Testing the POTS portion of ADSL.
At present, less than 10% of online households or businesses in the U.S. have high-speed DSL or cable Internet access. Based on current trends, however, the number of DSLlines installed worldwide should easily reach 100 million by 2005.
With the advent of ADSL, ordinary telephone lines are transmitting and receiving voice and full-duplex data at data rates up to 8 MHz. To accomplish this, a technique called discreet multitone (DMT), a frequency division multiplexing technique, is used in which the bandwidth of the telephone line (1.1 MHz) is split into 256 channels, each with a bandwidth of about 4 KHz. The first channel is assigned to carry POTS, and this channel must be carefully tested to ensure that the high-speed digital data on the other channels is not causing problems in the voice channel.
TESTING ANALOG DATA TRANSMISSION LINES
Using a voice frequency (VF) dedicated communication line for the transmission of data can result in signal distortion and degradation due to several causes. Among these are power loss, noise, poor frequency response and interference due to transient phenomenon. Measurement techniques and test specifications for dedicated VF lines have been published by AT&T for U.S. lines. Persons testing these lines should be aware of the test specifications and know how to measure the various forms of signal distortion. All examples here focus on testing AT&T lines.
(Note: Measurement techniques and test specifications for dedicated VF lines have been published by the CCITT/ITU for international lines. The differences in testing international lines according to those specifications appear at the end of this article.)
Power loss (or line loss) is one of the most common measurements made on VF lines. Loss is measured with a dBm meter and is referenced to the measured attenuation of a 1004 Hz test tone transmitted at a level of O dBm. At the time of installation of the VF transmission line, the installer adjusts the active line components to yield 16 dBm +/- 1 dBm attenuation for the O dBm test tone. According to AT&T, long-term readings for this loss measurement should be 16 dBm +/- 4 dBm. This last figure must be met during routine line-loss test measurements.
Figure 1A shows two arrangements for measuring line loss using an analog test set. Note that in 1 B, the attenuation measured is that of the complete round trip circuit.
[Figure 1 ILLUSTRATION OMITTED]
The gain/slope frequency response test is another common VF line measurement standardized by AT&T. For this test, first measure the attenuation at three specific frequencies: 404 Hz, 1004 Hz and 2804 Hz. Next, calculate the difference in attenuation between the lowest reading, i.e., the lowest attenuation and the other two readings. These two values give a good indication of the frequency response of the transmission line.
For example, consider a VF transmission line where the three values for loss are given in Table 1.
Frequency (Hz) Loss (dBm) 404 26 100 414 280 428
These readings are plotted on the frequency response curve for an unloaded VF line in Figure 2.
[Figure 2 ILLUSTRATION OMITTED]
In Table 1, note that the loss reading is 14 dBm at 1004 Hz. This reading becomes the reference level. Thus, the first slope is found by taking the difference between this value and 26 dBm, the loss reading at 404 Hz. This gives a slope of 26 - 14 = 12 dBm.
The second slope is found by taking the difference between the lowest reading of 14 dBm and the reading of 28 dBm at 2804Hz. Taking the difference between these two values yields a slope of 28 - 14 = 14 dBm.
These two slopes are plotted on the frequency response curve shown in Figure 2. Note how they approximate the actual curve. The gain/slope test is automated on certain analog test sets.
Taking the gain/slope measurements a step further, a much more accurate frequency response of the line can be obtained by measuring the line's attenuation at numerous points. Typically, readings are taken in 100 Hz steps in the range from 200 Hz to 3500 Hz. From these attenuation readings, an accurate graph of attenuation vs. frequency for the line under test can be obtained. Atypical curve for unloaded VF lines is shown in Figure 2. In general, the flatter the curve between 300 Hz and 3200 Hz, the better the voice channel and its use for transmitting data.
Noise is another common cause of signal degradation. In general, noise is defined as any unwanted disturbance generated within the electronic communications system, a phenomenon that is usually analyzed mathematically using statistical methods. Simple background noise in communications systems having no active elements, such as amplifiers or voice compandors, can be measured easily by using an analog test to monitor the line.
For accurate readings, the line must be properly terminated at both ends in order to eliminate signal reflections. At the test site, the line may be terminated by either a communications device or by the test set itself. When terminated by a communications device, the test set normally monitors the line. The Rx impedance of the test set should be set to a high impedance "bridge" mode so as not to impose any additional loading on the line. Noise measurements are in dBrn units (where O dBrn = -90 dBm) and represents the background noise level of the line.
When the communications system contains active elements, such as amplifiers and voice compandors, background noise measurements require the injection of a holding or test tone at the transmitter end of the test facility. This tone, a 1004 Hz sine wave at a level of O dBm, forces the compandors and amplifiers to operate as if normal data were being transmitted.
Figure 3 shows a test setup to measure the noise on a line. The test set at the transmitter end sends a 1004 Hz tone to the test set at the other end. The test tone activates all of the compandors and amplifiers on the line, simulating actual operating conditions. The test set at the remote end filters out the power of the 1004 Hz tone with a 1004 Hz notch filter in the receiver, so that only the noise is measured. The reading is taken in units of dBrn and indicates the true background noise level. Noise measured in this way is referred to as "notched noise." If desired, C-message weighting may also be applied to the received signal by switching in the C-message filter in series with the notch filter. When a C-message and notch filters are both used, the measurement is referred to as a "C-notched noise" measurement.
[Figure 3 ILLUSTRATION OMITTED]
In the past, C-message filtering was primarily used to test VF lines that were used only for voice communications. Today, it is also used to test VF lines that are used for data transmission.
SIGNAL-TO-NOISE (S/N RATIO)
In addition to background noise, the signal-to-noise (S/N) ratio is another standard indication of transmission line quality. In systems where the signal power is a fairly constant, known value, the S/N ratio is more meaningful than a simple background noise figure, because it indicates noise power with respect to signal power. For instance, a line may have a high noise level. If the signal level is much higher than the noise level, then the line will have a high signal-to-noise ratio and audible voice or error-free data communications, or both.
To measure the S/N ratio, first measure the level of the received 1004 Hz tone-plus-noise. Measuring the received signal without a notch filter does this. Next, switch in the notch filter to remove the test tone power, and measure the noise level. The difference between these two measurements is the S/N ratio. In general, an S/N ratio of 24 dBm or greater is acceptable.
The following shows why the above procedure provides an S/N ratio. The received signal contains both test tone signal (S) plus noise (N) and can be represented by the term (S+N). If we divide by the noise we get
Simplifying to separate terms yields
(S+N)/N = S/N + N/N = S/N + 1
Now, if the signal level is much greater than the noise level, then S/N will be much larger than one, so that we can say that S/N + 1 is approximately equal to S/ N. Therefore
(S+N)/N [nearly equal to] S/N
When measurements are given in dBm, division is accomplished simply by subtraction. Thus, subtracting the dBm reading for N from the dBm reading for S+N gives the signal-to-noise ratio, S/N (in dBm).
TESTING INTERNATIONAL LINES
When testing international circuits, the test specifications set forth by the CCITT/ ITU must be adhered to. The differences between these specifications and the AT&T specifications are as follows:
* A test tone at 820 Hz is used instead of 1004 Hz.
* A notch filter centered at 825 Hz is used instead of 1020 Hz.
* A four-tone gain/slope test is used instead of a three-tone gain/slope test. The four tones used are 300 Hz, 820 Hz, 2000 Hz and 3000 Hz.
* A psophometric filter is used instead of a C-message filter.
The shape of the psophometric and C-message filters are almost identical. Both have minimum attenuation at approximately 1004 Hz.
D'Antonio is president at International Data Science, Inc. (IDC), Warwick, RI.
Circle 278 for more information from IDS
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Technology Information|
|Date:||Sep 1, 2000|
|Previous Article:||Take your business mobile.|
|Next Article:||Latest testing and diagnostics products.|