Note: Descriptions are shown in the official language in which they were submitted.
~L23~3~3~
SUBSCRIBER LINE INTE~ACE
FIELD OF THE INVENTION
The present invention relates to telephone
line circuits and more particularly to a subscriber
line interface circuit for supplying the full range
of line supervision functions and regulated loop
current while at the same time reducing power dyes-
potion in the subscriber line interface circuit.
BACKGROUND OF THE INVENTION
Subscriber stations are normally connected to
the central office or telephone exchange by means of two
metallic conductors to form a subscriber's loop circuit.
Such loops often vary greatly in length and other kirk-
teristics resulting in different loop impedances as seen
from the central office. As a result, difficult problems
related to the signal and battery feed are experienced.
Battery feeds historically have used split
winding, highly balanced transformers and large keeps-
ions to feed do current to the subscriber's loop and
also to provide an arc. coupling on the wire. These
passive devices are bulky and heavy but generally provide
good performance. By incorporating various arrangements
of relays and some method of loop sensing, other special-
iced services are achieved: Hotel/motel, ground start,
and AN. More recently, many modifications in the form
of active device enhancement to transformers have been
tried so as to reduce their bulk and weight. These have
included floating feeds, negative inductance simulation,
do blocked inductors, etc. These methods generally
have preserved good performance but only partially
offset the size, weight, and cost factors of transformers
while incorporating, as a tradeoff, circuit complexity.
The most recent efforts, however, have been aimed at the
use of an all solid state interface circuit, commonly
Jo
- - \
1 3
called a SLICK The inherent assumption is of realizing
a custom integrated circuit so as to achieve the ultimate
goal of minimum size and weight. These factors are
extremely important since line circuits consume the
bulk of the space and weight in a moderate size telephone
exchange. The all solid state realizations of these
circuits have, however, been fraught with two general
problems:
(1) Difficult in meeting some transmission
specifications that the older passive device
approaches could meet, and
I Limited features, once beyond a simple
standard line interface the circuits would
need additional relays (a cost, size penalty)
or not be usable at all. Some of the specific
problems, one or more of which tend to appear
in recent SLIT circuits are:
(a) power dissipation at short loops is
excessive, leading to thermal problems
for the ICY
(b) excess COO. current drain beyond
that required for adequate transmission
performance and therefore excess engery
cost,
(c) longitudinal balance is deficient,
(d) limited features, such as lack of
(i) analog loop back for testing
(ii) ground start line
(iii) AN lines
(iv) loop current reversal for
hotel/motel.
SUMMARY OF THE INVENTION
This invention directly addresses each of the
deficiencies cited above. This can best be understood
1 3 8
by reference to the figures. Two Its are shown, split
by voltage requirements. The circuitry is shown this
way solely upon the assumption that final cost may be
minimized by separating high voltage (relatively larger
area of silicon) devices from low voltage (relatively
small) devices. To minimize excess power on short loops,
a pulse width modulator (Pal) of the "chopper" type is
employed to lower the effective battery voltage supplied
to the TIP, RING line driving amplifiers. This voltage
is made a function of the do loop current control
circuit so that just enough overhead voltage to keep
the amplifiers active is supplied above and beyond that
required to provide desired loop current. This minimizes
ohmic losses. The chopper itself, operating in a Class D
switching mode, operates in a near loss less manner. To
minimize excess COO. battery drain at short loops, a do
loop control circuit is employed in a feedback path,
whereby loop voltage is sensed, low pass filtered, analog
processed by a predetermined linear function, and the
output used to control the TIP, RING driving (transcon-
Dakotans) amplifiers. By proper choice of the feedback
function, a desirable loop current versus loop resistance
is obtained: a current between 55-65 ma at 100 ohms
loop which decreases smoothly to a minimum of 20 ma at
2000 ohm loop. A secondary output of this circuit controls
pulse width of the PAM described above.
It can be shown that for the type of TIP, RING
drive amplifiers chosen, longitudinal balance becomes
essentially a match of certain pairs of resistors, on a
ratio basis, which together with the operational amplifier
form the functional block which is a transconductance
amplifier. This is essentially a technology issue to
obtain the required ratio matching (0.1% or better), and
thin film resistors are envisioned. Longitudinal balance
~1238~;~
in the present of 60 ho common mode current, when those
current magnitudes exceed the steady-state do loop
current component, can only be maintained if the output
TIP, RING amplifiers can both source or sink current.
This implies a complementary output stage and further
imposes a restriction of Class A stage biasing so as
to minimize crossover distortion when current direction
reverses. Balanced source or sinking to the common mode
current only is achieved by a common mode amplifier.
Adequate amplifier gain-bandwidth must be available to
achieve desired balance at the upper frequency of interest
(about 3khz).
Synthesis of a complex ARC. input impedance is
achieved by feedback, wherein loop voltage is sensed,
hi-pass filtered with a time constant equal to 900 ohms
and 2.16 us, and used to control loop current.
Good OW return loss (match to desired 2 wire
input impedance) requires attention to the gain-bandwidth
of the intrinsic operational amplifiers and the choice
of resistor values which together with the Ox Amp form the
TIP, RING line driving transconductance amplifiers. This
is a circuit analysis/design tradeoff problem. Hybrid
balance (OW return loss) requires first good 2 wire input
impedance and then a careful accounting in the balance
networks for all poles zeros formed by amplifiers and
external components.
This invention, by design, incorporates detectors
for Ring Trip, AN, and ground start line conditions and
sufficient control features to support ground start, loop
current reversal, and line cutoff. For cutoff, both TIP
RING amplifiers are put in a high impedance state. This
eliminates the need for an external cutoff relay. (This
cutoff feature is needed prior to the new exchange initial
cutover at an existing older exchange site.) For ground
start, the TIP AMPLIFIER is put in a high impedance state.
I,
I I
- BRIEF DES~RIPTIQN OF THY DRAWING
Figure 1 shows the relative placement of Figures
2 and 3. Figures 2 and 3, when put along side each other,
show a detailed schematic of an embodiment of the subscriber
line interface circuit of the present invention. Figure 4
shows in greater detail the power supply circuit. Figure 5
is a chart showing the timing for the power circuit.
DESCRIPTION OF TIRE PREFERRED EMBODIMENT
The drawings, Figures 2 and 3, when placed along
side each other, as shown in Figure 1, show in schematic
form a subscriber line interface circuit with associated
circuitry for providing signal conversion between a two
wire bidirectional transmission line and a pair of unit
directional transmission paths.
The terminals for connection to the subscriber's
line are labeled To and Al. Direct current is supplied to
the line for operation of the subscriber's station from a
pair of line driver amplifiers labeled AT and AR. Long-
tudinal balance of the line is maintained by inserting 2
pairs of matched resistors R5, R6 and Roll, R13 into each
of the lines.
Longitudinal balance in the present of 60 cycle
power line interference currents which exceed the steady
state direct current loop component is maintained by having
the drive amplifiers capable of sourcing or sinking current.
To minimize crossover distortion in such an arrangement, the
amplifiers are each operated as Class ABE Resistors R6 and
Roll in cooperation with diodes Do, Do and Do and Do also
provide lightning protection by shunting excess current to
ground.
The line drive amplifiers are operated as current
drivers with a shunt voltage feedback to avoid possible
instability when connected to highly capacitive lines. The
two amplifiers AT and AR are symmetrical. The configuration
of these amplifiers is arranged to have a differential input
and a bidirectional output. In this manner current can be
either soured or sunk, depending upon the differences of
the inputs. Additionally, a common mode amplifier AC is
used to drive the inverting inputs of these amplifiers AT
and AR via resistors R4 and R17, respectively, as shown.
The output of amplifier AC is a function of the bias voltage
and the common mode voltage which is one-half of the sum of
the voltages at the To and Al leads taken from the junction
of resistors R7 and R8. The output of amplifier AC is con-
netted to the inverting inputs of amplifiers AT and AR in the configuration shown, and for a zero difference between
the positive inputs and the negative inputs of transconduct-
ante amplifiers AT and AR the voltage output will be one-
half of the "effective" battery voltage. Therefore, the
loop resistance load which "floats" between the To and Rlterminals will in the quiescent state (no loop current),
have voltages at the To and Al terminals such that the
"mid-point" voltage between resistors R7 and R8 will be
equal to one-half the effective source battery voltage.
The action of the common mode negative feedback AC is such
that common voltages on the To and Al terminals as induced
by a balance resistance source to ground will tend to be
offset in an equal manner by excursions at the transconduct-
ante amplifiers AT and AR outputs, thus providing a balanced longitudinal impedance to ground.
The common mode feedback only affects signals
common to the To and Al conductors and has no effect on
differential (voice) signals. A further advantage of the
overall configuration is that the common mode feedback is
taken after resistors R6 and Roll. Protection resistors,
R6 and Roll, limit lightning surges to a maximum current of
about AYE, where the current is shunted to ground or
battery by diodes Do J Do, Do or Do. This limits otherwise
3 I O
destructive potentials to a safe value. although the value
of resistors R6 and Roll in each leg To and Al is nominally
the same, over their life, the ratio matching could change
(due to lightning exposure) several percent and could never
be guaranteed to match closely. However, the feedback after
resistors R6 and Roll makes balance very insensitive to the
resistor R6 and Roll matching. The resistors R5 and R13
though are the most sensitive to affecting balance and
track closely.
To ensure greater reliability the circuit has
been designed for the placement of the components on two
circuit chips. The components that must interface with the
external line have a higher voltage rating and are located
on a first chip labeled Icily, while those components that
interface with the internal exchange circuitry are designed
with lower voltage rated components and are located on a
second chip labeled ISSUE. The difference in control levels
is corrected by providing a level shift circuit that trays-
fates the lower voltage inputs from ISSUE to the higher
voltages required by the components of Icily. The input
relative to ground potential is shifted to appear relative
to one-half of the effective battery voltage. This is done
to match the quiescent biasing of the amplifier AC.
A switching mode transistor, Al, is also present
on Icily because o-f voltage requirements. This transistor
with inductance Lo, capacitance Of, and catch diode Do
form a conventional "chopper" converter that converts the
exchange battery voltage to the output voltage US= (Vat),
where Don time duty cycle of Al. Additionally, control
inputs high impedance-HZT and power down-PDN are fed to both
output stage amplifiers AT and AR. The high impedance state
HUT allows implementation of the cutoff and ground start
functions. The power down feature is to conserve power.
The current IO, shown as a logic function of HUT and HER,
at gate Go, is switched in for ground start applications.
I I
To maintain control of the do loop current and
the proper arc. levels, the sideband differential signal
from amplifier PA is split into a "do" path and an "a"
path, by use of a low-pass filter consisting of R26 and C2
and hi-pass filters R28 and C3. A typical time constant for
the for low-pass filter is 10 my, which allows adequate
filtering for loop current control while still providing
fast enough response to detect dial pulsing (10 pus) trays-
itchiness. The high-pass filter serves two functions. One is
to simply couple the voice signals and block do including
do offset errors of the amplifiers) and the other is to
synthesize a complex input impedance at the TIP I RING, by
proper time constant choice. Input impedance synthesis
occurs because the voice signal is fed back thrum the summing
amplifier SUM to the line driving amplifiers AT and AR via
amplifier AD with overall negative feedback.
The do loop current control features two con-
trot characteristics designated "main" and "secondary",
plus input and output polarity swishing at the amplifiers.
The polarity switching amplifier provides either +1 or -1
voltage gain under logic control, depending on desired sub-
scriber loop current direction: normal or reverse. Since
the whole structure is in a feedback loop, from the line
conductors, amplifier DA, DC loop control, summing amp SUM,
and amplifier AD, simultaneously inverting signal polarities
to the do at the control block at input and output will
force a reversal of loop current direction. That is, toe
input signal sign to tip AT, ring AR amplifiers and the loop
voltage sign will both invert when loop current is reversed,
but signal sense is always positive thrum the control block.
The desired characteristic for loop current versus
loop resistance is a smoothly decreasing current from 55-65
ma at 100 ohms to 20 ma at 2000 ohms. The constraints on
curve shape are relatively loose, so as to still satisfy
~23~31~
transmission requirements, and a generally hyperbolic shape
is satisfactory. The importance of this is that the relation-
ship of loop current versus loop voltage is linear. This
form, being linear, is relatively easy to implement and is
the correct relationship to place in a feedback loop. Some
scaling is necessary because of loop gain factors. This
"main" characteristic controls loop current out to 2000 ohms.
The potential problem of amplifier clipping to
voice signals at long loops or low battery is possible.
Simply stated, there may not be enough battery voltage to
sustain desired loop current and keep the amplifiers in the
active region: the amplifier voltage "overhead". To pro-
vent this, a "secondary" characteristic is constantly computed
keeping the amplifier active region voltage constant as a
function of battery voltage.
The loop current is totally under control of
the do current control just described. The pulse width
modulator PAM is used to supply an effective battery
voltage, less than the real battery voltage, at short
loops to reduce power dissipation. The PUMA is slaved
to the voltage level as output from the loop control such
that the effective battery potential supplied to the amply-
liens AT and AR is equal to the loop voltage plus a constant.
The level of the constant voltage is chosen to be somewhat
larger than the amplifier overhead voltage. Thus only
necessary ohmic losses are dissipated as heat. This sire-
logy keeps maximum IT power under 1 watt as compared to the
normal 2 watts or greater otherwise.
The PAM also incorporates a novel "open loop"
battery voltage compensation as shown in Figure 4. A
voltage to current converter VCl generates a current
proportional to the exchange battery voltage. A second
voltage to current converter VC2 is slaved to the first
via resistor R30 to generate a second current. A current
I
mirror, CM, also generates a current, I, in an opposite
direction from that of the first current I. A logic
controlled switch, S, can interrupt the current I and
is controlled from a D flip-flop FF2, whose D input=GND
(logic 0). A zero is clocked to output Q at the check
pulse Folk positive edge. An asynchronous preset, PRY
sets Logic one at PRY set to zero, via the comparator;
CMPl, which occurs when the capacitor voltage Ye is less
or equal to zero. The overall action of these elements
is to synchronize the capacitor, C22, voltage, as shown, to
the clock Folk such that the capacitor starts a discharge
ramp with current I as Folk goes positive, and starts a
charge ramp with current I just after Ye goes to zero.
For any constant Vat and associated current I, and con-
slant clock frequency, vcl=vc2 and Ye goes to zero at exactly T0/2 (half-cycle). The effect of Vat is only
to change the slope of the capacitor voltage ramp. A
second comparator, CMP2, generates the duty cycle signal D
to transistor Al. The duty cycle is D=vth/vcpeak. The
voltage vth=(loop voltage +V2), scaled down.
Battery compensation is achieved as follows: For
capacitor C22, vcpeak=(K3)(Vbat)(T0)/(2C). Let I=(K3)(Vbat)
then vcpeak=(K3)(Vbat)(T0)/(2C). But, for the converter
output voltage=(inpu~ voltage)X(duty cycle). Thus Us=
(Vat), where D=Ton/T0. But Ton/T0=vth~vcpeak=D. Therefore,
Vs=(vth)(Vbat)(2C)/((K3)(Vbat)(T0))=(vth)(2C)/((K33)(T0)), and
is not dependent on Vat. The advantage of this "open loop"
Vat compensation over a closed loop feedback approach is
the fact that both a loop filter capacitor and associated
IT pin are eliminated. Without Vat compensation, voltage
V2 would have to be set substantially larger to cover Vat
variation and would result in needlessly higher IT power
dissipation. With this scheme, the PAM is swanked to an
external clock so that it can be phase locked to the 8khz
-10-
-
~38~3~
voice sampling frequency o-f a following CODE or equivalent
to eliminate inter modulation products.
Four different characteristics of the external
loop are sensed: (loop sense, AN, ground start, and ring
trip. Loop sensing involves detecting OFF hook vs. ON hook
states and dial pulsing. Since the loop is driven with
transconductance amplifiers, the voltage input to these
amplifiers is proportional to loop current. But loop current
is a predetermined function of loop resistance (the LOOP
CONTROL circuitry), so a simple threshold on the input
signal, taken from the DO loop control can be used. Thus
the threshold is related directly to loop resistance, and
set for about 2500 ohms equivalent. A small hysteresis is
added.
AN (Automatic Number Identification) in this
context means differentiating between two subscribers
sharing a two party line. One Jill have a standard tote-
phone set and the other a set where there is a do path
(less than 2600 ohms) to ground from the "mid-point".
This path will be inductive (so as not to load the voice
signal) and is provided either by ringer inductance,
specially arranged, or a separate inductor. Detection is
accomplished by sensing loop current imbalance resulting
from the path to ground when the AN marked subscriber
goes off hook. The imbalance current is directly related
to the difference of TIP, RING bias voltage TV, and the
common mode voltage, Yam'. The level shifted value of Yam',
that is Yam''. The difference voltage is compared to
threshold and a small hysteresis is added.
Ground start detection is employed for lines
toeing to a PAYBACKS using ground start signaling. In ground
start, the COO. side TIP feed is open circuit (hi-impedance),
and the COO. must sense external application of a resistive
ground (less than 400 ohms) at the RING lead on the opposite
end of loop. For the SLICK the task is to sense current
3 3
flow in the RING conductor alone, ignoring activity on the
TIP side. To accomplish this, amplifiers AT and AR are put
in HUT state while a constant current It, about ma, is
sunk to -50V at the junctions of R8 and R7 on the RING lead.
Then OR a SLIT is a function of RING current. To sense
RING voltage only, a scaled factor of loop voltage is sub-
treated from Yam and compared to a threshold value, plus
small hysteresis. This complete operation is performed by
a circuit using a differential comparator similar to that
for AN.
Ring trip detection is initiated when a signal
from the logic sets the ring trip latch which actuates the
ringing relay. The ring relay contacts connect ringing
generator and its ground to the loop via the R6 resistors.
The SLIT Icily is disconnected at contacts RCl and Rc2 to
prevent the higher ringing voltages from damaging Icily.
Detection of ringing current is achieved by summing the
signals vdiff and (Vrng~, as shown in the drawing which
then is proportional to ringing current. The ringing
current thrum the ringer has an arc. component only, since
the ringer is capacitively coupled to the line. when the
subscriber goes OFF hook, an additional do path thrum the
phone is established. Because the ringing generator is
biased to -50V, a do component will now appear in add-
lion to the arc. component. The overall function of the detector is to quickly sense less than 100 my) this new
condition and reset the ringing latch and thereby disco-
neat the ringing generator. That is, "ring trip" occurs.
The detector first sums vdiff and (Vrng) and
applies a gain of 10X to the resultant signal. This signal
is then limited to +0.75V swing, which creates a clipping
of the upper portion of the otherwise sinusoidal signal.
This signal is then low pass filtered to reduce the magnitude
of the arc. component, such that no detection occurs for ON
hook at comparator (Vth=lZ5mV). For OFF hook, the additional
3 3 O
do current raises the peak magnitude of the filtered signal
to trip the comparator and reset the latch. The filter kapok-
it or is biased to -100 my during ON hook so that the total
225 my excursion needed to reach Vth cannot occur for greater
than 5 my. This delay is needed to mask the operate time of
the ringing relay during which the "ring current" signal is
not valid. The circuit parameters allow trip times no greater
than 100 my.
In this hybrid, the voice signal present at the
receive port is transmitted to the subscriber line via the
TIP and RING current driver Ox Amps AT and AR. However a
portion of this signal is picked up by the differential
amplifier DUFF bridging the TIP and RING conductors and is
fed back toward the transmit port. The actual amount of
this signal that gets fed back depends Oil the impedance of
the subscriber line. In order to cancel these signals to
prevent echoes and system oscillations, a separate "leakage"
path from the receive port to *he transmit port is establish-
Ed This leak path shapes the frequency spectrum of the
received voice to be identical with that of the reflected
signal coming from the differential amplifier bridging the
TIP and RING conductors. These two signals are then summed
together 180 degrees out of phase and are used as the output
to the transmit port. Three different leak paths (Net,
NET, and NET) are presented that will cancel reflections
from three different classes of lines. The top two paths
match the SLICK in a four wire sense, to the R-C parallel
combinations assumed to represent loaded and non loaded
lines. The bottom path matches the standard termination
of 900 ohms and 2.16 us. Network selection is accomplished
by a logic controlled "three position switch". The "switch"
is implemented with solid state technology.
I
It will be obvious to those skilled in the art
that numerous modifications o-f the present invention can
be made without departing from the spirit of the invention
which shall be limited only by the claims appended hereto.
-14-
...
