Hi Frank,
You raise many good questions. In a limited amount of time let me try a few answers:
The AMI T1 signal has a spectrum peak at 772 kHz and therefore occupies considerably more bandwidth than you suggest, depending on how you define bandwidwidth.
The ADSL Forum was set up at least initially to be an ADSL booster, by people who had invested a lot of resources in the technology in hopes of finding a market for it. The asymmetry was required for the initial target application: video dialtone, which hasn't materialized.
Web browsing suggests an asymmetric need and has been used as a rationale for continuing to support a high degree of asymmetry. Fortunately the design of the signals on an ADSL system can be such that it can support symmetric transmission, as well, up to a point.
T1 signal performance in cables is controlled by wire gauge (attenuation) and near-end crosstalk (NEXT). Regenerator spacing is dictated by one or the other impairment, depending on the adjacency of transmit and receive signals in the cable. If the signals are segregated into non-adjacent binder groups, I believe attenuation controls spacing but there are application rules to to allow T1 signals to be put in adjacent and same binder groups with shorter regenerator spacing. The baseband T1 signal has energy in the voiceband. It was customary to segregate T1 signals in separate binder groups, I think, to reduce crosstalk into voice pairs but I'm not sure that was the primary reason.
ADSL signals can be designed to be frequency division duplex with non-overlapping spectrum for the two directions of transmission, thus eliminating self-NEXT as an impairment. Modulation techniques for ADSL favor passband approaches with minimal energy in the voice band.
At high enough frequencies far end crosstalk (FEXT)becomes a limiting impairment in twisted pair cables. VDSL with data rates as high as 60 Mb/s will suffer from self-FEXT.
Bellcore has done some excellent modelling and measurement work over the years on crosstalk, although the analysis has been based on limited data characterizing AT&T production cables of 30 years ago or more. You might be able to find someone there who is familiar with that work.
I don't think crosstalk has much to do with how tightly cables are "packed". Large and small cross-section cables are built up from the same kinds of binder groups. They just use greater or lesser numbers of them. I don't think that paper, pulp, or plastic insulation makes much difference in the balance of twisted pairs. They're all pretty good. Actually I should think that crosstalk would be a bigger issue in smaller cables, since there would be less flexibility in reassigning conflicting signals to pairs with lower coupling.
Single pair drop wire that I am familiar with is not twisted but, as it is single pair, there is no crosstalk issue. Multi-pair drop cables, called service wire, I think, use twisted pairs. Wiring inside houses has been largely unspecified since it was deregulated in 1976 by the FCC. I think that a lot of it has at least two pairs bundled together in a cable that used to be called station wire. I don't think the pairs are twisted, which could be a problem. Unfortunately in-place station wire is uncharacterized. Fortunately, the cost of augmenting it, if there is a problem, isn't too awful in most cases.
All of the issues you raise should be concerns for ADSL as they have been for T1 and HDSL before but they appear to be workable.
George Hawley
|
