73K222BL
V.22, V.21, Bell 212A, Bell 103
Single-Chip Modem with Integrated Hybrid
23
To avoid problems, power supply and ground traces
should be routed separately to the analog and digital
functions on the board, and digital signals should not
be routed near low level or high impedance analog
traces. The analog and digital grounds should only
connect at one point near the K-Series device
ground pin to avoid ground loops. The K-Series
modem ICs should have both high frequency and
low frequency bypassing as close to the package as
possible.
USING THE SERIAL MODE ON THE
73K2XXAL AND 73K222BL
A sensitivity to specific patterns being written to the
control registers in the 73K212/221/222AL and
73K222BL modem data pumps has been seen on
some parts when used in the serial control interface
mode. An alternating pattern followed by its
complement can cause the registers to not have the
intended data correctly written to the registers.
Specifically, if an alternating ..1010.. pattern is
followed by its compliment, ..0101.., the register may
instead be programmed with a ..0001.. pattern. After
analysis, it has been found that any normal
programming sequence should not include these
steps with one exception, and that is in DTMF
dialing. Since any random DTMF sequence could be
dialed, there is the potential for these patterns to
appear. For example, if a DTMF digit “5” , 0101 bin
is followed by a DTMF digit “0” , 1010 bin, some
parts will instead transmit a DTMF digit “8”, 1000 bin,
in its place. The solution to this problem is to always
clear the DTMF bits, D3-D0, between dialed digits.
This will not add additional time to dialing since there
is ample time between digits when the DTMF bits
can be cleared. Previously during the DTMF off time
the next digit would be loaded into the TONE
register. It is now recommended to first clear bits D3-
D0, then the next digit to be dialed is loaded into the
DTMF bits.
As mentioned earlier, under normal circumstances
these patterns would not be programmed for other
registers. If for some reason other registers are
programmed in such a way that an alternating
pattern is followed by its compliment, those bits
should be cleared before the complimentary pattern
is sent.
This method has been tested over the entire voltage
and temperature operating ranges. It has been
found to be a reliable procedure to ensure the
correct patterns are always programmed.
MODEM PERFORMANCE
CHARACTERISTICS
The curves presented here define modem IC
performance under a variety of line conditions while
inducing disturbances that are typical of those
encountered during data transmission on public
service telephone lines. Test data was taken using
an AEA Electronics’ “Autotest I” modem test set and
line simulator, operating under computer control. All
tests were run full-duplex, using a Concord Data
Systems 224 as the reference modem. A 511
pseudo-random-bit pattern was used for each data
point. Noise was C-message weighted and all
signal-to-noise (S/N) ratios reflect total power
measurements
similar
measurement specification. The individual tests are
defined as follows.
to
the
CCITT
V.56
BER VS. S/N
This test measures the ability of the modem to
operate over noisy lines with a minimum of data-
transfer errors. Since some noise is generated in the
best of dial-up lines, the modem must operate with
the lowest S/N ratio possible. Better modem
performance is indicated by test curves that are
closest to the BER axis. A narrow spread between
curves representing the four line parameters
indicates minimal variation in performance while
operating over a range of operating conditions.
Typically, a DPSK modem will exhibit better BER
performance test curves receiving in the low band
than in the high band.
BER VS. RECEIVE LEVEL
This test measures the dynamic range of the
modem. Because signal levels vary widely over dial-
up lines, the widest possible dynamic range is
desirable. The minimum Bell specification calls for
36 dB of dynamic range. S/N ratios are held
constant at the indicated values while the receive
level is lowered from a very high to very low signal
levels. The width of the “bowl” of these curves, taken
at the BER point, is the measure of dynamic range.