The two families utilize different types of transistors in their
construction,
Bipolar Junction transistors for TTL and Field Effect
Transistors are used for the CMOS. There are various types of both
transistors, but one common way to categorize them is wether they are N
or P devices, NPN and PNP for bipolar and NMOS and PMOS for the CMOS,
where the N and P refers to the element used to dope the silicon. The
MOS stands for Metalic Oxide Semiconductor, which is a very thin high
valued insulator that is used to create the the very low input current
of CMOS devices. The C in CMOS stands for complementary, which means
that both N and P transistors have been fabricated together to
implement the logic. This allows the rail to rail 0 - 5 volt output
swings achievable with CMOS.
A bipolar transistor can be considered as a current driven switch where
the FET can be viewed as a voltage controlled switch. This factor
comes into play in that it is responsible for the different HIGH and
LOW logic voltages and influences how and when TTL and CMOS devices can
be connected together.
While Bill's post indicated a number of advantages of CMOS, mostly with
regards to digital logic, when it comes to analog performce, such as an
amplifier or ADC circuit, bipolar (TTL) has some advantages. Most
notibly, bipolar based amplifiers and transistors can often times
achieve a higher gain and operate at a much lower noise level.