Re: Re[2]: [SI-LIST] : power supply filtering and bypassing

Ray Anderson (raymonda@radium.eng.sun.com)
Fri, 6 Mar 1998 13:59:59 -0800 (PST)

In reply to Jeff's posting (attached below since it never
made it to the list due to a wrong address), I've got a couple
comments and questions related ferrite inductor models.

Here is a typical ferrite bead model that I've extracted
utilizing an HP4194 Impedance analyzer.

----+------L-----+-----
| | L=18.47uH
+------C-----+ C=1.4pF
| | R=1.6K
+------R-----+

The problem I've found is that this model has a limited
frequency range over which the model to hardware correlation
is good. The frequency dependent nature of the ferrite material
makes this simple topology model diverge quite a bit from
reality at higher frequencies. Some of the special ferrite
blends that vendors use in EMI suppressors are quite non-linear
wrt frequency.(I've seen quite a delta as low as 10MHz on a
number of ferrite bead inductors). Also, it doesn't account
for saturation of the ferrite due to DC current.

I've never been able to get a model out of a ferrite vendor
that I felt was any better.

Does anyone have any comments and/or suggestions on ways and
means of extending the model to account for frequency and
current dependency? (or perhaps have run across any good
articles or papers addressing the issue)


-Ray Anderson
Sun Microsystems Inc.

> Date: Fri, 6 Mar 1998 16:14:46 -0500 (EST)
> From: "Jeff M. Gloudemans" <c22jmg@koptsv01.delcoelect.com>
> To: owner-si-list@silab.eng.sun.com
> Subject: Re: Re[2]: [SI-LIST] : power supply filtering and bypassing
>
> John,
>
> I've had good luck modelling ferrite beads as a parallel RLC. Usually I
> charcterize and extract the model myself. This way I can optimaze it over the
> frequency range of interest. At one time I did get some RLC models from TDK
for
> some of their surface mount ferrites.
>
> I'd probably have little fear of resonance effects if paralelling ferrites
since
> they are quite lossy.
>
> Jeff Gloudemans
> Delphi Delco Electronics Systems
>