Question 1: Should "Unused Contacts" be Tested?
Best practice is to test every point in every connector. Sometimes people will scrimp on the number of test points needed by populating only the contact positions in their fixturing with the pins actually used in the cable or harness to be tested. We call this a "Partially Populated" rather than "Fully Populated" test fixture. Here are reasons to choose between the two approaches.
|Partially Populated Connectors / Fixtures||Fully Populated Connectors / Fixtures|
Leaving unused contacts disconnected from test points is considered when:
There is no contact present in your connector cavity. If a contact is present in the Device Under Test (DUT), the risk of missing a defect might still be low if the contact is completely covered (such as an insertable contact in a MIL-style Cylindrical connector). Soldered contacts are exposed and can have solder bridges. They should not be left untested. Exposed contacts (soldered and IDC contacts) can be a risk for shorts and HV breakdown from loose strands. *
Including unused contacts in connectors is considered when:
You want 100% test coverage for all types of errors - Opens, Shorts or Mis-Wires, and Breakdown to all contacts.
|It is understood that a wire reported as missing during testing (OPEN) is either truly OPEN, or is actually mis-wired to an untested point.||You want to differentiate miswires from missing wires in error reports.|
|You want to save money by using testers with a smaller number of test points or want to reduce fixture labor and materials especially if the fixture will be dedicated to a particular assembly.||You want to save money by standardizing on a library of adapter cables that can be used on a variety of assemblies. Or, you want to easily adapt to modifications of assemblies because all points are "live."|
* Do shorts to unused contacts really matter?
While you may have complete information about a cable assembly, you seldom have complete information about the devices that a cable-assembly-to-be-tested "connects between." What is an unused contact in a connector may be "used" in the device. These contacts can carry signals or provide power but are not used in the specific application for which the cable is intended. If these unused signals are shorted by an attached cable to other connections in use, problems likely would occur in the device to which it connected.
Question 2: Should Connector Shells be Tested?
Plastic shells obviously don't need to be tested, but what about metal shells and even plastic shells with conductive coatings? If you have a drain wire or shield wire that connects from shell to shell, or shell to any other contact, then you definitely would connect a test point to the shell in order to test this connection.
What if there is no connection to the conductive (metal) shell?
Even if a metal shell has no connection to it, best practice would require shorts and IR/DWV testing to the shell since a defect could result in a short to a chassis or earth ground in the device at the end to which the cable connects. Conductive backshells substantially increase the risk of shorts by stray strands and insulation failures from insulation compression failures. See more on compression failures.
Some shells are not designed to have a predictable resistance value when mated. In these instances, an alligator clip can be used as a good alternative. At times the nuisance factor has been used to justify a no connection.
What if the shield connection is only to one end?
In the case of a shield or other connection to only one end a capacitance test should be used to determine that the shield is properly connected (hopefully your tester supports such capability). The metal shell at the other end should be connected to a test point to confirm that the shells are not connected to each other.
If a connection is made to the shell, connect a test point to the shell.
If the shell is conductive to the mating shell, still connect a test point to the shell unless you have evaluated your assembly and know it can't result in a failure.