RE: Some additional comments on PMD
The temperature dependence of the center wavelength for VCSELs at 1310nm is
essentially the same as for DFBs, i.e. about 0.1nm/C or smaller. They are
driven by the same mechanism, namely temperature dependence of the
refractive index which determines the cavity resonance and therefore the
lasing wavelength. More subtle differences may make the VCSEL temp
dependence slightly smaller.
Cheers,
Jack
-----Original Message-----
From: Rahn, Juergen (Juergen) [mailto:krahn@lucent.com]
Sent: Tuesday, May 08, 2001 8:25 AM
To: 'stds-802-3-hssg-serialpmd@ieee.org'; Rahn, Juergen (Juergen)
Subject: Some additional comments on PMD
Transmitter specification on spectral characteristics.
The triple tradeoff Idea is introduced considering spectral characteristics
and dispersion penalties in the current draft.
While in 1310 nm the tradeoff is specified in the 1500 nm domain this is
completely left open and an unspecified penalty can be compensated by
unspecified additional power.
Some thoughts about this
Traditionally the laser spectral characteristics of lasers are specified
depending on the laser type. For Multi-longitudinal-mode (MLM) or Fabry
Perot (FP) lasers the rms spectral width is defined/specified, for which the
Mode-partition-noise mechanism is the dominant effect determining the
chromatic dispersion penalty. For Single-longitudinal-mode (SLM) or DFB
lasers the-20 dB linewidth together with the Side Mode Suppression (SMS)
ratio is used to specify its spectral characteristics. Unfortunately (as
described in ITU Rec. G.957) there is not yet a reliable model to correlate
the actual dispersion penalty with the SLM laser spectral characteristics.
In IEEE many references are made to the use of VCSELs, which very often seem
to exhibit single-mode behaviour. Therefore it is very surprising that only
an rms width is used to specify the transmitter spectral behaviour, in
particular because rms width is pertinent to MULTI-longitudinal-mode lasers.
In Figure 52-5 and in Table 52-15 the socalled triple-trade-off
characteristics are mentioned, allowing an optimization of laser central
wavelength, laser RMS width and OMA.
This triple-trade-off, might be useful from a "theoretical" point of view,
but from a testing/conformance verification point of view it definitely is
just too complicated and will add substantially to the cost of the optics
instead of allowing the desired cost reduction.
One simple principle in optics is "simple specs allow simple and therefore
cheap testing".
While it is understood that widening the window of allowable devices is
important this only makes sense if in the field the correlation between the
parameters are evident. But, how can a transmitter in operation "know" which
the actual wavelength and rms width is, in order to adjust its OMA.
Here it is critical to understand the temperature characteristics of the
spectrum of VCSELs.
Both of MLM and SLM lasers the temperature dependence of the centre
wavelength is clearly understood (max values of about 0.5 nm/*C and 0.1
nm/*C respectively). What is this for VCSELs in the 1310 nm window?
It is our opinion that before there is sufficient experimental verification
of the triple-trade-off curves and correlation to practical use, care should
be taken to use it as a basis for specification.
Therefore it is suggested to have a fixed/single set of spectral
characteristics pertinent to the appropriate laser type.
Regards Juergen