Re: [BP] Recommended material and connectors
Ayal-
What Brian failed to state is the most basic of problems with regard to
this problem. When we do twisted pair or fiber transmission PHY standards
we target them at transmission over well-specified standardized
media.
No such luck with backplanes.
The equivalent standards for materials and transmission paths did not
exist when P802.3ap started.
I don't know if that situation has improved in the meantime but, even if
it has, there is a lot of catching up to do.
My guess would be that things haven't improved much because of Brian's
statement:
·
The standard will not
specify or even recommend particular connectors, materials or
construction methods.
I hope this helps.
Geoff
At 11:33 AM 8/22/2005 , Brian Seemann wrote:
Hi
Ayal,
You have asked a very simple, reasonable question. The amount of
time and energy that the 802.3ap contributors spent on this is
staggering. But over the past several years, we have learned a
great deal about the role of materials and connectors in high speed
signal integrity. A large amount of this work has been contributed
in IEEE 802.3ap.
Here are some points that address some aspects of your question:
·
In general, the industry
has moved forward significantly in what can be done with mainstream
materials, connectors and manufacturing processes.
·
The IEEE 802.3ap
standard specifies the performance of the transmitter and the
receiver. This is what will ultimately determine the channels
servable by the standard.
·
The channel model
(Section 69.3) is informative only, to give guidance to backplane
builders for what will likely work and not work with the
transceivers.
·
The standard will not
specify or even recommend particular connectors, materials or
construction methods.
·
Careful layout design
techniques are probably the most performance-effective and the most
cost-effective measures to ensure signal integrity.
·
There are 10Gbps-capable
connectors available from several manufacturers.
·
The performance of good
connectors can be completely obscured by poor layout design
methods.
·
There are multiple
contributions showing successful 10Gbps performance on materials such as
Nelco 4000-13, Nelco 4000-13SI, and Isola FR408. Again the standard
will not make any recommendations about materials.
·
In many cases on the
line cards, lower grade (higher loss) board material (such as 4000-6) has
actually been shown to be preferable to higher grade (lower loss)
material. This is because higher loss can reduce the Q (reflection
effect) of stubs.
·
The performance of good
material can be completely obscured by poor layout design methods.
·
Vias from top layer
traces are probably the most disruptive layout feature to signal
integrity.
·
Backdrilling vias to
eliminate the stub is viewed by many manufacturers as cost
feasible.
Others on this reflector can weigh-in with more insightful or alternative
perspectives.
Brian Seemann
From: Ayal Lior
[mailto:Ayal.Lior@tera-chip.com]
Sent: Monday, August 22, 2005 8:54 AM
To: STDS-802-3-BLADE@listserv.ieee.org
Subject: [BP] Recommended material and connectors
Hi,
I am new to this reflector and have a very basic question.
We would like to know what is the recommended material and connectors for
backplane to drive the 10GBase-KR.
Target application is chassis with maximum length of 40" with two
connectors.
Any feedback is appreciated.
Thanks,
Ayal Lior