FCC Part 68 Requirements for Connecting to the U.S. Phone Network

Many local telephone companies sprang up around the country after Mr. Bell's invention of the telephone in 1877. Gradually, AT&T purchased most of these to become the dominant carrier and equipment provider across the U.S. During this period AT&T developed uniform technical rules as well as a unique terminology ("Bellspeak") for certain aspects of networks. This terminology persists today in such terms as "metallic," "longitudinal" and "on-hook/off-hook."

The beginnings of a breakup in this effective monopoly began in 1956 with the Hushaphone and 1968 Carterphone cases that broke that monopoly, allowing other companies to interconnect with the telephone network. To protect its network, the telephone company supplied (and required others to use and pay for) an interface device known as the data-access arrangement, or DAA. Although the DAA performed much of the network-interface function and preserved the integrity of the network, it had its disadvantages. First, the user had to pay for the device, and second, the marketing of telephone equipment became awkward due to the necessity for the cooperation of the Bell System.

With the growth of data communications, the FCC saw a need for regulations to allow direct interconnection to the telephone network without the use of the DAA. After an involved rulemaking process, the Commission in 1976 promulgated the regulations now contained in Part 68 of Title 47 of the Code of Federal Regulations. Their purpose is "to provide for uniform standards for the protection of the telephone network from harms caused by connections of terminal equipment thereto." It is important to note that the purpose of the FCC rules is to protect the telephone network from harm, not the consumer: devices can be designed that will meet the requirements of Part 68 but still not work very well. Harmonization of FCC Part 68 rules with Canada's CS-03 took effect April 20, 1998. The details of grandfathering are still in question. The rules are also under review in light of new products and technologies. For example, 56K modems are restricted in speed by Part 68's signal power limits. This issue is under technical review.

Conceptually, the telephone system performs the simple task of providing a switched two-wire connection, with enough electronics to power a station set and amplify the audio signals for transmission. Although much has changed since the old days of manual and electromechanical switchboards, their overall function has remained constant, and, during the course of this evolution, compatibility has always been maintained with older station equipment. This historical legacy of compatibility explains the nature of some of today's interfacing specifications.

The standard two-wire telephone-set connection, referred to as the public switched telephone network (PSTN) or the plain old telephone service (POTS), is not the only type of service offered by the telephone companies. Four-wire services, utilizing separate pairs of wires for transmitting and receiving, afford improved fidelity; reverse battery services allow automated PBXs to act as localized central offices; and ground-start and "E and M" tie trunks offer more reliable methods of signaling than the conventional loop system. In addition to the "public" services, which enable the user to dial a variety of numbers, there are also a variety of "private" point-to-point services available in both two- and four-wire formats. Although all of the above services can be used by modems for audio-encoded data transmission, the data rates available are limited by the 4-kHz audio bandwidth provided. For higher-bandwidth service, digital services ranging in speed up to 1.544 Mb/sec are available, and are included in the FCC Part 68 attachment program.

While these services vary in form, most derive their basic operational characteristics from the plain old telephone network.

Each telephone customer or "subscriber" has a two-wire connection or "subscriber loop" to the local telephone office, known as the central office. Through a matrix-switching arrangement, each subscriber's telephone may be connected to any other telephone served by the central office. Communications between offices are performed through interoffice trunks. Channel banks are sometimes used to compress the voice or data signals from the subscriber loops on multiplexed interoffice trunks.

The subscriber loops are shown in more detail in Figure 1. The central office provides battery power, which is fed through a current-sensing relay coil. The dc resistance between the central office and the customer equipment can vary from 400 to 1740 ohms and is dependent on the length of the loop. Since the two-wire loop forms a transmission line, the ac impedance, controlled by the wire's spacing, is 600 ohms. Nominally, telephones match this 600-ohm impedance over the voice-frequency range and have a dc resistance of less than 200 ohms when connected to the line ("off-hook"). Power for the customer equipment can be drawn from the central-office battery, which will vary the voltage from 42.5 to 56.5 V. Loop current is generally in the 20 to 80 mA range.

Placing a Call

When the telephone receiver is on-hook, the telephone draws no current from the line. Lifting of the receiver causes a connection to be made, allowing current to flow from the central office. The central office detects this off-hook state and sends a dial tone, which in its most common form consists of a combination of a 350-Hz and a 440-Hz tone. When the user dials, the telephone set interrupts the loop current, causing a current sensor (relay) at the central office to change state. Alternating make-and-break contacts signal the central office that digits have been dialed, and depending on the number, the central office connects the subscriber loop either to an interoffice trunk or to another subscriber. If the loop sensor at the central office detects that the party being called has the phone off-hook, the calling party is sent back a busy tone (480 Hz plus a 620 Hz tone, at –24 dBm). If the party being called has the phone on-hook, he or she is sent a ringing signal that makes the phone ring (typically 20 Hz at 100 volts, superimposed over the central office's battery voltage). If the party being called lifts the receiver off-hook, the central office detects the current being drawn by the second subscriber's loop and stops, or "trips," the ringing signal. Direct connection is then established between the two subscribers.

Modern technology has displaced the relays and pulse dialers used in older telephone systems. Dual-tone multifrequency (DTMF) telephone-address signaling has largely replaced the old rotary-pulse dialers, and electronic switching systems (ESSs) have superseded the mechanical relay and crossbar switching. Nonetheless, as far as the user is concerned, the functioning and electrical parameters of the system are little changed from the days when it was built of relays and coils.

Based on these parameters, the FCC has specified a model for simulating the central office and the subscriber line. This simple model simulates both the dc and the audio-frequency characteristics of the telephone network.

The United States benefits from telecom regulations that are minimal by intent. Part 68 covers only those attributes of terminal equipment which protect the network equipment, signals or personnel from impairment or injury. Hearing-aid compatibility is an equal-access provision added by the Congress. Issues of equipment quality or functionality are left to the marketplace.

This is not the case in other jurisdictions around the world, where detailed requirements for interoperability are imposed upon the telecom equipment manufacturer. As a result, the United States has among the simplest and easiest telecom rules anywhere. There are only a few basic tests to be performed for the sake of Part 68 registration. These tests may vary in implementation a bit, however, depending upon the nature of the terminal equipment. An excellent resource for understanding the reasons behind the tests, and how to perform the tests, is found in the TIA document Part 68 Rationale and Guidelines. The document reference TSB 31B matches the Part 68 rules effective April 20, 1998. That document serves as the framework for what follows. It describes the basic tests as:

Part 68 specifies two tests to check for exposure of the network to hazardous voltages. The first is a conventional leakage test, in which a 50- or 60-Hz voltage is applied between the telephone connections (known as tip and ring) and both of the following:

Part 68 specifies signal-level maximums for three discrete reasons. First, some telephone lines cover such great distances that amplification is needed. Since amplifiers have limited dynamic range, the amount of signal placed on the line in the voice band (200–3995 Hz) must be limited. Second, in communicating from one central office to another, the telephone network will use frequency- and time-multiplexing schemes, and it is important that signals outside the voice band be limited (up to 6 MHz) since out-of-band signals may interact with the multiplexing to create spurious in-band signals via frequency aliasing. Finally, crosstalk between lines must be controlled.

Different limits apply to voice and data signals than to signals intended for dialing. The latter, known as network-control signals, may be in the form either of pulse dialing, which is not covered by FCC regulations, or of dual-tone multifrequency (DTMF) dialing.

The setup shown in Figure 2 can be used for measuring voice, data, or DTMF transmissions. The equipment under test is connected to a loop simulator that mimics its connection to the telephone network. Since different limits apply in different frequency ranges, a frequency-selectable bandpass filter is used.

Figure 2. Setup for measuring differential (metallic) signals placed on the network.

In the voice-frequency range, the RMS reading, integrated over the entire band, should be below –9 dBm (or 9 dB below 1 mW) into 600 ohms, averaged over a 3-second period. Moving to the 4 to 12 kHz range, voice or data signals should not be stronger than specified limits averaged over a 100-ms period. Above 12 kHz, signal levels are examined in every 8-kHz segment from 12 kHz to 6 MHz. In each of these 8-kHz segments, the signal should be below the limits indicated in the table, using the specified termination.

It is permissible to take measurements over a bandwidth wider than 8 kHz; while the 8-kHz range is the minimum required by the rules, a wider bandpass will speed testing.

Since 8 kHz is used for sampling by digitizing equipment, the band from 3995 to 4005 Hz is sensitive. For most applications, signals in this narrow frequency band must be less than –27 dBm. For DTMF signaling, the permissible level from 100 Hz to 4 kHz is less than 0 dBm averaged over a 3-second period. It is measured with the same test setup used for in-band voice or data transmission. No limits are specified for DTMF signals above 4 kHz.

In addition to prescribing limits for signal levels between tip and ring, known as metallic or differential voltages, Part 68 specifies limits between tip/ring and earth ground, known as longitudinal or common-mode voltages.

In the case of in-band signals (defined for longitudinal voltages as 100 Hz to 4 kHz), a high-pass filter is inserted in series with the test network with frequency characteristics as shown. The averaging period for these measurements is 100 milliseconds.

Devices that send voice signals or music over the network are required to be tested while driven with a 1 kHz, –13 dBm signal.

Anything connected to the phone lines that results in one side of the line's having a different impedance with respect to earth ground than the other will tend to cause differential voltages present in the environment to be converted to common-mode noise. This signal, which is usually in the form of 60-cycle noise, will show up as hum.

When unbalanced lines transmit signals coupled in multiconductor cables, they produce cross talk. In order to prevent this, the rules require any equipment attached to the telephone line to meet some minimum balance criteria.

For balance testing, an oscillator with a 600-ohm output impedance drives the primary of a transformer, which produces a differential voltage on the tip and ring lines. Capacitors are adjusted so that the voltage is nulled when a 600-ohm resistor is placed across tip and ring.

When the equipment under test is placed across the tip and ring lines, some imbalance will occur. This imbalance will cause a voltage to be built up across the 600-ohm resistor (for analog equipment), and this voltage can then be compared with VM. The ratio yields the transverse balance reading, which should exceed 60 dB below 1 kHz and 40 dB between 1 and 4 kHz. Similar balance values apply for digital equipment up to 1.544 Mb/sec.

Dc on-hook resistance is required to be greater than 5 megohms for an applied voltage of up to 100 V and greater than 30 kilohms for an applied voltage of up to 200 V (see Figure 3). The telephone companies use an automated insulation test system as part of their maintenance routine. If the dc on-hook resistance of an attached piece of equipment is too low, the line is interpreted as being damaged, and repair procedures are initiated. The dc resistance requirement prevents false service alarms.

Figure 3. Setup for on-hook dc impedance test. Caution: High voltage requires care.

Figure 4. Setup and limits for the ac on-hook impedance test. Dividing the applied voltage by the ac current yields the impedance. In a second test, a 100-kilohm resistor is connected to tip or ring; the voltage observed on the oscilloscope should fall by more than a factor of 2.

Since the central office also has to make the phone ring, the rules specify a minimum ac on-hook impedance for the ringer as well. During this test (Figure 4), a ringing signal is applied to the equipment under test.

Under the 1998 rules either of only two ringer frequencies A (20, 30 Hz) and B (15–68 Hz) are used. A minimum impedance must be met, depending on the ringer and frequency of 1400 ohms at 20 Hz and 1000 ohms at 30 Hz for ringer A, or 1600 ohms at 15–68 Hz for ringer B.

As part of the on-hook impedance limitations, the FCC specifies two additional tests. First, during the ringing test, dc current is also measured and must not exceed 3 mA. Second, the Commission requires a minimum "longitudinal" impedance of 100 kilohms from each tip and ring to ground. A convenient way of measuring this is shown in Figure 5. A 100-kilohm resistor from either tip or ring to earth ground is attached while the ringing test is performed; if the voltages measured from tip and ring to earth fall by more than a factor of 2, then the longitudinal impedance of the equipment under test is greater than the specified 100-kilohm limit.

Figure 5. Separating transmit from receive: the transmitted signal is fed simultaneously to the interface through a resistor and to a second inverting amplifier, causing cancellation.

These on-hook impedance tests are designed to protect the network from excessive current flow while on-hook. However, if several devices are connected to the same phone line, excessive current may still be drawn. To prevent this, a system known as ringer equivalence number (REN) is used. As a standard, the amount of current drawn by an old telephone (known as the 500 set) is assigned a REN of 1. A device that draws twice as much current has a ringer equivalence of 2, and so on. The ringer equivalences of parallel devices add, with the maximum ringer equivalence (i.e., the sum of all attached devices) for any installation being 5.

The REN is followed by a letter corresponding to the type of ring signal applied during testing. For example, if the highest REN calculated was .8, and the device was tested as a B-type ringer, the REN would be 0.8B.

Although Part 68 specifications are adequate to protect the network, they do not fully define all characteristics needed for optimal device design. For example, device dc off-hook resistance, which is not specified by Part 68, should be approximately 200 ohms, while off-hook ac impedance should match the network's impedance of 600 ohms. Additional information on standard practice may be found in TIA and ANSI standards and in Bellcore publications.

Telephone companies are understandably anxious that customer premise equipment should not upset their ability to bill for service. To this end, four rules have been included in Part 68 under the heading "Billing Protection."

The device is first subjected to drop tests. Unpackaged drop tests on equipment that is not usually carried by the customer are performed from heights of 6 inches on each resting face and 3 inches on all other faces and corners; equipment that is normally carried is dropped from a height of 30 inches in a random fashion, except for equipment used at head height, which is subject to 18 random drops from a height of 5 feet. All of these drop tests employ a concrete surface covered with 1/8 inch of asphalt tile.

If the equipment survives these tests, it is subjected to two voltage surges that simulate lightning and other transients occurring on power or telephone lines. There are two types of surge, A and B, to be applied. In between them, the terminal equipment must be checked for compliance and on-hook/off-hook functionality of the interface.The new Type B surge is a 9 x 720 ms pulse, 1000 V metallic and 1500 V longitudinal. One pulse of each polarity is applied. Then the Type A surge is applied (this is the pulse specified in the 1997 rules). Two 800-V surges are applied between tip and ring, with a 10-µs maximum front time and a 560-µs minimum decay time (to half crest), one of each polarity. Fifteen-hundred-volt longitudinal surges of each polarity are required between the telephone-line connections and earth ground, on each of the telephone lines individually and with both lines tied together. The maximum front and minimum fall times here are 10 and 160 µs, respectively. A last surge test is applied between the phase and the neutral wires of the ac line if the device uses ac power. Three surges of each polarity (six in all) are applied at a level of 2500 V, each having a 2-µs maximum front time and a 10-µs minimum decay time.

After environmental simulation, all of the tests described above (leakage, hazardous voltage, signal power, balance, on-hook impedance, and billing protection) are performed again. Any variations between readings must be explained in the registration application. Although devices may be damaged, they must still meet the "non-harm" requirement: for example, it is acceptable for a piece of equipment to fail after the Type A surge in the permanently on-hook state, but a unit that is permanently off-hook after this surge may not be registered.

Data rates from 2.4 Kb/sec ("subrate") to 1.544 Mb/sec (T-1 and ISDN) are regulated under Part 68 by applying tests very similar to voice band terminal equipment: signal power, transverse balance and environmental requirements apply. In addition, there are tests specific to data devices: pulse frequency, waveform templates, and signaling duration. Within the network, these digital signals are sometimes decoded and then re-digitized; therefore, the "encoded analog content" must be analyzed for the same parameters as for an analog phone: network signal levels (comparable to DTMF), non-network signals, and on-hook signal power.

From its beginnings, the telephone system has always been a two-wire full-duplex system, meaning that a single pair of wires carries signals in both directions. When data are transmitted, collisions will result that a computer cannot resolve—hence, the need for circuits known as hybrids (Figure 5), which cause cancellation of the transmitted signal and leave the telephone device free to receive signals coming down the telephone line. In Figure 5, the transmitting signal is fed to a follower, which drives resistor RH and the telephone-interface transformer. Since the telephone line is nominally 600 ohms, the transmitted signal is divided by 2 through this process. The signal VA is therefore the voltage received down the telephone line (VR) plus one half the transmitted voltage (VT). This combined signal is fed to a noninverting gain-of-2 amplifier, which simultaneously subtracts VT, causing cancellation of the transmit signal. Note, however, that the ability of the circuit to separate transmit from receive depends on the impedance of the telephone line. If it is exactly 600 ohms, signal rejection may be as high as 40 dB (factor of 100), though in most applications, loop rejection greater than 20 dB cannot be expected.

Up to this point, we have focused on a two-wire connection to the public switched telephone network. However, there are many other kinds of connections provided by local operating companies and regulated by the FCC.

The conventional two-wire circuit is one type of public switched line. Such lines allow calls to any other public telephone subscriber. There are also private lines, which usually implement a dedicated connection.

Public switched lines come in several basic forms. Loop-start lines are the kind we have explored: to start the process, the user draws loop current from the central office. There is also the more robust ground-start system, in which the subscriber initiates a call via an earth-grounding ring and is alerted to an incoming call either by the sight of earth ground in tip or by a conventional ring signal. Often used in PBXs, such systems prevent collisions between incoming and outgoing calls by speeding the recognition of alerting (ringing) signals. There are also direct inward dial (DID) lines, which reverse the roles of the customer and the central office. Here the central office does the dialing, and the customer's PBX receives the number of the person that the central office wants to ring. The PBX then connects the central office line to that individual's phone. This bypasses the switchboard attendant through the use of "direct dial" numbers.

Private lines also come in other flavors. In tie trunks, used to tie two PBXs together, the reliability of dialing and ringing is improved through the use of a separate "E and M" pair dedicated to alerting (ringing) and signaling.

"Standard" telephone lines are available in private-line form as well. Such a line is known as a ringdown tie trunk. If only a short connection is needed, one possible choice is a metallic channel—a pair of wires strung by the telephone company from point to point. Alternatively, a PBX can connect to a station in a remote location by using an off-premise station (OPS) line.

Finally, users of local-area networks may want to consider the telco's own offering, the local-area data channel (LADC).

To protect these various telephone facilities, Part 68 requires further tests in addition to those specified for two-wire loop-start lines. Such treatments, though similar to those discussed above, are beyond the scope of this article.

Notable changes that affect manufacturers include:

Although Part 68 appears to be a daunting document, it covers for the most part only harms to the network. Also, Part 68 contains the regulatory requirements for all terminal equipment: analog, digital, private-line, PBX, voice, and nonvoice. The telecom rules for many other countries are much more comprehensive—and, in many cases, are divided into separate standards for each type of terminal device. This makes them less accessible as well as more difficult to meet. The relative simplicity of the U.S. telecom connection requirements no doubt contributes to a vibrant domestic marketplace for telecom products and technologies.