random access memory

Random access memory is the fast, temporary storage that a processor uses while running. Unlike a hard drive or flash storage, RAM loses all its contents the moment power cuts off. A modern workstation might have 16 or 32 gigabytes of RAM; an industrial control system might use far less. The processor can read from or write to any address in RAM in roughly equal time, which is what "random access" means. This is different from sequential storage like magnetic tape, where you must wind through everything before you to reach what you want.

The key distinction is volatility. RAM is volatile memory: it requires continuous electrical power to maintain its contents. DRAM (dynamic RAM) is the most common type in general computing. It stores charge in tiny capacitors, one per bit, and must be refreshed thousands of times per second to prevent the charge from leaking away. SRAM (static RAM) uses bistable flip-flops instead and does not need refreshing, but costs more per byte and consumes more power. Industrial equipment often uses SRAM where reliability and speed matter more than cost. Specialized applications use NVRAM (nonvolatile RAM), which combines RAM speed with persistent storage, typically using battery backup or flash memory alongside a SRAM array.

The speed of RAM is measured in access time (typically nanoseconds) and bandwidth (gigabytes per second). Modern DRAM speeds are expressed in MHz or data rate (e.g., DDR4-3200 means data transfer at 3200 MHz). Latency, the delay between a read request and the appearance of data, matters for real-time systems. In manufacturing control, servo drives, and test equipment, slow RAM can introduce unacceptable jitter or missed timing windows. Industrial embedded systems often specify RAM timing explicitly in their datasheets.

Failure modes and environmental limits

RAM degrades under thermal stress. Most DRAM operates within 0 to 70 degrees Celsius; exceeding this shortens its life and increases soft error rates. Soft errors (single-bit flips caused by cosmic radiation or alpha particles) are rare in a single module but become statistically significant in large installations or aerospace applications. Error-correcting code (ECC) RAM detects and corrects single-bit errors in real time, critical for servers and mission-critical control systems. Parity RAM, a simpler scheme, only detects errors without correcting them.

In industrial settings, RAM must survive vibration, electrical noise, and temperature extremes. Military and aviation specifications (e.g., MIL-STD) often demand extended temperature ranges. Burst errors, multiple bits corrupted together, can occur from power supply transients or electromagnetic interference; proper decoupling capacitors and shielding mitigate these. RAM modules themselves are usually not field-replaceable in embedded systems; instead, the entire board is swapped. Desktop and server RAM uses standard form factors (DIMM, SO-DIMM), making replacement straightforward once the system is powered down and discharged.