Re: Question? 1300nm WWDM channel width
Eric,
I agree that we should write this spec to be addressable with 1310nm VCSELs as
well as DFBs as it provides a low-cost migration path. At higher temperatures,
the optical output power of a VCSEL tends to roll-over. To compensate, the
drive circuitry provides more DC current to the VCSEL. This can lead to higher
device temperatures. For this reason, I think we should plan on the 30 deg C
rise from ambient to die temperature or the 100 deg C temperature range just to
be safe. I believe this should be 0..100 deg C, however, for the reason Jeff
stated in his email.
As you pointed out, the maximum wavelength variation for VCSELs and DFBs is 0.1
nm/deg C (I agree that 0.085nm/C is typical). Based on these worst case
assumptions, the wavelength ranges become:
Temperature variation: 0.1nm/deg C x 100 deg C = 10 nm
If we provide a spec of +/- 2.5 nm for laser variation, the wavelength range for
the laser is 15 nm.
Is this managable for the filter design if we maintain the channel spacing at
24.5nm?
Rgds,
Bob
Eric Grann wrote:
> Jeff,
>
> I agree with what you are saying.
>
> Furthermore, the Typical spec for 1300nm DFBs and VCSELs are 0.085nm/*C.
> The Maximum spec is 0.1nm/*C. In determining the junction temperature range
> that the lasers will see, there are several factors to consider. The
> tolerance of the temperature measuring equipment of the laser vendor can
> introduce an error in the wavelength. The amount of current being applied
> to the laser directly effects the junction temperature and therefore the
> wavelength of the laser. I do not believe one can easily design a
> transceiver for operation from 0 to 70*C with only a 10*C margin. One
> should at least have a 20*C margin. A 30*C margin would be nice and safe.
>
> Using only the typical numbers:
> Temperature variation: 0.085nm/*C x (70*C+20*C) = 7.7nm
> Laser tolerance at a particular temperature: 11.4nm - 7.7nm = 3.7nm
>
> Therefore, the laser manufacutures have a +/- 1.8nm spec. A spec of +/- 3nm
> would significantly increase the yield and reduce the cost of the lasers.
>
> If we are trying to make a specification that is conservative and not on the
> edge, we should widen the channel width. With a 14nm channel width, the
> filter design and fabrication is still easy. Why should we complicate the
> overall design, reduce the yield of the laser manufacturing, and increase
> the costs of the lasers, when we can increase the channel width now very
> easily.
>
> Regards,
>
> Eric
> *************************************************
> Eric B. Grann, Ph.D.
> Chief Technical Officer
>
> Blaze Network Products Inc.
> 5180 Hacienda Drive
> Dublin, CA 94568
>
> phn: (925) 558-2100
> fax: (925) 558-2197
> email: grann@xxxxxxxxxxx
> website: http://www.blazenp.com
> *************************************************
>
> -----Original Message-----
> From: Jeff Porter (rgbn10) [mailto:j.porter@xxxxxxxxxxxx]
> Sent: Tuesday, October 31, 2000 5:57 PM
> To: Eric Grann
> Subject: Re: Question? 1300nm WWDM channel width
>
> > In order to achieve a final transceiver temperature specification
> > of 0-70*C, the lasers must have a slightly higher temperature range. To
> be
> > safe, I would recommend a range of (-5) to 95*C.
>
> I think Brian has you on the low end - don't really see a way
> to be lower than ambient (for very long).
>
> By claiming 80C range, is he claiming only 10C theta-J/A? Or is he
> counting on warming up to within 10C of final Tj "quickly enough",
> so just adds 10C to minimum 70C range for slow time constants and
> Tja variation (power variation, theta-variation?)
>
> -----Original Message-----
> From: brian_lemoff@xxxxxxxxxxx [mailto:brian_lemoff@xxxxxxxxxxx]
> Sent: Tuesday, October 31, 2000 11:53 AM
> To: stds-802-3-hssg@xxxxxxxx
> Subject: RE: Question? 1300nm WWDM channel width
>
> Eric,
>
> Having surveyed the temperature dependence, not only of our lasers, but also
> lasers from other vendors, a more reasonable number is 0.08 nm/degreeC,
> rather then the 0.1 nm/degree number that you quoted. In addition, with a 0
> to 70 degree ambient temperature requirement, there is no reason why the
> laser junction should see a 100 degree temperature range. Of course this is
> implementation dependent, but it is difficult to imagine much more than an
> 80 degree variation. This would use up 6.4 nm of the 11.4 nm window,
> leaving another 5 nm for tolerance. Most laser vendors are comfortable with
> a +/- 2.5 nm specification.
>
> Best Regards,
>
> Brian
>
> ***********************************************************************
> Brian E. Lemoff, Ph.D.
> Project Manager
> LAN/MAN Optical Technologies
> Agilent Laboratories
> 3500 Deer Creek Rd., MS 26M-9
> Palo Alto, CA 94304-1392
>
> phone: (650) 485-8957
> FAX: (650) 485-3626
> email: brian_lemoff@xxxxxxxxxxx
> ***********************************************************************
>
> > -----Original Message-----
> > From: Eric Grann [mailto:grann@xxxxxxxxxxx]
> > Sent: Tuesday, October 31, 2000 9:12 AM
> > To: stds-802-3-hssg@xxxxxxxx
> > Subject: Question? 1300nm WWDM channel width
> >
> >
> >
> > To all:
> >
> > I have been trying to get some feedback on the channel width
> > of the 1300nm
> > WWDM system. As I currently understand the specifications,
> > the bandwidth
> > of an individual wavelength channel is 11.4nm (spec of *
> > 5.7nm). At these
> > laser wavelengths, both DFBs and VCSELs have a temperature
> > coefficient of
> > 0.1nm/*C. In order to achieve a final transceiver
> > temperature specification
> > of 0-70*C, the lasers must have a slightly higher temperature
> > range. To be
> > safe, I would recommend a range of (-5) to 95*C. With this range, the
> > wavelength drift will account for 10nm of the channel
> > bandwidth, leaving
> > only 1.4nm for the tolerance of the laser fabrication. The
> > feedback I have
> > received from several laser vendors is, this will be very difficult,
> > resulting in low yields and increased costs for both DFBs and
> > especially new
> > 1300nm VCSELs.
> >
> > I would like to recommend that this specification be
> > increased to at least
> > 14nm (spec of * 7.0nm). With a 24.5nm channel spacing, this
> > will still
> > provide plenty of room for a relatively simply filter.
> >
> > I would like to get some feedback and response to this recommendation.
> >
> > Eric Grann, Ph.D.
> > Chief Technical Officer
> > Blaze Network Products
> >