of course, the process becomes highly cumbersome for large-capacity ROM chips.

Another approach is to have a reference ROM plugged into the test instrument along with the test ROM. The instrument reads data in each of the locations on the test ROM and then compares them with the data stored on the corresponding locations of the reference ROM.

Yet another method is to use a CHECKSUM. Checksum is a code that is stored in the last one or two locations of the ROM. It is derived from the addition of different data words stored in different locations of the ROM under test. For instance, if the data words stored in the first three locations are 11001001, 10001110, and 11001100, then the checksum up to this point will be 00100011. When the test instrument reads data in the test ROM, it creates its own checksum. It compares the checksum with the one already stored in the test ROM. If the two match, the ROM may be considered to be a good one. We have used the word 'may' because even wrong data can possibly lead to a correct checksum. However, if the checksums do not match, it is definitely a faulty ROM.