[SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs

  • From: Scott McMorrow <scott@xxxxxxxxxxxxx>
  • To: Lee Ritchey <leeritchey@xxxxxxxxxxxxx>
  • Date: Tue, 20 Oct 2009 11:18:57 -0400

I'd like to refine the discussion just a bit.  There are two separate 
things to consider regarding differential pair skew.
1)  Skew introduces differential to common-mode conversion and thereby 
reduces received signal differential amplitude. If you have ample margin 
this is not an issue.  The impact is easily measured, simulated, and 
estimated.

2) Differential skew upon entry into non well confined structures (vias, 
connectors ... etc) will cause common mode energy to be injected into 
planes, shields, and space.  This may or may not be a problem, but at 
the least constitutes a loss of signal amplitude available to the receiver.


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



Lee Ritchey wrote:
> Actually, I was being conservative at 100 mils.  A mismatch of 300 mils
> total is acceptable throughout the total path.  300 mils is roughly 50
> pSec.out of a 320 pSec bit interval.
>
>
>   
>> [Original Message]
>> From: Peterson, James F (EHCOE) <james.f.peterson@xxxxxxxxxxxxx>
>> To: <si-list@xxxxxxxxxxxxx>
>> Date: 10/20/2009 6:09:07 AM
>> Subject: [SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs
>>
>> One comment on length matching at 100 mils : a lot of times the board
>> we're looking at is only one section of the interface. There is often
>> two more boards involved (a backplane and endpoint). When we say 100
>> mils matched lengths, are we saying total matched length or at each
>> board (so total could be 300 mils in that case)? The thread below says
>> "matched at the receiver", which implies total, so the 100 mils should
>> be budgeted through 3 boards.
>>
>> Regards,
>> Jim Peterson
>> Honeywell
>>
>> -----Original Message-----
>> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
>> On Behalf Of steve weir
>> Sent: Tuesday, October 20, 2009 1:13 AM
>> To: icer world
>> Cc: Lee Ritchey; Paul Hurst; si-list@xxxxxxxxxxxxx; chundi srikanth;
>> Lambert Simonovich
>> Subject: [SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs
>>
>> 1) Matching much closer than 1/4 Tr/Tf offers little value by itself.  
>> 100mils translates to about 16ps skew.  That corresponds to a 64ns Tr/Tf
>>
>> for a 3.125G link which is a little to the outside, but it is not 
>> horrible.  I would rephrase Lee's advice as:  Don't take any 
>> extraordinary measures to match closer than 1/4 Tr/Tf as seen at the 
>> receiver.  The longer the link, the slower Tr/Tf will be.  There is a 
>> good treatment on this in Eric Bogatin's "Signal Integrity Simplified".
>>
>> If you get very tight matching truly for free, then fine.  But obsessing
>>
>> about mechanical match to tight tolerances does not improve the design.
>>
>> It can in fact harm it if the matching is done with dense serpentines 
>> which introduce their own dispersion and timing skew that is not 
>> properly accounted in some tools. 
>>
>> 2) Continuous return path is very important.  Diff tolerates obscenities
>>
>> like crossing moats, but at undesirable costs.  The best answer is don't
>>
>> interrupt the return path.
>>
>> 3) Surface ground guards are more often unintended resonators than of 
>> specific value.  See if the cross-talk can be satisfied with spacing.  
>> If it can't, then consider alternatives.
>>
>> 4) I am not clear on what you are trying to recommend.  Are you talking 
>> about termination at both ends of the link, or even and odd mode 
>> terminations?
>>
>> 5) An appropriate stitch density helps with EMC and signal integrity.  
>> Follow Bruce Archambeault's hierarchy on layer assignments and your life
>>
>> will be good:
>> i. Route on one layer that faces a contiguous plane.
>> ii. Switch between layers on either side of the same contiguous plane.
>> iii. Switch between layers that reference planes on the same DC 
>> potential that are adequately stitched together.  This is often 
>> misunderstood as needing to assign a return stitch via near each 
>> transition.  The point is to raise the resonant frequency of the 
>> structure sufficiently so that it won't be a problem to the signal. 
>> iv. If you must switch between layers that are stitched with bypass caps
>>
>> and planar capacitance.  Be wary of PDN resonances in the signal 
>> frequency range.  These will tend to occur at much lower frequency than 
>> cavity resonances.
>>
>> Steve.
>>
>> icer world wrote:
>>     
>>> 1) length matching must be consided  seriously and mismatch should be
>>>       
>> below 100mils;
>>     
>>> 2) generally speaking,the differential pairs impedance must keep
>>>       
>> 100ohm and   the ground return path must not be choped ;
>>     
>>> 3) the ground guarding trace should keep two times of differential
>>>       
>> trace width away from the differantial pairs,which can not influences
>> the       differential pairs impedance and avoid crosstalk issues;
>>     
>>> 4) using serial and parallel matching resistors simultaneously   for
>>>       
>> debug ;
>>     
>>> 5) changing layers is not expected,but if necessary ,you should place
>>>       
>> a groung via near the signal via ;
>>     
>>> the above is just an advice ,and you'd better do a simulatiom to
>>>       
>> decide the rules of matching resistors and crosstalk;
>>     
>>>
>>> ________________________________
>>> From: Lee Ritchey <leeritchey@xxxxxxxxxxxxx>
>>> To: Paul Hurst <paul@xxxxxxxxxxxxxxxxxxxxxx>; si-list@xxxxxxxxxxxxx
>>> Cc: chundi srikanth <chundis@xxxxxxxxx>; Lambert Simonovich
>>>       
>> <bertsimonovich@xxxxxxxxxx>
>>     
>>> Sent: Mon, October 19, 2009 11:46:36 PM
>>> Subject: [SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs
>>>
>>> Length matching does not need to be tighter than 100 mils.
>>>
>>> It is not necessary to back drill vias used to connect component pins
>>>       
>> to
>>     
>>> traces or to change layers.  The 0.6 pF or so capacitance does not
>>> adversely affect the signals at this data rate.
>>>
>>> It is not necessary to put "ground" vias next to routing vias as long
>>>       
>> as
>>     
>>> the PDS is well designed."
>>>
>>> "Guard"  traces have no value.  Proper spacing to avoid crosstalk
>>>       
>> does. 
>>     
>>> The "rule of thumb" for spacing given, while crude, is not far off.
>>>
>>> Lee Ritchey
>>>
>>>
>>>   
>>>       
>>>> [Original Message]
>>>> From: Paul Hurst <paul@xxxxxxxxxxxxxxxxxxxxxx>
>>>> To: <si-list@xxxxxxxxxxxxx>
>>>> Cc: chundi srikanth <chundis@xxxxxxxxx>; Lambert Simonovich
>>>>     
>>>>         
>>> <bertsimonovich@xxxxxxxxxx>
>>>   
>>>       
>>>> Date: 10/18/2009 5:45:08 PM
>>>> Subject: [SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs
>>>>
>>>> Srikanth,
>>>>
>>>> Additional to Bert's comments:-
>>>>
>>>> 1) Match the lengths of the pairs exactly
>>>> 2) Use a solid ground return
>>>> 3) Avoid stubs by design or by using stub-drilling
>>>> 4) If you have vias in the path use a small drill and a large
>>>>         
>> anti-pad and
>>     
>>>> place a ground via next to each signal via
>>>>
>>>> Regards
>>>>
>>>> Paul
>>>>
>>>> -----Original Message-----
>>>> From: si-list-bounce@xxxxxxxxxxxxx
>>>>         
>> [mailto:si-list-bounce@xxxxxxxxxxxxx]
>>     
>>>>     
>>>>         
>>> On
>>>   
>>>       
>>>> Behalf Of Lambert Simonovich
>>>> Sent: 16 October 2009 21:57
>>>> To: 'chundi srikanth'; si-list@xxxxxxxxxxxxx
>>>> Subject: [SI-LIST] Re: Routing guidelines for 3.12Gbps LVDS pairs
>>>>
>>>>
>>>> Srikanth,
>>>>
>>>> A rough rule of thumb is greater than 3 times trace width separation
>>>>     
>>>>         
>>> between
>>>   
>>>       
>>>> pairs. Using a 2D field solver will ultimately give you the crosstalk
>>>> coupling factor for the exact geometry in your stackup, and dictate
>>>>         
>> the
>>     
>>>> routing rules you need to follow to satisfy your noise budget. 
>>>>
>>>> Adding GND guarding will more often than not present more issues than
>>>>         
>> it
>>     
>>>> solves. You should stitch these guard traces at regular intervals
>>>> approximately 1/10 of the wavelength of the highest frequency
>>>>         
>> component of
>>     
>>>> the aggressor signal to avoid it from resonating and coupling noise
>>>>         
>> back
>>     
>>>> onto other adjacent traces. This further reduces routing density of
>>>>         
>> the
>>     
>>>> board. By the time you factor in the additional space of one trace
>>>>         
>> width
>>     
>>>> between the guard trace and Diff pair, plus the additional via
>>>>         
>> stitching,
>>     
>>>> you will find you are already at 3 times separation between pairs and
>>>>         
>> you
>>     
>>>> would gain back more real estate.
>>>>
>>>> Regards,
>>>>
>>>> Bert Simonovich
>>>>
>>>>
>>>>
>>>>     
>>>>
>>>> -----Original Message-----
>>>> From: si-list-bounce@xxxxxxxxxxxxx
>>>>         
>> [mailto:si-list-bounce@xxxxxxxxxxxxx]
>>     
>>>>     
>>>>         
>>> On
>>>   
>>>       
>>>> Behalf Of chundi srikanth
>>>> Sent: October-16-09 12:40 PM
>>>> To: si-list@xxxxxxxxxxxxx
>>>> Subject: [SI-LIST] Routing guidelines for 3.12Gbps LVDS pairs
>>>>
>>>> Hi Techies,
>>>> We have a 12-layer board in which we have lot of differential LVDS
>>>>         
>> pairs
>>     
>>>> operating at several hundreds MHz. And we have SERDES signals
>>>>     
>>>>         
>>> (differential
>>>   
>>>       
>>>> TX & RX pairs) operating at CPRI rate-4 i.e., 3Gbps. So can you just
>>>>         
>> share
>>     
>>>> me some inputs on exactly what are the guidelines to be followed
>>>>         
>> while
>>     
>>>> routing these signals. And is GND gaurding between the differential
>>>>         
>> pairs
>>     
>>>> improve the SI?Please share or refer me any documents in which i can
>>>>         
>> get
>>     
>>>> Good information on High-Speed design guidelines.
>>>>
>>>> Thanks
>>>>
>>>> With Best Regards
>>>> Srikanth
>>>>
>>>>
>>>> ------------------------------------------------------------------
>>>> 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
>>>   
>>>
>>>
>>>   
>>>       
>> -- 
>> Steve Weir
>> IPBLOX, LLC
>> 150 N. Center St. #211
>> Reno, NV 89501
>> (866) 675-4630 Business
>> (707) 780-1951 Fax
>> http://www.ipblox.com
>> A Teraspeed Consulting Group LLC Partner
>>
>> Teraspeed Consulting Group
>> 121 North River Drive 
>> Narragansett, RI 02882 
>> (401) 284-1827 Business 
>> (401) 284-1840 Fax 
>>
>> Teraspeed Labs
>> 13610 SW Harness Lane
>> Beaverton, OR 97008
>> (503) 430-1065 Business
>> (503) 430-1285 Fax
>> A Teraspeed Consulting Group LLC Partner
>>
>> http://www.teraspeed.com
>>
>> This e-mail contains proprietary and confidential intellectual property
>> of IPBLOX LLC
>> ------------------------------------------------------------------------
>> ------------------------------
>> Teraspeed(R) is the registered service mark of Teraspeed Consulting
>> Group LLC
>>
>> ------------------------------------------------------------------
>> 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
>   
>
>
>   

-- 
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



------------------------------------------------------------------
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: