[SI-LIST] Re: Ground vias and the land of ID ("It Depends")
- From: Scott McMorrow <scott@xxxxxxxxxxxxx>
- To: "Loyer, Jeff" <jeff.loyer@xxxxxxxxx>, "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
- Date: Wed, 22 Oct 2014 10:35:37 -0400
Jeff
Very nice work. Exactly what is predicted. It gets better with
differential vias, due to field cancellation, but when you are looking at
TX to RX via aggression at the receiver side on a platform designed for
maximum reach (highest loss) you need something better than -60 dB
isolation. That's 1 mV/ V at 10-to13 Gbps. For 28G we need even better
than that. 500 uV/V should be the target I try to design to a NEXT target
of -70 dB for Tx to Rx aggression and a FEXT target of -55 to -60 dB for
Rx-to-Rx and Tx-to-Tx aggression. If designed cleverly, the additional
grounds do not need to take up much additional routing room.
In your Single ended simulation, it looks like you are using a pretty fast
rise time, but of course, DDR signals often have very fast rise times.
With no adherence to grounding you can see 4% crosstalk. Seems small,
until you realize that there can be multiple simultaneous aggressors in the
vicinity, all in phase, that aggress against a victim. With 40 mV/V of
crosstalk it does not take long before you start accumulating 100's of mV
of noise ( and the associated jitter) at a receiver. I am currently working
a design where a top ground was used, but at the memory modules, there were
no through vias to attach the top ground to the underlying planes, and
there were no ground stitch vias within a fairly large area between the
controller and the module. Interesting things happen. Needless to say,
there was a 50% bounce in a quiescent signal under SSO conditions, part of
which was crosstalk, part of which was true GND plane bounce.
BTW, if you hit the no-xtalk ground via design with a fast enough edge
rate, you will see ringing. In your case, it is most assuredly out of
band, but it is quite possible to design via patterns, and ball patterns
that have built in resonances, in spite of the grounds.
Scott
Scott McMorrow
Teraspeed® Consulting - A Division of Samtec
16 Stormy Brook Rd
Falmouth, ME 04105
(401) 284-1827 Business
http://www.teraspeed.com
On Wed, Oct 22, 2014 at 8:55 AM, Loyer, Jeff <jeff.loyer@xxxxxxxxx> wrote:
> Oops - bad link. Please try this one.
> https://www.filesanywhere.com/fs/v.aspx?v6e6587616373beac6b
>
>
> Jeff Loyer
>
>
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
> On Behalf Of Loyer, Jeff
> Sent: Wednesday, October 22, 2014 5:45 AM
> To: si-list (si-list@xxxxxxxxxxxxx)
> Subject: [SI-LIST] Ground vias and the land of ID ("It Depends")
>
> At the risk of finding myself embroiled in a furball, I thought I'd see if
> I could clear up (in my own mind anyway) some of the effects of ground
> vias. I ran 3-D simulations on a pair of single-ended signal vias with
> various configurations of ground vias around them and compared impedance,
> NEXT, and FEXT. The results strongly suggest that nearby ground vias are a
> good investment to reduce via crosstalk. Of course, if you can absorb the
> extra crosstalk, it's a moot point. And ground vias are often hard to put
> where we want them, since they impede routing on every layer, so we seldom
> have the luxury of all the vias we might like (the "no xtalk" configuration
> in the study).
> Here's a link to the study.
> https://www.filesanywhere.com/fs/v.aspx?v<6e658761616dab72a2
>
> Note: I'm leaving the presentation a bit vague (leaving out any
> conclusions, many exact dimensions, and not stipulating simulation
> assumptions) on purpose - this should not be regarded as any formal
> "report-out". It's only a "quick-and-dirty" peek at the issue. Others may
> duplicate the simulation with different results, though that seems
> improbable - I don't see anything leading me to believe they are
> fundamentally flawed. They are not overly complex topologies, so I invite
> others to replicate the simulations if the results seem questionable.
>
> And, of course, exact results will depend on design specifics (stackups,
> via dimensions, etc.). Your mileage may vary...
>
> Cheers,
> Jeff Loyer
>
>
> ------------------------------------------------------------------
> 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 forum is accessible at:
> http://tech.groups.yahoo.com/group/si-list
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
>
> 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 forum is accessible at:
> http://tech.groups.yahoo.com/group/si-list
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
>
> 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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
