[SI-LIST] Re: Relevance of Common Mode Return Loss

  • From: "Jeon, Tae-Kwang" <Tae-Kwang.Jeon@xxxxxxx>
  • To: "Eric Bogatin" <eric@xxxxxxxxxxxxxxx>, "David Instone" <dave.instone@xxxxxxxxxx>, <olaney@xxxxxxxx>
  • Date: Tue, 20 May 2008 12:23:18 -0600

Hi,

Some industrial committees already have had the SCDxx spec documented in
the standard. You may want to take a look at the FibreChannel, SATA, SAS
and OIF. Although 4G FC has no SCDxx, 8G FC adapted the SCDxx for both
Tx and Rx. As far as I know, the impedance balance spec specified in
SATA in the past has been replaced by SCD spec and the impedance
imbalance spec shown in SAS before has been replace by SCD too from the
most recent revision. OIF left the SCD spec table blank at the moment,
however, it seems they are considering the spec. I'm not quite sure why
PCIE does not have the spec, but it is interesting that all standards
that I mentioned earlier have the data rate of at least 6Gb/s for the
spec. Who knows if the next generation of PCIE spec would have the SCD
spec.

BTW, when it comes to the return loss, we must consider not only the
termination resistor but both the load capacitance and the parasitic
capacitance because the zero frequency and the pole frequency are
determined by RC after all. Therefore, if you have bigger capacitance,
the pole frequency will get lowered which may result in failure to meet
the spec.

Regards,
TK

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On Behalf Of Eric Bogatin
Sent: Tuesday, May 20, 2008 8:01 AM
To: 'David Instone'; olaney@xxxxxxxx
Cc: si-list@xxxxxxxxxxxxx; eric@xxxxxxxxxxxxxxx
Subject: [SI-LIST] Re: Relevance of Common Mode Return Loss

Guys-
=20

If there is so much concern about the presence of the common signal on
the
interconnect, and its rattling around, shouldn't there be a spec on the
amount of common signal allowed at the receiver?

=20

Anyone have a feel for what a reasonable value is, before it starts to
affect the jitter, as Steve Weir pointed out?

=20

Lynn Greene suggested that any common signal present could be converted
back
to diff and screw up the diff signal. If the source of the comm. signal
is
from asymmetries in the interconnect, then isn't this the first order
problem to fix?

=20

Shouldn't there be a spec on the SCD21 performance of the interconnect
to
evaluate the conversion of the differential signal into common signal?
This
is the first order problem, the second order one being the conversion of
"rattling around" common signal back into differential signal, further
screwing up the diff signal.

=20

Steve Weir suggests that a spec for the SCC11 and SCC21 of an
interconnect
is related to the fact it is easy to do, not that it is the most
reasonable
approach. It is sort of like the joke where the punch line is, "because
the
light is better over here" (only if I am publicly encouraged will I
provide
the rest of the joke)

=20

I am still trying to understand the importance of the SCC11 and SCC21
spec,
as opposed to a spec on the magnitude of the common signal, or on SCD21
or
SCD11.

=20

Does anyone have any insight on the discussions that went on at the
committee meetings for the specs?

=20

Thanks

=20

--eric

=20

=20

=20

*******************************************************

Eric Bogatin

Signal Integrity Evangelist

Bogatin Enterprises

Setting the Standard for Signal Integrity Training

26235 W 110th Terr

Olathe, KS  66061

e: eric@xxxxxxxxxxxxxxx

v: 913-393-1305

cell: 913-424-4333

f: 913-393-0929

www.BeTheSignal.com

=20

San Diego: EPSI, BBDP, July 28-31, 2008

San Jose, SICT, Aug 12-13

San Jose, EPSI, BBDP, Sept 29-Oct 2

***********************************************

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On
Behalf Of David Instone
Sent: Tuesday, May 20, 2008 4:16 AM
To: olaney@xxxxxxxx
Cc: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Relevance of Common Mode Return Loss

=20

Orin,

As Lynne pointed out in her posting on this subject, part of  the common


mode energy can get converted to differential.  Surely then, reflecting=20

the common mode back to the Tx, even if the Tx is a good CM match, would


give it twice the opportunity to interfere with the differential.  How=20

much this matters of course depends on how imbalanced the differential=20

lines are and how imbalanced the signal is.=20

Centre tapping the differential terminating resistor, at the Rx, to=20

ground only fully terminates the common mode if there is little coupling


between the lines, if they are coupled then three resistors are=20

required, 1 from each line to gnd to terminate the even mode and 1=20

across the lines, which in parallel with the other two terminates the=20

odd mode,  how necessary this is depends on how tightly coupled the=20

lines are.    Currently SATA and PCIe implementations that I have seen=20

have the Rx termination inside the chip which makes the 3 resistor=20

termination difficult to achieve.  This makes a common and differential=20

return loss specification at the Rx relevant as it enables the traces=20

and cable to be designed to match the termination in both modes.

=20

=20

Regards

Dave Instone

+44 (0)1235 824963

=20

OXFORD SEMICONDUCTOR LIMITED

25 MILTON PARK

ABINGDON

OXFORDSHIRE

OX14 4SH

Registered in England no 2733820

Registered Address: As above=20

=20

=20

=20

olaney@xxxxxxxx wrote:

> Since we don't want them, common mode signals can be treated=20

> differently than for differential mode.  Given that common mode is=20

> undesirable, at the transmit end we often use a deliberate mismatch=20

> (CMC) to reflect this signal back to the transmitter.  This energy can


> be absorbed by the transmitter if there is an adequate common mode=20

> backmatch, or it can be left to ring between the driver and choke if=20

> that is considered harmless.  At the receiver, the intent of providing


> a common mode termination is simply to prevent unwanted CM energy from


> returning up the line, giving it an additional chance to radiate. =20

> If the common mode signal is terminated before the differential signal


> passes through a CMC to reach the DM termination at the receiver, then


> the best of both worlds is achieved: the CM signal is both absorbed=20

> and suppressed.  The receiver common mode range becomes much harder to


> violate.  For coding with a zero at DC (accepts AC coupling), a center


> tapped inductor is an easy way to provide the CM termination:

> =20

> Orin Laney

> =20

> =20

=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:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:    =20
                http://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
 =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:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://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: