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

Re: 8b/10b and EMI



Hi Ed

All those that I know in fibre channel using GBICs has had problems when
the idle signal is sent.   I do not know yet whether 8B/10B is the culprit, but
idle signal sure seems to have accentuated the problem compared to a random data stream.

Has anyone done a spectral analysis of 46b/66b vs 8b/10b with idle
characters continuous?

Thanks

ron miller

Edward Chang wrote:

Tom, Rick and all:

If the only reason to scramble the 8B/10B code is to minimize the
probability of EMI emission caused by the occasional, repetitive IDLE
signal, we may have to ask ourselves a question: have we done enough home
work to prove it is required?  Even a simple circuit, it is not free.

So far, in the real industry-wide installations, no one has the 8B/10B IDLE
EMI problem.  Furthermore, no one has proved that 8B/10B IDLE signals will
cause EMI problem for 10 GbE in an enclosed environment.

Regards,

Edward S. Chang
NetWorth Technologies, Inc.
EChang@xxxxxxxxxxxxxxxx
Tel: (610)292-2870
Fax: (610)292-2872

-----Original Message-----
From: owner-stds-802-3-hssg@xxxxxxxx
[mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of Rick Walker
Sent: Thursday, March 16, 2000 11:03 PM
To: stds-802-3-hssg@xxxxxxxx
Subject: Re: 8b/10b and EMI

Dear Tom,

Tom Truman <truman@xxxxxxxxxx> writes:
> If 8b/10b were to be scrambled, then it would appear to me that all it
> is providing at the XAUI interface is packet delineation and some
> error monitoring capability.  I imagine that each lane would need a
> separate scrambler/descrambler, initialized to different states so
> that the transitions across the lanes are uncorrelated.  Synchronizing
> these scramblers, and deskewing the lanes would require some thought
> -- it isn't difficult, but it isn't as straightforward as the
> "alignment column" proposed for HARI.

It's not as bad as you think.

The scrambling is done *prior* to 8b/10b encoding, so that the full
run-length and DC-balance properties are preserved.

The scramblers would be randomized by the data itself, and no special
effort would be required to de-correlate them.

> At that point, the 25% overhead of the 8b/10b scheme seems to be a
> staggering price to pay for delineation and error monitoring -- why
> not start with scrambling, at a lower baud rate, and make the overall
> design problems simpler?

Because the data is scrambled *prior* to coding, the benefits of 8b/10b
are not lost.  The net result is that the spectral properties are improved
at the cost of some added circuitry.
--
Rick Walker

-- 
Ronald B. Miller  _\\|//_  Signal Integrity Engineer
(408)487-8017    (' 0-0 ') fax(408)487-8017                 
     ==========0000-(_)0000=========== 
Brocade Communications Systems, 1901 Guadalupe Parkway, San Jose, CA  95131
rmiller@xxxxxxxxxxx,  rbmiller@xxxxxxxxxxxx