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

Re: Equalization and benefits of Parallel Optics.

It is quite possible to get rid of skew between fibers as a factor that limits
distance. I speak with some authority here, since our standards committee
(NCITS T11.1) has a 12-channel parallel 6.4 Gbit networking standard 
(HIPPI-6400) that can compensate for up to 10 nanoseconds of skew.
This is done dynamically, with timing adjusted every 10 usecs.
Works great. We don't care much about skew any more. Jitter is our big
distance limiter since we have a discreet clock and our jitter budget must be
divided between two signals (clock and data).

However, that said, I don't see why skew is much of a concern for 
parallel optics that consist of individual serial streams, each with an
embedded clock. The maximum (allowed) skew can be calculated and
logic in the receiver must be sized appropriately to sort out the data
streams and pass on properly sequenced data.

----- Original Message ----- 
From: <NetWorthTK@xxxxxxx>
To: <NetWorthTK@xxxxxxx>; <Ali.Ghiasi@xxxxxxxxxxx>; <stds-802-3-hssg@xxxxxxxx>; <bradley.booth@xxxxxxxxx>
Sent: Friday, July 28, 2000 12:37 AM
Subject: Re: Equalization and benefits of Parallel Optics.

> Further Comment:
> The parallel technique will add more skew than a single fiber to further 
> restrict the distance and cost.
> Regards, 
> Ed Chang
> NetWorth Technologies, Inc.
> << 
>  Ghiasi:
>  If you have 75% support for 6-PMDs to include parallel-interconnect, I will 
>  vote for parallel, since I support all VCSEL technologies.  Otherwise, 
> 5-PMDs 
>  is sufficient already.
>  The Parallel approach is mainly for up to 20 meter connections.  It is not 
>  designed for 100 meter to go through ducked, or underfloor pipe 
>  installations, because a ribbon fiber is not jacketed enough for those 
> rough, 
>  punishing pulling environment.  Furthermore, at the patch panel connections, 
>  the fibers are all single (duplex fibers) fibers, but not 4-parallel 
> (duplex) 
>  fibers.  For a parallel fiber to connect to an existing single (duplex) 
> fiber 
>  at the patch panel, one has to perform field termination, to which a 
> parallel 
>  fiber is not designed for due to the tight tolerance of spacing between 
>  adjacent channels.  Normally, the parallel ribbon fiber cable is factory 
>  terminated only.  
>  However, if the parallel fibers are used just as a jumper cable to 
>  interconnect closely located nodes -- 5 meter, 10 meter,-- the ribbon cable 
>  can do the job.  Then, how about the serial 850 nm approach, which is 
>  cheaper, and easier eventually to reach more than 20 meters?
>  I was a member of OETC consortium in early 1991, which promoted the parallel 
>  interconnect in industry with the blessing from ARPA.  The project failed 
>  several years later due to the lack of interest from industry.  The reason 
>  was too expensive, difficult in termination and alignment, and expensive 
>  ribbon cables.
>  I was a big fun for, the industry first commercial parallel interconnect, 
>  OCTOBUS.  I tried very hard to implement to my company's equipment.  After 
>  several years, the product never reach production stage, and was canceled, 
>  due to the lack of interest from industry.  The reason was the same as OETC.
>  There was only two ribbon cable suppliers and was expensive that time.  The 
>  factory only termination was very inconvenient for users.  It implies there 
>  is no flexibility in modifying the cable lengths, when an equipment, or 
>  terminals are rearranged to a different location.  One has to go back to 
>  order new set of cables?
>  For last 10 years, parallel interconnect was highly valued; however, it was 
>  never motorized as a contender for the top interconnect solutions.  I hope 
> it 
>  will this time?
>  Regards, 
>  Ed Chang
>  NetWorth Technologies, inc.
>   >>