[SI-LIST] Re: Ground clearance at connector vias

Jim,

We can get into a back and forth discussion on this but will probably 
not come to a consensus.  There are many different ways to resolve this 
issue.  But I will maintain that if there were no coupling to planes, no 
parallel plate mode conversion will occur.  The fields will remain well 
attached to the via structure.  Unfortunately, fixed pitch PTH vias 
remove one of the design variables from the equation and leave only the 
antipads for tuning.  Yes, I am aware of the effect of the PTH via stub 
and of the advantages of backdrilling (however I am unsure of the 
ability for customers to be able to perform testing after connector 
insertion in a manufacturing environment, where backside bed of nails 
testers are normally used.)  It is also true that with careful sizing of 
the antipads, some of the effect of the via stubs can be tuned out and 
peaked, by creating other resonant structure that form antiresonances 
with the via stub.

My work has been performed through simulation and measurement. 
 Simulation is done with 3D FDTD fullwave simulators which take into 
account the broadband characterisitics of the entire structure.  I am 
not a "fan" of the lumped element modeling approach that FCI has used in 
the past and has published.

As for parallel plate mode conversion, there will always be conversion 
when fields which are attached to vias pass through any set of parallel 
plates. (The vias are always orthogonal to the planes, even in an 
infinitely long transmission line scenario.)  In fact, as the antipad 
relief tends towards infinity  you have the case of a true parallel wire 
transmission line structure embeded in a dielectric media.  As the 
planes approach the vias, and coupling occurs, then there is a deviation 
from a true 2-line structure.  Removing the plane from between the via 
reduces this coupling and returns the structure to something that more 
closely approximates a 2-wire differential transmission line.  This can 
be easily seen in fullwave simulations.

I advocate returning differential backplane via structures to something 
closer to the ideal 2-line, and take this into consideration when 
actually designing the connector and backplane, instead of staying with 
the 2mm pitches of old.


best regards,

scott

-- 
Scott McMorrow
Teraspeed Consulting Group LLC
2926 SE Yamhill St.
Portland, OR 97214
(503) 239-5536
http://www.teraspeed.com





Jim G Roberts wrote:

>Hi Scott,
>        Much of our reseach was joint with FCI using there 10GHz test bed and
>simulators.
>        If I understand correctly, you are indicating that the genaration of 
> the
>parallel plate-mode is due to the anti-pad shape.
>My experience is that the major source of paralle-plate mode comes from 
>bending the
>transmission path. This will occur at the backplane connector since it is
>orthognal.
>The best way of reducing it is to ensure that the planes are shorted as close 
>to
>the pin/via as possible.
>An added problem for press-fit which we use, is that dependent on the layer in 
>use,
>you will end up with a stub due to the length of the pin.
>Mostly this can be seen as additional capacitance but it is also an antennae 
>which
>can launch other modes.
>One of the solutions is to back drill, the other is to ensure that you exit the
>connector at the bottom layer minimising the length of the stub.
>--
>Regards,    __________   James G Roberts
>           /___  ____ |  jrobert@xxxxxxxxxxxxxxxxxxxxx
> Jim      __  / /___/ /  jgroberts@xxxxxxxxxx
>         / /_/ /---| |   Room: BE436, Hilversum
>         \____/    /_/   Tel: +31 35 687 4308 Fax: 5976
>
>Scott McMorrow wrote:
>
>  
>
>>Jim,
>>
>>I'm not exactly sure how your research was performed.  However, I do
>>agree that one must take into account the via spacing and the antipad
>>sizing. The reason for placing an oval antipad (or two circular antipads
>>connected by a rectangular region which works better in CAD systems) is
>>to eliminate any parallel plate mode power injection into the plane
>>cavities.  Yes, it is possible to create a narrow band impedance match
>>with some coupling to the planes, but at the sacrifice of impedance
>>mismatch at higher frequencies.  It is generally better to eliminate the
>>planar power injection with larger antipads that encompass the entire
>>differential via.  This will reduce return loss over a very large
>>frequency range.  An optimization can then be performed with a 3D full
>>wave field solver that uses the via spacing and antipad size as
>>variables and seeks to minimize return loss.  An optimized design with
>>an antipad region encircling both vias can be designed to have better
>>broadband return loss than an optimized design with individual antipad
>>regions around each differential via.  Now, for press fit connectors,
>>the PTH via spacing  cannot be altered, but for surface mount
>>connectors, there is a great latitude of spacing that is possible.
>>
>>regards,
>>
>>scott
>>
>>--
>>Scott McMorrow
>>Teraspeed Consulting Group LLC
>>2926 SE Yamhill St.
>>Portland, OR 97214
>>(503) 239-5536
>>http://www.teraspeed.com
>>
>>Jim G Roberts wrote:
>>
>>    
>>
>>>Hi Scott,
>>>       When you say it is "much better" are you saying that you can achieve a
>>>larger reduction in capacitance than with two circular.
>>>       Along the same lines would it not also be attained by a rectangular 
>>> cut
>>>out which maybe easier to attain in CAD than an oval which will cost many
>>>segments.
>>>       I would also like to add that it is only necessary to reduce the
>>>parastic capacitance in line with the inductance of the pin/via to achieve 
>>>the
>>>correct impedance. Research in Backplane conducted at Lucent with upward of 
>>>30
>>>layers has shown this to be acheivable with standard antipads for 50 ohm (100
>>>ohm balanced) but not for 75 ohm.
>>>
>>>--
>>>Regards,    __________   James G Roberts
>>>          /___  ____ |  jrobert@xxxxxxxxxxxxxxxxxxxxx
>>>Jim      __  / /___/ /  jgroberts@xxxxxxxxxx
>>>        / /_/ /---| |   Room: BE436, Hilversum
>>>        \____/    /_/   Tel: +31 35 687 4308 Fax: 5976
>>>
>>>Scott McMorrow wrote:
>>>
>>>
>>>
>>>      
>>>
>>>>Christian,
>>>>
>>>>The recommendation is for an oval antipad that encompasses both vias of
>>>>a differential pair, not just one.  And, yes, it is much better than two
>>>>circular antipads around each via.
>>>>
>>>>best regards,
>>>>
>>>>scott
>>>>
>>>>--
>>>>Scott McMorrow
>>>>Teraspeed Consulting Group LLC
>>>>2926 SE Yamhill St.
>>>>Portland, OR 97214
>>>>(503) 239-5536
>>>>http://www.teraspeed.com
>>>>
>>>>Christian Schuster wrote:
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>>>Fabrizio and all the others:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>I have a question regarding ground clearances for signal vias in a
>>>>>>backplane connector, like the Molex/Teradyne HSD.  Connector and
>>>>>>semiconductor manufacturers recommend using an oval ground clearance
>>>>>>around the differential pairs, to minimize reflections.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>I completely agree that increasing antipad sizes is one way to reduce
>>>>>parasitic via capacitiance. What makes me wonder is the fact that
>>>>>oval shapes are proposed. Is that a mechanical/space requirement?
>>>>>
>>>>>If not: has anybody ever seen a difference in via transmission between a
>>>>>circular and an oval antipad of the same area?
>>>>>
>>>>>
>>>>>Regards,
>>>>>
>>>>>Christian Schuster
>>>>>IBM
>>>>>
>>>>>
>>>>>
>>>>>------------------------------------------------------------------
>>>>>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 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
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>------------------------------------------------------------------
>>>>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 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
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>------------------------------------------------------------------
>>>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 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
>>>
>>>
>>>
>>>
>>>      
>>>
>>------------------------------------------------------------------
>>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 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
>>
>>    
>>
>
>
>
>  
>




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