A manufacturer of DC power supply units (PSU) used a voltmeter and shunts to measure load currents of its prototypes. The meter was calibrated regularly but the shunts were not -- the reason (wrongly) being that "they're passive so they don't need it". One shunt got "cooked" through misuse, increasing its resistance. This meant the PSU delivered lower power than assumed and results of thermal tests were over-optimistic. When the error was finally found, it took months to requalify a series of PSUs before they could actually be put into production.