Professional Documents
Culture Documents
Optical Receiver Performance Evaluation: Q-Factor in The Presence of ISI
Optical Receiver Performance Evaluation: Q-Factor in The Presence of ISI
Optical Receiver Performance Evaluation: Q-Factor in The Presence of ISI
Rf V(t)
(A/W)
In
BWTIA CDR
3
1.0E+00
V(t) 1.0E-02
V1 1.0E-04
BIT-ERROR-RATE
1.0E-06
V1 - VISI
1.0E-08
VTH
1.0E-10
V0 + VISI 1.0E-12
V0 1.0E-14
1.0E-16
0 1 2 3 4 5 6 7 8
QBER
Figure 2. Worst-case amplitude-noise distribution in the presence of ISI Figure 3. BER vs. QBER is plotted based on equation 3.
facilitates estimation of the ISI penalty on optical sensitivity.
4
-10 -15
-11 MAX3970 (Rf = 0.5kΩ) -16
-12 -17
OPTICAL SENSITIVITY (dBm)
Figure 4. Receiver optical sensitivity is shown for 10Gbps TIA ICs, Figure 5. This graph shows the receiver optical sensitivity for 2.5Gbps
assuming the SONET minimum extinction-ratio requirement. TIA ICs when each is used with a limiting amplifier of
different sensitivity.
5
3
TOPEN
OPTICAL-POWER PENALTY (dB) 2.5 (BER = 10-12)
2
DJ/2 DJ/2
Q=0
1.5
1
TOPEN
0.5
RJP-P/2 RJP-P/2
0 Q=7
0 0.1 0.2 0.3 0.4 0.5
INTERSYMBOL INTERFERENCE (ISI) 0 T b /2 Tb
Figure 6. The relationship between ISI and optical power penalty in an Figure 7. This illustrates the relationship between CDR minimum eye
ideal case is illustrated. opening and the Q-factor.
The minimum-required TIA input current is related to ISI is RJRMS, the total deterministic jitter is DJP-P, and the
according to: CDR minimum required eye opening is T OPEN at a
2 x QBER x NTOTAL specified BER, then the timing Q-factor is defined as:
IP-P = (Eq 11)
(1 - ISI)
Tb - TOPEN - DJP-P
Q= (Eq 12)
The ISI penalty is defined as the difference in optical 2 x RJRMS
sensitivity in the presence of ISI, as compared to an ideal
case when ISI = 0. The calculated result is shown in RJP-P = 2QBER x RJRMS (Eq 13)
Figure 6. The optical power penalty is 0.46dB for 10%
For more information, refer to application note HFAN-
ISI distortion.
04.0.2 Converting between RMS and Peak-to-Peak Jitter
Finally, the total optical power penalty in dB is the sum of at a Specified BER. (www.maxim-ic.com/AN337)
the ISI penalty and the overall random-noise penalty.
The relationship between the CDR minimum eye-opening
CDR jitter-tolerance penalty requirement and the Q-factor is illustrated in Figure 7. To
achieve a specified BER, the corresponding minimum
As the signal goes through the receiver amplifier chain to QBER can be found from Figure 3.
the limiting stage, the amplitude noise is converted into
When the jitter frequency at the CDR input is higher than
timing jitter at the data midpoint crossing. Random and
the PLL bandwidth, the CDR jitter tolerance (noted as
deterministic jitter are generated due to the existence of
JTP-P) is related to TOPEN as:
random noise, limited bandwidth, passband ripple,
group-delay variation, AC-coupling, and nonsymmetrical JTP-P = (Tb - TOPEN) (Eq 14)
rise/fall times. The combination of these jitter
components decreases the eye opening available for To avoid degrading the optical sensitivity, the CDR high-
error-free data recovery. Consequently, CDR jitter- frequency jitter tolerance should satisfy:
tolerance capability is another critical factor for
determining optical sensitivity. JTP-P ≥ 2 x QBER x RJRMS + DJP-P (Eq 15)
CDR jitter tolerance is a measure of how much peak-to-
peak jitter can be added to the incoming data before Depending on the slope of the edge transition, the random
errors occur due to misalignment of the data and jitter RJRMS is generated from the additive white noise at
recovered clock. For a PLL-based CDR design, a signal transitions. Assuming that the signal rise/fall time
minimum data-eye opening is required, which is (20% to 80%) before limiting is symmetrical and equal to
determined by the clock-to-data sampling position, the tr, the random jitter can be estimated by:
retiming flip-flop setup/hold times, and the phase tr
RJRMS = (Eq 16)
detector characteristics. Assuming that the random jitter (VP-P/NRMS) x 0.6
6
Here tr is dependent on the overall receiver small-signal Conclusion
bandwidth BWTOTAL. Assuming a first-order lowpass filter:
To estimate optical-receiver sensitivity, it is necessary to
0.22
tr ≈ (Eq 17) consider error sources in both amplitude and timing. This
BWTOTAL article shows how the amplitude and timing-error sources
At the optical receiver input, it is assumed that the TIA is separately affect the overall receiver BER with practical
linear before the limiting amplifier. Therefore, random device implementations. Using these guidelines, optical
jitter can be expressed as a function of the peak-to-peak receiver performance can now be accurately predicted to
current to the total RMS noise ratio at TIA input: choose the proper TIA, limiting amplifier, and CDR. In
tr reality, the optical input is not an ideal signal, because it
RJRMS = (Eq 18) suffers random noise from the transmitter as well as ISI
(IP-P/NTOTAL) x 0.6
from fiber dispersion. When a stressed optical signal is
Using Equation 18, the random jitter at the limiting- received, the same approach presented in this article can
amplifier output is plotted in Figure 8 as a function of be used for estimating the signal Q-factor and, therefore,
TIA input current to noise ratio. determining the BER.
The CDR jitter-tolerance penalty on optical sensitivity can
be estimated by combining equations 15 and 18, then
solving for IP-P as:
2 x QBER x tr
IP-P = (Eq 19)
(JTP-P - DJP-P) x 0.6/NTOTAL
0.06 -11
-12
0.05
-13 DJ = 0.2UI
OPTICAL SENSITIVITY (dBm)
tr = 0.4UI
RANDOM JITTER (UIRMS)
0.04 -14
-15 DJ = 0.15UI
0.03
-16
0.02 -17
-18
0.01 tr = 0.3UI
DJ = 0.1UI
tr = 0.2UI -19
0 -20
10 15 20 25 30 35 40 10 15 15 20 25 30 35 35 35 35 40
TIA INPUT CURRENT-TO-NOISE RATIO (IP-P/NTOTAL) CDR HIGH-FREQUENCY JITTER TOLERANCE (UIP-P)
Figure 8. Random jitter and input-current-to-noise ratios are shown for Figure 9. The relationship between optical sensitivity and CDR jitter
rise times of 0.2UI, 0.3UI, and 0.4UI. tolerance is illustrated for deterministic jitter tolerances of
0.1UI, 0.15UI, and 0.2UI.