This article looks at the various test situations you may encounter and gives our recommendations for the appropriate low voltage testing.
This is justified when testing does not find sufficient defects to justify the additional cost in testing. This assumes the process of manufacturing assemblies can be expected to stay in statistical control.”No testing” is justified also when the assembly is simple and visual inspection can consistently find defects.
Manual testing “ringing-out” an assembly
This is economically justified when quantities are small enough and labour is sufficiently inexpensive that fixturing for test represents a significant cost of production.
Eliminating testing for shorts
This is a risky approach. This is usually done to reduce test time. It should only be allowed when there is no reasonable chance a short can occur and the cost of a short is very small. THIS FEATURE IS NOT AVAILABLE on Cirris testers due to the risk of passing cables with shorts. We have developed very high speed methods of testing for shorts. The time used to test for shorts is small.
Single pass tests that continue-as-soon-as-good
This should not be used on assemblies that may have intermittent open defects since this allows operators to “wiggle and jiggle” a defective assembly until it tests “good.” The preferred solution when intermittent connections may be present would be rapid repetitive testing.
Crimp force monitoring
Automated crimp-force-monitoring can find defects in wires to contact crimps that an electrical resistance test will miss.
Expected increases to connection resistance
The resistance of a wired connection increases when wires are longer, wire diameter is smaller (gauge is a larger number), and temperature is higher.At 20oC (68oF) resistance for 28-gauge, 7-strand tinned copper wire is .0649 ohms per foot (.0373 for #26, .0233 for #24, .0147 for #22, .0103 for #20, .0059 for #18, and .0037 for #16). A temperature change of 2.5oC (4.6oF) changes resistance by about 1 percent.
Test fixturing adds to the overall resistance the tester measures for a connection. Each wire being tested has two fixturing wires connected to it. So the fixture resistance is often double what you might expect. A 10-foot, 28-gauge ribbon cable will have .0649 x 10 = .649 ohms of resistance per conductor. If this ribbon cable is used between the tester and the fixture, the path out and then back from the device under test adds 2 x .649 = 1.298 ohms per measurement. Also, connectors used in fixturing will add from .05 to .5 ohms per connector pair. These values can be much higher if fixture connector contacts are worn.
Compensating for fixture resistance with “Tare Values”
Raising the minimum acceptable resistance thresholds for connections can compensate for fixture resistance. Automatic setting of this tare value without careful measurement and calculation can easily create invalid tests. Also, accuracy is degraded when tare resistance is a significant portion of the DUT resistance. Accuracy degrades by the proportion that fixture resistance is to the DUT resistance [accuracy of measurement = (accuracy of measuring equipment and variation in contact resistance) x (1 + fixture resistance/device under test resistance)]. Five percent accuracy on a .5 ohm cable with 1 ohm of fixture resistance subtracted would degrade accuracy increasing the tolerance to 5% x 1.5 / .5 = 15%. Variation in contact resistance of any fixture connector will exaggerate these effects.
Eliminating fixture resistance with 4-wire testing
Fixture resistance can be eliminated with 4-wire testing. Usually accuracy is maintained even when fixture resistance is much larger than the DUT resistance
Detecting broken strands
One or two broken strands of a stranded wire does not create detectable changes in resistance. The effects are much smaller than the variation from the connector contacts, wire length, and temperature during measurement. Several broken strands typically change the wire’s resistance less than 0.1%.