Twisted Pair Testing
Twisted pairs introduce complexity that can make assemblies more difficult to build and test. However, there are some helpful guidelines that can lead to better overall results.
Why Twisted Pairs?
First, it’s helpful first to understand why twisted pairs are used. A signal (current) on a wire generates a magnetic field that can cause interference on adjacent wires. This interference is called “crosstalk” and it negatively affects performance. The longer the wires and the higher the frequency of the signals, the more crosstalk occurs. Twisted pairs, two wires that are twisted together, are part of a system used to reduce crosstalk to and from adjacent wires. To begin, signals are sent down the paired wires so that when one wire becomes positive the other wire becomes negative by the same amount. This is called differential signaling. Neighboring wires close to this pair will be affected by crosstalk equal to the sum of the two signals.
Therefore, if this sum is zero (or nearly zero) then the effects of crosstalk are reduced or eliminated. However, if straight instead of twisted wires were used in the pairs, adjacent wires could be closer to the positive or negative wires for their entire length, which would greatly reduce the benefit of differential signaling. The twist addresses this problem by varying the distance between the positive and negative wires and their neighbors over the length of the cable.
Twisted Pair Manufacturing Defects
Some characteristics of twisted-pair performance, such as return loss, which is tested at high frequencies, cannot be measured on cable / harness testers. Instruments that perform such testing are designed specifically for this narrow purpose and are unable to perform other types of testing at which Cirris testers excel. In short, the reason for this is that design priorities for cable / harness tester often conflict with the capabilities needed in a tester that performs specialized high frequency measurements.
However, several Cirris units, including the CH2, Easy-Touch Pro , and 4250, can test for the common manufacturing defects that affect the performance of twisted-pair assemblies without the need for dynamic analysis using high frequencies. As in other wired assemblies, continuity and isolation testing can detect opens and shorts. Additionally, these Cirris units can perform twisted pair testing that checks for consistent twists and for split pairs. A split pair error occurs when one wire from each of two different pairs gets swapped identically on both ends of the cable. The result is a cable that will pass a standard continuity test, but will have serious crosstalk problems, and will most likely not perform adequately at specified data rates.
Correct
Split Pair
Learning and Testing
When performing a learn of an assembly that includes twisted pairs, the tester generates a square wave signal (noise) on a wire we’ll call Signal+ (S+) and measures the interference (crosstalk) induced on other wires in the assembly. The wires are sorted according to the level of measured interference. The two wires with the highest interference levels are identified as “Signal-” (S-) Candidates 1 and 2. The tester then applies differential square waves (waves that are 180 degrees out-of-phase) to the S+ and S- Candidate 1 while it measures the interference on a “listen line” in this case, S- Candidate 2. As the square waves are 180 degrees out-of-phase, they should cancel each other out if the two wires are part of a twisted pair. Therefore, if the listen line is quiet, then the S+ and S- Candidate 1 are a twisted pair consisting of S1+ and S1-. If the listen line is noisy, then the test is repeated with the roles of S- Candidates 1 and 2 reversed. Depending on the results of the second test, the roles of S+ and S-, and the listen line may be switched until the pair is identified.
During testing, the twisted pair measurements are made in a similar way using differential square waves and a listen line. The process is completed quickly on properly assembled devices. However, when a device fails, a process like the learn is used to identify the failed pairs and as a result the test will take longer to complete.
Tips for Testing Twisted Pairs
The nature of twisted pairs can make it difficult to achieve the desired results. A few tips may help.
- Twisted pairs shorter than 5 feet (1.5 meters) can make it difficult to perform accurate measurements and, therefore, may produce incorrect or erratic results. Depending on circumstances:
- Make the twisted pairs twice as long as needed and test from both ends, then cut the cable in half. The result will be two cables which have both been through twisted pair testing and only need continuity and isolation testing after terminating the second end on each.
- As twisted pair testing can be performed when only one end of the cable is terminated, twisted pair wire on a reel can be tested. Ensure one end of the spool of cable is terminated. After testing for twisted pairs cut the cable to length from the reel and terminate the other end. Using standard continuity and isolation testing will now guarantee that the cable has been terminated correctly. The additional length of the cable on the reel will also give the tester more margin while measuring for the presence of twisted pairs and will produce more repeatable results.
- If not connecting directly to an adapter on a Cirris benchtop tester (Easy-Touch Pro / 4250), it is best practice to keep the interface as short as possible and to use test cables assembled with high quality twisted pairs following the same patterns found in the device under test.
- Coaxial cables can be used as a substitute for twisted pairs in the test fixture. In this case, all twisted pair points must be connected through an individual coaxial cable center conductor and the shields of all the coax wires must be connected together and grounded to the tester chassis.
Poor quality wire
Poor quality twisted pairs, including any of the following conditions, can complicate testing:
- Each pair is not shielded separately
- The wires have less than 2 twists per inch
- There is a general twisting lay to the bundle every foot for the length of the cable.
Cirris Testers Capable of Performing Twisted-Pair Testing
Cirris testers that can perform the twisted pair testing described here, include:
A Final note
A Cirris twisted pair test is not a substitute for CAT5, CAT6, or other types of RF testing and performance certification. A Cirris twisted pair test only ensures that the quality of the twists is consistent, and that the cable is free of split pairs. To check performance characteristics (propagation, impedance, frequency, crosstalk, packet loss) you will need a different type of instrument (example : http://www.flukenetworks.com/datacom-cabling).
