[SI-LIST] Re: Damping of Cavity-Mode Resonances in PCB
- From: Scott McMorrow <scott@xxxxxxxxxxxxx>
- To: Jory McKinley <jory_mckinley@xxxxxxxxx>
- Date: Tue, 22 Nov 2011 18:51:15 -0500
Jory,
Lets be clear on the nomenclature. We are not dealing with coupling to the
universe (radiation), we are dealing with cavity coupling. Radiation is a
different phenomena.
When you say one eighth wavelength radial, do you mean that every ground
via is one eighth wavelength from the signal via and each neighboring
ground via, such that there are 6 ground vias surrounding each signal via
at 60 degree increments? If that is the case, then I'd say that whether or
not you have suppressed a cavity resonance is dependent upon the radial
area not covered , rather than the wavelength criteria you specify. I see
these resonances all the time in precision measurements.
Scott
On Nov 22, 2011 5:30 PM, "Jory McKinley" <jory_mckinley@xxxxxxxxx> wrote:
> Hello Scott,
> Agree, one return Via to signal transition does not provide enough
> radiation suppression and might excite the cavity, we also found this out
> through design and simulation of a package with multiple PLL power planes.
> We found that radial Via stitching within the 1/8 wavelength constraint
> was enough to contain the z-direction energy in the cavity. The coupling
> between power plane and signal/Via stitching did not change only the
> potential for cavity excitation. Did not read the paper yet but glancing
> over it is appears to be consistent with our findings.
> -Jory
>
> ------------------------------
> *From:* Scott McMorrow <scott@xxxxxxxxxxxxx>
> *To:* Jory McKinley <jory_mckinley@xxxxxxxxx>
> *Cc:* "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
> *Sent:* Tuesday, November 22, 2011 4:39 PM
> *Subject:* [SI-LIST] Re: Damping of Cavity-Mode Resonances in PCB
>
> Jory
> If you mean 1/8 wavelength in enough directions to suppress coupling
> to the cavity, then I might agree. But if you mean to say that only one
> ground stitch via 1/8 wavelength away from the signal transition via is
> necessary to suppress cavity modes, then you are sorely mistaken. You
> may want to take a look at last year's DesignCon paper that I co-wrote
> with Jason Miller of Oracle and Ashley Rebelo of LSI.
>
>
> http://www.electrical-integrity.com/Paper_download_files/DC11_8-TA3_Miller-paper.pdf
>
> I can tell you that since then we have specifically looked at cavity
> mode excitation in exquisite detail, through both full-wave simulations
> and TRL de-embedded measurements. I'm sitting here looking at overlaid
> measurements and simulations of a 60 micron thick cavity specifically
> designed to have a 12 GHz resonance (wavelength of 83.33 ps) with two
> single-ended nets and vias transitioning through it at a 4 mm spacing.
> The cavity is formed by a bounding box of ground stitch vias. Two
> measured configurations exist. One configuration where there are no
> additional grounds within the cavity. In the other configuration one
> ground stitch via is placed adjacent to each signal via transition at a
> 500 micron pitch. The material has an Er of 3.5 at this frequency,
> making the stitch via less than 3.2 ps away. My calculation says that
> this is approximately 1/26 wavelength away.
>
> If I understand you correctly, you say that the signal vias with stitch
> grounds at 500 micron spacing will not excite the cavity mode at 12
> GHz. My understanding of theory says that there will be cavity mode
> excitation, because coupling to the cavity is not fully contained.
> Modeling and measurements have thoroughly confirmed that the cavity mode
> exists, and is excited by the single-ended via transitions. This cavity
> mode excitation can be seen by measuring crosstalk between the two 4 mm
> spaced differential signals, with subsequent peaking of measured near
> end and far end crosstalk at 12 GHz (about a 20 dB boost from the
> crosstalk noise floor prior to the resonance.) There are additional
> cases we have tested where crosstalk "skips" through a board, aided by a
> regular array of identical cavities.
>
>
> 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
>
>
> On 11/22/2011 3:49 PM, Jory McKinley wrote:
> > Hello Scott,
> > If we keep the return or stitching Via that traverse with signal Via
> > in the z-direction to close enough so as to prevent noise from
> > radiating then by definition we have prevented the cavity from
> > resonating for each such transition, this goes to your (1). Now I
> > contend that I will provide a design guideline or constraint that
> > states place return stitching below 1/8 of the highest energy content
> > possible, this ensures I will not create unwanted resonances. I have
> > done this with very good success.
> > Now, if you are talking about direct coupling noise this is different,
> > isolation cannot be achieved by stitching Via's and require
> > spacing/layout considerations.
> > I have worked on one recent system where we required -100dB of noise
> > isolation and through altering layout, beading we got there.
> > -Jory
> > ------------------------------------------------------------------------
> > *From:* Scott McMorrow <scott@xxxxxxxxxxxxx>
> > *To:* si-list@xxxxxxxxxxxxx
> > *Sent:* Tuesday, November 22, 2011 2:02 PM
> > *Subject:* [SI-LIST] Re: Damping of Cavity-Mode Resonances in PCB
> >
> > Jory
> > If a signal on a via passes through multiple planes of different types,
> > the return path is broken by definition, irrespective of whether or not
> > it eventually ends up at the same reference plane on exit that it was
> > attached to upon entry. If we are trying to isolate the digital ground
> > referenced signals from the analog ground, then the solution is to
> > eliminate cavity excitation by creating a barrier of ground vias around
> > the signal via transition designed to:
> >
> > 1. provide a continuous reference path in 3 dimensions
> > 2. minimize leakage to the cavity.
> >
> > The number of grounds and spacing required for effective isolation is
> > ultimately dictated by the noise requirements of the analog system, not
> > by some arbitrary fraction of a wavelength. I've seen analog systems
> > that require as much as 120 dB of isolation. Any design such as this
> > starts with an analysis of the noise isolation requirements.
> >
> > 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
> >
> >
> > On 11/22/2011 1:31 PM, Jory McKinley wrote:
> > > Hello Cris,
> > > Make sure that your signal paths follow the appropriate return path
> > (digital signals coupled with digital ground and analog signals
> > coupled with analog ground). If you have done this correctly and Via
> > stitched below 1/8 of a wavelength for all z-axis transitions you will
> > not excite any of these cavity resonances assuming not other signals
> > cut through these planes. Now, if your signal return jumps reference
> > for some poor design reason then you will have potential resonance
> > issues in which case redesign your signal returns. In general you can
> > either push your plane resonances above or below critical frequencies,
> > by shape and/or thin dielectric. Increasing your plane shape will
> > shift your resonance lower and decreasing shape will shift resonance
> > higher. You will have the added benefit of resonance peak suppression
> > with a thinner dielectric.
> > >
> > > -Jory
> > >
> > >
> > >
> > > ________________________________
> > > From: Cristian Gozzi<cristian.gozzi@xxxxxxxxx
> > <mailto:cristian.gozzi@xxxxxxxxx>>
> > > To: Brad Brim<bradb@xxxxxxxxxxx <mailto:bradb@xxxxxxxxxxx>>
> > > Cc: steve weir<weirsi@xxxxxxxxxx <mailto:weirsi@xxxxxxxxxx>>;
> > si-list@xxxxxxxxxxxxx <mailto:si-list@xxxxxxxxxxxxx>
> > > Sent: Tuesday, November 22, 2011 11:51 AM
> > > Subject: [SI-LIST] Re: Damping of Cavity-Mode Resonances in PCB
> > >
> > > Hi folks
> > > thanks to everybody for your feedback
> > >
> > > Let me describe better my application!
> > >
> > > I have a backplane with one Digital_GND over one Analog_GND
> > >
> > > of course these 2 nets are separated each others, but due to big plane
> > > shape dimension I got resonance modes starting from 220 MHz
> > >
> > > I don't think I can use decoupling capacitor in this case, so I
> > guess I can
> > > only work with plane shape dimension and dielectric thickness
> > >
> > > My idea is to move/shift at higher frequency these modes.
> > > I don't want to couples these low frequency modes to digital signals
> > that
> > > operates at 400-800 Mbps
> > >
> > > Thanks for any further comments
> > >
> > > Regards
> > > Cris
> > >
> > > 2011/11/21 Brad Brim<bradb@xxxxxxxxxxx <mailto:bradb@xxxxxxxxxxx>>
> > >
> > >> hi Steve,
> > >>
> > >> I did not say it failed to dampen a resonance, I only said it does not
> > >> serve
> > >> to eliminate the resonance. I am familiar with Chris's applications
> > and in
> > >> many cases he may not be able to apply the commercially available thin
> > >> dielectrics, let alone one less than 5um thickness, as cited from
> > Istvan's
> > >> paper. Our comments are each equally correct and equally applicable to
> > >> Chris.
> > >>
> > >> cheers,
> > >> -Brad
> > >>
> > >>> -----Original Message-----
> > >>> From: si-list-bounce@xxxxxxxxxxxxx
> > <mailto:si-list-bounce@xxxxxxxxxxxxx>
> > >>> [mailto:si-list-bounce@xxxxxxxxxxxxx
> > <mailto:si-list-bounce@xxxxxxxxxxxxx>] On Behalf Of steve weir
> > >>> Sent: Monday, November 21, 2011 12:03 PM
> > >>> To: si-list@xxxxxxxxxxxxx <mailto:si-list@xxxxxxxxxxxxx>
> > >>> Subject: [SI-LIST] Re: Damping of Cavity-Mode Resonances in PCB
> > >>>
> > >>> Brad, thin dielectrics most definitely do suppress cavity
> > >>> modal resonances. They do so by reducing the Q. Inductance
> > >>> goes down, but skin resistance remains fixed, and for any dK
> > >>> and tangent loss coefficient dielectric loss increases.
> > >>> There are a number of papers out there that show these
> > >>> characteristics. Istvan published a paper circa
> > >>> 2004 that showed for dK of 4 and tangent loss of 0.02 that
> > >>> around 0.1 -
> > >>> 0.2 mils thickness, cavity Q falls to one. Already having
> > >>> served its
> > >>> primary purpose, an empty beer can still makes a terrific
> > >>> resonant cavity, very thin cavities are quite another matter.
> > >>>
> > >>> Best Regards,
> > >>>
> > >>>
> > >>> Steve.
> > >>> On 11/21/2011 11:42 AM, Brad Brim wrote:
> > >>>> hi Chris,
> > >>>>
> > >>>> By "cavity modes" do you mean power plane resonances? If
> > >>> so, is your
> > >>>> objective to (a) drive down device impedance, (b) drive down
> > >>>> emissions, (c) both? Is your goal to only change plane shapes,
> > >>>> stackup, etc. or are you also able to modify decap/emicap
> > >>> placement and selection?
> > >>>> You cannot eliminate all resonance from a plane cavity of
> > >>> electrically
> > >>>> large size with only shape or stackup changes. The modes will have
> > >>>> different spatial distribution as you change the shape but
> > >>> you cannot eliminate them.
> > >>>> As you remove plane cavity area you will reduce spectral
> > >>> mode density.
> > >>>> The big question is how much plane cavity are you able to
> > >>> eliminate?
> > >>>> Using thinner dielectrics does nothing to eliminate resonances. It
> > >>>> certainly raises capacitance but its most significant
> > >>> contribution is
> > >>>> to reduce inductance (not just for spreading but also for
> > >>> mounting caps).
> > >>>> If you are trying to eliminate resonances by adding caps it is
> > >>>> interesting to note that the best solution is not a
> > >>> sequential process
> > >>>> such as: (1) simulate to find the first resonance peak, (2)
> > >>> add a cap
> > >>>> near the resonance peak, repeat (1-2). This process yields reduced
> > >>>> resonances. But if you are able to add multiple caps
> > >>> simultaneously by
> > >>>> considering broadband impedance across the entire PDN, then
> > >>> you find
> > >>>> better solutions with a smaller set of caps.
> > >>>>
> > >>>> cheers,
> > >>>> -Brad
> > >>>> Sigrity
> > >>>>
> > >>>>> -----Original Message-----
> > >>>>> From: si-list-bounce@xxxxxxxxxxxxx
> > <mailto:si-list-bounce@xxxxxxxxxxxxx>
> > >>>>> [mailto:si-list-bounce@xxxxxxxxxxxxx
> > <mailto:si-list-bounce@xxxxxxxxxxxxx>] On Behalf Of Cristian Gozzi
> > >>>>> Sent: Monday, November 21, 2011 7:52 AM
> > >>>>> To: si-list@xxxxxxxxxxxxx <mailto:si-list@xxxxxxxxxxxxx>
> > >>>>> Subject: [SI-LIST] Damping of Cavity-Mode Resonances in PCB
> > >>>>>
> > >>>>> Hi SI expert
> > >>>>> I'm looking for a good article/paper that describe techniques to
> > >>>>> reduce or suppress cavity modes in PCB, just by working on plane
> > >>>>> shape dimension, pwr to gnd dilectric thickness etc.
> > >>>>>
> > >>>>> Thanks in advance for your suggestion
> > >>>>>
> > >>>>> regards
> > >>>>> Cris
> > >>>>>
> > >>>> ------------------------------------------------------------------
> > >>>> To unsubscribe from si-list:
> > >>>> si-list-request@xxxxxxxxxxxxx
> > <mailto: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
> > <mailto: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
> > >>>>
> > >>>> 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
> > >>> www.ipblox.com <http://www.ipblox.com>
> > >>>
> > >>> (775) 299-4236 Business
> > >>> (866) 675-4630 Toll-free
> > >>> (707) 780-1951 Fax
> > >>>
> > >>>
> > >>> ------------------------------------------------------------------
> > >>> To unsubscribe from si-list:
> > >>> si-list-request@xxxxxxxxxxxxx
> > <mailto: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
> > <mailto: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
> > >>>
> > >>> 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
> > <mailto: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
> > <mailto: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
> > >>
> > >> 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 <mailto: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 <mailto: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
> > >
> > > 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 <mailto: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 <mailto: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
> > >
> > > 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 <mailto: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 <mailto: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
> >
> > 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
>
> 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
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
