[SI-LIST] Re: Do you really ship products at BER 10e-xx ?
- From: "Tom Waschura" <tom_waschura@xxxxxxxxxxxxxxxxxxxxx>
- To: <si-list@xxxxxxxxxxxxx>
- Date: Tue, 12 Apr 2005 17:58:57 -0700
Andy,
Errors in backplanes are typically correlated to the data patterns because
of the high frequency rolloff and nulls in the traces, vias, connectors,
etc. This means that bit errors are pattern dependent and that your error
location probabilities are therefore only as predictable as your
data--typically not predictable at all (although coding could be used to
optimize this). A single-bit FEC (Fire code, hamming code) is great for
boosting random error rate; however, errors on a backplane are not random.
This might mean that you could have "killer patterns" that cause system
failures....
Tom
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Andy Pedler
Sent: Tuesday, April 12, 2005 3:36 PM
To: Bradley.S.Henson@xxxxxxxxxx
Cc: Chris.Cheng@xxxxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Do you really ship products at BER 10e-xx ?
This is actually right-on topic with a design problem that I'm
investigating. Here's what I require, and maybe someone can suggest
something.
I need a relatively high-speed serial link; let's say 1 Gbps, but if I
can run 2.5 Gbps it will save me cost in another part of the design.
I'd like to run over a backplane, but the design may simply be
board-to-board connectors. It could also be 1-2 foot cables (perhaps
Infiniband type cables). It's a theoretical exercise at this point.
But I can certainly live with 1 Gbps. I can add forward error
correction into my data that is traversing this link, so I can live with
an occasional *single* bit error that comes along once in a blue moon.
But my system will crash and burn if the receiver ever gets a continuous
stream of errors. So I would be happy with a predictable BER of even
1E-7 or 1E-9, so long as the errors are single bit and correctable. But
even 1E-20 is bad if the errors show up in huge numbers all at once.
When I've talked to serdes vendors about how they define BER, I've been
told that these serial links typically operate error free, but every so
often for whatever reason (Chris's cosmic ray), a PLL might get just out
of sync and have to re-lock, and when that happens you get a ton of
errors all at once. Obviously, that will kill my system.
I've built chassis systems with 1 Gbps backplanes and run them for weeks
at a time without recording any errors. But that still doesn't make me
extremely confident that I would *never* see a problem. This system
would have to run for months at a time, and a hiccup would cause a lot
of problems.
Any thoughts?
Andy Pedler - Greenfield Networks
Henson, Bradley S wrote:
> This could make an interesting topic. I have to say that in general, I
> have noticed the same trend: Links work so well the BER is hard to
> determine (lots of test time or link-stress)-or- the links are totally
> messed up. However, I did get called in to troubleshoot a Fibre channel
> application that was just marginal on some of the links. By that I mean
> they would almost make the spec 1E-12 BER sometimes, but usually fell
> short. Some days they operated considerably poorer than 1E-12, but not
> pure garbage.=20
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ
List technical documents are available at:
http://www.si-list.org
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ
List technical documents are available at:
http://www.si-list.org
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
