Design Note HFDN-18.0 (Rev.1; 04/08)
Maxim Integrated Products
Page 8 of 9
).
(
5
k
2
V
2
.
1
5
APC_set
APC_set
diode)
(monitor
av
R
I
R
I
×
=
Ω
+
=
(17)
where I(R
APC_set
) is the current flowing through
R
APC_set
. Average optical power output from the laser
is controlled.
The stability of the APC feedback loop depends on a
combination of the bias-integrator time constant and
the other gains and poles around the loop. In
particular, it depends on the laser-to-monitor current
gain. Current shunt #23 is provided to reduce the
loop gain when the laser-to-monitor current gain is
too large. For laser gains less than 0.005, either
connect MD_X to ground or leave it unconnected;
for laser gains greater than 0.02, connect MD_X to
MD; for laser gains between 0.005 and 0.02, use
either arrangement.
4.2.1
Mark-Density Compensation
In the very long term, the data input contains equal
numbers of 0s and 1s, and the average optical power
output is truly the average of the powers that
correspond to data_0 and data_1. However, in the
shorter term there may be a local excess of either 0s
or 1s. Said differently, the very-long-term mark
density is 50%, but the short-term mark density is
not 50%.
Any APC loop attempts to hold the average power
constant, and therefore adjusts the powers that
correspond to data_0 and data_1, up or down
depending on the local mark density. The rate at
which this adjustment takes place is set by the APC
loop time constant. One common approach to
reducing this undesirable effect is to make the APC
time constant very long, but this has the
disadvantage of slowing the response to other
changes. The MAX3865 uses a different technique,
mark-density compensation.
When the local mark-density of the data is 50%, the
average output from the mark-density multiplier is
zero. However, when the data consists locally of an
excess of 1s, the average output from the mark-
density multiplier goes positive and adds to the bias
reference current. The numerical details in Figure 4
are such that this increase in effective reference
current compensates exactly for the local increase in
feedback current from the monitor diode, provided
the current, monitor-diode I
AMC
, which originates in
#15, is programmed equal to the difference between
the data_0 and data_1 currents in the monitor diode.
(This requires monitor-diode I
AMC
to be programmed
equal to the monitor-diode modulation current.)
There is then no error output applied to the input to
integrator #14, and no shift in the laser bias current.
4.3 AMC Mode
In the AMC mode, the second input to subtractor #9
is enabled, also the second input to multiplier #19.
The current maximum I
mod
programmed in #7 takes
on the significance of an upper bound to the peak-
peak modulation current output—exactly like the
arrangement for bias current described in Section
4.2. Current control I
mod
subtracts away from
maximum I
mod
to give the actual I
mod
in the laser.
Multiplier #19 generates a pilot-tone current and
adds this to both the bias and modulation output
currents. There is, therefore, a pilot-tone component
of current in the monitor diode.
When the AMC loop has settled to equilibrium, the
pilot-tone component of feedback current from the
monitor diode must be equal to the pilot-tone
reference current that originates in #15, #16 and #20.
If, for example the monitor-diode feedback current is
momentarily too large, the resulting difference
reduces the modulation current (and hence pilot-tone
current) in the laser. The pilot-tone current in the
monitor diode, and hence peak-peak modulation
current, is controlled:
).
(
5
k
2
V
2
.
1
5
AMC_set
AMC_set
diode)
(monitor
p
p
R
I
R
I
×
=
Ω
+
=
−
(18)
where I(R
AMC_set
) is the current flowing through
R
AMC_set
. Peak-peak optical power output from the
laser is controlled.
In summary, the average optical power output and
extinction ration in the AMC mode are given by:
,
)
(
5
)
k
2
(
V
6
monitor
APC_set
monitor
APC_set
monitor
diode)
(monitor
av
av
A
R
I
A
R
A
I
P
×
=
Ω
+
=
=
and
(19)
