[SI-LIST] Re: plane-to-plane decoupling

  • From: Scott McMorrow <scott@xxxxxxxxxxxxx>
  • To: Chris Cheng <Chris.Cheng@xxxxxxxx>
  • Date: Tue, 16 Sep 2008 11:47:31 -0400

A) I'm a real believer that if you're going to do the sort of silliness 
that requires diff-pairs to cross split planes, the results should be 
tested with simulation and lab measurements.  To be conservative, I'd 
make sure that one cap is adjacent to each side of a diff pair, so N + 1 
caps are required, where there are N differential pairs.  It would be 
simple enough to put a board like that in an EMI chamber, selectively 
depopulate caps, and see what the incremental results are with different 
ratios.  No matter what, I'd always have at least two caps on the left 
and right of the diff pairs crossing the split to close off the split 
and contain the slot mode.
The other solution that can also provide significant radiated common 
mode attenuation is Lee's approach using thin dielectrics.  Thin 
dielectrics between power and ground will decrease the impedance of the 
power/ground plane pair, thus increasing loss/attenuation of the common 
mode that is left to rattle around.  This approach does nothing to 
effect transfer of common mode from one side of the split to the other, 
since there is no path for this to occur. (The slot cuts off that 
path).  Bypass capacitors will help to shunt lower frequency common mode 
rattles, but have no impact on GHz noise, which will rattle on until 
they are dissipated through loss or radiation.

B)  With a two layer board, placing an in-line return path capacitor 
across a split is trivial. With a multi-layer board, the capacitors 
would be placed to the side of the diff pairs. One might want to try to 
alter the geometry of the split to guide the return path across the 
capacitor bridges on either side of the diff pair.  (I suspect that 
there are some modifications that could be made which might be 
beneficial.) The closer the split is to the capacitors in the z-axis, 
the better.  Yes, the loop inductance will be higher, but it can be made 
lower than the shunt path. There is an advantage to guiding some of the 
common mode across the split, if the receiver can terminate common mode, 
even if imperfectly.  If the receiver does not have any common mode 
termination, then who cares?

C) nope.  Who cares if common mode is reflected back.  The differential 
mode is still transmitted across the split with negligible loss for real 
splits.


Scott McMorrow
Teraspeed Consulting Group LLC
121 North River Drive
Narragansett, RI 02882
(401) 284-1827 Business
(401) 284-1840 Fax

http://www.teraspeed.com

Teraspeed® is the registered service mark of
Teraspeed Consulting Group LLC



Chris Cheng wrote:
> Couple of questions I am curious :
>
> a) The original question refer to multiple diff pairs. Assuming a series AC 
> cap is good enough for 1/2 of a pair. Does that mean it is good enough for 
> multiple diff pairs ? If not, what is the correct ratio ?
> b) The series AC cap is inline with the signal path. How does one put a 
> return plane AC cap inline with the signal path ? If the cap is offset to the 
> side, what does the loop inductance impact will be ? Extrapolate that to 
> multiple pairs per cap, how does the loop inductance look like ?
> c) Finally, do we really care that much of the common mode return ? How does 
> the differential mode impact look like ?
>
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx
> [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of Vinu Arumugham
> Sent: Monday, September 15, 2008 4:02 PM
> To: leeritchey@xxxxxxxxxxxxx
> Cc: Scott McMorrow; Michael Rose; si-list
> Subject: [SI-LIST] Re: plane-to-plane decoupling
>
>
>
> Lee, Scott,
> I don't see the difference whether you want to look at the capacitor as
> a series or shunt element. A decoupling capacitor may look like a shunt
> element when it is part of a PDN but that PDN could serve as a series
> element in the signal return path. So, if a capacitor is acceptable in
> an AC coupling role in the signal path, the same capacitor should be
> acceptable as part of a PDN that is a return path for that signal. In
> other words, think of it as an AC coupling capacitor for the return path
> instead of the signal path (US Patent 7262974).
>
> For this application, the capacitor only needs to support ~10mA of
> switching current at 1.5GHz, and a few tens of mV drop across its
> impedance would be acceptable.
>
> Thanks,
> Vinu
>
> Lee Ritchey wrote:
>   
>> Scott,
>>
>> I suspect you are right, but the thread was about decoupling power
>>     
> planes.
>   
>>
>>     
>>> [Original Message]
>>> From: Scott McMorrow <scott@xxxxxxxxxxxxx>
>>> To: <leeritchey@xxxxxxxxxxxxx>
>>> Cc: Vinu Arumugham <vinu@xxxxxxxxx>; Michael Rose <mrose@xxxxxxxxxxxx>;
>>>
>>>       
>> si-list <si-list@xxxxxxxxxxxxx>
>>
>>     
>>> Date: 9/15/2008 12:53:27 PM
>>> Subject: [SI-LIST] Re: plane-to-plane decoupling
>>>
>>> Lee
>>> I believe  that Vinu is speaking of using a capacitor as a series pass
>>> element, rather than as a shunt element.
>>>
>>> Scott
>>>
>>> Scott McMorrow
>>> Teraspeed Consulting Group LLC
>>> 121 North River Drive
>>> Narragansett, RI 02882
>>> (401) 284-1827 Business
>>> (401) 284-1840 Fax
>>>
>>> http://www.teraspeed.com
>>>
>>> Teraspeed. is the registered service mark of
>>> Teraspeed Consulting Group LLC
>>>
>>>
>>>
>>> Lee Ritchey wrote:
>>>
>>>       
>>>> Vinu,
>>>>
>>>> Can you show me a capacitor that works at 3.125 Gb/S for decoupling?
>>>>
>>>> Lee
>>>>
>>>>
>>>>
>>>>     ----- Original Message -----
>>>>     *From:* Vinu Arumugham <mailto:vinu@xxxxxxxxx>
>>>>     *To: *Scott McMorrow <mailto:scott@xxxxxxxxxxxxx>
>>>>     *Cc: *Lee Ritchey <mailto:leeritchey@xxxxxxxxxxxxx>; Michael Rose
>>>>     <mailto:mrose@xxxxxxxxxxxx>; si-list
>>>>         
> <mailto:si-list@xxxxxxxxxxxxx>
>   
>>>>     *Sent:* 9/15/2008 12:24:44 PM
>>>>     *Subject:* Re: [SI-LIST] Re: plane-to-plane decoupling
>>>>
>>>>     Scott,
>>>>
>>>>     I was not suggesting that capacitors connecting split planes were
>>>>     a "clean" solution. I just wanted to point out that Lee's
>>>>     statement, "There are no capacitors that work at 3.125 Gb/S for
>>>>     decoupling.", is not entirely true.
>>>>
>>>>     Thanks,
>>>>     Vinu
>>>>
>>>>
>>>>     Scott McMorrow wrote:
>>>>
>>>>         
>>>>>     Vinu
>>>>>
>>>>>     Not quite.  As long as there is a ground plane underneath, and
>>>>>     close to, the capacitor, some return path energy will get
>>>>>     across.  But, there is a mismatch in impedance between the
>>>>>     capacitor and plane, and here is still an inductive loop for the
>>>>>     return energy to get to the capacitor.  Because of this, quite a
>>>>>     bit of the common mode energy will be reflected back into the
>>>>>     near end power/ground plane cavity.
>>>>>
>>>>>     regards,
>>>>>
>>>>>     Scott
>>>>>
>>>>>     Scott McMorrow
>>>>>     Teraspeed Consulting Group LLC
>>>>>     121 North River Drive
>>>>>     Narragansett, RI 02882
>>>>>     (401) 284-1827 Business
>>>>>     (401) 284-1840 Fax
>>>>>
>>>>>     http://www.teraspeed.com
>>>>>
>>>>>     Teraspeed. is the registered service mark of
>>>>>     Teraspeed Consulting Group LLC
>>>>>
>>>>>
>>>>>
>>>>>     Vinu Arumugham wrote:
>>>>>
>>>>>           
>>>>>>     Lee,
>>>>>>     The capacitor used for AC coupling on the signal path, should be
>>>>>>
>>>>>>             
>> able
>>
>>     
>>>>>>     work just as good if it were placed on the return path as a
>>>>>>
>>>>>>             
>> decoupling
>>
>>     
>>>>>>     capacitor for that signal.
>>>>>>     Each signal trace will of course need a dedicated capacitor to
>>>>>>
>>>>>>             
>> avoid
>>
>>     
>>>>>>     additional crosstalk.
>>>>>>
>>>>>>     Thanks,
>>>>>>     Vinu
>>>>>>
>>>>>>     Lee Ritchey wrote:
>>>>>>
>>>>>>
>>>>>>             
>>>>>>>     Michael,
>>>>>>>
>>>>>>>     There are no capacitors that work at 3.125 Gb/S for decoupling..
>>>>>>>
>>>>>>>               
>> The way to
>>
>>     
>>>>>>>     provide this path is by placing the planes close to each other..
>>>>>>>
>>>>>>>               
>> I use 3
>>
>>     
>>>>>>>     mils all of the time for this purpose.  Works greast!
>>>>>>>
>>>>>>>     Lee Ritchey
>>>>>>>     Speeding Edge
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>>     [Original Message]
>>>>>>>>     From: Michael Rose <mrose@xxxxxxxxxxxx>
>>>>>>>>     To: SI-List <si-list@xxxxxxxxxxxxx>
>>>>>>>>     Date: 9/15/2008 10:01:17 AM
>>>>>>>>     Subject: [SI-LIST] plane-to-plane decoupling
>>>>>>>>
>>>>>>>>     I looking for some suggestions regarding decoupling between
>>>>>>>>
>>>>>>>>                 
>> co-planar
>>
>>     
>>>>>>>>     plane splits. I'm working on a backplane with a number of
>>>>>>>>
>>>>>>>>                 
>> 3.125Gbps diff
>>
>>     
>>>>>>>>     pairs. I've specified a dual stripline stackup assigned as
>>>>>>>>
>>>>>>>>                 
>> follows:
>>
>>     
>>>>>>>>     1 - P
>>>>>>>>     2 - G
>>>>>>>>     3 - S
>>>>>>>>     4 - S
>>>>>>>>     5 - P
>>>>>>>>     6 - G
>>>>>>>>     and so on
>>>>>>>>
>>>>>>>>     Some diff pairs on L4 will cross power plane splits (actual
>>>>>>>>
>>>>>>>>                 
>> different
>>
>>     
>>>>>>>>     power sources and loads) and I wanted to provide an effective
>>>>>>>>
>>>>>>>>                 
>> AC path
>>
>>     
>>>>>>>>     for any common-mode return currents. I was thinking about
>>>>>>>>
>>>>>>>>                 
>> placing some
>>
>>     
>>>>>>>>     nearby decoupling caps from plane-to-plane across the split.
>>>>>>>>                 
> Do
>   
>> you
>>
>>     
>>>>>>>>     think it would be better to decouple from plane-to-ground on
>>>>>>>>
>>>>>>>>                 
>> both sides
>>
>>     
>>>>>>>>     to steer the current through the L6 ground layer? L5 and L6
>>>>>>>>                 
> are
>   
>> already
>>
>>     
>>>>>>>>     coupled through the inter-plane capacitance (they're about
>>>>>>>>
>>>>>>>>                 
>> 4mils apart).
>>
>>     
>>>>>>>>     Which will provide a lower inductance path?
>>>>>>>>
>>>>>>>>     Mike
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>> ------------------------------------------------------------------
>>
>>     
>>>>>>>>     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
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>             
>>> ------------------------------------------------------------------
>>> 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
>
>
>
>
> This email and any attachments thereto may contain private, confidential, and 
> privileged material for the sole use of the intended recipient. Any review, 
> copying, or distribution of this email (or any attachments) by others is 
> strictly prohibited. If you are not the intended recipient, please contact 
> the sender immediately and permanently delete the original and any copies of 
> this email and any attachments thereto.
> ------------------------------------------------------------------
> 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: