A rectifier stack of 1927 on a hardwood workbench in a railway signal workshop in Pennsylvania, fifteen centimetres long, about twenty round copper discs two centimetres across, each dark red to black on one face, with matt grey lead discs and larger spacer washers between them, all threaded on a steel rod with brass nuts at both ends, beside it a lead-acid cell in a glass jar with clamps, a signal relay in an oak case, a tin tray of loose copper discs, a hand in a blue work sleeve tightening a brass nut with a small spanner, daylight from a high factory window above, grey steel shelving behind.
What it was used for
The copper-oxide rectifier is a dry rectifier: copper with a layer of cuprous oxide formed on it that passes current readily in one direction only. Per the handbook Metallic Rectifiers (1957), L. O. Grondahl and P. H. Geiger of the Union Switch and Signal Company discovered the copper-oxide effect in November 1920 while trying to build a relay with no moving parts and no contacts. In service the rectifier charged the batteries of fire alarm and railway signal installations and of switchgear control, supplied DC to passenger lift control circuits, and drove electromagnets, solenoids and motor fields. A self-cooled stack was rated at about 0.1 A DC per square inch at 35 degrees Celsius; the cells worked at 2 to 10 volts, 8 volts on average.
What was new
Grondahl's patent US 1,640,335 (filed 7 January 1925, granted 23 August 1927, assignee Union Switch and Signal Company) claims in 20 numbered claims the combination of copper with cuprous oxide formed on the metal itself as a unidirectional device, plus the bolted stack with heat-radiating washers; it describes forming the oxide by heating in an oxygen-containing atmosphere to about 1000 degrees Celsius and the application to AC-DC conversion and battery charging. The barb: in 1932 the assignee filed a disclaimer that withdrew several claims entirely and limited others to copper/cuprous-oxide combinations rectifying over a relatively wide area.
What the sources leave open: Three kinds of date must be kept apart: discovery November 1920 (per the 1957 book), patent filing 7 January 1925, grant 23 August 1927. The book states the interval since the discovery as 31 years, which does not fit its year of publication; so only the date November 1920 is used here, not the arithmetic. No original publication by Grondahl and Geiger was opened.