Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

[10GMMF] Laser Non-Idealities for TP2

Title: Laser Non-Idealities

On behalf of Andre - I recall that a purpose of this work is to provide a checklist to help ensure that TP2 specs and tests capture the matters of concern.
As we discussed, the information below is for use by the TP2 group. 

Thank you,


Andre Van Schyndel, Ph.D.

Senior Systems Architect

Subsystem Development

Bookham Technology plc

Tel: (613) 270-4074


A discussion on the Oct.14 2004 TP2 call suggested it would be useful to have a list of the non-ideal characteristics of a laser and some indication of how we might account for them in the LRM compliance tests.  The following information is meant for use by the group to this end.  This is not meant to be definitive but rather, a heuristic set of characteristics to help in defining TP2 tests.  I have left out quite a few sources of non-ideality that are probably not important for our purposes, and perhaps some of those listed may not be important.  Input/modification to this list is encouraged.

For the purposes of the following discussion, we will define an ideal laser as one which produces a single, constant wavelength optical output power proportional to the input electrical current reduced by a threshold current.  Our working definition of a laser non-ideality is any characteristic which describes the laser's deviation from this ideal behaviour.

The typical values given are for 10G FPs and DFBs.  I encourage those with greater VCSEL experience to fill in the appropriate numbers.


The laser has a threshold current under which there is very little optical output (see spontaneous emission below), and over which the optical power is very nearly linear with applied current offset by this threshold current.  At the threshold current, the optical power changes by orders of magnitude over a narrow range of current.  Due to the dynamics of laser transients, rise times and fall times tend to slow if some part of the modulation current goes below threshold.  In addition when operating below threshold, the optical output is dominated by stimulated emission broadening the output spectrum.  On the other hand, the extinction ratio is higher if the "zeros" correspond to lower current.  As a compromise generally, DMLs are biased to operate such that the minimum current is just above the threshold current (to maximize the extinction ratio and minimize laser dynamics effects which reduce optical rise and fall times).

Typical values:

Threshold current: 10-20ma (~30C), 20-50ma (~85C)


There is an intrinsic oscillation between carrier density and photon density which manifests itself as a damped oscillation whenever the current is changed quickly.  This is most visible on the rising edge as an optical power overshoot followed by a damped oscillation.  Relaxation oscillations are far less predominant on the falling edge.  The intensity oscillations are mirrored by oscillations in the optical wavelength through the chirp characteristics described later.

Typical values for the rising edge:

Intensity overshoot: 0-100%

Relaxation oscillation frequency: 5GHz-15GHz

Damping time constant: 50ps-500ps


There are different mechanisms governing the rise and fall times of the laser which generally result in asymmetric slew rates.  The optical rise time is usually faster than that of the applied current and the fall time is usually slower than that of the applied current.

Typical values (assuming a 10G driver):

Maximum Optical Rise time: 10-30ps (assuming the electrical rise time is slower than the optical rise time)

Minimum Fall time: 30-80ps  (assuming the electrical fall time is faster than the optical fall time)


In an FP laser, there are typically 10 or so modes contributing (separated in wavelength by half the inverse of the cavity optical-length) In a DFB, sideband suppression reduces the optical power in all but the main mode by about 40dB. Typical values:

Spectral width (including all modes): 2nm (this is essentially the spectral width of FPs)

Line width (width of each mode): 0.5-10MHz (this is essentially the spectral width of DFBs)


The wavelength of the laser changes with the optical intensity.  Small percentage changes in the wavelength are often expressed as changes in the optical frequency.

Typical value:

Change in optical frequency: 1-25 GHz/mW


The wavelength of the laser changes with the rate of change in optical intensity.

Typical value:

Change in optical frequency: 0.3-3 GHz/mW/ns


The wavelength of the laser changes through a long series of ones (zeros) due to heating (cooling) and thermal expansion (contraction) of the laser cavity.

Typical value:

Change in optical frequency: 0.2-0.5 GHz/UI


There is a current saturation level above which the output power is no longer proportional to the current.

Typical value:

Saturation threshold optical power: 10-20mW


As the current is increased above threshold but below power saturation, the power in the central mode becomes a greater fraction of the total power (the gain for the side modes saturates).  This results in a sharpening of the output spectrum as the output power increases.  Above power saturation, the relative power in the central mode reaches a maximum and then decreases.

Typical value:

Saturation threshold optical power: 10mW to 0.05mW over modes 0 to +/-4


RIN is the ratio of the output intensity fluctuation to the output intensity.  Although the spectrum is not flat (it peaks at the relaxation oscillation frequency), for digital systems it is usually averaged over the bandwidth of interest (typically 0.2-12GHz for 10GB/s lasers) and specified in dB/Hz.  It is generally inversely proportional to the cube of the output power, although at high powers it does not fall as fast.

Typical value:

RIN is typically -130dB/Hz for an FP and -150dB/Hz in a DFB.


Whenever there is current through the laser there is spontaneous emission.  Indeed below threshold current, spontaneous emission dominates the output.  Spontaneous emission behaves as broad spectrum noise, hundreds of nm wide.

Typical value:

Integrated spontaneous emission power: 100 uW (irrespective of drive current over 10GB/s time scales)


Mode partition noise is caused by power shifting between a laser's multiple modes.  In an FP, there are typically 10 or so modes contributing (separated in wavelength by the half the inverse of the cavity optical-length).  Large fractions (0-100%) of the power in an FP mode can move to another on time scales from picoseconds to nanoseconds however, the total power remains essentially constant Mode partition noise refers to the increase in noise in any particular mode due to this effect.  If the total power in all the modes is measured, the mode partition noise reduces to the RIN.  In a DFB, sideband suppression reduces mode partition noise by about 40dB.


Due to the properties of the materials used in the active region, FP and DFB lasers emit light linearly polarized in the plane of the active region (TE mode only).  No effects due to polarization non-idealites are expected for FPs and DFBs.