BBH Central IconBBH Central Home Page
  CENTRAL home  |   BBHL home About/Contact Us  |   Subscribe  |   Index by Topic  
You are here: central > Broadband Home Labs > Home Networking > Homeplug > Test Procedure
Revised 6/14/03

HomePlug Test Procedure

We first tested HomePlug products based on the Intellon A1 chip from three companies - Asoka USA, Phonex Broadband, and ST&T xNetworks; we did full tests of Ethernet bridge products from all three companies and USB adapters from Asoka and ST&T. We later tested ST&T's U22 USB "wall adapter" based on a Cogency chip, and ST&T's newest M53 Ethernet bridge and U23 USB adaptor based on the latest Intellon X1 chip.

We performed quantitative measurement of network speed by creating a diagnostic -- HONEST: HOme NEtworking Speed Test -- to precisely time the transfer of a large file between PCs. This gives transfer rates substantially lower than the "raw" network speed (HomePlug claims 14 Mbps), but we believe it provides a realistic speed estimate and comparison between different networking technologies.

We chose fourteen outlets around the house for testing, including several far from the main electrical panel at the opposite end of the house. Our house has a subpanel for an addition, and we chose several outlets operating from circuits on that subpanel. We chose test circuits on different phases so we could test the ability of HomePlug to "cross" phases. See the test locations page for a diagram of our house showing the location of each tested outlet.

Our testing model was as follows:

  • We used one HomePlug Ethernet bridge as the "master" adapter. We plugged it into the Ethernet switch that connects our Internet gateway, server and desktop PCs. After several false starts (see below), we found a "good" outlet for the master adapter.
  • We used a "mobile" adapter connected to a portable PC which we carried around the house and plugged into each of the test outlets.
  • For a few tests, we also used a third adapter connected to one of our desktop PCs and plugged into its outlet.
  • For each test on the portable and desktop PCs, we used our HONEST test program to measure the transfer speed of a large file from our server to the test PC. Thus we were measuring the file "download" speed from our server. We ran HONEST at least twice at each test outlet.

Finding a "good" outlet for the master adapter

Our first effort was to find a “good” outlet for the master adapter. The first outlet we chose (location 3 on the diagram) was on the circuit strip where we plug in a computer monitor, many of our networking devices (Ethernet switches, Internet gateway) and other electronic devices (telephone chargers, portable battery charger, copying machine).

We ran a series of tests and found that we could not use HomePlug to communicate from many of the test outlets. After calling several vendors to understand why the performance was so bad, it became clear that our obvious choice for the master adapter was in fact a badly "impaired" outlet - the equipment plugged into the outlet was generating noise that got in the way of proper communications. (It's not the electrical load that counts but the kind of powerline noise generated by the devices; we were told that little "brick" transformers are among the worst offenders -- and we have three of them plugged into the circuit strip.)

We then looked around to find a better choice for the master. Fortunately, our house has Category 5 cabling to most rooms, so we were able to find another outlet that was fairly close to the electrical panel, and had a Cat 5 jack nearby (location 1 on the diagram). We connected the master adapter to that outlet and used our Cat 5 cabling to connect back to the main Ethernet switch in another room. (By the time we were finished with our tests, we found that another outlet in the same room as the Ethernet switch would probably have worked just as well - it was on a different circuit which appeared to be unimpaired.)

Two-node Networks

We tested eight different two-node networks with HomePlug Ethernet bridge and USB adapters. Our test sequence was as follows:

TestMaster AdapterRemote Adapter
1Asoka PlugLink Ethernet Bridge PL9610-ETHAsoka PlugLink Ethernet Bridge PL9610-ETH
2Phonex QX-201 NeverWire 14 Ethernet BridgePhonex QX-201 NeverWire 14 Ethernet Bridge
3Asoka PlugLink Ethernet Bridge PL9610-ETHST&T iPower Point EtherFast 10/100 Bridge M51
4Asoka PlugLink Ethernet Bridge PL9610-ETHAsoka PlugLink USB Adapter PL9710-USB
5ST&T iPower Point EtherFast 10/100 Bridge M51ST&T iPower Point PLC USB Adaptor U21
6ST&T iPower Point EtherFast 10/100 Bridge M51ST&T iPower Point PLC USB Adaptor U22
7ST&T iPower Point EtherFast 10/100 Bridge M53ST&T iPower Point EtherFast 10/100 Bridge M51
8ST&T iPower Point EtherFast 10/100 Bridge M53ST&T iPower Point PLC USB Adaptor U23

Results of these tests are given on the HomePlug Results page.

Mixed-Vendor Networks

We also attempted to create two- and three-node networks mixing different company's adapters. We had no problem mixing the Asoka and ST&T products; they appeared to interoperable without problems.

But mixing the Phonex units with the others was difficult, because they use different methods to set "advanced security encryption":

  • When we set up the Phonex units, we followed the procedure in the "Quickstart Guide" to set "advanced" security - the user-friendly procedure involves pushing the SECURE button on one Phonex unit for five seconds and pushing the SECURE on the other Phonex unit for one or two seconds.
  • The Asoka and ST&T adapters use an "Encryption Management Utility" on the installation CD to set advanced security after the adapter is connected to a PC. We chose not to run this utility but rather to leave the adapters set with the default security code.

Since the Phonex master adapter was set with "advanced" security, we wondered how to set the Asoka or ST&T units to the same encryption code. After a lot of discussion with the vendors, and quite a while spent downloading and trying without success to install a program that "might" work, we concluded that this was essentially not possible for a consumer.

So we set the Phonex master back to the "default" encryption code (by pushing the SECURE button for 15 seconds!) and it worked fine with the other units. We ran tests at a few outlets to confirm that performance was similar to what we'd already seen, and chose not to run full test sequences. Since all the units are based on the same Intellon chip, we believe that a full test would give results similar to our prior tests.

Phonex has done a great job of creating a more "user-friendly" system than the others we tested (for example, their Quick Start guide says: "Instant Install: Plug Neverwire 14 units into outlets near the computers ... that you are going to network together. Plug Ethernet cables into Neverwire 14 and computers... . Your Neverwire 14 units are now networked together!"). Unfortunately, their approach seem to have compromised interoperability to make it easy for users to install the system. The challenge for HomePlug is how to deliver both ease of install and interoperability.

Mixed-Vendor Three-node Networks

We tried to set up three-node networks mixing Ethernet bridges from different vendors, and started getting inconsistent results; we saw some long delays (half a minute or more) between transmissions and several timeouts.

From discussions with the vendors, we learned that the first generation of HomePlug Ethernet bridges has a problem when a network has more than two Ethernet Bridge adapters. Intellon told us that this is a known problem in their first-generation "A1" chip used in all the products we tested first.

Adapter manufacturers have taken different approaches to work around this problem: Phonex addresses it within the adapters, while Asoka and ST&T do so in the software installed on networked PCs. These approaches are not compatible with each other and do not appear to interoperate: we were unable to get any three-adapter network to work properly. Networks built with a single vendor's adapters and installed following the vendor's instructions might not manifest the problems we saw with multi-vendor networks.

Intellon told us that this problem has been resolved in the next generation of chips, which should appear in products later this year.


Next: Homeplug Test Results