AW: OMA vs. OMA/2 poll and some more importand issues with OMA
Hi,
For me the issue if OMA or OMA/2 is defined in the tables is as such of
minor importance (in my view). The most important issue is that the current
spec does not allow a simple verification if an interface/ optical span is
in spec or out of spec. So for me every spec would be fine that is
understandable , ensure consistence between transmitter and receiver specs,
and allow easy path budget and receiver power calculation. In this context I
still have doubts if OMA is the correct means to achieve all of this and add
some further considerations:
Introduction.
It is recognized that for 10Gbit/s operation of in particular directly
modulated lasers, widening the "window" of operation compared to the
"conventional" set of parameters is very important for achieving the lowest
cost by increasing yield of 10G transmitters. Parameters/requirements which
should be considered for relaxation are the transmitter extinction ratio and
the eye mask specifications.
The introduction of the Optical Modulation Amplitude (OMA) parameter was
intended to achieve the above mentioned objective.
Looking at the latest draft of the 10GbE specification, this window of
operation has been widened, however at the expense of an easy verification
of conformance and sacrifice of an in principle transparent optical power
budget specification, traditionally used for optical interface
specifications.
As testing of conformance is one of the key additives of cost of optical
devices (very often underestimated), it is suggested to carefully
(re)consider the changes introduces.
Transmitter specification on power/OMA/extinction ratio.
As stated above the minimum extinction ratio of 6 dB currently in use in ITU
G.691 for directly modulated transmitters in the 1310 nm could prove too
stringent, not allowing a substantial amount of usable transmitters. To this
extent the OMA principle was introduced to allow a widening of the range of
usable devices. Initially the extinction ratio minimum limit was completely
removed which however would have lead to unacceptably low extinction ratios.
E.g. the minimum OMA spec of 477 uW with a max Pav of +1 dBm (1.25 mW) would
imply a minimum extinction ratio of 68% or 1.7 dB. Therefore a minimum
extinction ratio of initially 3 dB and later 4 dB was introduced.
Now the transmitter power and associated modulation setting is specified by
max average power (+1 dBm or 1.26 mW), minimum OMA of 477 uW (leaving out
the "correlation" with spectral characteristics for the time being) and a
minimum extinction ratio of 4 dB.
In practice this means that a minimum OMA spec of 477 uW is valid between
-6.2 and -2.5 dBm average power and that a minimum ER spec 4 dB is valid
between an average power of -2.5 and +1 dBm. One of the reasons to introduce
the OMA spec is to allow settings well above laser threshold current. This
means that average powers of -5 dBm or lower will most likely not be used
because those would imply an extinction ratio of better than 8.5 dB. Even at
-4 dBm average power an extinction ratio of 6 dB minimum is implied. Higher
minimum extinction ratios are not considered practical.
If this is the case then there is no reason to completely abandon the
"ITU-style" of power budget by specifying "only" an average power range and
a minimum extinction ratio. The real request is to allow lower extinction
ratios, which makes sense. So instead of changing the complete way of
specification one could just add a minimum OMA spec to the minimum
extinction ratio spec and the same result of increasing the transmitter
yield is achieved.
So one proposal could be to specify an output power range of -4/+1 dBm with
a minimum extinction ratio of 4dB AND a minimum OMA of 477 uW.
Receiver specification.
The introduction of OMA at the receiver is dramatic from a verification
point of view. While measuring OMA (and also Extinctio Ratio) at the
transmitter is already not quite accurate, this is even more difficult at
the receiver, because of the much lower power levels (at least 10dB). The
only accurate way to derive a receiver sensitivity (at a certain BER) is to
measure BER-curves (BER vs. Power Level), making use of automated testing in
combination with inter/extrapolation methods, using reference transmitters
with (pre)defined pulse patterns, eye mask conditions, extinction ratio,
etc.
A final question is , if "regular" receiver sensitivity is "informative"
what is the value of specifying it?
Regards Juergen Rahn
> ----------
> Von: Del Hanson[SMTP:d_hanson@tripath.com]
> Gesendet: Montag, 23. April 2001 16:31
> An: stds-802-3-hssg-serialpmd@ieee.org
> Betreff: RE: OMA vs. OMA/2 poll
>
>
> Serial PMDers:
>
> I prefer option a). However, as I state in my D3.0 comments, the OMA
> concept
> continues to have problems in referencing an optical power budget for link
> analysis.
>
> Regards,
> Del Hanson
>
> -----Original Message-----
> From: David Kabal [mailto:dkabal@picolight.com]
> Sent: Friday, April 20, 2001 12:09 PM
> To: stds-802-3-hssg-serialpmd@ieee.org
> Subject: OMA vs. OMA/2 poll
>
>
>
> Serial PMDers:
>
> I volunteered to set up an informal poll (based on the lack of
> participation
> at the last Serial PMD conference call: we lacked quorum) on whether to
> specify in Clause 52, for example, Receive Sensitivity for each of the
> PMDs,
> in OMA or OMA/2.
>
> Background:
> Originally, all values in the tables and receive sensitivity were OMA/2.
> As
> of D3.0, we voted to change everything to OMA, but the triple trade off
> tables still have both OMA and OMA/2 values listed.
>
> Discussion:
> There has been some discussion on this on the reflector, see thread:
> http://www.ieee802.org/3/10G_study/public/serial_adhoc/email/msg00228.html
>
> I propose that we have a single solution for all instances in Clause 52.
>
> a) specify everything in OMA
> b) specifiy everything in OMA/2
> c) specify everything in both (very difficult to read)
>
> Please post your comments and input to the reflector.
>
> Cheers,
> Dave
> ------
> David Kabal
> Picolight
>
> Phone: 303-530-3189 ext. 272
> Fax: 303-527-4968
>