The Speech Network

The speech network - also known as the "hybrid" or the "two
wire/four wire network" - takes the incoming signal and feeds it
to the earpiece and takes the microphone output and feeds it down
the line. The standard network used all over the world is an LC
device with a carbon microphone; some newer phones use discrete
transistors or ICs.
4
One of the advantages of an LC network is that it has no
semiconductors, is not voltage sensitive, and will work
continuously as the voltage across the line is reduced. Many
transistorized phones stop working as the voltage approaches 3 to
4 Volts.
When a telephone is taken off the hook, the line voltage
drops from 48 Volts to between 9 and 3 Volts, depending on the
length of the loop. If another telephone in parallel is taken
off the hook, the current consumption of the line will remain the
same and the voltage across the terminals of both telephones will
drop. Bell Telephone specifications state that three telephones
should work in parallel on a 20 mA loop; transistorized phones
tend not to pass this test, although some manufacturers use ICs
that will pass. Although some European telephone companies claim
that phones working in parallel is "technically impossible," and
discourage attempts to make them work that way, some of their
telephones will work in parallel.
While low levels of audio may be difficult to hear, overly
loud audio can be painful. Consequently, a well designed
telephone will automatically adjust its transmit and receive
levels to allow for the attenuation - or lack of it - caused by
the length of the loop. This adjustment is called "loop
compensation." In the United States, telephone manufacturers
achieve this compensation with silicon carbide varistors that
consume any excess current from a short loop (see fig. 2).
Although some telephones using ICs have built-in loop
compensation, many do not; the latter have been designed to
provide adequate volume on the average loop, which means that
they provide low volume on long loops, and are too loud on short
loops. Various countries have different specifications for
transmit and receive levels; some European countries require a
higher transmit level than is standard in the United States so a
domestically-manufactured telephone may suffer from low transmit
level if used on European lines without modification.
Because a telephone is a duplex device, both transmitting
and receiving on the same pair of wires, the speech network must
ensure that not too much of the caller's voice is fed back into
his or her receiver. This function, called "sidetone," is
achieved by phasing the signal so that some cancellation occurs
in the speech network before the signal is fed to the receiver.
Callers faced with no sidetone at all will consider the phone
"dead." Too little sidetone will convince callers that they're
not being heard and cause them to shout, "I can hear you. Can
you hear ME?" Too much sidetone causes callers to lower their
voices and not be heard well at the other end of the line.
A telephone on a short loop with no loop compensation will
appear to have too much sidetone, and callers will lower their
voices. In this case, the percentage of sidetone is the same,
but as the overall level is higher the sidetone level will also
be higher.