[SI-LIST] Re: AC coupling of SATA and SAS I/O
- From: Alief <saifj_m@xxxxxxxxx>
- To: "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>, "T.K. Jeon" <tkjeon@xxxxxxxxxxx>, RichardWard <richard.ward@xxxxxx>
- Date: Tue, 12 May 2009 19:30:30 -0700 (PDT)
Dear all,
There apparently is one impairment that AC coupling does introduce : some
increase in jitter.
The AC coupling causes the signal to droop : the percentage droop is approx
(tw/RC)*100
where tw is the time between data transitions; and,
the added jitter is equal to (tr*tw)/(2*RC)
where tr is the transition time of the signal.
AFN,
Alief
--- On Sat, 5/9/09, Alief <saifj_m@xxxxxxxxx> wrote:
From: Alief <saifj_m@xxxxxxxxx>
Subject: [SI-LIST] Re: AC coupling of SATA and SAS I/O
To: "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>, "T.K. Jeon"
<tkjeon@xxxxxxxxxxx>, "RichardWard" <richard.ward@xxxxxx>
Date: Saturday, May 9, 2009, 4:37 AM
Hi Richard,
Thanks v much for your response and sorry for being a little tardy in getting
back - I was out of town all day and have just got back.
I understand now where you were coming from in terms of your comments regarding
to 0V bias.
Also, I appreciate your comments about hot swap involving uncommon risk -
mainly due to the "human intervention" factor. I too like using AC coupling
and agree that the AC caps can be your friends in preventing transients,
failures, etc. The only thing I would like to add is that having worked with
different types of caps and having been burned by cap failures I wonder about
the reliability of the coupling caps : is it comparable to the reliability of
the silicon? Since these caps are in the signal path and are not redundant
(like bypass caps etc can be); does anyone know of reliability info for these
caps ?
Thanks,
Alief
--- On Fri, 5/8/09, Ward, Richard <richard.ward@xxxxxx> wrote:
From: Ward, Richard <richard.ward@xxxxxx>
Subject: [SI-LIST] Re: AC coupling of SATA and SAS I/O
To: "Alief" <saifj_m@xxxxxxxxx>, "si-list@xxxxxxxxxxxxx"
<si-list@xxxxxxxxxxxxxxx>, "T.K. Jeon" <tkjeon@xxxxxxxxxxx>
Date: Friday, May 8, 2009, 7:24 AM
Hi Alief,
My initial statement was directed at one of the comments below:
"If your transmitter has 0V DC bias, then I guess you don't need to implement
AC coupling".
I read this as a statement of 0V common-mode from the Tx.
A large portion of the Serdes out there and used for these standards are VTT
referenced or GND referenced swings and given the supply ranges for the type of
systems these go into (often somewhere between 0-0.8V to 0-5V), I don't believe
the statement above (or I misunderstand it and someone will correct me).
We're agreed on the function of the cap - but the prevous thread seemed to
contradict that (to me anyway).
The point about the differently referenced systems was for interoperability
yes, but that fact that it allows a wider range of Si to work with those
standards if ac caps are there is a benefit to all - and therefore usually a
good thing for a standard to have.
For hot-swap, one of the protections the circuit designer needs to put in is
for "things to go wrong" and short. This is a failure mode. It may be a
transient, but it may be someone, somehow, managed to stuff the wrong connector
in the slot - my children are experts at that. We could argue this isn't
strictly hot-swap, but it is a precaution many circuit designers take and label
under the hot-swap banner. The diff pairs, being the most sensitive signals are
usually the most difficult to protect. For EM breakdown limits, time is
obviously one of the major parameters. So if it's a short on a high-speed
signal that's ac coupled...we've blocked the major dc component and therefore
reduced the time and the EM risk. That was the only point I was making there -
caps can be your friends in a failure mode.
Regards,
Richard
________________________________
From: Alief [saifj_m@xxxxxxxxx]
Sent: Thursday, May 07, 2009 11:36 PM
To: si-list@xxxxxxxxxxxxx; T.K. Jeon; Ward, Richard
Subject: Re: [SI-LIST] Re: AC coupling of SATA and SAS I/O
Hi Richard,
Thanks a bunch for your response.
I'm really not sure why you found it necessary to state that "AC caps aren't
used to set the Rx bias". Of course, the AC cap does not set the Rx bias. In
fact, how can one expect an AC cap to set a bias ? What the AC cap does is to
remove the common mode level not set it; and then, of course, that permits the
Rx to set the common mode wherever it chooses to.
You have made good points about how ac coupling helps differently referenced
systems work together - I would classify that under interoperability..
Those are all good points worth noting. However, I am afraid I am not aware of
any long term effects of hot swap. Doesn't hot swap only cause transients ?
Assuming there are no application mismatches or design marginalities/errors how
could hot swap result in long term effects ? Would greatly appreciate your
expounding on that. Regards,
Alief
--- On Thu, 5/7/09, Ward, Richard <richard.ward@xxxxxx> wrote:
From: Ward, Richard <richard.ward@xxxxxx>
Subject: [SI-LIST] Re: AC coupling of SATA and SAS I/O
To: "Alief" <saifj_m@xxxxxxxxx>, "si-list@xxxxxxxxxxxxx"
<si-list@xxxxxxxxxxxxxxx>, "T.K. Jeon" <tkjeon@xxxxxxxxxxx>
Date: Thursday, May 7, 2009, 9:48 PM
Hi Alief,
My view is a little different from that below.
AC caps aren't used to set the Rx bias, but to allow the Rx to set it's own
bias point. This is especially useful for rack-to-rack systems where the common
mode cannot be well controlled. It also has the advantages of:
- allowing VTT referenced and GND referenced systems to work together
- allowing a single Serdes to cover multiple standards
- allowing newer (restricted supply) devices to work with legacy devices
- ... (I'm sure someone on this list has written a book or two on this...)
The hot-swap ability doesn't really demand dc blocking caps, but they give
advantages. "hot-swap" is a bit of a nebulous term too, often.
One of the risks of hot-swap is shorting the connections (signal-signal or
signal-supply). Some devices will be designed to withstand the short-term
current only (not long-term lifetime degradation). Having the caps means a
long-term effect, actually becomes a short-term one, therefore EM
considerations are less restrictive on the circuit design side.
I believe PCIE defined them at the Tx for this (shorting) reason (whereas a
majority of non-pluggable applications put them at the Rx side, why? well it
depends).
Regards,
Richard
-----Original Message-----
From:
si-list-bounce@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-bounce@xxxxxxxxxxxxx>
[mailto:si-list-bounce@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-bounce@xxxxxxxxxxxxx>]
On Behalf Of Alief
Sent: 07 May 2009 12:41
To:
si-list@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list@xxxxxxxxxxxxx>;
T.K. Jeon
Subject: [SI-LIST] Re: AC coupling of SATA and SAS I/O
Thank you for your response T.K.
Yes, a 0V DC bias would establish an automatic reference (in theory); thus it
would be akin to DC coupling.
However, I am hesitant to make my common mode 0V; because of issues with ground
loops, ground path parasitics, etc and also because the receiver would see
negative voltages.
I realize that this is to first order a biasing issue and allows for
interoperability.
However, are there other considerations ? I have heard/read that it might
permit hot swappability. Is this true. If I had a known common mode level
doesn't that make it easier for hot swapping ?
Also, are there other issues which AC coupling helps with ?
Thanks a bunch in advance,
Alief
--- On Thu, 5/7/09, T.K. Jeon
<tkjeon@xxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=tkjeon@xxxxxxxxxxx>>
wrote:
From: T.K. Jeon
<tkjeon@xxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=tkjeon@xxxxxxxxxxx>>
Subject: RE: [SI-LIST] AC coupling of SATA and SAS I/O
To: "Alief"
<saifj_m@xxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=saifj_m@xxxxxxxxx>>,
"si-list@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list@xxxxxxxxxxxxx>"
<si-list@xxxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list@xxxxxxxxxxxxxx>>
Date: Thursday, May 7, 2009, 6:55 PM
Alief,
AC coupling is needed to maintain the correct DC bias for receivers. If your
transmitter has 0V DC bias, then I guess you don't have to implement AC
coupling.
TK
-----Original Message-----
From:
si-list-bounce@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-bounce@xxxxxxxxxxxxx>
[mailto:si-list-bounce@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-bounce@xxxxxxxxxxxxx>]
On Behalf Of Alief
Sent: Thursday, May 07, 2009 10:46 AM
To:
si-list@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list@xxxxxxxxxxxxx>
Subject: [SI-LIST] AC coupling of SATA and SAS I/O
Greetings everyone,
I am currently looking at SATA and SAS SerDes designs. It looks like the data
transmission is AC coupled. Is this a standards requirement ?
Could anyone enlighten me on why SATA/SAS specifies only AC coupled links [if
that is in fact the case]; i.e. what was the thinking/justification for
restricting transmission to ac coupled only.
Also, do the other standards : Gbit Ethernet, PCI Express, RapidIO, etc. have
similar restrictions/requirements ?
Thanks v much in advance,
Alief
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=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@xxxxxxxxxxxxxx<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-request@xxxxxxxxxxxxx>
with 'help' in the Subject field
List technical documents are available at:
http://www.si-list..net<http://www.si-list.net/>
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<http://us.mc435.mail.yahoo.com/mc/compose?to=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<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-request@xxxxxxxxxxxxx>
with 'help' in the Subject field
List technical documents are available at:
http://www..si-list.net<http://www.si-list.net/>
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<http://us.mc435.mail.yahoo.com/mc/compose?to=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<http://us.mc435.mail.yahoo.com/mc/compose?to=si-list-request@xxxxxxxxxxxxx>
with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net<http://www.si-list.net/>
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 technical documents are available at:
http://www.si-list.net
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 technical documents are available at:
http://www.si-list.net
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 technical documents are available at:
http://www.si-list.net
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:
