In a recent conversation with my friend and colleague, Ed Rousselot from Megger Corporation, he stated that he agreed with me on the process of provisioning and maintaining the copper infrastructure for bandwidth, but he felt that a significant amount of information could be gleaned about the health of the physical layer from DSL testing.
Here is what Ed wrote to me:
In the early 1970s I was trying, with little success, to get analog subscriber carrier working in Arkansas, and an old wire chief gave me some advice that is as true now as it was then. He slowly shook his head and said, “You can’t make chicken soup out of chicken poop”. And, sadly, that particular copper plant was akin to chicken poop.
If, however, the analog subscriber carrier was chicken soup from a can with only 5 noodles and 3 pieces of chicken, then VDSL is homemade soup that is chock full of chicken and wholesome veggies and noodles. While these technological wonders, from analog subscriber carrier to VDSL, work over twisted-pair copper, it has to be good twisted-pair copper.
And, in some ways, our plant is not even in as good shape now as it was in the 1970s. Don, as you point out, changes such as ready-access plant and cut-to-clear troubleshooting have allowed our plant to deteriorate. Luckily we have the testing technology to get our plant up to “chicken soup” standards.
There are the “standard” tests that look at the copper, also called “Layer One” or the “Physical Layer” to tell us if our pair is good. These test voltage, resistance, leakage, and longitudinal balance/stress. The tests also include the presence of load coils, impulse noise, and length (open meter). If we fail any of these, there are diagnostic tests that lead us to where and why the first test failed. These are resistive fault locator (RFL), wide and narrow band spectrum analyzers, and the Time Domain Reflectometer (TDR).
Modern DSL test sets use the same chipsets as do the manufacturers of DSL equipment. This gives the test sets the capability to look at the DSL signal, at Layer Two (Sync) and Layer Three (Surf). We do not have to worry about what these are called, or about the differences between layers and the different methods of achieving the task of a particular layer. What we need to know is that each layer communicates with the layers directly above and below it.
The chipset keeps records of its performance. We can use these records as another window into the condition of the copper pair (Layer One). Of course the DSL has to be working at least to some degree in order for us to be able to use these windows, so they certainly do not eliminate the need for the copper tests listed above. In fact, they point only to which of the copper tests we should use to verify the cause of the suspected problem.
Two of the Layer Two attributes of the DSL signal that the chipsets measure that are important to us are bits-per-tone and signal-to-noise ratio.
Without going too deeply into it, the VDSL signal is broken up into sub-channels or tones. Each of these tones can carry a maximum of 15 bits. If all is well, the tones will carry the same number of bits, and the maximum data throughput is achieved. If all is not well, some of the tones may carry no bits while others do carry 15 bits. At what frequencies this happens can help us determine the cause. In other cases, the number of bits carried by all or most of the tones is reduced, which we can also analyze.
Bits-per-tone is displayed in a bins graph which has the number of bits in each tone on the vertical (y) axis, and the frequencies, the tones, increasing along the horizontal (x) axis.
Dips in bits-per-tone indicate interference. Checking the frequency of the interference often makes it possible to identify its cause or source.
- Very low or non-existent bits-per-tone in the high frequency band are indicators of a long or too-long loop.
- If there is a broad, major dip in bits-per-tone, the most likely troubles are bridged taps, wet sections, or other changes in impedance.
- If they are narrow, the degraded throughput is most likely due to a steady-frequency noise influence.
- If bits-per-tone numbers are low across the entire bandwidth, the cause is likely DC faults such as shorts or grounds.
Signal-to-Noise Ratio (SNR)
This is simply the ratio of the power of the desired signal being received to that of an undesired signal (noise). The goal is to have high signal power and low noise power; a high signal to noise ratio. Signal-to-noise ratio is displayed in a graph with the ratio in dB on the vertical (y) axis and frequency increasing along the horizontal (x) axis. Oftentimes the SNR and bits-per-tone are displayed on the same graph with the frequency being common.
If data throughput is reduced, is the cause low signal power or high noise power or, perhaps, both? The answer can help us find the source of the reduced throughput. A look at the signal-to-noise graph shows this.
If the signal-to-noise ratio is falling as the frequency increases, the circuit is likely too long and is beyond the reach of the DSL service. When a circuit is beyond the reach you get 2 things: an upset customer, and an upset field technician who can’t fix it.
Troubleshooting already-installed VDSL will almost always require looking at the copper. But as long as some Telcos have spent the money on test sets that allow their technicians to look at Layers One and Two, these are some simple indicators that can be used to help determine what the problem with the copper might be.
I said that the chipsets in the VDSL equipment keep records. It likely won’t be too long before the equipment itself will have the interface necessary to access these records, making the ability to look at Layers One and Two, which is currently in some tests sets, redundant and an unneeded expense.
Thanks Ed, this information is invaluable to the field technician. An empty bin or tone on the bits per tone graph indicates that there is an interferer or disturber at that frequency. The spectrum analyzer will indicate the interfering frequency and its amplitude.
Interferers are external frequencies such as AM, short wave, and ham radio frequencies. They can be reduced with proper bonding and grounding. Disturbers such as T1 and HDSL must be moved to another sub-unit in the cable. The field technician cannot fix that — it falls on engineering.
I hope this column is helpful, and if not, please don’t be shy about telling me why, and/or what topics you would like me to cover in the coming months. You can reach me at firstname.lastname@example.org or 831.818.3930.
Ed Rousselot started his career in the telephone industry at Southwestern Bell in 1971. He worked for various telephone companies until 1984, when he took a position with a communications test equipment manufacturer. He has been in the industry since then serving in both technical sales and support positions. For the past several years, Rousselot has served Megger as a Senior Applications Engineer. For more information, visit http://megger.com.