Power Supply Design Minimizes Computer Memory Errors

ARINC 413A and RCTA/DO160C indicate that maximum input power interruptions may range from 0.05 to 1.0 seconds, as well as various other power anomalies.  To date power supplies and their relationship to hard drive and flash drive failures have received little too no attention, or have concentrated on traditional mechanical and / or environmental factors.

White traditional failure modes are the largest cause of failures, close examination of computer hardware shows that a significant number of failures occur when power is interrupted to hard disk and flash drives during memory write cycles.

Often when power fails during these write cycle the file allocation table (FAT) can be destroyed, necessitating the manufacturer to perform a low level format.

Experience has shown that when power supplies with an uninterruptible feature such as a hold-up capacators or batteries are utilized in conjunction with hardware and software data back-up, as well as, firmware data write interrupts, drive reliability and service life are greatly increased. Figure 1 shows a suggested primary power interrupt sequence that hardware and software designers must take into account.  An example would be for a non-critical system T₁ would be equal to 0.4 seconds and that by time T₂ all data would be backed-up and all files would be closed.  As for critical systems sufficient back-up must be guaranteed so as not to interfere with normal operations.

Figure 2 is a system block diagram showing all the major elements of a UPS design.

• T₀ = Input begins voltage drop
• T₁ = Hardware interrupt to computer bus (T>DO160C minimum)
• T₂ = Output drops in input not restored
• T₃ = Input restored
• T₄ = Input stabalized, output voltage restored
• T₅ = Power good signal sent to re-booted computer