The extract I downloaded for Richmond County, Virginia (county code 1590, not Richmond City) has no observations in the 1940 census in either the 1% or 100% sample. It is present (for the same variables extracted) in the 1920 and 1930 1% and 100% samples. There are images of enumeration forms on the National Archives 1940 Census site:
What’s the next step I should take regarding this issue?
For the 1940 data, it appears that the data for Richmond County, VA has been lumped in with the data for Richmond City, VA. I did a quick tab on the 1940 100% count data for the VA counties, and the result for Richmond City appears to be the sum of the Richmond County + Richmond City pop counts (based on data from IPUMS NHGIS).
Roanoke County and Roanoke City are also lumped together in 1940.