[SI-LIST] Re: Importance of Package Height
- From: "Istvan Novak" <istvan.novak@xxxxxxxxxxxxxxxx>
- To: <bdewitt@xxxxxxxxxxxxx>, <si-list@xxxxxxxxxxxxx>
- Date: Thu, 28 Mar 2002 09:03:13 -0500
Brent,
You are correct, the primary parameters here are the dimensions inside the
capacitor (this is something the capacitor vendors could supply) and the
geometry and dimensions of the rest of the loop external to the capacitor
(dependent on the user).
It was correctly pointed out in several mails in this thread that the loop
inductance could be broken down into three major elements: L_cap,
L_vias_pads, and L_plane (plus the mutuals among them). As the capacitor
vendor is responsible for L_cap, but the capacitor construction also has an
impact on the mutual values, where one half of the geometry (the external
hook-up geometry) is a'priori unknown to the capacitor vendor, asking for
inductance values from the part vendor does not seem feasible. They might
be able to supply L_cap (partial self inductance of the capacitor body), but
they have no way of knowing our external connections, which in turn
determine some of the mutuals.
My preference is to break the loop at the interface where the vertical via
connections - coming from the capacitor - meet the planes. We can make the
connection with one or more pair of vias, but either way, one (set of) via
will connect to the plane being closer to the capacitor, the other (set of)
via goes through an antipad to the plane being further away. As a first
approximation, assuming that the antipad opening is not a very big portion
of the entire loop size, we may say that the closer plane will separate the
two halves of the loop such that the mutuals in our original loop inductance
will be dependent on the capacitor-pad-via-closer_plane geometry, but not
being affected by the further plane. As long as this assumption is valid,
the loop can be uniquely broken down to two inductance values: L_plane, and
what I call L_attached. L_attached contains L_cap, L_vias_pads and the
mutuals.
We still should not expect the capacitor vendors to give us L_attached, but
it works the other way around: if we know the internal geometry of the
capacitor, for each connection geometry in our application, we could
determine L_attached by field-solver simulations. Or, if we have sample
parts, we dont need to know the internal geometry of the capacitor,
L_attached can be determined by measurements for each of our application
geometries.
The benefit of this definition is that L_attached is highly independent of
the plane separation and on the horiozontal location on the planes (L_plane,
as was mentioned in the previous mails, is dependent primarilty on the plane
separation, and as second order effects, it is also dependent on frequency
and the horizontal location on the planes).
The above L_attached definition is also convenient for simulation purposes,
as for a given capacitor and hook-up geometry it yields a 'constant' model,
independent of the planes and horizontal locations on the planes. By adding
the simulation model of the planes, the composite model (with the same
'constant' L_attached model) will correctly capture the variation of loop
inductance and variation of resonance frequency as we change the plane
separation and/or move the capacitor horizontally on the planes.
The above L_attached model has been verified by hardware correlations and
was found to be satisfactorily accurate even for aggrassive geometries,
yielding <100pH loop inductance. If necessary, any frequency-dependent
nature of L_attached can also be captured in this model, again, independent
of the plane separation and horizontal location on the planes.
Istvan Novak
SUN Microsystems
----- Original Message -----
From: "Brent DeWitt" <bdewitt@xxxxxxxxxxxxx>
To: <zabinski.patrick@xxxxxxxx>; <si-list@xxxxxxxxxxxxx>
Sent: Wednesday, March 20, 2002 10:58 PM
Subject: [SI-LIST] Re: Importance of Package Height
>
> It appears there is a strong argument for the capacitor vendors to supply
a
> _mechanical_ model of their devices. Precise dimensions of layers and
> dielectric constants/dissipation factors would allow anyone to fit to
their
> own model. Is there a flaw in this thought?
>
> Brent
>
------------------------------------------------------------------
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 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:
