BOSTON — The Cat5e Ethernet cable that runs through the walls of your home may quite adequate for your needs. But, in plenty of locations even 1 Gbit/s speeds aren't enough. Take, for example, public locations such as coffee shops, offices, or shopping mall food courts, which must support dozens or even hundreds of users.
While 10 Gbit/s Ethernet has become commonplace in datacenters, it's not backwards compatible with 1 Gbit/s signaling. For locations that must serve many users, typically over Wi-F1, installing cable capable of handling 10 Gbit/s signals between switches, routers, and wireless access points simply costs too much.
Mike Klempa, UNH-IOL
That's where IEEE 802.3bz and NBase-T comes in. These standards define signaling for 2.5 Gbit/s and 5 Gbit/s Ethernet over existing Cat5e and Cat6 cables. With many chips, boards, and routers, switches, and wireless access points ready to come to market, compliance and interoperability tests are important. After all, products from every manufacturer needs to communicate with everyone else's products. To meet that need, the University of New Hampshire Interoperability Lab (UNH-IOL) has added 2.5 and 5 Gbit/s Ethernet testing to its long list of testing technologies. EE Times spoke to UNH-IOL's Mike Klempa (pictured) about the service on February 21 about how the lab tests these speeds.
"The IEEE 802.3bz specification grew out of the NBast-T standard," said Klempa. "We offer testing for both. We're a member of the NBase-T Alliance and products must play nice with both technologies. They both run at the same data rates, but there are some differences that can trip you up if your products aren't tested for both."
This Ethernet test bed is one of many test beds at UNH-IOL in Durham, New Hampshire.
According to Klempa, anyone can replace their 1 Gbit/s network equipment with the higher speeds without changing wiring. Both speeds can run at distances to 100 m. You need not go to Cat6a cables, which support 10 Gbits/s. But, because of bandwidth and far-end crosstalk (FEXT) issues, products must negotiate speed and will drop from 5 Gbit/s to 2.5 Gbits/s when necessary. "The 2.5G and 5G physical layer (PHY) circuits is based on 10G, simply halved and quartered," said Klempa.
Another new feature for the PHY is ability to reduce signal power for short cable lengths. Lowering the power reduces FEXT, which improves performance. That feature is what makes it compatible with Cat5e cable as where 10 Gbit/s Ethernet requires Cat6a cable. The figure below shows where the 2.5 Gbit/s and 5 Gbit/s signaling fits into the overall Ethernet scheme.
2.5 Gbit/s and 5 Gbit/s take their places among the family of Ethernet speeds. Source: Wikipedia
Conformance testing is similar to 10 Gbit/s testing. Measurements include S-Parameters on the transmission channel with a network analyzer and power-spectral density measurements on the incoming signal at the receiver. In the time domain, tests use a real-time oscilloscope to display eye diagrams and make jitter measurements. Error rates are specified in terms of frame errors, a rate of 7.8E-9.
In addition to compliance testing, the UNH-IOL hosts plugfests. The most recent took place in October 2016 with the Ethernet Alliance and NBase-T Alliance to prove the technology works across both implementations. Another plugfest should take place in the third question of 2017.
—Martin Rowe covers test and measurement for EE Times and EDN. Contact him at firstname.lastname@example.org