RE: 2.5 GBS vs 10 GBS on same fiber
I am passing along this exchange I had with Dr. Andrew Chraplyvy regarding
non-linearities. While his response is not detailed, it provides assurance
that non-linearities will not be a concern for 10GbE PMDs. He later
confirmed that four-photon mixing is also not an issue, once I relayed that
the wavelengths were separated by about 25 nm.
I don't think that any of the three cases will suffer degradations due to
nonlinearities. If you look very carefully you might see a hint of
four-photon mixing products (similar to third order intermodulation) in case
3 but not enough to impact the system performance.
"Kolesar, Paul F (Paul)" wrote:
> Hi Andy,
> This email correspondence mentions
> your name. The author, Bob Dhalgren, is responding to questions
> non-linearities and sites you as an expert. He asked me to relay
this to you
> and get your advice. The systems we are talking about are those
> defined for 10 Gigabit Ethernet. There are three in draft that
> singlemode fiber.
> 1. 10.3125 Gbaud serial at 1310 nm with a max transmit power
of 1 dBm
> and a max distance of 10 km.
> 2. 10.3125 Gbaud serial at 1550 nm with a max tx power of 4
dBm and a
> max distance of 40 km.
> 3. 4-wavelengths near 1300 nm system that used a max of -2.5
> wavelength (3.5 dBm total). Each wavelength operates at 3.125
> distance of 10 km.
> What non-linearity effects might we need to be concerned with, if
> Paul Kolesar
> 949 1255
From: Bob Dahlgren [SMTP:bob@SVphotonics.com]
Sent: Wednesday, December 13, 2000 12:58 PM
To: 'Ron Miller'
firstname.lastname@example.org; Paul Kolesar (E-mail); 'Robert Dahlgren'
Subject: RE: 2.5 GBS vs 10 GBS on same fiber
All materials fiber included are non linear to a small degree.
Glass has no 2nd-order nonlinearity because of symmetry reasons.
However there is a minute dependence of the refractive index as a
function of optical power, such that n = no + (3/8no)*RE[Chi3]*I
Where no is the linear refractive index, and RE[Chi3] is the
real part of the 3rd-order nonlinear susceptability tensor.
This effect is called the optical Kerr effect (OKE). This should
not be confused with the "electro-optic Kerr effect" which is the
DC-Term so to speak of the OKE.
The real part of the nonlinear coefficient [Chi3] of SMF
is in the range of 10^-19 to 10^-20 inverse Watts, according
to my reference. For a 1 mW pulse, and no = 1.5, the Kerr
effect would be a shift of the refractive index by a few
parts in 10^-23 upward. From 1.5 to 1.500000000000000000000015
Mind you, the Kerr effect is very fast, it tracks the optical
power instantaneously at the _optical_ frequency, which is
about 200 THz for 1550 nm light.
While I am not an expert in impairments on communication
systems, these temporal perturbations in refractive index
will manifest themselves in all kinds of curious ways under
HIGH POWER DENSITIES:
Self-Focussing of light beams. This should not be an issue in
guiding media like fiber - only in bulk media. This is a major
issue in laser fusion and ultrafast laser research.
Self-Phase modulation (SPM) is when the phase gets diddled from
an intensity-modulated optical field. Under the right conditions,
SPM can induce chirp, that will cause pulse boadening like
the chromatic dispersion. There can be some interplay between
SPM and chromatic dispersion with very strange effects at high
powers. Dont know about Kerr + DMD :-0
Cross-Phase modulation (XPM) is similar to SPM, but it involves
more than one optical signal in the same medium of different
wavelengths. If two or more pulses are simultaneous, intensity
of one can perturb the refractive index seen by the pulses.
This has ramifications in WDM systems.
Four-Wave Mixing will actully generate signals at new wavelengths
which correspond to the various permutations of the sums and
differences of the optical frequencies preesnt. My impression is
that this is one the main impairments of DWDM systems. FWM and XPM
are not issues in serial or parallel optics.
Brillouin effect. This is a interaction of the optical wave, an
optically-induced acoustic wave (dont ask), and the backscattered
"Stokes" wave. This causes the backscattered light to have an
optical frequency downshifted by about 23 MHz at 1550 nm. This
only occurs at very high powers.
Raman Effect. This is an interaction between light and nuclei
which have been excited to a vibrational state. The forward-
scattered Raman-shifted light is extremely broadband and
by about 12 THz. You can take advantage of the Raman gain and
make an amplifier or laser without Erbium or rare-earth doping!
As to the bottom line, how much "OKE penalty" we will have?
For single-mode fiber, these effects come into play after many km
of fiber, and at powers well above the power levels in our draft
standard. However at what point we will see additional eye closure
for the different PHYs I cannot say at this time. There is an expert
at Lucent by the name of Dr. Andrew Chraplyvy at Holmdel who would
be able to talk to the specifics of our situation (low channel
low powers, etc) if Agilent, Blaze, and others havent already
them for the CWDM case. Maybe a tutorial or presnetation might be in
order - Paul could you talk to Andrew i think he has a tutorial.
I would guess for MMF these effects are further reduced, due to the
low optical power density in the core region.
See texts by Iannone "Nonlinear Optical Communication Networks" and
Agrawal "Nonlinear Fiber Optics."
Robert Dahlgren, President, Silicon Valley Photonics
PO Box 60638, Sunnyvale, CA 94088 USA
> -----Original Message-----
> From: email@example.com
> [mailto:firstname.lastname@example.org]On Behalf Of Ron Miller
> Sent: Sunday, December 10, 2000 9:30 PM
> To: Robert Dahlgren
> Cc: 'email@example.com';
> Subject: 2.5 GBS vs 10 GBS on same fiber
> Hi Bob
> At the standards meeting ieee802.3 met and I volunteered to ask
> you, our expert in the optical field the following question.
> Please give it a whirl.
> Durring the final session IEEE session on Friday there was a
> duscussion on
> Single mode Fibre for 2.5 GBS and 10 GBS usage. Jonathan
> Thatcher Was trying to get
> some concurrence in the discussion but many red flags were being
> Thrown up with people looking at the fibre as being a
> non-linear medium.
> I refuted them and explained that the edge rate was the only
> Also that the amplitude and phase for a 1 GBS or 10 GBS would
> be the same
> Because the edge rate was what could make a difference.
> However, I also
> volunteered to get Bob Dalhgren our Consultant to fill in
> further details.
> Ron Miller, Signal Integrity Engineer,
> Brocade Communicaitons
> 408-487-8017, page 888-354-0690