Semi-automatic production testing of spark plugs

The spark plug has been in existence for nearly 150 years, with few significant design changes over that time. Similarly, spark plug production testing systems have evolved slowly - mainly because there have been simple, yet ingenious, tests which offered adequate 'Go / No go' testing. This paper describes a new method of spark plug testing, featuring elementary detection and diagnosis of faults. Spark voltage waveforms are captured and classified using a neural network. This paper follows up initial work reported in some other recent publications by the Author, presenting the prototype testing system and the method and results of an initial factory-based test.