[SI-LIST] Re: Interference from planar magnetics

Ray,

Ott is correct.  I transcribed it incorrectly in my notes to myself a long 
time ago, and old habits die hard.  2oz Cu is close to 1 MHz.

Regards,


Steve.
At 10:03 AM 1/28/2003 -0800, Ray Anderson wrote:


>Looking at equation 6-11b and table 6-2 in Henry Ott's text
>"Noise Reduction Techniques in electronic Systems" he gives the
>equation for skin depth as:
>
>         depth(in inches) = 2.6/sqrt(f*ur*or)
>
>Where the depth is defined as "the distance required for the
>wave to be attenuated to 1/e or 37% of the original value"
>
>For a few common materials, or and ur:
>
>Material        Relative Conductivity (or)      Relative Permeability (ur)
>--------        --------------------            ----------------------
>Cu                      1.00                    1
>Al                      0.61                    1
>Steel (SAE1045)         0.10                    1000
>
>
>For copper ur (relative permeability) and or (relative conductivity)
>both equal 1 so the equation reduces to
>
>         depth = 2.6/sqrt(f)
>
>Using this equation the skin depth for copper is .0026 inches at 1 MHz
>
>table 6-2 from Ott's book summarizes the results as:
>
>Freq            Cu              Al              Steel
>------          -----           ------          --------
>60 Hz           0.335           0.429           0.034
>100 Hz          0.260           0.333           0.026
>1 kHz           0.082           0.105           0.008
>10 kHz          0.026           0.033           0.003
>100 kHz         0.008           0.011           0.0008
>1 MHz           0.003           0.003           0.0003
>10 MHz          0.0008          0.001           0.00008
>
>
>Looks like there is a discrepancy between sqrt(F) and F in Ott's
>formula and the one Steve presented. Who's correct ?
>
>
>-Ray Anderson
>Sun Microsystems Inc.
>
>
>
> >
> >
> >
> >Larry, it works out to be just about 1MHz in 1oz Cu.  The formula is depth
> >in inches = sqrt( 2.6/F )  At 1MHz it comes out 1.6mils which is close
> >enough for government work to 1.4mils.
> >
> >Regards,
> >
> >
> >Steve.
> >At 06:50 PM 1/27/2003 -0800, Larry Smith wrote:
> >
> >>Actually, the copper planes are a good magnetic shield at high 
> frequency but
> >>poor at low frequency.  The transition region occurs when skin depth is
> >>equal to the copper plane thickness.  The skin depth is the depth to which
> >>magnetic fields penetrate the copper (or at least diminish to 1/e of their
> >>value on the surface).  At high frequency where skin depth is much less 
> than
> >>copper thickness, very little magnetic field penetrates the copper.
> >>
> >>As I recall, the skin depth of copper is about the same as the plane
> >>thickness at 1 MHz (but I cannot remember if this is for 1 oz or half
> >>oz copper).  The copper planes will not shield the fundamental switching
> >>frequency but will shield the higher harmonics from reaching buried
> >>transmission lines.
> >>
> >>regards,
> >>Larry Smith
> >>Sun Microsystems
> >>
> >>Dennis Schmitz wrote:
> >> >
> >> > "I could, of course put a large copper shape (with plenty of vias 
> tied to
> >> > ground planes below) beneath the VRM to block the electric fields from
> >> > penetrating into the lower wiring planes.   While I believe that 
> this would
> >> > affect the magnetic field (since it affects the electric field), it
> >> will not
> >> > block it!."
> >> >
> >> > I think you'll find that copper is a very effective magnetic shield at
> >> those
> >> > frequencies. The principal is that the magnetic fields induce 
> currents that
> >> > cancel out the field on the far side of the copper plane. The frequency
> >> > components where the magnetic fields are no longer effectively 
> cancelled by
> >> > copper are so large that your diffpairs won't even even notice it's 
> there.
> >> > You'll get some coupling on the single-ended stuff, though.
> >> >
> >> > The rule then is that the copper plane needs to be as large as you 
> can make
> >> > it, preferably the entire top surface of the board (and be a ground 
> plane).
> >> > The higher the conductivity the better, so use 1 oz (or even 2 oz 
> copper if
> >> > you can get your board shop to do it). The GHz stuff should be ok too.
> >> >
> >> > If you have sensitive analog signals, you'll have trouble, though.
> >> >
> >> > Dennis
> >> >
> >> > -----Original Message-----
> >> > From: si-list-bounce@xxxxxxxxxxxxx
> >> > [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of Bob Welte
> >> > Sent: Friday, January 24, 2003 4:31 PM
> >> > To: si-list@xxxxxxxxxxxxx
> >> > Subject: [SI-LIST] Interference from planar magnetics
> >> >
> >> > Hello (once again)
> >> > I would like to get some insight, (or references) on the effects of 
> having
> >> > planar magnetics (from DC/DC converters) on a digital board in close
> >> > proximity to the digital traces.
> >> > Specifically, I need to determine the effects that the fields from the
> >> > planar magnetics of the DC/DC VRM might have on various types of digital
> >> > signals.
> >> > The digital signals fall into two classes:  High frequency (1.25Ghz) 
> which
> >> > are differential, and low voltage (1V amplitude), and low frequency 
> (<200
> >> > Mhz), unbalanced, but with various voltages (some are 3.3V, 2.5, and 
> 1.8V
> >> > signals).
> >> > The VRMs are relatively low voltage/ high current (2.5 and 1.8V at 10-25
> >> > Amps).  The VRMs typically have switching frequencies not much 
> greater than
> >> > a couple Mhz.  The magnetics of the VRM are not "part of the board" (ie
> >> > board traces do not make up any portion of the inductor/transformer
> >> > windings).
> >> > Ideally, I would like to be able to route the digital signals near (and
> >> > even under) the VRM.   I could, of course put a large copper shape (with
> >> > plenty of vias tied to ground planes below) beneath the VRM to block the
> >> > electric fields from penetrating into the lower wiring planes.   While I
> >> > believe that this would affect the magnetic field (since it affects the
> >> > electric field), it will not block it!.
> >> > As voltages get lower, and VRMs need to be closer to the ASICs, 
> where board
> >> > real estate is at a premium, I thought this topic would have wide 
> interest.
> >> > Any thoughts?
> >> >
> >> > Bob Welte
> >> > IBM Microelectronics
> >
>------------------------------------------------------------------
>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: