Hi,
I have no particular knowledge here, but would be interested to hear opinions.
The poor quality of in-house wiring is often blamed for low xDSL sync speeds. Consumers are advised by providers to install ADSL filters or "splitters" on every piece of voiceband equipment, and to site the xDSL modem as close as possible to the NTE5 demarcation point.
But maybe that tells only part of the story. There could be other things that the customer can do to improve his sync speed, such as changing the overall topology of his in-house wiring.
The paper below is from Westell. [1] It was published in 2003 when consumer ADSL was in relative infancy. Today, the same technologies are still in use, albeit in newer forms, and the problems with those technologies persist, too.
The Westell paper addresses the effect of bridge taps in consumer wiring. That's what specifically interests here.
The bridge tap, also known as a "branch" or a "lateral" is a well documented problem for telcos. It is where there is a non-terminated cable pair that is 'tapped' onto a subscriber line. The tap introduces impedance mismatches at the splice point of the tap, and at the end point of the tap. These mismatches cause signal reflections. These reflections can impair the performance of the subscriber line. For a line used to provide an xDSL service, the bridge tap is likely to cause notches in the spectral response. These notches reduce the number of 'bins' available in the tone map. This will inevitably affect the overall line sync speed.
Manufacturers of test equipment to detect bridge taps provide some useful background to the problem. [2] [3] [4]
However, bridge taps are not just a problem for telcos, they can also exist in home wiring. In particular, the tap is known to cause reflections of the strong upstream signal from the CPE xDSL modem in the home. These signal reflections from the tap are picked up as noise by the sensitive receiver in the modem. This impairs the receiver's ability to recover the much weaker downstream signals from the DSLAM.
It is claimed that the topology of the consumer wiring can directly contribute to the problem of bridge taps.
Intuitively, we assume that it is always better to site the xDSL modem as close as possible to the NTE5, to limit the effect of induction noise and insertion loss in the consumer wiring. A longer cable run, we are told, will result in greater signal attenuation, additional noise and a lower data throughput. Yet there may be other factors to consider.
The topology of the in-house wiring may also have a bearing. If the consumer wiring is in a star configuration, where the modem is attached to one of the star 'points', then the other points to the star could have a bridge tap effect on the xDSL signal. Each point of the star introduces its own reflection-loss.
It is said that installing an ADSL splitter at the end of each of those star points should dampen those reflections. Others however state that the splitter is designed only to perform a low-pass filtering. Any damping effect it has on signal reflections is entirely perfunctory. Those reflected signals will still be present to some degree in the home wiring.
Westell's paper is really a sales pitch for the company's 'new' power spectral density shaping algorithm. The algorithm, Westell says, can mitigate the worst effects of a bridge tap. As such, the claims in the paper come with the usual caveat for any commercial vested interest. In particular, the figures showing attenuation (aka insertion loss) measurements due to bridge taps look distinctly exaggerated.
Westell's measurements vary quite significantly from those published by researchers with a purely academic interest. In his book,
Home Networking Basis, Walter Chen publishes his own graphs of measured insertion losses from artificially introduced bridge taps of some 15000 ft. [5] Chen points out that taps of that length would never be found in consumer wiring. Chen measures the reflection-losses at some 2dB-3dB per tap.
Whatever the scale of the bridge tap effect, the topology of consumer wiring evidently can introduce a bridge tap into the home network.
Further, it's not just the star wiring formation that can pose a problem with bridge taps. The daisy-chain topology is implicated too. In this wiring layout, the cable from the NTE5 is run in a single line, and where each phone socket is picked up along the cable run.
We might assume in the daisy-chain formation that the best place to site the xDSL modem is at the very beginning of the cable run, right next to the NTE5. Yet it is possible that the cable running on from the modem to the last socket in the run will itself serve as a bridge tap. And that could introduce the same problems of signal reflection. Ironically, to overcome the bridge tap problem in a daisychain topology, it might actually be better to site the modem at the far end of the cable run!
cheers, a
[1]
http://www.at2.com/downloads/documents/westell/westell_2200white.pdf[2]
http://documents.exfo.com/appnotes/anote233-ang.pdf[3]
http://www.syrus.ru/files/pdf/ssmtt/APP-xDSL-4B.pdf[4]
http://www.jdsu.com/ProductLiterature/sctpsbridgedtap_an_tfs_tm_ae.pdf[5]
http://flylib.com/books/en/2.92.1.16/1/