[SI-LIST] Re: matching differential impedance for edge-coupled lines
- From: "Jian X. Zheng" <jian@xxxxxxxxxx>
- To: <si-list@xxxxxxxxxxxxx>
- Date: Fri, 15 Jun 2001 14:46:44 -0700
Hi, Perry and Steve:
1. Matching a differential line with ground reference is in fact very
simple. We can terminate each line to ground with Rs and terminate the Rp
across the two lines as shown below. The Rs and Rp should such that:
Rdiff = Rodd * 2 = 2 * Rs * Rp / ( 2 * Rs + Rp )
Reven = Rs ;
In the other word, if you choose: Rs = Reven and Rp = 2 * Reven * Rodd /
Reven + Rodd ), you will get perfect match for both differential signal and
even mode circuit.
============================ +
/ \
/ \
/// \\\ Rp
/ Rs \
/ \
= ============================ -
G /
/
///
/ Rs
/
=
G
2. For a differential line without a ground reference, you can not connect
the Rs to the ground, and you can not use the above circuit. However, for
such a case, there is no even mode. Even mode exists only when you have a
ground reference. When the ground reference is removed, the even mode is not
propagating. There is no need to match it.
Thanks!
-----------------------------------------------------------------------
Jian-X. Zheng, Ph.D
Zeland Software, Inc., 48890 Milmont Drive, 105D, Fremont, CA 94538, U.S.A.
Tel: 510-623-7162, Fax: 510-623-7135, Web: http://www.zeland.com
---------------------------------------------------------------------
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx
> [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of Perry Qu
> Sent: Friday, June 15, 2001 2:12 PM
> To: si-list@xxxxxxxxxxxxx
> Subject: [SI-LIST] Re: matching differential impedance for edge-coupled
> lines
>
>
>
> Hi! Steve:
>
> I understand that closer spacing between the two trace bring even
> mode impedance
> higher. However, for differential application, can we just
> terminate the line for
> odd mode signal and ignore the even mode ?
>
> Thanks
>
> Perry
>
> "S. Weir" wrote:
>
> > Perry,
> >
> > Tighter coupling between the lines is an advantage within pairs, but it
> > means that the even mode impedance has to be raised above 50 ohms. This
> > means a different layer, or narrower line widths than your single-ended
> > lines. No matter what you do, you are unlikely to get coupling
> above 15%.
> >
> > Regards,
> >
> > Steve
> > At 03:05 PM 6/15/01 -0400, you wrote:
> > >Hi!
> > >
> > >There has been a few threads about odd/even mode impedance and I would
> > >like to throw in another one:
> > >
> > >Basically there are two types of differential pair for edge-coupled
> > >lines, one is loosely-coupled, with trace spacing much wider than trace
> > >width/height above ground; I find from my 2D field solver that only in
> > >this case you can obtain a 100 ohm differential impedance (or 50 ohm
> > >Zodd as mentioned in TDA's paper); To match the diff pair, you can
> > >either terminate each line with 50 ohm to ground or a 100 ohm across
> > >plus and minus;
> > >
> > >The second type is a tightly-coupled diff pair, with trace spacing
> > >equivalent to trace width and trace height against ground
> plane. In such
> > >case, I can't get a 100 ohm differential impedance (or 50 ohm Zodd) in
> > >my calculation with a realistic stackup and manufacturing constaints.
> > >Zodd is more likely in the region of 40-45 ohm range. To match the diff
> > >pair in this case, I need to terminate each line against ground with
> > >Zodd or 2*Zodd across plus and minus. (I did this with HSPICE on a very
> > >simple topology: Ideal voltage source (50ps rise time step pulse) ->
> > >transmission line -> load). This is intuitive since we can imagine that
> > >there is a virtual ground between plus and minus.
> > >
> > >My question:
> > >
> > >In reality, I saw many diff pair done in second case because of the
> > >space limit on already very dense board. In such case, I think instead
> > >of using standard 100 ohm termination (e.g., LVDS application),
> > >shouldn't we terminate the diff pair with smaller values to match the
> > >differential signal ?
> > >
> > >Thanks in advance
> > >
> > >Perry Qu
> > >
> > >
> > >
> > >
> > >
> > >
> > >------------------------------------------------------------------
> > >To unsubscribe from si-list:
> > >si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> > >For help:
> > >si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> > >
> > >List archives are viewable at:
> //www.freelists.org/archives/si-list
> > >Old 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
> > For help:
> > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >
> > List archives are viewable at:
> //www.freelists.org/archives/si-list
> > Old 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
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
> Old 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
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at: //www.freelists.org/archives/si-list
Old list archives are viewable at: http://www.qsl.net/wb6tpu
Other related posts:
