Digital Signal Processing Reference
In-Depth Information
Ireland with Newfoundland and could not be financially viable unless it operated at
speeds similar to dry land telegraph cables. Unfortunately, signal integrity problems
caused the Queen's 99-word inaugural message to take nearly 17 hours for trans-
mission instead of several minutes as had been expected [2].
The roots of this nineteenth-century disaster are both managerial and technical. The
underlying management fault was that the investors and managers did not possess a
rudimentary understanding of telegraphy. Because of this they hired a chief electri-
cian lacking a thorough background in electrical engineering or telegraphy theory.
This decision was at the root of the cable's technical flaws. Without this knowledge
the chief electrician was unable to understand the latest technical developments and
instead relied on his intuition to design the cable and instrument set. These intuitive
designs were appealing to management but inappropriate for the task.
Some outside experts (most notably Professor William Thompson, later to be
Lord Kelvin) had questioned the cable's electrical design and the electrical equip-
ment to be used with it. With his theoretical background, Thompson knew that
the electrical characteristics of a long submarine cable were significantly different
from the open air cable that had earlier been used in proof-of-concept experiments,
and he correctly believed that the submarine cable would distort high-speed Morse
code pulses. He preferred another cable design and suggested transmitting and re-
ceiving equipment with very different electrical characteristics than the equipment
designed by the chief electrician. Because of management's weak technical back-
ground, they were unable to ask the kinds of probing, focused questions needed to
resolve the technical debate between Thompson and the chief electrician.
After several false starts, the intuitively designed but electrically flawed cable
was finally laid, and direct current from a battery in Ireland was detected by instru-
ments in Newfoundland. This low-frequency test seemed to vindicate the design,
but it was soon discovered that high-speed pulses could not be properly received. In
fact, signaling was only possible at speeds that were far too slow to be profitable.
Management was unable to properly judge the merits of the various approach-
es to debug suggested by staff. Instead of using the sensitive transmitting and re-
ceiving instruments purposed by Thompson (which we now know would have
significantly improved the signaling speed), they authorized the chief electrician
to increase the voltage of the transmitted pulses. The misguided idea was that the
higher voltage would “push” the pulses down the cable with greater force, allow-
ing signaling to occur at high speeds. In fact, this did not improve the signaling rate
because the fundamental problem was pulse distortion caused by the cable's large
RC delay, not signal strength.
The large RC delay was the critical difference between the dry-land cables
used during the proof-of-concept experiments and the submarine cable. The dry-air
cables used air as insulation, but the submarine cable used a waterproof insulation
that greatly increased the capacitance. We will return to this effect in Section 1.7.
The large RC delay so distorted the pulses that only the slowest of them could
be properly received. Modern SI engineers would say that intersymbol interference
was created by the way in which the cable attenuated high frequencies (the high-
speed Morse pulses) but allowed low frequencies (slow speed pulses, including
