DSL's dirty little secret–revisited

Sun, 08/31/2003 - 8:00pm
Jeffrey Krauss, President of Telecommunications and Technology Policy
About five years ago, I wrote a column about the possibility that digital emissions from digital subscriber lines (DSL) could cause interference to AM radio reception. But maybe I got it backwards, because lately I've been reading about interference into DSL, caused by AM radio, ham radio transmitters, light dimmer switches, street lights and other devices. Five years ago, asynchronous DSL was having trouble penetrating the market because there were two different incompatible technologies: Carrierless Amplitude Phase modulation (CAP) and Discrete Multi Tone modulation (DMT). Since that time, the ITU has done a lot of work on standardization, and in 1999, this resulted in two standardized versions of DMT. One, known as G.DMT or G.992.1, supports data rates of 6 Mbps downstream and 640 kbps upstream. The other, known as G.Lite or G.992.2, supports 1.5 Mbps downstream and 512 kbps upstream. Now, I didn't get it totally wrong in that 1998 column. A review of the ITU-T Study Group 15 document files shows that there was indeed concern about egress interference from ADSL lines. While unshielded twisted pair wiring that is properly twisted is unlikely to emit serious levels of RF, the problem is that wiring installed and owned by the customer might not be properly twisted. A good explanation of those egress issues can be found at Both G.Lite and G.DMT transmit subchannels that are 4.3125 kHz wide. G.Lite uses 25 upstream subchannels, occupying the band from 26 to 138 kHz, and 122 downstream subchannels, occupying the band from 138 kHz up to about 665 kHz. G.DMT uses the same upstream channels as G.Lite, and 249 downstream subchannels, from 138 kHz up to about 1.1 MHz. Thus, G.Lite pretty much avoids the 535 to 1700 kHz AM broadcast band, but G.DMT runs smack into it. One study finds that RF interference from AM broadcast stations can reduce the ADSL data carrying capacity as much as 79 percent on longer phone loops, or can even kill the connection entirely. (For details, see Tracking down the interference problem can be a nightmare. The Internet newsgroup comp.dcom.xdsl contains some examples. One poor fellow had his DSL connection die every few minutes, but only late at night and early in the morning. He finally tracked it down to two problems–one in the streetlight across the street, and one in a neighbor's dimmer switch. Another fellow is sure his problem is caused by high power transmissions from a nearby ham (amateur) radio operator, but there are no ham bands below 1800 kHz (160 meter band). On the other hand, on a ham message board, there is a report by a ham transmitting with 100 watts on 80 meters (3500-3750 kHz) that he is causing interference not only into his own DSL modem, but into the next door neighbor's as well. So evidently, a strong RF signal can overload DSL, even if it is not on the same frequency. Here is a link to the American Radio Relay League's suggestions for dealing with ham interference to telephone lines: Other things being equal, co-frequency transmissions are more likely to cause interference than transmissions separated by a few megahertz. A review of the FCC's Table of Frequency Allocations shows that there are aeronautical and maritime radiobeacons and other radionavigation transmitters in the bands used by ADSL. So anyone near an airport or Coast Guard station might have a problem trying to use DSL. Another big ADSL interference problem seems to be crosstalk within the telephone plant itself. Hundreds of wire pairs are packed into the cables that are strung from one pole to the next. Because of their physical proximity, signals from one pair of wires can leak into the next pair of wires. The FCC investigated this issue a few years ago, but decided it was a telephone industry problem that did not need government regulation. In 2002, G.992.1 and G.992.2 evolved into G.992.3 and G.992.4, known as ADSL2. These achieve data rates up to about 8 Mbps. Early in 2003, the ITU-T adopted a standard for the newest member of the ADSL family, G.992.5, known as ADSL2plus. ADSL2plus, with a data rate up to 12 Mbps or double the capacity of G.992.1, occupies a spectrum band that is also doubled, up to 2.2 MHz. While this increases the number of interfering signals, the new standard is supposed to include signal processing schemes to deal with the interference environment. And most recently, there was an announcement from Alcatel, Broadcom and nine other vendors, supporting a version of DMT for very high bit rate DSL (VDSL). It is claimed that this will carry video as well as data, at rates up to 70 Mbps downstream. VDSL uses much higher frequencies to achieve these data rates. And it is subject to correspondingly higher numbers of potential interference sources. Back in 1998, I wrote that DSL was facing a number of challenges, including both lack of standards and interference issues. Interference issues still exist, and the lack of standards has evolved into another common telecom problem: too many standards. Have a comment? Contact Jeff via e-mail at:

Share This Story

You may login with either your assigned username or your e-mail address.
The password field is case sensitive.