U.S. patent application number 15/224663 was filed with the patent office on 2018-02-01 for multi-data rate, burst-mode transimpedance amplifier (tia) circuit.
The applicant listed for this patent is Avago Technologies General IP (Singapore) Pte. Ltd. Invention is credited to Georgios Asmanis, Faouzi Chaahoub, Kartikeya Gupta, Rahul Shringarpure.
Application Number | 20180034432 15/224663 |
Document ID | / |
Family ID | 61005513 |
Filed Date | 2018-02-01 |
United States Patent
Application |
20180034432 |
Kind Code |
A1 |
Shringarpure; Rahul ; et
al. |
February 1, 2018 |
MULTI-DATA RATE, BURST-MODE TRANSIMPEDANCE AMPLIFIER (TIA)
CIRCUIT
Abstract
A burst-mode TIA circuit for use in PON receivers is provided
that supports multiple data rates, has high receiver sensitivity,
wide dynamic range, and that performs burst-mode synchronization
very quickly. The multi-rate burst-mode TIA circuit has a
high-speed data path that has low input-referred noise. Based on
the chosen data rate at which the multi-rate burst-mode TIA circuit
will operate, the rate select switch selects an appropriate
feedback resistor of the resistive feedback network.
Inventors: |
Shringarpure; Rahul;
(Hayward, CA) ; Asmanis; Georgios; (Lake Forrest,
CA) ; Chaahoub; Faouzi; (San Jose, CA) ;
Gupta; Kartikeya; (San Jose, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Avago Technologies General IP (Singapore) Pte. Ltd |
Singapore |
|
SG |
|
|
Family ID: |
61005513 |
Appl. No.: |
15/224663 |
Filed: |
July 31, 2016 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H03F 2203/45288
20130101; H03F 2200/78 20130101; H04B 1/16 20130101; H03F 3/45278
20130101; H04B 10/616 20130101; H03G 3/3084 20130101; H03F
2203/45212 20130101 |
International
Class: |
H03G 3/30 20060101
H03G003/30; H04B 10/61 20060101 H04B010/61; H04B 1/16 20060101
H04B001/16; H03F 3/45 20060101 H03F003/45 |
Claims
1. A transimpedance amplifier (TIA) circuit for use in a receiver
used to receive asynchronous data sent in bursts over optical
networks, the TIA circuit being capable of operating at multiple
data rates, the TIA circuit receiving an TIA input signal at an
input terminal of the TIA circuit and outputting a TIA output
signal from an output terminal of the TIA circuit: a data rate
selection circuit that is configured to be switched between
multiple settings to select the data rate at which the TIA circuit
operates; a direct current (DC) offset correction loop configured
to apply a DC offset correction at the TIA input terminal, wherein
the DC offset correction loop comprises: a capacitor bank
comprising a plurality of capacitors and at least a first switch
that can be activated and deactivated; a direct current (DC)
injecting/bleeding circuit electrically coupled to the TIA input
terminal and to the capacitor bank and comprising at least a second
switch that can be activated and deactivated and at least a first
blood transistor; and a level detection circuit that selectively
activates and deactivates the first and second switches based on a
TIA output signal outputted from the output terminal to control
said at least a first bleed transistor to inject current into or
shunt current away from the input terminal.
2. (canceled)
3. The TIA circuit of claim 1, wherein the data rate selection
circuit includes a bank of switchable resistor.
4. The TIA circuit of claim 3, wherein the switchable feedback
resistor bank comprises: at least first and second resistors
connected in parallel; and at least a first transistor connected in
series in between the first and second resistors.
5. The TIA circuit of claim 4, wherein the bank of switchable
resistor comprises: at least a third resistor connected in parallel
with the first and second resistors; at least a second transistor
connected in series in between the second and third resistors; and
a third resistor connected in parallel with the first, second and
third resistors.
6. (canceled)
7. The TIA circuit of claim 1, wherein the DC offset correction
loop further comprises: a charge pump electrically coupled to the
capacitor bank, the DC injections/bleeding circuit and to the level
detection circuit, the charge pump charging and discharging the
capacitors of the capacitor bank.
8. The TIA circuit of claim 7, wherein the capacitor bank further
comprises at least a third switch can be activated and deactivated
by the level detection circuit.
9. The TIA circuit of claim 8, wherein the DC injecting/bleeding
circuit includes at least a fourth switch that can be activated and
deactivated by the level detection circuit, and wherein the second
and fourth switches are shunt transistors.
10. The TIA circuit of claim 9, wherein the charge pump is a
dual-polarity charge pump.
11. The TIA circuit of claim 1, further comprising: an automatic
gain control (AGC) loop configured to automatically control a gain
of the TIA circuit.
12. The TIA circuit of claim 11, wherein the AGC loop includes the
level detection circuit, the level detection circuit detecting a
peak-to-peak amplitude of the TIA output signal at the output
terminal of the TIA circuit and causing the gain of the TIA circuit
to be varied based on the peak-to-peak amplitude.
13. The TIA circuit of claim 12, wherein the level detection
circuit comprises: M comparators that are triggered when the TIA
output signal crosses M threshold levels, respectively, where M is
a positive integer that is greater than or equal to 2, the M
comparators outputting M signals representing M bits, respectively,
such that an M-bit current signal is output by the level detection
circuit; an M-bit current-mode digital-to-analog converter (DAC)
that converts the M bits into an analog current signal and outputs
the analog current signal; and a current-to-voltage converter that
converts the analog current signal output from the M-bit DAC into
an analog voltage signal and outputs the analog voltage signal, and
wherein the level detection circuit causes the gain of the TIA
circuit to be adjusted based on the analog voltage signal output
from the current-to-voltage converter.
14. The TIA circuit of claim 11, further comprising: a reset
circuit that resets the TIA circuit if a loss of signal (LOS)
indication is detected at the input of the TIA circuit.
15. The TIA circuit of claim 14, wherein the reset circuit
comprises: a zero-level detector that detects when the TIA input
signal received at the input terminal of the TIA circuit is at a
zero power level and outputs a pulse; and a counter that counts a
number of pulses output by the zero-level detector, wherein the
reset circuit outputs a reset signal that resets the TIA circuit
when the count exceeds a predetermined number.
16. The TIA circuit of claim 15, wherein the counter comprises a
free-running ring oscillator and an N-bit synchronous counter,
where N is a positive integer greater than or equal to 1.
17. The TIA circuit of claim 15, wherein the counter comprises an
N-bit ripple counter, where N is a positive integer greater than or
equal to 1.
18. A transimpedance amplifier (TIA) circuit for use in a receiver
used to receive asynchronous data sent in bursts over optical
networks, the TIA circuit having an input terminal that receives a
TIA input signal and an output terminal that outputs a TIA output
signal, the TIA circuit comprising: an automatic gain control (AGC)
loop configured to automatically control a gain of the TIA circuit,
wherein the AGC loop includes a level detection circuit having an
input that is electrically coupled to the output terminal of the
TIA circuit, the level detection circuit detecting a peak-to-peak
amplitude of the TIA output signal M comparators that are triggered
when an output signal of the TIA circuit crosses M threshold
levels, respectively, where M is a positive integer that is greater
than or equal to 2, the M comparators outputting M signals
representing M bits, respectively, such that an M-bit current
signal is output by the level detection circuit; an M-bit
current-mode digital-to-analog converter (DAC) that converts the M
bits into an analog current signal and outputs the analog current
signal; and a current-to-voltage converter that converts the analog
current signal outputted from the M-bit DAC into an analog voltage
signal and outputs the analog voltage signal, and wherein the level
detection circuit causes the gain of the TIA circuit to be adjusted
based on the analog voltage signal outputted from the
current-to-voltage converter.
19. The TIA circuit of claim 18, further comprising a data rate
selection circuit, wherein the data rate selection circuit includes
a switchable feedback resistor bank.
20. The TIA circuit of claim 19, further comprising: a DC offset
correction loop comprising: a capacitor bank comprising a plurality
of capacitors; and a charge pump electrically coupled to the
capacitor bank for charging and discharging capacitors of the
capacitor bank.
21. (canceled)
22. The TIA circuit of claim 18, further comprising: a reset
circuit comprising: a zero-level detector that detects when the TIA
input signal received at the input terminal of the TIA circuit is
at a zero power level and outputs a pulse; and a counter that
counts a number of pulses output by the zero-level detector,
wherein the reset circuit outputs a reset signal that resets the
TIA circuit when the count exceeds a predetermined number.
Description
TECHNICAL FIELD OF THE INVENTION
[0001] The invention relates to transimpedance amplifier (TIA)
circuits. More particularly, the invention relates to a multi-data
rate, burst-mode TIA circuit for fiber to the home (FTTH)
applications
BACKGROUND OF THE INVENTION
[0002] A typical optical receiver (Rx) includes at least one
photodiode that detects an optical signal and converts it into an
electrical current signal and at least one transimpedance amplifier
(TIA) that converts the electrical current signal into an
electrical voltage signal. The photodetector, which is typically a
P-intrinsic-N (PIN) photodiode, produces an electrical current
signal in response to light detected by the photodetector. The TIA
converts this electrical current signal into an output voltage
signal having some gain, commonly referred to as transimpedance
gain. The TIA circuit typically includes several control loops for
improving performance, such as a direct current (DC) offset
correction loop, an automatic gain control (AGC) loop, and a TIA
feedback impedance adjustment loop.
[0003] FTTH Passive Optical Networks (PON) using Packet-based
systems require burst-mode receivers that amplify input data
instantaneously for each asynchronous packet. Currently, burst-mode
PON receivers have the two standards: Institute of Electrical and
Electronics Engineers (IEEE) Ethernet Passive Optical Network (IEEE
EPON) and International Telecommunication Union (ITU-T)
Gigabit-capable Passive Optical Networks (GPON). The IEEE EPON
standard uses data rates of 1.25 Gigabits per second (Gb/s) and
10.05 Gb/s, whereas GPON uses 2.5 Gb/s and 9.95 Gb/s.
[0004] There are multiple challenges in designing burst-mode PON
Receivers. One of the challenges involves receiver (Rx)
Sensitivity. PON networks suffer from higher losses and hence
require higher optical power budgets. As a result, the Rx
sensitivity requirement is less than -28 decibel-milliwatts (dbm)
at a data rate of 1.25 Gbps. To achieve such low Rx sensitivity
values, the TIA needs to either have very low input-referred noise
or very high gain, both of which are difficult requirements to
meet.
[0005] Another challenge in designing burst-mode PON receivers is
that they need to have a wide dynamic range (typically greater than
around 22 db for 10 G EPON) due to the higher optical power budget
requirement. Hence, the TIA gain has to be decreased under overload
conditions when the input power is high and increased when the
input power is very low (at sensitivity).
[0006] Another challenge in designing burst-mode PON receivers is
enabling the receiver to perform burst-mode synchronization very
quickly. Burst-mode PON receivers are packet based, which means
they receive fixed-length asynchronous data bits. Each packet is
broken down into preamble bits and payload bits. Rx synchronization
must be performed during the preamble bit period, which is very
short, e.g., less than 100 nanoseconds (ns) in case of GPON. Hence,
a very fast synchronization scheme is needed that is capable of
settling the TIA output to the correct amplitude before the output
is sampled and retimed by the Clock and Data Recovery (CDR) circuit
that follows the TIA.
[0007] Yet another challenge in designing today's burst-mode PON
receivers is that they have to be capable of supporting multiple
data rate operations. For example, the EPON burst-mode receiver can
be required to provide an upstream burst data rate of 10.0 Gbps and
a downstream burst data rate of 1.25 Gbps.
[0008] Accordingly, a need exists for a burst-mode TIA circuit that
meets all of these design challenges.
BRIEF DESCRIPTION OF THE DRAWINGS
[0009] FIG. 1 illustrates a block diagram of a TIA circuit in
accordance with an illustrative embodiment.
[0010] FIG. 2 illustrates a block diagram of the level detection
circuit shown in FIG. 1 in accordance with an illustrative
embodiment.
[0011] FIG. 3 is a timing diagram demonstrating the manner in which
the TIA circuit shown in FIG. 1 operates when operating at a data
rate of 1.25 Gbps.
[0012] FIG. 4 illustrates a block diagram of the reset block shown
in FIG. 1 in accordance with an illustrative embodiment.
[0013] FIG. 5 is a timing diagram demonstrating the manner in which
the reset block shown in FIG. 4 operates.
DETAILED DESCRIPTION
[0014] In accordance with illustrative embodiments, a burst-mode
TIA circuit for use in PON receivers is provided that supports
multiple data rates, has high receiver sensitivity, wide dynamic
range, and that performs burst-mode synchronization very quickly. A
few illustrative embodiments of the burst-mode TIA circuit will now
be described with reference to the FIGS. 1-5, in which like
reference numerals represent like elements, features or
components.
[0015] In accordance with an illustrative, or exemplary,
embodiment, the multi-rate burst-mode TIA circuit has a high-speed
data path that is made up of only first and second gain stages,
which leads to the TIA circuit having very low input-referred
noise. The first gain stage is a single-ended resistive feedback
network that includes a rate select switch. Based on the chosen
data rate at which the multi-rate burst-mode TIA circuit will
operate, the rate select switch selects an appropriate feedback
resistor of the resistive feedback network. The second gain stage
is a single-to-differential fixed gain stage that converts the
single-ended output signal of the first gain stage into a
differential signal having an appropriate common-mode voltage for
providing a seamless interface with the output transmission
line.
[0016] The multi-rate burst-mode TIA circuit has two feedback
loops. The first feedback loop is an AGC loop that regulates the
first gain stage based on the amplitude level of the input signal
to the first gain stage. In accordance with the illustrative
embodiment, the AGC loop comprises a plurality of Schmitt
Trigger-type Hysteresis comparators that function as respective
level detectors for detecting respective different amplitude levels
of the input signal to the first gain stage. The appropriate level
detector is triggered when the amplitude level of the input signal
reaches a certain voltage level. The outputs of the level detectors
are provided to a current-mode digital-to-analog converter (DAC),
which sets a gate voltage of a metal oxide semiconductor field
effect transistor (MOSFET)-based variable feedback resistor in the
first gain stage, thereby regulating the transimpedance gain of the
first gain stage. The second feedback loop is a DC offset
correction loop. In accordance with an illustrative embodiment,
this loop comprises a current-boosting charge pump circuit that
either sources or sinks an appropriate current into an integrating
capacitor in the DC offset correcting feedback loop to correct the
DC bias of the input signal received at the input of the first gain
stage. The two feedback loops operate independently of one another
with the AGC loop having a shorter settling time than the settling
time of the DC-offset correction loop to allow very fast burst mode
synchronization to be achieved.
[0017] As used in the specification and appended claims, the terms
"a," "an," and "the" include both singular and plural referents,
unless the context clearly dictates otherwise. Thus, for example,
"a device" includes one device and plural devices.
[0018] It should be noted that when an element is referred to
herein as being "connected to" or "coupled to" or "electrically
coupled to" another element, it can be directly connected or
coupled, or intervening elements may be present.
[0019] FIG. 1 illustrates a block diagram of a multi-data rate
burst mode TIA circuit 100 in accordance with an illustrative
embodiment. The TIA circuit 100 includes a first gain stage 101
that is a single-ended variable gain stage and a second gain stage
102 that is a single-ended-to-differential fixed gain stage. The
first gain stage 101 includes first and second hetero-junction
bipolar transistors (HBTs) T.sub.1 and T.sub.2, respectively, a
resistive feedback network comprising a bank of switchable
resistors, R.sub.F1, R.sub.F2 and R.sub.F3 connected in parallel
with one another and with a MOSFET M.sub.AGC., a load resistor RL,
and a rate select circuit 103. The switchable resistors R.sub.F1,
R.sub.F2 and R.sub.F3 are connected in series with two other
MOSFETs M.sub.1 and M.sub.2 having gates that are connected to a
rate select switch 103. Because there are two PON standards (i.e.
EPON and GPON) that are currently used in FTTH PON networks, the
illustrative embodiment of the TIA circuit 100 is configured to
support multiple data rates, namely, 1.25 Gbps/10 Gbps for the EPON
standard and 2.5 Gbps/9.95 Gbps for the GPON standard. By using the
rate select switch 103 to activate or deactivate the MOSFETs
M.sub.1 and M.sub.2, the feedback resistance provided by the
parallel arrangement of resistors R.sub.F1, R.sub.F2 and R.sub.F3
is varied to the appropriate resistance for the corresponding
bandwidth. High data rates require a higher TIA bandwidth and hence
a lower feedback resistance is selected across the first gain stage
101 for higher data rates. Conversely, lower data rates require a
lower TIA bandwidth, and hence a higher feedback resistance is
selected for lower data rates. Moreover, at lower data rates the
data bursts have a lower input optical modulation amplitude (OMA),
and therefore the first gain stage 101 should have a very high
sensitivity. In those cases, the higher feedback resistance (i.e.,
higher transimpedance gain) helps to reduce input-referred noise at
the input of the first gain stage, thereby improving
sensitivity.
[0020] In accordance with the illustrative embodiment shown in FIG.
1, the first gain stage 101 has three feedback resistors, R.sub.F1,
R.sub.F2, R.sub.F3 and two MOSFET switches, M.sub.1 and M.sub.2.
However, the first gain stage 101 could have a different number of
resistors and switches. For example, increasing the number of
resistors and switches would allow the TIA circuit 101 to operate
at greater number of data rates. When the M.sub.1 and M.sub.2
switches are both OFF, the first gain stage 101 operates at the
1.25 Gbps (low) data rate with a gain of R.sub.F1. When the M.sub.1
switch is turned ON and the M.sub.2 switch is turned OFF, the first
gain stage 101 operates at the 2.5 Gbps (medium) data rate with a
gain of R.sub.F1.times.R.sub.F2/(R.sub.H+R.sub.F2). When the M1 and
M2 switches are both ON, the first gain stage 101 operates at the
10 Gbps (high) data rate with a gain of
R.sub.F1.times.R.sub.F2.times.R.sub.F3/(R.sub.F1.times.R.sub.F2+R.sub.F2.-
times.R.sub.F3+R.sub.F1.times.R.sub.F3). The second gain stage 102
is a single-ended-to-differential fixed gain differential pair of
transistors T.sub.3 and T.sub.4. The second gain stage 102 has a
small-signal gain of -gm.times.RL/2 as only one half of the
differential pair receives the input signal output from the first
gain stage 101, where RL is the load resistance of the second gain
stage 102. The other half of the differential pair is set to a
fixed reference voltage, V.sub.REF.
[0021] As indicated above, the multi-rate burst-mode TIA circuit
100 has a DC offset correction loop and an AGC loop. In accordance
with this embodiment, both of the loops are connected to a common
reset pin 119, which is used to restore the internal nodes of the
loops to their default state, as will be described below in more
detail. The AGC loop regulates the transimpedance gain of the first
gain stage 101 by increasing or decreasing the feedback resistance
provided by resistive feedback network of the first gain stage 101.
When the amplitude of the input signal applied to the base of the
HBT T.sub.1 is large, the AGC loop decreases the feedback
resistance and when the amplitude of the input signal applied to
the base of the HBT T.sub.1 is small, the AGC loop increases the
feedback resistance. The AGC loop comprises a level detection
circuit 112, an M-bit current-mode digital-to-analog converter
(DAC) 113, where M is a positive integer that is greater than or
equal to 2, a current-to-voltage converter 114, and the MOSFET
M.sub.AGC of the first gain stage 101.
[0022] The level detection circuit 112 detects the level of the
differential signal output from the second gain stage 102 and
outputs a corresponding M-bit value to the DAC 113, which converts
the M-bit value into an analog current signal. The analog current
signal is converted into an analog voltage signal, which is applied
to the gate of MOSFET M.sub.AGC. The MOSFET M.sub.AGC functions as
a variable resistor having a resistance that varies based on the
magnitude of the gate voltage. By varying this resistance, the gain
of the first gain stage 101 is varied.
[0023] In accordance with an illustrative embodiment, the level
detection circuit 112 comprises M alternating current (AC)-coupled
level detectors 112a and 112b. Instead of using a peak or envelope
detector for this purpose, as is common in AGC loops in typical TIA
circuits, and which have slower settling times, preferably M
Schmitt trigger hysteresis comparators are used for this purpose
due to their faster settling times. For illustrative purposes, it
is assumed that M=2. The appropriate level detector 112a and 112b
is triggered when the differential output signal is at a particular
voltage level. The outputs of the level detectors 112a and 112b are
converted via the DAC 113 and the current-to-voltage converter 114
into the gate voltage of MOSFET M.sub.AGC, thereby causing the
transimpedance gain of the first gain stage 101 to be varied
accordingly. Thus, the feedback resistance of the first gain stage
101, which has previously been fixed by the rate select circuit 103
for a given data rate, is regulated by the AGC loop by modifying
the gate voltage of MOSFET M.sub.AGC.
[0024] Comparators are typically capable of detecting high speed
signal amplitudes very fast, but they suffer from meta-stability.
Meta-stability is a problem that occurs in latching comparators
when the input is near the comparator decision point. The problem
occurs when the comparator takes more time to switch to a valid
output state than is available in the sample interval. In
accordance with the illustrative embodiment, by using hysteresis
comparators 112a and 112b, which introduce memory effect in the
form of .DELTA.V, reduced meta-stability is achieved.
[0025] Before describing the DC offset correction loop, an
illustrative embodiment of the level detection circuit 112 will be
described with reference to FIG. 2. FIG. 2 illustrates a block
diagram of the level detection circuit 112 having a finite state
machine (FSM) 120 and M AC-coupled level detectors
112.sub.1-112.sub.M, each of which comprises a hysteresis
comparator 121.sub.1-121.sub.M, respectively, and an asynchronous
ripple counter 122.sub.1-122.sub.M, respectively. Each of the M
hysteresis comparators 121.sub.1-121.sub.M has a fixed hysteresis
level, |.DELTA.V.sub.K| that is different from all of the other
fixed hysteresis levels. Each ripple counter 122.sub.1-122.sub.M is
an N-bit asynchronous ripple counter that changes state on every
input transition, where N is a positive integer. Each ripple
counter 122.sub.1-122.sub.M comprises N latches 123.
[0026] The outputs of the comparators 121.sub.1-121.sub.M are fed
to the clock inputs of the respective asynchronous ripple counters
122.sub.1-122.sub.M. The outputs of the ripple counters
122.sub.1-122.sub.M are fed to the FSM 120. When N transitions are
detected by the FSM 120 for a given one of the level detectors
112.sub.1-112.sub.M, the output of the ripple counter
122.sub.1-122.sub.M that is fed to the FSM 120 is a logic 1. This
causes the FSM 120 to set the output of the corresponding
comparator 121.sub.1-121.sub.M to a logic 1 and to confirm the
input signal amplitude level by sending the M bits to the
current-mode DAC 113 (FIG. 1). The current-mode DAC 113 outputs the
corresponding analog current signal to the current-to-voltage
converter 114 (FIG. 1), which generates the gate voltage for MOSFET
M.sub.AGC. Thus, based on the input OMA levels, the gain of the
first gain stage 101 is adjusted accordingly.
[0027] In the example where M=2, when the outputs of the level
detectors 112a and 112b are both logic 0, the gain of the first
gain stage 101 is set to a high gain. When the outputs of the level
detectors 112a and 112b are logic 1 and logic 0, respectively, the
gain of the first gain stage 101 is set to a medium gain. When the
outputs of the level detectors 112a and 112b are both logic 1, the
gain of the first gain stage 101 is set to a low gain.
Additionally, as will be described below in more detail, the M bits
output by the FSM 120 are also used by the DC offset correction
loop to regulate the DC offset correction.
[0028] With reference again to FIG. 1, in accordance with an
illustrative embodiment, the DC offset correction feedback loop
comprises a charge pump 115, a switchable integrating capacitor
bank 116 and a DC current injecting/bleeding circuit 117. The
switchable integrating capacitor bank 116 comprises a plurality of
MOSFET switches M.sub.3, M.sub.4, M.sub.5 and M.sub.6 and a
plurality of capacitors C.sub.1, C.sub.2 and C.sub.3. The DC
current injecting/bleeding circuit 117 comprises a plurality of
MOSFET switches M.sub.7 and M.sub.8 and a plurality of DC bleed
transistors M.sub.9, M.sub.10 and M.sub.11.
[0029] The M bits that are output from the FSM 120 are received by
the switchable integrating capacitor bank 116 and by the DC current
injecting/bleeding circuit 117. Those bits are used to selectively
activate/deactivate the MOSFET switches M.sub.3-M.sub.8.
Selectively activating/deactivating the MOSFET switches
M.sub.3-M.sub.6 causes an appropriate feedback capacitance to be
set in the switchable integrating capacitor bank 116 based on the
average input OMA level detected by the level detection circuit
112. Selectively activating/deactivating the MOSFET switches
M.sub.7 and M.sub.8 causes DC current to be injected into or
shunted away from the base of HBT T.sub.1 through one or more of
the bleed transistors M.sub.9, M.sub.10 and M.sub.11, depending on
whether the corresponding MOSFET switches M.sub.7 and/or M.sub.8
are activated/deactivated. Activation/deactivation of the MOSFET
switches M.sub.7 and M.sub.8 also causes the feedback resistance of
the DC offset correction loop to be varied. The switchable
integrating capacitor bank 116 sets the DC voltage at the gates of
the bleeding transistors M.sub.9-M.sub.11. The bleeding transistors
M.sub.9-M.sub.11 shunt the input DC current to ground when there is
a common mode DC offset voltage at the output of the second gain
stage 102.
[0030] To meet the short burst synchronization timing requirement
for EPON and GPON networks, the DC loop settling time should be
less than 100 ns. This can be achieved by reducing the resistance
and capacitance of the DC offset correction loop, but doing so
would result in a lower cut-off frequency at the output of the
second gain stage 102. To avoid that, the charge pump 115 is used.
The Charge pump is a dual polarity (i.e., current steering) charge
pump that either charges or discharges the integrating capacitors
C.sub.1-C.sub.3 to source or sink current. The charge pump 115
includes a dual-polarity boost circuit (not shown) for this
purpose. The charge pump 115 receives the analog current signal
output from the current-mode DAC 113 and the differential voltage
signal output by the second gain stage 102. The signal output from
the DAC 113 is asserted whenever the DC offset voltage is greater
than a user-specified amount and is de-asserted when the DC offset
voltage is less than the user-specified amount. The boost circuit
helps to reduce the DC offset correction loop settling time and
also saves power because it is only engaged whenever the DC offset
is greater than the user-specified amount and is otherwise
disengaged, which is most of the time.
[0031] The reset block 118 outputs a reset pulse to the internal
nodes of the AGC feedback loop and the DC offset correction
feedback loop to cause their internal nodes to be reset to initial
conditions. The reset pulse can also be externally generated, as
indicated by reset bypass pin 119. As indicated above, FTTH PONs
communicate data in the form of asynchronous data packets. These
packets resemble bursts of data with guard bands in between data
packets where there is no data. Because the data packets come at
different data rates and varying OMA levels, they are not DC
balanced. For this reason, a reset pulse is provided either
externally or on-chip to reset the internal node voltages of the
feedback loops to initial conditions. This helps reduce the loop
settling times and enables short burst synchronization times to be
achieved.
[0032] FIG. 3 is a timing diagram demonstrating the manner in which
the TIA circuit 100 operates when operating at a data rate of 1.25
Gbps. The top waveform 301 corresponds to an input burst of burst
mode data packets received at the input of the first gain stage
101. The units are in milliamperes (mA) on the vertical axis and
time in microseconds on the horizontal axis. The waveform 310
corresponds to the reset pulse. The units for waveform 310 are in
volts (V) on the vertical axis and time in microseconds on the
horizontal axis. The waveform 320 corresponds to the output of the
first level detector 112a. The waveform 330 corresponds to the
output of the second level detector 112b. The units for waveforms
320 and 330 are in volts (V) on the vertical axis and time in
microseconds on the horizontal axis. The waveform 335 corresponds
to the gain of the first gain stage regulated by AGC feedback loop.
The waveform 340 corresponds to the DC offset correction applied at
the gate of HBT T.sub.1 by the DC offset correction loop. The units
for waveform 340 are in millivolts (mV) on the vertical axis and
time in microseconds on the horizontal axis. The waveform 350
superimposed on waveform 340 corresponds to the boost current
applied by the boost circuit of the charge pump 115. The units for
waveform 350 are in microamperes (.mu.A) on the vertical axis and
time in microseconds on the horizontal axis. The waveform 360
corresponds to the differential signal output from the second gain
stage 102. The units for waveform 360 are in mV on the vertical
axis and time in microseconds on the horizontal axis.
[0033] With respect to the waveform 301, the input signal to the
TIA circuit 100 is at -28 dbm between data bursts and at -2 dbm
during data bursts. The rest pulse is only asserted immediately
before and after each data burst and is otherwise deasserted. In
this example, the outputs of each of the level detectors 112a and
112b result in the gain shown in waveform 335. The settling time of
the AGC feedback loop corresponding to the amount of time that
passes from when the overshoot in the gain occurs to when the gain
settles to its steady state is only about 20 ns. With respect to
waveforms 340 and 350, it can be seen the DC boost causes the DC
offset correction loop to settle within about 36 ns. The
differential output waveform 360 tracks the input burst waveform
301, but takes about 56 ns to settle. The amount of time that it
takes the differential output waveform 360 to settle is the sum of
the DC offset correction loop settle time and the AGC loop settle
time. This amount of time corresponds to the synchronization time
of the TIA circuit 100, which is well within the 100 ns
synchronization time required by the specifications for PON
networks.
[0034] FIG. 4 illustrates a block diagram of the reset block 118
shown in FIG. 1 in accordance with an illustrative embodiment.
Because the settling times of the DC offset correction and the AGC
feedback loops are required to be very small in order to meet the
stringent synchronization specifications defined for PON networks,
having precise initial voltage or current conditions in the
internal nodes of these loops at the start of a burst can shorten
the loop settling time. For this reason, the TIA circuit 100
preferably includes the reset block for generating a reset pulse
that restarts the loops from known initial conditions. Also,
typical receiver optical subassemblies (ROSAs) are small and have
only a few input/output pins. In such cases, there is typically no
area left in the ROSA for a reset pin. As a result, an on-chip
reset pulse generation circuit is needed inside of the TIA circuit
chip. The reset block 118 provides such a solution. Additionally,
the reset block 118 is also capable of detecting long sequences of
zero bits, which allows the TIA circuit 100 to determine that it is
receiving erroneous data.
[0035] In accordance with this illustrative embodiment, the reset
block 118 comprises a level detector 201, an N-bit synchronous
counter 202 and a ring oscillator 203. The level detector 201 is
nearly identical to the level detectors 112.sub.1-112.sub.M shown
in FIG. 2. The level detector 201 has a hysteresis comparator 211,
an asynchronous ripple counter 212 and a FSM 220. The asynchronous
ripple counter 212 is an N-bit asynchronous ripple counter that
changes state on every input transition, where N is a positive
integer. The N-bit asynchronous ripple counter 212 comprises N
latches 213. The N-bit synchronous counter 202 comprises N latches
214 and an AND gate 215. The output of the comparator 211 is fed to
the clock input of the N-bit asynchronous ripple counter 212. The
output of the ripple counter 212 is fed to the FSM 220. When N
transitions are detected by the FSM 220, the output of the ripple
counter 212 that is fed to the FSM 220 is a logic 1, which
indicates that a string of N logic 0s has been detected. This
causes the FSM 220 to set the output of the comparator 211 to a
logic 1 so that the output of the asynchronous ripple counter 212
remains at a logic 1. The logic 1 output by the comparator 211
corresponds to the asserted reset pulse received at the node
connected to pin 119 (FIG. 1).
[0036] The output of the ring oscillator 203 and the output of the
comparator 211 are inputs to the AND gate 215. When both of these
inputs are logic 1s, the output of the AND gate 215, which is used
as the clock signal for the latches 214 of the N-bit synchronous
counter 202, also goes high. Therefore, when the output of the
comparator is a logic 1, then once the clock signal output by the
ring oscillator 203 has transitioned a predetermined number of
times, the output of the N-bit synchronous counter 202 becomes a
logic 1, which causes an OR gate 218 to set the output of the
comparator 211 to logic 0.
[0037] In accordance with this illustrative embodiment, the ring
oscillator 203 is a 3-to-6 CMOS inverter-based ring oscillator that
generates a clock at a very low frequency (F.sub.CLK) compared to
the received data rate. After the FSM 220 causes the output of the
comparator 211 to be asserted, thereby causing the reset pulse to
be asserted, the clock signal generated by the ring oscillator 203
is sent to the N-bit synchronous counter 202. The synchronous
counter counts N clock pulses and then sends a signal to the OR
gate 218 to cause the output of the comparator 211 to be
de-asserted, which de-asserts the reset pulse. Thus, the reset
pulse is of duration N/F.sub.CLK, during which the feedback loops
are reset to the initial states.
[0038] The level detector 201 is disabled during data bursts, but
it can erroneously trigger a reset pulse if a data burst has a long
series of 0 bits. Typically, the data in a PON network is encoded
such that long strings of 0 or 1 bits are not allowed in the data
packets. Therefore, if the data packet has a long string of 0 bits
that causes the reset pulse to be asserted, then this implies that
the received data has bit errors. The only time period during which
the TIA circuit 100 receives a long string of 0 bits is during the
Guard or Dead time. Hence, the transition count contained in the
FSM 220 confirms if the detected zero level is occurring during the
Guard or Dead time.
[0039] FIG. 5 is a timing diagram demonstrating the manner in which
the reset block 118 operates. Waveform 401 corresponds to an input
burst of burst mode data packets output as the differential output
signal of the second gain stage 102 and input to the reset block
118. Waveform 402 corresponds to the clock signal generated by the
ring oscillator 203. Waveform 403 corresponds to the disable signal
output from the OR gate 218 to the comparator 211. Waveform 404
corresponds to the enable signal output from the FSM 220 to the
comparator 211. Waveform 405 corresponds to the reset pulse
generated by the reset block 118. The vertical axis represents
voltage and the horizontal axis represents time.
[0040] It can be seen from waveform 403 that at any time other than
during the Guard or Dead time between packet bursts, the disable
signal output from the OR gate 210 to the comparator 211 is
asserted, thereby preventing the rest pulse from being asserted
during that time period. Therefore, the reset pulse represented by
waveform 405 is always deasserted while the disable signal is
asserted. During the Guard or Dead time, the disable signal is
deasserted, and therefore the reset signal can be asserted during
the Guard or Dead time. After the FSM 220 has counted N 0 bits
while the disable signal is deasserted, the enable signal
represented by waveform 404 is asserted, which results in the reset
pulse represented by waveform 405 being asserted. Once N clock
pulses (waveform 402) of the ring oscillator 203 have occurred
after the reset pulse and enable signal have been asserted, the
enable signal and the reset pulse are deasserted. At the end of the
Guard or Dead time, the disable signal (waveform 403) is again
asserted.
[0041] It should be noted that although the illustrative
embodiments have been described with reference to a few
illustrative embodiments for the purpose of demonstrating the
principles and concepts of the invention. Persons of skill in the
art will understand how the principles and concepts of the
invention can be applied to other embodiments not explicitly
described herein. It should also be noted that the circuits and
method described above with reference to FIGS. 1-5 are merely
examples of suitable circuit configurations and methods that
demonstrate the principles and concepts of the invention. As will
be understood by those skilled in the art in view of the
description being provided herein, many modifications may be made
to the embodiments described herein while still achieving the goals
of the invention, and all such modifications are within the scope
of the invention.
* * * * *